Biochemistry of Lipids_ Lipoproteins and Membranes

Document Sample
Biochemistry of Lipids_ Lipoproteins and Membranes Powered By Docstoc
New Comprehensive Biochemistry

                Volume 20

               Generul Editors

             A. NEUBERGER

            L.L.M. van DEENEN

     Amsterdam London New York Tokyo
    Biochemistry of Lipids,
 Lipoproteins and Membranes



       Lipid and Lipoprotein Research Group, Faculty of Medicine,
328 Heritage Medical Research Centre, Edmonton, Aha., Canada, T6G 2S2

           Amsterdam * London New York Tokyo     -
0 1991 Elsevier Science Publishers B.V. All rights reserved

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in
any form or by any means, electronic, mechanical, photocopying, recording or otherwise, with-
out the prior permission of the Publisher, Elsevier Science Publishers B.V., Copyright & Per-
missions Department, P.O. Box 521, 1000 AM Amsterdam, The Netherlands.

No responsibility is assumed by the Publisher for any injury and/or damage to persons or
property as a matter of products liability, negligence or otherwise, or from any use or operation
of any methods, products, instructions or ideas contained in the material herein. Because of the
rapid advances in the medical sciences, the Publisher recommends that independent verifica-
tion of diagnoses and drug dosages should be made.

Special regulations for readers in the USA - This publication has been registered with the
Copyright Clearance Center Inc. (CCC), Salem, Massachusetts. Information can be obtained
from the CCC about conditions under which photocopies of parts of this publication may be
made in the USA. All other copyright questions, including photocopying outside of the USA,
should be referred to the Publisher.

ISBN 0 444 89321 0 (hardback)
ISBN 0 444 89384 9 (paperback)
ISBN 0 444 80303 3 (series)

This book is printed on acid-free paper.

Cover photo from a painting by S.J. Thomas, giving the artist’s view of biological membranes

Published by:
Elsevier Science Publishers B.V.
P.O. Box 21 1
1000 AE Amsterdam
The Netherlands

Sole distributors for the USA and Canada:
Elsevier Science Publishing Co. Inc.
655 Avenue of the Americas
New York, NY 10010

Printed in The Netherlands

This is the second edition of this advanced textbook. The name has been changed
from ‘The Biochemistry of Lipids and Membranes’ to ‘Biochemistry of Lipids, Lipo-
proteins and Membranes’ to reflect the increased coverage of lipoproteins. There are
two major objectives of this book. One is to provide an advanced textbook in the field
of lipid, lipoprotein and membrane biochemistry. The second is to provide a clear
and current summary of these research areas for scientists presently working in these
and related fields.
   Biochemistry has matured to the point that advanced textbooks in the various sub-
categories are required. This book should satisfy that need for the fields of lipid,
lipoprotein and membrane biochemistry. The chapters are written for students who
have taken an introductory course in biochemistry. We assume that students are fa-
miliar with the basic principles and concepts of biochemistry and have a general
background in the subject.
   The second objective relates to the need for a general reference and review book for
scientists in the lipid, lipoprotein and membrane research area. There are many excel-
lent reviews available of the various topics covered by this book and many of these
reviews are cited in the appropriate chapters. However, the second edition of this
book remains unique in that it provides a current, readable, and critical summary of
these areas of research. This book should allow active scientists to become familiar
with recent developments in areas of lipid metabolism related to their research inter-
ests. This book should also help clinical researchers, medical students and residents
keep abreast of developments in basic science that are important for subsequent clini-
cal advances.
   We have not attempted to cover the general area of structure and function of
biological membranes since that subject is covered in a number of excellent books.
The first chapter does address basic principles of membrane structure required for an
understanding of the subsequent chapters.
   We have limited the number of references cited and emphasised review articles.
However, readers may wish access to the primary literature in some instances. Thus,
we have introduced a novel approach to literature citation suggested by Charles
Sweeley. In some of the chapters reference has been made to published work by citing
the name of the senior author and the year in which the work was published. This
should allow the reader to find the original citation via a computer search.
   The editors and contributors assume full responsibility for the content of the vari-
ous chapters. We would be pleased to receive comments and suggestions about this
   Finally, the editors and contributors are indebted to the many other people who
have made this book possible. In particular, we extend our thanks to Brenda Struk,
Linda Lang, Carolina Landolt, Joseph Casey, Jeffrey Charuk and David Williams.
                                                             Dennis and Jean Vance
                                                           Edmonton, Alta., Canada,
                                                                        April, 1991

     For all students of lipid, lipoprotein and membrane biochemistry
                         - past, present and future.

                                                List of contributors
Konrad Bloch, 363
Department of Chemistry, Harvurd University, 12 Oxford Street, Cambridge, M A
02138, U.S.A.
Pierre Borgeat, 297
Le Centre Hospitulier de I’Universiti. Luval, 2705 Boulevard Laurie6 Que., GI V 4G2,
David N. Brindley, 171
Department o Biochemistry, und Lipid and LQoprotein Research Group, 328 Heritage
Medical Reseurch Centre, University of Albertu, Edmonton, Altu., T6G 2S2, Cunadu.
Harold W. Cook, 141
Departments of Pediatrics and Biochemistry, and Atlantic Research Centrefor Mental
Returdation, Dulhousie University, Hulifux, N. S., B3H 4H7, Cunadu.
John E. Cronan, 43
Departments of Microbiology und Biochemistry, University of Illinois, Urhuna, I L
6 1801, U.S.A.
Pieter R. Cullis, 1
Depurtment of Biochemistry, University of British Columbia, Vancouver, B. C., V6T
I W5, Canudu.
Roger Davis, 403
Hepatohiliary Research Center, University of Colorado Heulth Sciences Centre, Box
B158, 4200 E. Ninth Avenue, Denver, CO 80262, U.S.A.
Peter A. Edwards, 383
Departments of Biological Chemistry und Medicine, University of Califbrnia at LOS
Angeles, Los Angeles CA 90024, U.S.A.
Christopher J. Fielding, 427
Curdiovascular Research Institute, I315 M , University o California Medical Centre,
Sun Francisco, CA 94143, U . S . A .
Phoebe E. Fielding, 427
Curdiovusculur Research Institute, 131.5 M , University of California Medical Centre,
Sun Francisco, CA 94243, U.S.A .
Frank A. Fitzpatrick, 297
Department of Pharmacology, University of Colorado Health Sciences Center, Denver,
CO 80262. U.S.A .
Alan G. Goodridge, 111
Department of’ Biochemistry, University of Iowa, Bowen Science Bldg., Iowa City, I 0
52242, U.S.A.
Michael J. Hope, 1
Department of Biochemistry, University of British Columbia, Vancouver, B. C., V6T
I W5, Canada.

Suzanne Jackowski, 43
Department of Biochemistry, St. Jude Children ’s Research Hospital, 332 North Lau-
derdale, Memphis, T N 38101, U.S.A.
Reinhart A.F. Reithmeier, 525
Department of Medicine and M R C Group in Membrane Biology, Room 7307, Medical
Sciences Bldg., University of Toronto, Toronto, Ont., M5S lA8, Canada.
Charles 0. Rock, 43
Department of Biochemistry, St. Jude Children’s Research Hospital, 332 North Lau-
derdale, Memphis, TN 38101, U.S.A.
Wolfgang J. Schneider, 461
Department of Molecular Genetics, University of Vienna, Vienna, A-I010 Austria.
Horst Schulz, 87
Department of Chemistry, City College of’CUNY; Convent Avenue at 138 Street, New
York, N Y 10031, U.S.A.
William L. Smith, 297
Department of Biochemistry, Michigan State University, East Lansing, M I 48824,
U.S,A .
Fred Snyder, 241
Medical Sciences Division, Oak Ridge Associated Universities, P.0. Box 117, Oak
Ridge, TN 37831, U.S.A.
Charles C.Sweeley, 327
Department of Biochemistry, Michigan State University, East Lansing, M I 48824,
Dennis E. Vance, 205
Department of Biochemistry, and Lipid and Lipoprotein Research Group, 328 Heritage
Medical Research Centre, University of Alberta, Edmonton, Alta., T6G 2S2, Canada.
Dennis R. Voelker, 489
Department of Medicine, National Jewish Hospital, 1400 Jackson Street, Denver, CO
80206, U.S.A.
Moseley Waite, 269
Depariment of Biochemistry, Bowman Gray School of Medicine, Wake Forest Uni-
versity, Winston Salem, NC 27103, U.S.A.

Preface . . . . . . . . . . . . . . . . . . . . . . . .                                                                            v

List o contributors
      f                  . . . . . . . . . . . . . . . . . . . .                                                                  vii

Chapter 1. Physical properties and functional roles of lipids in membranes
Pieter R . Cullis and Michael J . Hope . . . . . . . . . . . . . .                                                                 i

1.   Introduction and overview . . . . . . . . . . .                            .      .     .       .      .      .       .       1
2.   Lipid diversity and distribution . . . . . . . . . .                       .       .     .      .      .       .       .      3
     2.1. Chemical diversity of lipids . . . . . . . . .                        .      .     .       .      .      .       .       3
     2.2. Membrane lipid compositions . . . . . . . .                           .       .     .      .      .      .        .      5
     2.3. Transbilayer lipid asymmetry . . . . . . . .                          .      .     .      .      .      .       .        7
3. Model membrane systems . . . . . . . . . . .                                   .      .      .      .      .      .        .    8
     3.1. Lipid isolation and purification . . . . . . . .                      .      .     .       .      .      .       .       9
     3.2. Techniques for making model membrane vesicles . . .                   .      .     .      .      .      .        .       9
     3.3. Techniques for making planar bilayers and monolayers .                .      .     .       .      .      .       .      13
     3.4. Reconstitution of integral membrane protein into vesicles .           .      .     .      .       .      .       .      13
4. Physical properties of lipids . . . . . . . . . . .                           .      .     .      .       .      .       .     15
    4.1. Gel-liquid-crystalline phase behaviour . . . . . .                      .      .     .       .      .      .        .    15
    4.2. Lipid polymorphism . . . . . . . . . . .                                 .      .     .      .      .       .       .    19
     4.3. Factors which modulate lipid polymorphism . . . .                     .      .     .       .      .      .       .      24
    4.4. The physical basis of lipid polymorphism . . . . .                     .      .     .       .      .      .       .      26
5 . Lipids and the permeability properties of membranes . . . .                 .      .      .      .      .      .        .     28
     5.1. Theoretical considerations . . . . . . . . .                          .       .     .      .      .      .        .     28
     5.2. Permeability of water and non-electrolytes . . . . .                   .      .     .      .      .       .       .     29
     5.3. Permeability of ions . . . . . . . . . . .                             .      .      .      .      .      .       .     30
6. Lipid-protein interactions . . . . . . . . . . .                               .      .     .       .      .       .       .   31
     6.1. Extrinsic proteins . . . . . . . . . . .                               .      .     .      .       .      .       .     32
    6.2. Intrinsic proteins . . . . . . . . . . . .                             .      .     .      .       .      .       .      33
7 . Lipids and membrane fusion . . . . . . . . . .                               .      .     .       .      .      .        .    34
     7.1. Fusion of model systems       . . . . . . . . .                       .      .      .      .      .      .       .      34
     7.2. Fusion of biological membranes       . . . . . . .                    .      .     .      .      .      .       .       36
8 . Model membranes and drug delivery . . . . . . . .                           .      .     .       .      .      .       .      38
9 . Future directions . . . . . . . . . . . . . . .                                .     .     .      .     .      .              40
References . . . . . . . . . . . . . . . . . .                                    .      .     .      .     .      .              40

Chapter 2. Lipid metabolism in procaryotes
S . Jackowski. J.E. Cronan Jr. and C.O. Rock                    .   .   .   .   .                                                 43

I.   The study of bacterial lipid metabolism    .   .   .       . . . . . . . . . . .                                             43
2.   Historical introduction . . . . .          .   .   .       . . . . . . . . . . .                                             43
3.   An overview of phospholipid metabolism    in E. coli       . . . . . . . . . . .                                             44
4.   Genetic analysis of lipid metabolism .     . . . . . . . . . . . . . .                                                       44
5.   Membrane systems of E . coli . . .         .   .       .   .   .   .   .   .      .     .      .       .      .       .      48

6.   Lipid biosynthetic pathways in E. coli . . . . . . . . . . . .                    . . .                                 50
     6.1. Initiation of fatty acid biosynthesis . . . . . . . . . . .                   . . .                                50
     6.2. Elongation of acyl chains        . . . . . . . . . . . . .                       .           .            .        51
     6.3. Product diversification . . . . . . . . . . . . . .                          .           .            .            52
     6.4. Transfer to the membrane . . . . . . . . . . . . .                              .            .                .    53
     6.5. Diversification of polar head groups . . . . . . . . . .                      .           .               .        54
     6.6. Central role of acyl carrier protein . . . . . . . . . . .                   .           .            .            54
7. Lipopolysaccharide biosynthesis          . . . . . . . . . . . . .                    .           .              .        58
8. Degradative pathways . . . . . . . . . . . . . . . .                                  .            .              .       59
     8.1. Phospholipdses . . . . . . . . . . . . . . . .                                . .                         .        59
     8.2. Thioesterases . . . . . . . . . . . . . . . . . .                             . .                                  62
     8.3. Fatty acid oxidation in bacteria . . . . . . . . . . . .                       . .                            .    63
9 . Phospholipid turnover . . . . . . . . . . . . . . . .                                      .            .            .   63
     9.1, The diacylglycerol cycle . . . . . . . . . . . . . .                          .           . .                      63
     9.2. The 2-acylglycerolphosphoethanolaminecycle . . . . . . . .                    .           . .                      65
10. Inhibitors of lipid metabolism . . . . . . . . . . . . . .                            .          . .                     66
     10.1. 3-Decynoyl-N-acetylcysteamine. . . . . . . . . . . .                           .           . .                    66
     10.2. Cerulenin . . . . . . . . . . . . . . . . . . .                                .          .                       66
     10.3. Thiolactomycin . . . . . . . . . . . . . . . .                                  .          . .                    67
11 . Regulation of fatty acid composition in E. co/i . . . . . . . . . .                    .           .            .       68
     11.1. Role of 3-hydroxydecanoyl-ACP dehydrdse . . . . . . . . .                    .           .               .        68
     11.2. Role of 3-ketoacyl-ACP synthase I . . . . . . . . . . .                      .           .               .        69
     11.3. Role of3-ketoacyl-ACP synthase I1 . . . . . . . . . . .                          .           .               .    70
     11.4. Factors affecting fatty acid chain-length distribution . . . . . .          .            .           .            72
     11.5. Synthesis of cyclopropane fatty acids . . . . . . . . . .                   .            .           .            72
     1 I .6 . Importance of the fatty acid composition of the membrane phospholipids   . . .                                 73
12. Regulation of phospholipid biosynthesis in E. coli . . . . . . . . .                . . .                                74
     12.1. Control of fatty acid positional distribution . . . . . . . . .              . . .                                74
     12.2. Regulation of total phospholipid synthesis . . . . . . . . .                . . .                                 76
     12.3. Coordination of phospholipid synthesis with cellular metabolism . . .        .           .               .        77
     12.4. Regulation of phospholipid head group composition . . . . . .                .              .            .        77
     12.5. E. co/i tolerates variations in phospholipid composition   . . . . .         .              .            .        78
     12.6. Outer membrane lipoprotein . . . . . . . . . . . . .                         .           .           .            79
13. Lipid metabolism in bacteria other than E . coli     . . . . . . . . .              . . .                                80
     13.1. Bacteria lacking unsaturated fatty acids      . . . . . . . . .              . . .                                80
     13.2. Bacteria containing phosphatidylcholine       . . . . . . . . .              . . .                                81
     13.3. Bacteria synthesizing unsaturated fatty acids by an aerobic pathway . .      . . .                                81
     13.4. Bacteria with a multifunctional fatty acid synthase . . . . . . .            .           .               .        81
     13.5. Bacteria with intracytoplasmic membranes . . . . . . . . .                   . . .                                82
     13.6. Other bacterial oddities . . . . . . . . . . . . . .                          . . .                               82
     13.7. Lipids of nonbacterial (but related) organisms . . . . . . . .               . . .                                83
14. Future directions . .                             . . . . . . . . . .               .                                    83
References . . . . .                                  . . . . . . . . . .               .                                    84

Chuptev 3. Oxidation offatty acids
Horst Schulz . . . . . . . .                                                                                                 87

I.   The pathway ofg-oxidation: a historical account .      . . . . . . . . . . .                                            87
2.   Uptake and activation of fatty acids in animal cells   . . . . . . . . . . .                                            89

3.   8-Oxidation in mitochondria     . . . . . . . . . .                  . . . . . . .                                                                   91
     3.1. Mitochondria1 uptake of fatty acids . . . . . . .              . . . . . . .                                                                    91
     3.2. Enzymes ofp-oxidation in mitochondria     . . . . .            . . . . . . .                                                                    93
     3.3. 8-Oxidation of unsaturated and odd-chain fatty acids . .       . . . . . . .                                                                    96
     3.4. Regulation of mitochondrialp-oxidation    . . . . .            . . . . . . .                                                                    98
     3.5. Inhibitors of mitochondria1 8-oxidation   . . . . .            . . . . . . .                                                                   100
4 . 8-Oxidation in peroxisomes . . . . . . . . . . .                      . . . . . . .                                                                  101
5 . Fatty acid oxidation in E. coli . . . . . . . . . .                  . . . . . . .                                                                   104
6. Inherited diseases of fatty acid oxidation . . . . . . .              .       .           .               .           .           .           .       106
7 . Future directions . . . . . . . . . . . . . . .                          .           .           .           .               .       .               108
References . . . . . . . . . . . . . . . . . .                               .           .       .               .               .       .               109

Chapter 4. Fatty acid svnthesis in eucaryotes
Alan G. Goodridge . . . . . . . . . . . . . . . . . . . .                                                                                                111

1.   Introduction . . . . . . . . . . . . . . . .                            .           .           .               .           .           .           Ill
2.   Signals in blood that mediate the effects of diet . . . . .         . . . . . . .                                                                   112
3.   Which enzymes regulate fatty acid synthesis? . . . . . .             . . . . . . .                                                                  114
4.   Regulation of substrate supply . . . . . . . . . .                  . . . . . . .                                                                   115
     4.1. Production of pyruvate from glucose . . . . . .                . . . . . . .                                                                   115
     4.2. Production ofcitrate from pyruvate . . . . . . .                . . . . . . .                                                                  115
     4.3. Production of NADPH . . . . . . . . . .                         . . . . . . . 117
5 . Regulation of the catalytic efficiency OF acetyl-CoA carboxylase .   . . . . . . . 117
     5.1.   A key regulatory reaction . . . . . . . . .                  . . . . . . .                                                                   117
     5.2. Structure and reaction mechanism . . . . . . .                  . . . . . . .                                                                  118
     5.3. Regulation by citrate      . . . . . . . . . .                 .       .           .               .               .       .           .       118
     5.4. Regulation by long-chain fatty acyl-CoA      . . . . .         .           .           .               .           .           .           .   119
     5.5. Regulation by covalent modification . . . . . .                .       .           .               .           .           .           .       121
6 . Fatty acid synthase . . . . . . . . . . . . . .                       . . . . . .                                                                    125
     6.1. Animal fatty acid synthase: the component reactions . .        . . . . . . .                                                                   125
     6.2. Animal kitty acid synthase: thc subunits are identical . .     .       .           .               .               .       .           .       127
     6.3. Animal fatty acid synthase: structural organization . .        .       .           .               .               .           .           .   128
     6.4. Comparison of yeast and animal fatty acid synthases . .        . . . . . . .                                                                   129
7. Regulation of enzyme concentration       . . . . . . . .               . . . . . . .                                                                  130
     7.1, Messenger R N A levels regulate enzyme synthesis rates . .         .           .               .               .           .           .       130
     7.2. Transcription is usually the regulated step . . . . .          .       .           .               .               .           .       .       131
     7.3. Regulation in cells in culture . . . . . . . .                 .     . . . . . .                                                               134
            7.3.1. 3T3-Ll cells - a pre-adipocyte cell line . . .        .     . . . . . .                                                               134
            7.3.2. Hepatocytes in maintenance culture . . . .            .     . . . . . .                                                               135
8. Future directions . . . . . . . . . . . . . . .                           . . . . . .                                                                 137
References . . . . . . . . . . . . . . . . . .                               .           .           .               .           .           .           138

Chapter 5 . Fatty acid desaturution and chain elongation in eucaryotes
Hurold W Cook . . . . . . . . . . . . . . . . . . . . .                                                                                                  141

I.   Introduction . . . . . . . . . . . . . . . . . . . . . .                                                                                            141
2.   Historical background . . . . . . . . . . . . . . . . . . .                                                                                         143
3.   Chain elongation of long chain fatty acids . . . . . . . . . . . . . .                                                                              144

     3.1. The microsomal elongation system . . . . . . . . . . . . . .                            i46
     3.2. The mitochondria1 elongation system . . . . . . . . . . . . .                           147
     3.3. Functions of the two elongation systems . . . . . . . . . . . .                         148
4 . Formation of monounsaturated fatty acids by oxidative desaturation . . . . . . .              148
    4.1. Nomenclature to describe double bonds . . . . . . . . . . . .                            148
     4.2. Characteristics of the monoene-forming desaturation enzymes . . . . . . .               149
     4.3. Modification of A9 desaturase activities in vitro . . . . . . . . . .                   151
     4.4. Dietary and hormonal regulation of A9 desaturase . . . . . . . . . .                    152
     4.5. Formation of monounsaturated fatty acids in plants . . . . . . . . .                    154
5. Formation of polyunsaturated fatty acids . . . . . . . . . . . . . .                           154
     5.1. Characteristics and restrictions in animal systems . . . . . . . . . .                  154
     5.2. Essential fatty acids-a contribution of plant systems . . . . . . . . .                 155
     5.3. Families of fatty acids and their metabolism     . . . . . . . . . . .                  157
             5.3.1. The (n-6) family    . . . . . . . . . . . . . . . .                           157
             5.3.2. The (n-3)family     . . . . . . . . . . . . . . . .                           159
             5.3.3. Competition between the (11-61 and (n-3) families . . . . . . . .             160
             5.3.4. The (11-91 family   . . . . . . . . . . . . . . . .                           161
             5.3.5. The (n-7)family     . . . . . . . . . . . . . . . .                           161
     5 4 Dietary and hormonal alterations of polyunsaturated acid synthesis . . . . .
      ..                                                                                          161
     5.5. Two or more double bonds in plants . . . . . . . . . . . . .                            163
6 . Unsaturated fatty acids with trans double bonds . . . . . . . . . . . .                       164
7 . Abnormal patterns of distribution and metabolism of long chain saturated and unsaturated
     fatty acids . . . . . . . . . . . . . . . . . . . . . .                                      165
     7.1. Essential fatty acid deficiency . . . . . . . . . . . . . . .                           165
     7.2. Zinc deficiency . . . . . . . . . . . . . . . . . . .                                   166
     7.3. Other clinical disorders . . . . . . . . . . . . . . . . .                              166
     7.4. Relationship to plasma cholesterol . . . . . . . . . . . . . .                          166
8 . Future directions . . . . . . . . . . . . . . . . . . . . .                                   167
References . . . . . . . . . . . . . . . . . . . . . . . .                                        168

Chapter 6. Metabolism of triacylglycerols
David N . Brindley . . . . . . . . . . . . . . . .                                                171

1.    Introduction . . . . . . . . . . . . . . . . . . . . . .                                    171
2.    Biosynthesis of triacylglycerols . . . . . . . . . . . . . . . . .                          173
      2.1.   Biosynthesis of phosphatidate . . . . . . . . . . . . . . .                          173
      2.2. Conversion of phosphatidate to triacylglycerol . . . . . . . . . . .                   178
      2.3. Conversion of monoacylglycerol to triacylglycerol . . . . . . . . . .                  179
3.    Digestion. absorption and transport of lipids . . . . . . . . . . . . .                     181
      3.1. Digestion of lipids . . . . . . . . . . . . . . . . . .                                181
      3.2. Absorption of lipids from the small intestine . . . . . . . . . . .                    183
      3.3. Formation of chylomicrons and VLDL . . . . . . . . . . . . .                           184
      3.4. Partitioning of fatty acids between the portal blood and the lymphatic system . .      186
4.    Control of triacylglycerol synthesis . . . . . . . . . . . . . . . .                        187
      4.1. Control of phosphatidate synthesis in the liver . . . . . . . . . . .                  188
      4.2. Control of the conversion of phosphatidate to triacylglycerol in liver . . . . .       189
      4.3. Diacylglycerol as a precursor of triacylglycerol, phosphatidylcholine and phosphati-
             dylethanolamine . . . . . . . . . . . . . . . . . . .                                193
5.    Metabolism of triacylglycerols when the action of insulin is high . . . . . . . .           195
6.    Triacylglycerol metabolism in conditions of metabotic stress . . . . . . . . .              198
7. Future directions . . . . . . . . . . . . . .                           . . . . . . .                200
References . . . . . . . . . . . . . . . . .                               . . . . . . .                201

Chapter 7. Phospholipid metabolism and cell signulling in eucaryotes
Dennis E. Vance . . . . . . . . . . . . . . . . . . . . .                                               205

1.   Introduction . . . . . . . . . .                 . . . . . . . . . . . .                    205
2.   Phosphatidylcholine biosynthesis . . .       .    .   .   .   .   .    .   .   .   .   .   .206.
     2.1. Historical background . . . .              . . . . . . . . . . . . . 206
     2.2. Choline transport and oxidation  .         . . . . . . . . . . . . . 207
     2.3. Enzymes of the CDP-choline pathway         . . . . . . . . . . . . . 207
     2.4.     Phosphatidylethanolamine-N-methyltransferase . . . . . . . . . . 208
3.   Regulation of phosphatidylcholine biosynthesis       . . . . . . . . . . . . 209
     3.1. The rate-limiting reaction . . . . . . . . . . . . . . . . 209
     3.2. The translocation hypothesis       . . . . . . . . . . . . . . . 210
     3.3. Regulatory mechanisms . . . . . . . . . . . . . . . . . 210
              3.3.1. Fatty acids are important regulators of phosphatidylcholine biosynthesis in
                       cultured cells . . . . . . . . . . . . . . . . . 211
              3.3.2. Diacylglycerol may also regulate phosphatidylcholine biosynthesis . . . 21 I
              3.3.3. Phosphorylation-dephosphorylation reactions . . . . . . . . 212
              3.3.4. Feedback regulation . . . . . . . . . . . . . . . 212
              3.3.5. Cholinephosphotransferase       . . . . . . . . . . . . . 213
     3.4. Substrate channeling . . . . . . . . . . . . . . . . . . 213
4.   Sphingomyelin biosynthesis . . . . . . . . . . . . . . . . . . 214
5.   Phosphatidylserine biosynthesis . . . . . . . . . . . . . . . . . 215
     5 . I . Historical developments and biosynthesis . . . . . . . . . . . . 215
     5.2. CHO mutants and regulation . . . . . . . . . . . . . . . 217
6.   Phosphatidylethanolaminebiosynthesis . . . . . . . . . . . . . . . 218
     6.1.     Historical background . . . . . . . . . . . . . . . . . 218
     6.2. Enzymes of the CDP-ethanolamine pathway . . . . . . . . . . . 218
     6.3. Regulation of the CDP-ethanolamine pathway . . . . . . . . . . . 219
              6.3. I . Regulation at the cytidylyltransferase reaction . . . . . . . . 219
              6.3.2. Diacylglycerol . . . . . . . . . . . . . . . . . 220
     6.4. Phosphatidylserine decarboxylation and the relative importance of the various
              pathways for phosphatidylethanolamine biosynthesis . . . . . . . . . 221
     6.5. N-acyl-phosphatidylethanolamine . . . . . . . . . . . . . . 221
     Pol ygl ycerophospholipids . . . . . . . . . . . . . . . . . . 222
     7 . I . Historical developments and biosynthetic pathways      . . . . . . . . . 222
     7.2. Enzymes and subcellular location . . . . . . . . . . . . . . 224
     lnositol phospholipids . . . . . . . . . . . . . . . . . . .                                224
     8.1. Historical developments . . . . . . . . . . . . . . . . . 224
     8.2. Biosynthetic enzymes . . . . . . . . . . . . . . . . . . 225
     Phospholipids as precursors of cellular second messengers . . . . . . . . . 226
     9.1. Discovery of the phosphatidylinositol cycle . . . . . . . . . . . . 226
     9.2. Degradation of phosphatidylinositol-4, 5-bisphosphate by phospholipase C . . . 227
     9.3. Metabolism of the inositol phosphates . . . . . . . . . . . . . 228
     9.4. Function of inositol phosphates . . . . . . . . . . . . . . . 229
     9.5. Diacylglycerol and protein kinase C . . . . . . . . . . . . . . 230
     9.6. Phosphatidylcholine cycles and formation of diacylglycerol . . . . . . . 231

10. Phospholipid catabolism and remodeling of the acyl substituents . . . . . . . . 231
11. Phospholipid biosynthesis in yeast . . . . . . . . . . . . . . . . 233
12. Phosphonolipids . . . . . . . . . . . . . . . .                            . . . . .              234
     12.1, Historical perspective   . . . . . . . . . . .                     . . . . . .             234
     12.2. Biosynthesis . . . . . . . . . . . . . . .                          . . . . .              235
13. Glycosyl phosphatidylinositols for attachment of cell surface proteins   .    .   .   .   .   .   236
14. Future directions . . . . . . . . . . . . . . . .                            . . . .      .       238
References . . . . . . . . . . . . . . . . . . .                                 . . . .      .       238

Chupter 8. Metabolism. regulation. und .function of ether-linkedglycerolipids und
their biouctive species
Fred Snyder . . . . . . . . . . . . . . . . . . . . . .                                               241

1.    Introduction . . . . . . . . . . . . . . . . . . . . . .                                        241
2.    Nomenclature . . . . . . . . . . . . . . . . . . . . .                                          241
3.    Historical events . . . . . . . . . . . . . . . . . . . . .                                     243
4.    Analytical approaches . . . . . . . . . . . . . . . . . . .                                     245
5.    Physical properties . . . . . . . . . . . . . . . . . . . .                                     247
6.    Occurrences in nature . . . . . . . . . . . . . . . . . . .                                     248
7.    Biologically active ether lipids . . . . . . . . . . . . . . . . .                              249
8.    Biosynthesis . . . . . . . . . . . . . . . . . . . . . .                                        250
      8 . I . Ether lipid precursors . . . . . . . . . . . . . . . . .                                250
                8.1. I . Acyl-CoA reductase . . . . . . . . . . . . . . .                             250
                8 . I .2. Dihydroxyacetone-P acyltransferase . . . . . . . . . . .                    250
      8.2. Ether lipids in membranes . . . . . . . . . . . . . . . .                                  252
                8.2.1. Biosynthesis of the ether bond . . . . . . . . . . . . .                       252
                8.2.2. Biosyntiiesis o f t h e alkyl analogue of phosphatidic acid . . . . . .        254
                8.2.3. Biosynthesis of complex neutral glycerolipids and phospholipids with 0-alkyl
                          bonds . . . . . . . . . . . . . . . . . . .                                 254
                8.2.4. Biosynthesis of plasmalogens . . . . . . . . . . . . .                         255
      8.3. Bioactive ether lipids . . . . . . . . . . . . . . . . . .                                 256
9. Catabolism . . . . . . . . . . . . . . . . . . . . . .                                             258
      9.1. Ether lipid precursors . . . . . . . . . . . . . . . . .                                   258
                9.1 . I . Long-chain fatty alcohols . . . . . . . . . . . . . .                       258
                9.1.2. Dihydroxyacetone-P and acyldihydroxyacetone-P . . . . . . . .                  258
      9.2. Ether lipids in membranes . . . . . . . . . . . . . . . .                                  259
                9.2.1. Alkyl cleavage enzyme . . . . . . . . . . . . . . .                            259
                9.2.2. Plasmalogenascs          . . . . . . . . . . . . . . . .                       260
                9.2.3. Phospholipascs and lipases . . . . . . . . . . . . . .                         260
      9.3. Bioactive ether lipids . . . . . . . . . . . . . . . . . .                                 262
10. Regulatory controls . . . . . . . . . . . . . . . . . . . .                                       263
1 1 . Functional roles . . . . . . . . . . . . . . . . . . . . .                                      264
      I 1 . I . Membrane components . . . . . . . . . . . . . . . . .                                 264
      11.2. Cell mediators (activities and mechanisms ofaction) . . . . . . . . .                     265
12. Future directions . . . . . . . . . . . . . . . . . . . . .                                       266
References . . . . . . . . . . . . . . . . . . . . . . . .                                            267

Chapter 9 . Phospholipases
Moseley Waite . . . . . . . . . . .                            .   .   .   .   .    .     .     .      .             269

I.   Overview . . . . . . . . . . . .                      .   .   . . .       .    .     .      .      .      .     269
     1.1.   Definition of phospholipases . . . . .         .   .   . . .       .     .     .     .      .      .     269
     I .2. Assay of phospholipases . . . . . .             .   .   . . .       .     .     .     .      .       .    271
     I .3. lnteraction of phospholipases with interfaces   .   .   . . .       .    .      .     .      .      .     272
2. The phospholipases . . . . . . . . .                    .   .   . . .       .    .     .     .      .      .      276
     2.1. Phospholipase A?       . . . . . . .             .   .   . . .        .    .     .      .      .      .    216
     2.2. Phospholipase B and lysophospholipases .         .   .   . . .        .     .     .      .      .      .   278
     2.3. Phospholipase A?       . . . . . . .             . . . . . . . . . . .                                     280
     2.4. Phospholipase C . . . . . . . .                  . . . . . . . . . . .                                     287
     2.5. Phospholipase D . . . . . . . .                  . . . . . . . . . . .                                     290
     2.6. Phospholipases in signal transduction . .        . . . . . . . . . . .                                     292
3. Future directions . . . . . . . . . .                   . . . . . . . . . . .                                     292
References .       . . . . . . . . . . .                   . . . . . . . . . . .                                     294

Chapter 10. The ricosunoids: cyclooxygenase. lipoxygenase. und epoxygenase
puth wuys
 William L . Smith. Pierre Borgeat and Frank A . Fitzputrick . . . . . . . 291

I.   Introduction . . . . . . . . . . . . . . . . . .                                 .     .      .     .           297
2.   Prostanoids . . . . . . . . . . . . . . . . . .                                  .    .      .     .            298
     2.1. Structures and nomenclature . . . . . . . . . .                            .      .      .      .      .   298
     2.2. Prostanoid chemistry . . . . . . . . . . . . .                             .     .      .      .       .   299
     2.3. Prostanoid biosynthesis . . . . . . . . . . . .                            .     .      .      .      .    299
     2.4. Arachidonate release . . . . . . . . . . . . .                             .     .      .      .      .    299
     2.5. Prostaglandin endoperoxide formation . . . . . . . .                      .     .      .      .      .     300
     2.6. Physico-chemical properties of PGH synthase . . . . . .                    .     .      .      .       .   302
     2.7. PGH synthase and nonsteroidal anti-inflammatory drugs . . .               .     .       .      .      .    302
     2.8. Anti-inflammatory steroids . . . . . . . . . . .                           .     .      .      .       .   303
     2.9. PGH synthase active site       . . . . . . . . . . .                       .     .      .      .       .   304
     2.10. Regulation of PGH synthase gene expression . . . . . .                   .     .       .      .      .    304
     2.1 1 . PGHZmetabolism . . . . . . . . . . . . .                                .     .       .      .      .   305
     2.12. Catabolism: prostanoids as local hormones . . . . . . .                  .     .      .      .      .     306
3.   Hydroxy- and hydroperoxy-eicosaenoic acids and leukotrienes . . .              .     .      .      .      .     308
     3.1. Introduction and overview . . . . . . . . . . .                           .     .      .      .      .     308
     3.2. Mechanism of leukotriene biosynthesis in human neutrophils . .            .     .      .      .      .     309
     3.3. The enzymes of the 5-lipoxygenase pathway    . . . . . .                   . . . . .                       312
     3.4. Regulation of leukotriene synthesis . . . . . . . . .                      . . . . .                       313
     3.5. The metabolism of lipoxygenase products . . . . . . .                     . . . . .                        315
     3.6. Biological activities of leukotrienes . . . . . . . . .                   .     . . . .                    317
4.   Epoxygenase products . . . . . . . . . . . . . .                               .     . . . .                    318
     4.1.    Introduction . . . . . . . . . . . . . . . .                            .     . . .                     318
     4.2. Structures. nomenclature, and biosynthesis . . . . . . .                   .     . . . .                   318
     4.3. Occurrence of epoxyeicosatrienoic acids . . . . . . .                     .     . . . .                    321
     4.4. Metabolism of epoxygenase metabolites of arachidonic acid . .             . . . . .                        321
     4.5. Biological actions of 'epoxygenase' derived EpETrEs and HETrEs             . . . . .                       322

5.   Future directions . . . . . . . . . . . .              . . . .       .    .    .   .    .       322
     5. I . Cyclooxygenase metabolites . . . . . .        . . . .       .      .    .    .    . .    322
     5.2. Lipoxygenase metabolites . . . . . .            . . . .       .     .    .    .    . .     323
     5.3. Epoxygenase metabolites      . . . . . .        . . . .       .     .    .    .    . .     323
References . . . . . . . . . . . . . . .                   . . . .       .     .   .    .    .       323

Chapter 11 . Sphingolipids
Charles C. Sweeley . . . . . . . . . . . . . . . . . . . .                                           327

1.   Introduction . . . . . . . . . . . . . . . . . .                              .    .    .   .   321
2.  Chemistry and distribution . . . . . . . . . . . . . .                         .    .    .   .   330
     2.1. Sphingoid bases . . . . . . . . . . . . . . .                            .    .    .   .   330
    2.2. Ceramides         . . . . . . . . . . . . . . . .                         . . . .           332
    2.3. Glycosphingolipids . . . . . . . . . . . . . .                            . . . .           333
            2.3.1. Neutral glycosphingolipids . . . . . . . . . .                  . . . .           333
            2.3.2. Acidic glycosphingolipids . . . . . . . . . .                   . . . .           334
   Gangliosides . . . . . . . . . . . . .                        . . . .           334
   Phosphorus-containing glycosphingolipids . . . . .            .    .    .   .   336
   Sulfatoglycosphingolipids . . . . . . . . . .                 .    .    .   .   337
3. Biosynthesis of sphingolipids . . . . . . . . . . . . .                         .    .    .   .   337
     3.1. Sphingoid bases and ceramide . . . . . . . . . . .                       .    .    .   .   338
     3.2. Sphingomyelin . . . . . . . . . . . . . . .                              .    .    .   .   340
     3.3. Sulfatoglycosphingolipids . . . . . . . . . . . .                        .    .    .   .   340
     3.4. Neutral glycosphingolipids . . . . . . . . . . . .                       .    .    .   .   341
            3.4.1. Cell surface glycosyltransferases . . . . . . . .               .    .    .   .   345
     3.5. Gangliosides . . . . . . . . . . . . . . . .                             .    .    .   .   345
4. Turnover of sphingolipids . . . . . . . . . . . . . .                           .    .    .   .   347
     4.1. Lysosomal metabolism of sphingolipids . . . . . . . . .                  .    .    .   .   348
            4.1.1. Sphingomyelin . . . . . . . . . . . . .                         . . . .           348
            4 . I .2. Ceramide . . . . . . . . . . . . . .                         . . . .           349
            4.1.3. Glycosphingolipids . . . . . . . . . . . .                      . . . .           349
     4.2. Glycosphingolipid storage diseases . . . . . . . . . .                   . . . .           350
            4.2. I . Tay-Sachs disease and related disorders . . . . . .           . . . .           351
   a-Subunit mutations . . . . . . . . . . .                     .    .    .   .   352
   B-Subunit mutations     . . . . . . . . . . .                 .    .    .   .   353
   Defective G,, activator protein . . . . . . . .               .    .    .   .   353
            4.2.2. Fabry’s disease and Schindler’s disease     . . . . . .         .    .    .   .   353
5. Regulation of glycosphingolipid metabolism . . . . . . . . .                    .    .    .   .   354
     5.1. Developmental changes . . . . . . . . . . . . .                          .    . . .        355
            5 .1.1. Glycosphingolipid markers in leukocyte differentiation . .     .    . . .        355
            5.1.2. Early embryonic markers . . . . . . . . . .                     . . . .           356
     5.2. Oncogenic transformation and immunomodulation by shed gangliosides       . . . .           356
6. Biological activities and functional roles of sphingolipids                                       357
     6.1. Adherence of microorganisms to cell surface glycosphingolipids . .       . . . .           357
     6.2. Modulation of signal transduction . . . . . . . . . .                    . . . .           358
            6.2.1. Glycosphingolipids . . . . . . . . . . . .                      . . . .           358
            6.2.2. Sphingoid bases . . . . . . . . . . . . .                       . . . .           359
7. Future directions . . . . . . . . . . . . . . . . .                             . . . .           359
References . . . . . . . . . . . . . . . . . . . .                                 . . . .           360

Chapter 12. Cholesterol: evolution of structure and function
Konrad Bloch . . . . . . . . . . . . . . . .                                                                                                                363

I.   Natural occurrence of sterols . . . . . . . . . .                                       .           .       .        .         .           .      . 363
2.   Metabolic and precursor functions of the sterol molecule . . .                      .           .       .        .         .        .              .364
3.   Sterol patterns . . . . . . . . . . . . . . .                                               . .            .          .      .       .              366
4.   Sterols and membrane function . . . . . . . . . .                                    .        .          .           .      .        .           . 368
     4.1. Function of sterols in animal membranes . . . . .                              .        .          .        .        .        .           . 368
     4.2. Sterol auxotrophs . . . . . . . . . . .                                          .        .          .        .        .        .           . 369
     4.3. Metabolic regulation in the membrane environment . .                           .        .          .        .        .        .           . 370
     4.4. Fluidity control . . . . . . . . . . . .                                        .        .          .        .         .        .           . 371
5. Regulatory roles for sterols in membranes . . . . . . .                                .        .          .        .        .        .           . 372
     5.1. Fungi and invertebrates . . . . . . . . . .                                      .        .          .        .        .        .           . 372
     5.2. Sterol effects on hormone and neurotransmitter receptors .                     .           .       .        .        .        .            . 373
     5.3. Sterols and fusion of viral membranes . . . . . .                              . . . . . .                                                . 374
     5.4. Capping of surface immunoglobulin        . . . . . .                           . . . . . .                                                  . 375
     5.5. Sterol effects on enzyme activities . . . . . . .                              .           .       .        .        .        .           . 375
     5.6. Effects of cholesterol on phospholipid synthesis . . .                         . .                 .        .        .        .            . 376
6. Evolution of the sterol structure . . . . . . . . . .                                  . .                 .        .        .        .            . 377
7. Future directions . . . . . . . . . . . . . . .                                          . .                .        .       .        .               380
References . . . . . . . . . . . . . . . . . .                                             . .                .        .       .        .                381

Chapter 13. Regulation of sterol biosynthesis and isoprenylation of proteins
Peter A . Edwards . . . . . . . . . . . . . . . . . . . . .                                                                                                 383

I.   Introduction . . . . . . . . . . . . .                        .        .        .           .       .        .       .         .       .           383
2.   Regulation of cholesterol synthesis . . . . . .          .        .        .        .           .       .        .         .           .           .
3.   Regulation of cholesteryl ester synthesis . . . .        .        .        .        .           .       .         .        .        .           . 387
4.   Transcriptional control of cholesterol synthesis . . .   .        .        .        .           .       .        .        .        .           . 389
5. Posttranscriptional regulation . . . . . . .               .        .        .        .           .       .         .        .        .           . 390
6. Identification of the physiological regulators . . .       .        .        .        .           .       .        .        .         .           . 392
7 . Regulation of bile acid synthesis     . . . . . .          .        .        .        .           .        .        .        .        .           . 394
8 . Isoprenylation of proteins . . . . . . . .                .      . . . . . . . .                                                                . 396
9. Future directions . . . . . . . . . . . .                       . . . . . . . . .                                                                    399
References . . . . . . . . . . . . . . .                           . . . . . . . . .                                                                    400

Chapter 14. Lipoprotein structure and secretion
Roger A . Davis . . . . . . . . . . . . . . . . .                                                                                                           403

1.   Introduction . . . . . . . . . . . . . . .                                   . . .                        .        .       . .   403
2.   Structure and function of plasma lipoproteins . . . . .                     . . .                        .        .        . . . 404
     2.1. Chylomicrons and very low density lipoproteins . .                    . . .                        .        .        . . . 404
     2.2. Structure: surface components . . . . . . .                           . . .                        .        .        . . . 405
     2.3. Core components . . . . . . . . . . .                                 .        .           .       .        .        .        .           .       408
     2.4. Low density lipoproteins: structure . . . . . .                       .        .           .        .       .            .        .           . 408
            2.4.1. Origin of low density lipoproteins . . . .                   .        .           .       .        .        .        .           .       409
            2.4.2. High density lipoproteins: structure and function            . . . . . . . .                                                             409
            2.4.3. Discoidal high density lipoproteins . . . .                  .        .           .       .        .        .        .           .       409

3.   Assembly of nascent lipoproteins . . . . . . . .
     3.1. Triacylglycerol-rich lipoproteins . . . . . . . . . . . . . . .
            3.1 .1. Role of lipids . . . . . . . . . . . . . . . . .
     3.2.   Core lipids . . . . . . . . . . . . . . . . . . . .
            3.2.1. Role of apolipoproteins . . . . . . . . . . . . . . .
            3.2.2.    Apolipoprotein B structure . . . . . . . . . . . . . .
     3.3. Vitellogenin: the primordial apolipoprotein . . . . . . . . . . . .
     3.4. Very low density lipoprotein assembly: ejection of a ball of fat out of the endoplasmic
            reticulum      . . . . . . . . . . . . . . . . . . . .                                     417
     3.5.   Filling the hydrophobic core: a thermodynamic dilemma . . . . . . . .                      422
     3.6.   High density lipoprotein synthesis and secretion . . . . . . . . . .                       422
4 . Future directions . . . . . . . . . . . . . . . . . . . .                                          423
References . . . . . . . . . . . . . . . . .                                                           424

Chuppter 15. Dynamics of lipoprotein transport in the circulutory system
Phoebe E. Fielding und Christopher J . Fielding . . . . . . . .                                        427

I.      Overview . . . . . . . . . . . . . . . . . . . . . . .                                         427
2.      Lipoprotein lipase and the initial metabolism of lipoprotein triacylglycerol . . . . .         430
        2.1. Initial events . . . . . . . . . . . . . . . . . . . .                                    430
        2.2.    Lipoprotein lipase structure . . . . . . . . . . . . . . . .                           431
        2.3. Synthesis and transport of lipoprotein lipase . . . . . . . . . . .                       431
        2.4.    The lipoprotein lipase reaction and the role of apolipoprotein C-2 . . . . . .         432
        2.5.    Lipoprotein lipase activity at the vascular surface . . . . . . . . . .                434
        2.6.    Regulation of lipoprotein lipase activity by products of lipolysis . . . . . .         435
        2.7.    Physiological regulation of lipoprotein lipase . . . . . . . . . . .                   436
        2.8. Congenital lipoprotein lipase deficiency . . . . . . . . . . . . .                        437
3.      Hepatic triacylglycerol lipase and the later metabolism of lipoprotein triacylglycerol . . .   437
        3.1. Introduction . . . . . . . . . . . . . . . . . . . .                                      437
        3.2. Structure of hepatic triacylglycerol lipase . . . . . . . . . . . .                       437
        3.3.    Acyl group hydrolysis and exchange catalyzed by hepatic triacylglycerol lipase . .     438
        3.4.    Hepatic triacylglycerol lipase and the hydrolysis of lipoprotein lipids . . . . .      438
        3.5.    Regulation of hepatic triacylglycerol lipase activity . . . . . . . . . .              439
        3.6.    Human hepatic triacylglycerol lipase deficiency . . . . . . . . . .                    440
        3.7.    Phospholipid exchange protein and its activity . . . . . . . . . . .                   440
4.      Lccithin:cholesterol acyltransferase and the initial metabolism of lipoprotein cholesterol .   440
        4.1.    Introduction . . . . . . . . . . . . . . . . . . . .                                   440
        4.2.                                                                                  .
                ‘Forward’ cholesterol transport by very low density lipoprotein and its product low
                density lipoprotein . . . . . . . . . . . . . . . . . .                                444
        4.3.    Origin of the major high density lipoprotein proteins . . . . . . . . .                445
        4.4. Classification of plasma and lymph high density lipoprotein . . . . . . .                 446
        4.5.    Apoprotein transfers in high density lipoprotein . . . . . . . . . .                   446
        4.6. Origin of plasma lecithin:cholesterol acyltransferase . . . . . . . . .                   447
        4.7.    Structure of plasma 1ecithin:cholesterol acyltransferase . . . . . . . .               447
        4.8.    Mechanism of plasma 1ecithin:cholesterolacyltransferase activity . . . . . .           448
        4.9. Role of apo A-I in the plasma 1ecithin:cholesterol acyltransferase reaction . . .         449
        4.10. Transfer of cholesterol among high density lipoprotein species . . . . . . .             450
        4.1 1 . Physiological regulation of plasma lecithin:cholesterol acyltransferase activity . .   452
        4.12. Studies of plasma 1ecithin:cholesterol acyltransferase deficiency . . . . . .            452

5.   Cholesteryl ester transfer protein and the later metabolism of lipoprotein   cholesterol .            .   452
     5.1. Introduction . . . . . . . . . . . . . . . .                                 . .     .
                                                                                             452   .
     5.2. Structure of cholesteryl ester transfer protein . . . . . .             . . . . . 454
     5.3. Mechanism of cholesteryl ester transfer protein activity . . .          . . . . . 454
     5.4. Quantitation of cholesteryl ester transfer    . . . . . . .              . . . . . 456
     5.5. Physiological regulation of cholesteryl ester transfer protein . .      .        .       .   .   . 456
     5.6. Congenital cholesteryl ester transfer protein deficiency . . . .        .        .       .   .   . 457
     5.7. Cholesteryl ester transfer protein inhibitor protein . . . . .          . . . . .                    451
6. Summary and future directions . . . . . . . . . . . .                           . . . . .                   457
References . . . . . . . . . . . . . . . . . . . .                                  . . . .                    458

Chupter 16. Removal of lipoproteinsfrom plasmu
Wolfgang J . Schneider . . . . . . . . .                      .   .   .   .   .    .       .       .   .   . 461

1.   Introduction . . . . . . . . . . . . . . . . . . . . . .                                                  461
2.   Removal of low density lipoprotein from the circulation . . . . . . . . . .                               464
     2.1. Receptor-mediated endocytosis . . . . . . . . . . . . . . .                                          464
     2.2. The low density lipoprotein receptor pathway . . . . . . . . . . .                                   465
            2.2. 1 . Familial hypercholesterolemia: clinical consequences of low density lipo-
                     protein receptor dysfunction . . . . . . . . . . . . .                                    466
            2.1.2. Biochemical findings in cultured fibroblasts from familial hypercholest-
                     erolemia homozygotes . . . . . . . . . . . . . . .                                        467
     2.3. Biosynthesis and structure of the low density lipoprotein receptor . . . . . .                       467
            2.3.1. The low density lipoprotein receptor protein . . . . . . . . .                              468
            2.3.2. The ligand binding domain          . . . . . . . . . . . . .                                469
            2.3.3. The EGF precursor homology domain . . . . . . . . . .                                       470
            2.3.4. The third domain . . . . . . . . . . . . . . . .                                            470
            2.3.5. The membrane anchoring domain . . . . . . . . . . . .                                       470
            2.3.6. The cytoplasmic tail        . . . . . . . . . . . . . . .                                   471
     2.4. Molecular defects in low density lipoprotein receptors of patients with familial hyper-
            cholesterolemia . . . . . . . . . . . . . . . . . . .                                              471
            2.4.1. The gene for the human low density lipoprotein receptor . . . . . .                         471
            2.4.2. Four groups of low density lipoprotein receptor mutations . . . . .                         472
   Class I : no delectable precursor . . . . . . . . . . . .                                 472
   Class 2: slow or absent processing of precursor . . . . . . . .                           472
   Class 3: abnormal ligand binding . . . . . . . . . . . .                                  473
   Class 4: internalization defective . . . . . . . . . . . .                                474
3 . Catabolism of chylomicrons . . . . . . . . . . . . . . . . .                                               475
4 . High density lipoprotein as a transport vehicle . . . . . . . . . . . . .                                  477
5 . Athcrosclerosis . . . . . . . . . . . . . . . . . . . . .                                                  418
     5.1. Uptake and modification of low density lipoprotein in the intinla . . . . . .                        478
     5.2. Scavenger receptors . . . . . . . . . . . . . . . . . .                                              480
6 . Lipoprotein transport in the laying hen . . . . . . . . . . . . . . .                                      481
7. Low density lipoprotein metabolism by rat serosal mast cells . . . . . . . . .                              483
8. Future directions . . . . . . . . . . . . . . . . . . . . .                                                 485
References . . . . . . . . . . . . . . . . . . . . . . . .                                                     485

Chapter 17 . Lipid assembly into cell membranes
Dennis R . Voelker . . . . . . . . . . . . . . . . . .                                                489

I.   Introduction . . . . . . . . . . . . . . . . . . . . . .                                         489
2.   The diversity of lipids . . . . . . . . . . . . . . . . . . .                                    489
3.   Methods to study intra- and inter-membranelipid transport . . . . . . . . .                      492
     3.1. Fluorescent probes . . . . . . . . . . . . . . . . . .                                      494
     3.2. Spin labeled analogs . . . . . . . . . . . . . . . . . .                                    494
     3.3. Asymmetric chemical modification of membranes . . . . . . . . . .                           495
     3.4. Phospholipid transfer proteins . . . . . . . . . . . . . . .                                496
     3.5. Rapid plasma membrane isolation . . . . . . . . . . . . . .                                 497
     3.6. Organelle specific lipid metabolism . . . . . . . . . . . . . .                             498
4 . Lipid transport processes . . . . . . . . . . . . . . . . . .                                     498
     4.1. Intramembrane lipid translocation and model membranes . . . . . . . .                       498
     4.2. Intramembrane lipid translocation and biological membranes . . . . . . .                    502
            4.2.1. Procaryotes . . . . . . . . . . . . . . . . . .                                    502
            4.2.2. Eucaryotes . . . . . . . . . . . . . . . . . .                                     503
   Transbilayer movement at the endoplasmic reticulum . . . . . . .                 503
   Transbilayer movement of phosphatidylcholine in erythrocytes . . . .             504
   Transbilayer movement of phosphatidylcholine in nucleated cells . . .            505
   ATP dependent transbilayer movement of aminophospholipids at the plasma
                     membrane of eukaryotic cells . . . . . . . . . . . . .                           506
     4.3. Intermembrane lipid transport . . . . . . . . . . . . . . .                                 507
            4.3.1. Transport in procaryotes . . . . . . . . . . . . . .                               507
            4.3.2. Transport in eucaryotes . . . . . . . . . . . . . .                                510
   Phosphatidylcholine    . . . . . . . . . . . . . . .                             510
   Phosphatidylethanolamine . . . . . . . . . . . . . .                             512
   Phosphatidylserine . . . . . . . . . . . . . . . .                               513
   Sphingolipids . . . . . . . . . . . . . . . . .                                  515
   Cholesterol . . . . . . . . . . . . . . . . . .                                  517
   Phospholipid transfer proteins and phosphatidylinositol transport . . .          518
5. Future directions . . . . . . . . . . . . . . . . . . . . .                                        521
References . . . . . . . . . . . . . . . . . . . . . . . .                                            522

Chapter 18. Assembly of proteins into membranes
Reinhart A.E Reithmeier . . . . . . . .                                                               525

1.   Organization ofmembrane proteins       . . . . . .         .   .   .   .   .   .    .    .    . 525
     1.1. Classification of membrane proteins . . . .           .   .   .   .   .   .    .    .    . 525
     1.2. Membrane protein structure and energetics . . .       .   .   .   .   .   .    .    .    . 528
     1.3. Assembly of membrane proteins . . . . .               .   .   .   .   .   .    .    .    . 530
2.   Secretion o proteins and the signal hypothesis . . . .
                f                                               .   .   .   .   .   .    .    .    . 532
     2.1. The Palade secretion pathway . . . . . .              .   .   .   .   .   .    .    .    . 533
     2.2. The Blobel signal hypothesis . . . . . . .            .   .   .   .   .    .    .    .    . 533
     2.3. In vitro translation and translocation systems . .    .   .   .   .   .   .    .    .    . 535
     2.4. The Milstein experiment: secreted proteins are made   with an amino terminal signal
            sequence . . . . . . . . . . . .                    . . . . . . . . . 538
     2.5. Signal sequences . . . . . . . . . .                  . . . . . . . . . 540

3.   The targeting and translocation machinery . . . . . . . . . . . . . .                    541
     3.1. The signal recognition particle . . . . . . . . . . . . . . .                       542
     3.2. The signal recognition particle receptor . . . . . . . . . . . . .                  543
     3.3. Translocation components . . . . . . . . . . . . . . . .                            544
     3.4. Ribosome-binding proteins . . . . . . . . . . . . . . . .                           545
     3.5. Signal peptidase . . . . . . . . . . . . . . . . . . .                              546
4 . Biosynthesis of type I simple membrane proteins . . . . . . . . . . . .                   546
     4.1. IgM and the relationship between the biosynthesis of secreted proteins and simple
            membrane proteins . . . . . . . . . . . . . . . . . .                             546
     4.2. Vesicular stomatitis virus glycoprotein and hemagglutinin . . . . . . . .           549
     4.3. Loop models and insertion into the lipid bilayer . . . . . . . . . .                551
     4.4. Evidence for a loop structure for insertion of signal sequences . . . . . . .       553
5. Biosynthesis of type I1 simple membrane proteins . . . . . . . . . . . .                   554
     5.1.   Ovalbumin, a secreted protein made without a cleaved signal sequence . . . .      556
     5.2. Asialoglycoprotein receptor . . . . . . . . . . . . . . . .                         557
     5.3. Sucrase-isomaltase . . . . . . . . . . . . . . . . . .                              557
6 . Biosynthesis of cytochrome P-450 and cytochrome b, . . . . . . . . . . .                  558
     6.1. Cytochrome P-450 . . . . . . . . . . . . . . . . . .                                558
     6.2. Cytochrome b, . . . . . . . . . . . . . . . . . . .                                 559
7. Biosynthesis of complex membrane proteins . . . . . . . . . . . . .                        559
     7 1. Artificial membrane proteins . . . . . . . . . . . . . . .                          559
     7.2. Band 3, the anion transport protein of the erythrocyte membrane . . . . . .         561
     7.3. Glucose carrier . . . . . . . . . . . . . . . . . . .                               561
     7.4. Rhodopsin . . . . . . . . . . . . . . . . . . . .                                   561
     7.5. Ca2' ATPase and calsequestrin . . . . . . . . . . . . . . .                         562
8. Glycosylation of proteins . . . . . . . . . . . . . . . . . .                              562
     8. I . N-glycosylation . . . . . . . . . . . . . . . . . . .                             562
     8.2. Processing of the oligosaccharide chain . . . . . . . . . . . . .                   563
     8.3. 0-glycosylation . . . . . . . . . . . . . . . . . . .                               566
9. Attachment of lipid to proteins . . . . . . . . . . . . . . . . .                          566
     9.1, Fatty acylation . . . . . . . . . . . . . . . . . . .                               566
     9.2. Phosphatidylinositol anchors          . . . . . . . . . . . . . . .                 567
10. Protein folding and exit from the endoplasmic reticulum . . . . . . . . . .               567
     10.1. Protein folding . . . . . . . . . . . . . . . . . . .                              567
     10.2. Disulfide formation . . . . . . . . . . . . . . . . . .                            568
     10.3. Assembly of multisubunit systems . . . . . . . . . . . . . .                       569
     10.4. Exit from the endoplasmic reticulum . . . . . . . . . . . . .                      570
     10.5. KDEL, an endoplasmic reticulum retention signal . . . . . . . . . .                570
11. Transport and targeting of proteins . . . . . . . . . . . . . . . .                       571
     11.1. Vesicles move proteins between organelles . . . . . . . . . . . .                  571
     11.2. Role of GTP-binding proteins . . . . . . . . . . . . . . .                         573
     11.3. Lysosomal targeting . . . . . . . . . . . . . . . . . .                            575
     11.4. Protein sorting in epithelial cells . . . . . . . . . . . . . . .                  576
12. Future directions . . . . . . . . . . . . . . . . . . . . .                               577
References . . . . . . . . . . . . . . . . . . . . . . . .                                    577

Index . . . . . . . . . . . . . . . . . . . . . . . . .                                       579
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemistry o Lipidv. Lipoproteins und Mrmhrcmrs
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                  1

                                                                                   CHAPTER I

          Physical properties and functional roles of
                                lipids in membranes
                                       PIETER R. CULLIS’ and MICHAEL J . HOPE2

Biochemistry Departnient und ‘Division of Dermatology, Department oj’hfedicine, University of
                                                  British Columbiu, Vancouver, B. C., ~ a n a d a

1. Introduction and overview

Biological membranes contain an astonishing variety of lipids. As detailed through-
out this book, generation of this diversity requires elaborate metabolic pathways.
The lipid compounds representing the end products of these pathways must bestow
significant evolutionary advantages to the cellular or multicellular systems in which
they reside, implying particular functional roles for each component. However, clari-
fication of the functional roles of individual lipid species has proven a difficult prob-
lem. Here we present a synopsis of the physical properties of lipid systems and indi-
cate how they may relate to the functional capacities of biological membranes.
   The major role of membrane lipids has been understood in broad outline since the
early experiments of Gorter and Grendell [ l]? who extracted lipids from the erythro-
cyte membrane and measured the areas these lipids were able to cdver as a monolayer
at an air-water interface. Although a number of unwarranted assumptions were
made in the analysis of these data, the errors fortunately compensated for one anoth-
er and led to the correct conclusion that the erythrocytes contained sufficient lipid to
provide a bilayer lipid matrix surrounding the red blood cell. This bilayer lipid or-
ganization. which provides a permeability barrier between exterior and interior com-
partments. has remained a dominant theme in our understanding of the organization
and function of biological membranes. Subsequent observations that such bilayers
are fluid, allowing rapid lateral diffusion of lipid and protein in the plane of the
membrane, and that membrane proteins are often inserted into and through the lipid
matrix. have further contributed to our present understanding of membranes, result-
ing in the Singer and Nicholson [2] fluid mosaic model, a refined version of which is
shown in Fig. 1.
   The ability of lipids to assume the basic bilayer organization is dictated by a unify-
ing characteristic of membrane lipids - namely, their amphipathic character, which
is indicated by the presence of a polar or hydrophilic (water loving) head group re-

Fig. 1. The topography of membrane protein, lipid and carbohydrate in the fluid mosaic model of a typical
eucaryotic plasma membrane. Phospholipid asymmetry results in the preferential location of phosphati-
dylethanolamine and phosphatidylserine in the cytosolic monolayer. Carbohydrate moieties on lipids and
proteins face the extracellular space. dyl represents the transmembrane potential, negative inside the cell.

gion and nonpolar or hydrophobic (water hating) region. The chemical nature of
these hydrophilic and hydrophobic sections can vary substantially. However, the
lowest-energy macromolecular organizations assumed in the presence of water have
similar characteristics, where the polar regions tend to orient toward the aqueous
phase, while the hydrophobic sections are sequestered from water. In addition to the
familiar bilayer phase, a number of other macromolecuIar structures are compatible
with these constraints, as indicated later in this chapter. It is of particular interest that
many naturally occurring lipids prefer nonbilayer structures in isolation.
   The fluidity of membranes depends on the nature of the acyl chain region compris-
ing the hydrophobic domain of most membrane lipids. Most lipid species in isolation
can undergo a transition from a very viscous gel (frozen) state to the fluid (melted)
liquid-crystalline state as the temperature is increased. This transition has been stud-
ied intensively, since the local fluidity, as dictated by the gel or liquid-crystalline na-
ture of membrane lipids, may regulate membrane-mediated processes. However, at
physiological temperatures most, and usually all, membrane lipids are fluid; thus, the
major emphasis of this chapter will concern the properties of liquid-crystalline lipid
systems. As indicated later, the melted nature of the acyl chains depends on the pres-

ence of cis double bonds, which can dramatically lower the transition temperature
from the gel to the liquid-crystalline state for a given lipid species.
   The ability of lipids to self-assemble into fluid bilayer structures is consistent with
two major roles in membranes: establishing a permeability barrier and providing a
matrix with which membrane proteins are associated. Roles of individual lipid com-
ponents may therefore relate to establishing appropriate permeability characteristics,
satisfying insertion and packing requirements in the region of integral proteins
(which penetrate into or through the bilayer), as well as allowing the surface associa-
tion of peripheral proteins via electrostatic interactions. All these demands are clearly
critical. An intact permeability barrier to small ions such as Na' , K', and H', for
example, is vital for establishing the electrochemical gradients which give rise to a
membrane potential and drive other membrane-mediated transport processes. In ad-
dition, the lipid in the region of membrane protein must seal the protein into the
bilayer so that nonspecific leakage is prevented and an environment appropriate to
a functional protein conformation is provided.
   In summary, membrane lipids satisfy demands related to membrane structure,
fluidity, and permeability, as well as protein association and function. These aspects
will be dealt with at length; however, before a coherent discussion is possible, a basic
overview of lipid diversity in membranes, a study of the methods employed to isolate
individual components, and a discussion of the physical properties of lipids are essen-
tial. These will comprise the bulk of Sections 2 4 . More extended discussions of bio-
membranes and the roles of lipids can be found in the excellent text by Gennis [3].

2. Lipid diversity and distribution
The general definition of a lipid is a biological material soluble in organic solvents,
such as ether or chloroform. Here we shall discuss the diverse chemistry of the sub-
class of lipids which are found in membranes. This excludes other lipids which are
poorly soluble in bilayer membrane systems, such as fats (triacylglycerols) and cho-
lesterol esters.

2. I . Chemical diversity o lipids

The major classes of lipids found in biological membranes are summarized in Fig. 2.
We shall discuss most of these compounds in depth at various points in this book; we
present only a brief synopsis here. In eucaryotic membranes the glycerol-based phos-
pholipids are predominant, including phosphatidylcholine, phosphatidylethanola-
mine, phosphatidylserine, phosphatidylinositol, and cardiolipin. Sphingosine-based
lipids, including sphingomyelin and the glycosphingolipids, also constitute a major
fraction. The glycolipids, which can also include carbohydrate-containing glycerol-
based lipids (found particularly in plants), play major roles as cell-surface-associated

    /""'     0           - CH$H&H3

                                         fiH I
                         - CH?- CH
                                    coo -

                         - CH,
                          HC   OH
                         - CH2

                           H        DH

                               OH   OH

Fig. 2 . The structure of the phospholipid molecule distearoylphosphatidylcholine in the liquid-crystalline
state is represented schematically. Head groups for the other major classes of phospholipid are also shown.
The glycerol moiety of a cardiolipin is esterified to two phosphatidic acid molecules.

antigens and recognition factors in eucaryotes (Chapter 1 1). The physical properties
of glycolipids have not been extensively characterized and will not be discussed in this
chapter. Cholesterol is also a major component of eucaryotic membranes, particu-
larly in mammalian plasma membranes, where it may be present in equimolar pro-
portions with phospholipid.
   In most procaryotic membranes, phosphatidylcholine is not usually present
(Chapter 2); the major phospholipids observed are phosphatidylethanolamine, phos-
phatidylglycerol, and cardiolipin. In plant membranes on the other hand, lipids such
as monogalactosyl and digalactosyl diacylglycerols can form the majority compo-
nents of membranes such as the chloroplast membrane.
   These observations give some impression of the lipid diversity in membranes, but
it must be emphasized that this diversity is much more complex. Minority species
such as sulfolipids, phospholipids with phosphorylated head groups, and lysolipids
abound. Furthermore, each lipid species exhibits a characteristic fatty acid composi-
tion. In the case of glycerol-based phospholipids, for example, it is usual to find a

saturated fatty acid esterified at the 1 -position of the glycerol backbone and an unsat-
urated fatty acid at the 2-position. Also, in eucaryotic membranes it is usual to find
that phosphatidylethanolamine and phosphatidylserine, for example, are more un-
saturated than other phospholipids. In order to give a true impression of the molecu-
lar diversity of phospholipids in a single membrane, we list in Table I the fatty acid
composition of phospholipids found in the human erythrocyte membrane. From this
table and other analyses [4]it is clear that the number of different molecular species
of phospholipids in a membrane can easily exceed 100.
   The lipid composition of membranes can vary dramatically among different cells
or organelles. In addition, different sides or monolayers of the same membrane can
contain different lipid species. These different compositions are indicated in the fol-
lowing sections.

2.2. Membrune lipid compositions

The lipid compositions of several mammalian membrane systems are given in Table
I1 (see also Chapter 17). Dramatic differences are observed for the cholesterol con-
tents. Plasma membranes such as those of myelin or the erythrocyte contain approx-
imately equimolar quantities of cholesterol and phospholipid, whereas the organelle
membranes of endoplasmic reticulum or the inner mitochondria1 membrane contain

Gas chromatographic analyses of the fatty acid chains i n human red cell phospholipid

Chain length     Total              Sphingomyelin    Phosphatidyl-     Phosphatidyl-    Phosphatidyl-
and              phospholipids                       choline           ethanolarnine    serine

I6:o"            20.1               23.6             31.2              12.9              2.1
I8:O             17.0                5.7             11.8              11.5             37.5
18:l             13.3               +                18.9              18.1              8. I
1x2               8.6                   +            22.8               7.1              3. I
20:o             +                      1.9           +                +                +
20:3              1.3               ~                    1.9            1.5              2.6
22:o              I .Y               9.5               I .Y             I .5             2.6
20:4             12.6                1.4               6.7             23.1             24.2
23:O             +                   2.0             +                 +                +
24:O              4.1               22.8              +                +                 +
22:4              3.1               -                 +                 1.5              4.0
24: 1             4.8               24.0             +                 +                +
225               2.0               ~                +                  4.3               3.4
2216              4.2               ~                  2. I             8.2              10.1

Thedata are expressed as weight % ofthe total. + Denotes that the concentration did not exceed 1% of the
total. Reproduced with permission of Van Deenen and de Gier (1974).
"This code indicates the number of carbon atoms in the chain and the number of double bonds.

The lipid composition of various biological membranes

Lipid                             Erythrocyte”       Myelin”         Mitochondriab        Endoplasmic
                                                                     (inner and           reticulum”
                                                                     outer membrane)

Cholesterol                       23                 22               3                     6
Phosphatidylethanolamine          18                 15              35                    17
Phosphatidylcholine               17                 10              39                   40
Sphingomyelin                     18                  8              -                      5
Phosphatid ylserine                 7                 9               2                     5
Cardiolipin                       -                  -               21
Glycolipid                          3                28              ~

Others                             13                 8              ~                     27

The date are expressed as weight % of total lipid.
“Human sources.
bRat liver (see Table I in chapter 17 for more complete lipid analysis of intracellular organelles).

only small amounts of cholesterol. This cholesterol distribution correlates well with
the distribution of sphingomyelin. Cholesterol may have a ‘fluidizing’ role in mem-
branes containing sphingomyelin, which is relatively saturated.
   Cardiolipin is almost exclusively localized to the inner mitochondria1 membrane,
and it has been suggested that cardiolipin is required for the activity of cytochrome
c oxidase, the terminal member of the respiratory electron-transfer chain. In general,
the lipids of more metabolically active membranes are considerably more unsaturat-
ed, as indicated in Table 111.
   It is interesting to note that the lipid composition of the same membrane system in
different species can also vary significantly. The rat erythrocyte membrane, for ex-
ample, contains low levels of sphingomyelin and elevated levels of phosphatidyl-
choline with respect to the human erythrocyte. In the bovine erythrocyte, this distri-
bution is reversed, with high sphingomyelin, and low phosphatidylcholine contents.

Double-bond composition of phospholipids of various membranes

Myelin                                0.5
Erythrocyte                           1.o
  reticulum                           1.4
  (inner)                             I .5
Nerve synapse                         >2

2.3. Transbilayer l@id asymmetry

The inner and outer leaflets of membrane bilayers may exhibit different lipid com-
positions [5]. Several different species of membranes have been investigated with re-
spect to lipid asymmetry; however, the plasma membrane of human erythrocytes re-
mains the most thoroughly investigated.
   The results obtained indicate that most membranes display some degree of lipid
asymmetry. The use of impermeable probes that react with the primary amines of
phosphatidylethanolamine and phosphatidylserine on only one side of the membrane
has shown that the majority of the amino-containing phospholipids of the erythro-
cyte are located on the inner monolayer. Combinations of chemical probes and phos-
pholipase treatments indicate that in a normal red blood cell all the phosphatidyl-
serine is located in the inner monolayer, whereas approximately 20% of the phospha-
tidylethanolamine can be detected at the outer surface, with 80% confined to the in-
ner monolayer. The outer monolayer consists predominantly of phosphatidylcho-
line, sphingomyelin, and glycolipids. Fig. 3 summarizes the transbilayer lipid dis-
tributions obtained for various mammalian cell membranes and viral membranes
derived from animal-cell plasma membranes. A common feature is that the amino-
containing phospholipids are chiefly limited to the cytosolic side of plasma mem-
                                     Outer Monolayw

                                    6                 D
                                                          SP'   DP



                                                                            - 40
                                                                            - 30
                                                                            - 10

                                     Inner monom

Fig. 3 . Phospholipid asymmetry in plasma membranes. (A) Human erythrocyte membrane, (B) rat liver
blood sinusoidal plasma membrane, (C) rat liver continuous plasma membrane, (D) pig platelet plasma
membrane, (E) VSV envelope derived from hamster kidney BHK-21 cells. See Table VI for phospholipid
branes. It is interesting that the information available for organelle membranes sug-
gests that phosphatidylethanolamine and phosphatidylserine are also oriented
toward the cytosol. A general feature of plasma membrane asymmetry is that the
majority of phospholipids that exhibit a net negative charge at physiological pH
(phosphatidylserine and phosphatidylinositol-phosphatidylethanolamine is only
weakly anionic) are limited to the cytosolic half of the bilayer. Certain proteins
appear to be involved in maintaining this asymmetry. Treatment of erythrocytes with
diamide, which induces cross-linking of the cytoskeletal protein spectrin, results in
the appearance of phosphatidylserine in the outer monolayer. Pathological red blood
cells known to have lesions associated with cytoskeletal proteins also exhibit a partial
breakdown of asymmetry, with an increased exposure of phosphatidylserine and
phosphatidylethanolamine on the outer half of the bilayer and an equivalent transfer
of phosphatidylcholine to the inner monolayer.
   These experiments suggest a possible interaction between cytoskeletal proteins and
membrane phospholipids to generate and maintain asymmetry. It has recently been
demonstrated, however, that some phospholipids will redistribute across the bilayer
of protein-free model membrane systems in response to transmembrane pH gra-
dients. Phosphatidylglycerol and phosphatidic acid, for example, will diffuse to the
inner monolayer of large unilamellar vesicles that exhibit an interior pH that is basic
with respect to the external pH [6].Similar responses to transmembrane proton gra-
dients would be expected to occur in vivo. On the other hand, an aminophospholipid
translocase (see also Chapter 17) has been identified in a number of plasma mem-
branes which appears to be responsible for the movement of phosphatidylethanol-
amine and phosphatidylserine across the bilayer. This ATP dependent ‘lipid pump’
activity has also been found in organelle membranes but oriented such that the ami-
nophospholipids are transported from the inner monolayer to the outer monolayer,
which is consistent with their phospholipid asymmetry [7].
   The functional importance of lipid asymmetry is not clear but could be related to
prevention of exposure of phosphatidylserine at the outer surface of a normal cell,
which has been suggested to be a signal of senescence [8].Alternatively, phosphati-
dylethanolamine and phosphatidylserine may be required to maintain a fusion com-
petent surface for endocytosis and organelle fusion (see Fig. 15 and Devaux, 1991).

3. Model membrane systems

The physical properties and functional roles of individual lipid species in membranes
are exceedingly difficult to ascertain in an intact biological membrane due to the
complex lipid composition. In order to gain insight into the roles of individual com-
ponents, it is necessary to construct model membrane systems that contain the lipid
species of interest. This requires three steps, namely, isolation or chemical synthesis
of a given lipid, construction of an appropriate model system containing that lipid,

and subsequent incorporation of a particular protein if understanding the influence
of a particular lipid on protein function is desired. By this method specific models of
biological membranes can be achieved in which the properties of individual lipid
components can be well characterized.

3.1. Lipid isolation undpurGcation

A variety of techniques has been developed for isolation of lipids from membranes
[9]. These differ according to the particular source and type of lipid being isolated. A
procedure commonly employed for the preparation of erythrocyte phospholipids is
illustrated in Fig. 4.A first step common to most procedures is to disrupt the mem-
brane in a solvent system which denatures and precipitates most of the protein and
solubilizes the lipid component. The Bligh and Dyer procedure is perhaps most often
employed and involves incubation of the membrane system in a chloroform-
methanol-water ( I :2:0.8) (v:v:v) mixture, which forms a one-phase system. The sub-
sequent addition of chloroform and water tr, the mixture containing the extracted
lipids results in a two-phase system where the lower (chloroform) phase contains
most membrane lipids.
   Column chromatography is usually subsequently employed for isolation of indi-
vidual lipid species. A solid phase such as silicic acid, DEAE cellulose, aluminum
oxide, or carboxymethyl cellulose is used, depending upon the lipid being isolated,
and lipids are eluted using mixtures of solvents with difrerent polarities, such as chlo-
roform and methanol. Thin-layer chromatography is generally used for lipid iden-
tification and for ascertaining purity. All these separation techniques rely upon the
different partitioning characteristics of lipids between the stationary phase surface
and mobile solvent phase for different solvent polarities. The exact nature of the
binding of lipid to the solid phase is not well understood but appears to involve both
electrostatic and hydrophobic interactions. Carboxymethyl cellulose and DEAE cel-
lulose are often used for separation of anionic lipids.
   High-pressure preparative liquid chromatography enables the rapid purification of
large quantities of natural lipids. Analytical high pressure liquid chromatography
techniques are well-developed for the rapid separation of phospholipids by head
group and acyl chain composition. Reversed-phase chromatography, where the sta-
tionary phase is hydrophobic and the mobile phase hydrophilic, is particularly use-
ful. The solid support is usually coated with hydrocarbon chains of a defined length
(and consequently of regulated hydrophobicity), and the mobile phase is hydrophilic.
This technique is particularly useful for separating single lipid classes according to
their acyl chain length and degree of unsaturation.

3.2. Techniquesfor making model membrane vesicles

Once lipids have been isolated, purified, and chemically characterized, their proper-

                                 i   I                      i


                                                      SPM       Base

         A PSrlch l a l IS
                  r cm
     eluted lrm a CM cellulose
      mmhard in c h b d m
Fig. 4.An outline of the procedure for extracting and purifying the phospholipid species of human erythro-
cytes. In step A, red cells are extracted using the Bligh and Dyer protocol. Denatured hemoglobin precipi-
tates at this stage and is readily removed by centrifugation. Two-dimensional thin-layer chromatography
is used to identify all the phospholipid species in the total lipid extract: ( I ) cholesterol, (2) phosphatidyleth-
anolamine (PE), (3) phosphatidylcholine (PC), (4) phosphatidylserine (PS) and (5) sphingomyelin (SPM).
Step B utilizes preparative liquid chromatography (LC) to obtain pure PC, PE and SPM. The PC and SPM
fractions are readily separated using chloroform/methanol/water (60:30:4, v/v) to elute the lipid from the
silica gel column. PE is further purified by passing the lipid once more through the column using chloro-
form/methanol/water/25% ammonium hydroxide (60:30:1:1, v/v). In step C an impure PS fraction, ob-
tained from the passes outlined above, is purified by elution from carboxymethyl (CM) cellulose using a
continuous gradient of &70% methanol in chloroform. For phospholipid nomenclature, see Table VI.
Acid refers to the thin-layer plate running solvent chloroform/methanol/acetic acid/water (60:30:8:3. v/v)
and base, to chloroform/methanol/25% ammonium hydroxiddwater (90:54:6:5, vlv).

ties as membrane components can be studied. For this purpose a number of tech-
niques have been developed for producing model membranes from lipids [101. Prep-
aration of the simplest model system involves the straightforward hydration of a lipid
film by mechanical agitation, such as vortex mixing. In the case of bilayer-forming
lipids, this hydration results in a macromolecular structure which is composed of a

series of concentric bilayers separated by narrow aqueous spaces [l 11. Such struc-
tures are usually referred to as liposomes or multi-lamellar vesicles (MLVs) and have
been used for many years as models for the bilayer matrix of biological membranes.
Their use is mostly restricted to physical studies on bilayer organization and the mo-
tional properties of individual lipids within a membrane structure. MLVs are not
ideal models for the study of other aspects of lipids in membrane structure and func-
tion, mainly because as little as 10% of the total lipid of a MLV is contained in the
outermost bilayer. As a result, methods have been sought by which unilamellar
(single bilayer) model membranes can be obtained either directly or from MLVs.
    Small unilamellar vesicles (SUVs) can be made from MLVs by subjecting the
MLVs to ultrasonic irradiation or by passage through a French press. However, their
small size limits their use in model membrane studies. Typically, diameters in the
range 2 5 4 0 nm are observed. The radius of curvature experienced by the bilayer in
S W s is so small (Fig. 5) that the ratio of lipid in the outer monolayer to lipid in the
inner monolayer can be as large as 2:l. As a result of this curvature, the packing
constraints experienced by the lipids perturb their physical properties in comparison
with less highly curved systems. This restricts the use of SUVs for physical studies on
the properties of membrane lipid. Moreover, the aqueous volume enclosed by the
SUV membrane is often too small to allow studies of permeability or ion distribu-
tions between the internal and external aqueous compartments.
   A more useful membrane model is the large unilamellar vesicle (LUV) system,
where the mean diameter is larger, and the distribution of lipid between the outer and
inner monolayers is closer to 1 : 1. The most common procedures for producing LUVs
result in unilamellar vesicles with diameters ranging from 50 to 500 nm. These prepa-
rative procedures usually include the use of detergents or organic solvents, although
LUVs can be produced directly from MLVs.
    Procedures that employ detergents vary depending upon the type of detergent;
however, the principle is the same. Lipids are solubilized by the detergent of choice
(such as cholate or octylglucoside); then the detergent is removed either rapidly by
dilution or gel filtration, or slowly by dialysis. As the detergent concentration decrea-
ses, the lipids adopt unilamellar vesicular structures. The vesicle size can be con-
trolled to some extent by the rate at which detergent is removed.
   A number of methods employing organic solvents exist for preparation of LUVs
[ 121. The lipid is first solubilized in an organic solvent which is subsequently diluted
by aqueous buffer. The largest unilamellar vesicles are produced by injection proce-
dures whereby lipid is dissolved in ether or ethanol, then slowly injected into aqueous
buffer. An alternative protocol employing organic solvent is called the reverse phase
evaporation procedure, which involves making an emulsion of lipid (dissolved in
ether or mixtures of other organic solvents) and aqueous buffer. The organic solvent
is carefully removed under partial vacuum, which gives rise to hydrated lipid in the
form of a thick gel. This gel can be diluted and sized by extrusion through polycar-
bonate filters of defined pore sizes to give LUVs.

        Curvature and some characteristicsof large unilamellarand Small unilamellarVBSiCleS

                   500 nm



                  I      IMJOM
                       (mole ratio)

                          0 36

                                          &I per pmol)


                                                             No phospholipid


                                                                 80 x l(r
                                                                                    No vesicles
                                                                                    per pmol
                                                                                    d lipid

                                                                                     1I   x

                                                                                     76 x 100

       500                  096                17                22x105              2 7 x 10"

Fig. 5. The curvature and some characteristics of large unilamellar vesicles (LUV) and small unilamellar
vesicles (SUV). LUVs typically have diameters in the range 100-500 nm. SUVs prepared by sonication can
be as small as 25 nm in diameter. The radius of curvature for each vesicle size is shown in proportion. The
ratio of lipid in the inner monolayer (IM) compared with lipid in the outer monolayer (OM) gives an
indication of the packing restrictions in bilayers with a small radius of curvature. The trapped volume
refers to the volume of aqueous medium enclosed per micromole of phospholipid. The calculations were
made assuming a bilayer thickness of 4 nm and a surface area per phospholipid molecule of 0.6 nm'.

   A technique that is gaining increasing popularity involves the direct extrusion un-
der moderate pressures (5500 psi) of MLVs through polycarbonate filters of defined
pore size. This technique can generate LUVs with size distributions in the range of
50-200 nm, depending on the pore size of the filter employed [lo]. This technique has
numerous advantages in that it does not require detergents or solvents, which are
difficult to remove. Also, it can be applied to all lipids which adopt liquid crystalline
bilayer structures, including long chain saturated lipids. Finally, the technique is
rapid, straightforward and convenient, allowing LUVs to be prepared in 10 min or

3.3. Techniquesfor making planar biluyers and monolayers

Planar bilayers (also known as black lipid membranes) are favorite model mem-
branes of electrophysiologists interested in current flow across a bilayer. They are
formed by dissolving phospholipids in a hydrocarbon solvent and painting them
across a small aperture (approximately 2 mm in diameter) which separates two
aqueous compartments. The solvent tends to collect at the perimeter of the aperture,
leaving a bilayer film across the center. The electrical properties of the barrier are
readily measured employing electrodes in the two buffered compartments. It is also
possible to incorporate some membrane proteins into the film, if the protein can be
solubilized by the hydrocarbon. With this technique, ion channels have been recon-
stituted and voltage-dependent ion fluxes recorded. The most serious problem of
black lipid membranes is the presence of the hydrocarbon solvent, which may change
the normal properties of the lipid bilayer being studied. More recent techniques avoid
some of these problems [13].
   In monolayer systems, amphipathic lipids orient at an air-water interface. The re-
sult is a monolayer film which, in the case of phospholipids, represents half of a bi-
layer, where the polar regions are in the aqueous phase and the acyl chains extend
above the buffer surface. Such films can be compressed and their resistance to com-
pression measured. The study of compression pressure versus surface area (occupied
by the film) yields information on molecular packing of lipids and lipid-protein inter-
actions. Perhaps the best-known result of monolayer studies is the condensation ef-
fect of cholesterol and phospholipid, in which the area occupied by a typical mem-
brane phospholipid molecule and a cholesterol molecule in a monolayer is less than
the sum of their molecular areas in isolation. This phenomenon provides a strong
indication of a specific interaction between this sterol and membrane phospholipids
[14]. For more information on monolayer studies, see Chapter 9.

3.4. Reconstitution o integral membrane protein into vesicles

An important step, both for the study of membrane protein function and for the
building of simple but more representative biological membranes, is the insertion of
purified integral membrane proteins into well-defined lipid model membranes. A
large variety of membrane proteins have been reconstituted [ 153. For the purpose of
discussing the salient features of reconstitution techniques, we shall use the example
of cytochrome c oxidase from bovine heart mitochondria. This integral membrane
protein, which has been purified and is relatively well characterized, spans the inner
mitochondria1 membrane and oxidizes cytochrome c in the terminal reaction of the
electron-transfer chain.
   Purified integral proteins such as cytochrome oxidase maintain a functional con-
formation when solubilized in detergents. The goals of reconstitution can be sum-
marized as follows. First, the protein must be inserted into a bilayer of desired lipid

Fig. 6. (A) Rotary-shadowed freeze-fracture micrograph of beef heart cytochrome c oxidase reconstituted
into a vesicle of dioleoylphosphatidylcholine by the cholate dialysis procedure, at a protein to lipid ratio of
1 :15 (w/w). (B) Unidirectionally shadowed freeze-fracture micrographs of cytochrome c oxidase reconsti-
tuted at protein to lipid ratios of <1:5000 (w/w). Each particle has been shown to represent one dimer of
cytochrome c oxidase and is approximately 10 nm in diameter. The bar represents 100 nm (see Madden
(1984) for details of the reconstitution procedure).

composition. This insertion is commonly achieved by solubilizing the lipid in deter-
gent, mixing the solubilized lipid and protein together, then removing the detergent
by dialysis. Second, the reconstituted systems must have constant lipid to protein
ratios between vesicles. Most reconstitution procedures give rise to heterogeneous
systems, where vesicles contain various amounts of protein. Column chromatog-
raphy techniques can be employed to obtain systems exhibiting uniform lipid to
protein ratios [15]. Finally, the systems should have asymmetric protein orientation.
In contrast with the intact biological membrane, the protein in reconstituted systems
is not necessarily inserted with a well-defined asymmetric orientation. In the case of

reconstituted cytochrome oxidase systems, for example, oxidase-containing vesicles
can exhibit protein orientations in which the cytochrome c binding sites are on the
outside or the inside. Asymmetric protein orientation can be achieved by reconstitu-
tion at low protein to lipid ratios such that most vesicles contain one or zero protein
molecules. Populations containing only one oxidase molecule per vesicle with well-
defined transmembrane orientations of the oxidase can subsequently be achieved by
ion-exchange or affinity column chromatography, as illustrated in Fig. 6 .
   In some cases asymmetric incorporation of other proteins can be achieved by dif-
ferent procedures. Erythrocyte glycophorin, for example, has a large carbohydrate-
containing region which is normally localized on the exterior of the red cell. Recon-
stituted systems can be obtained by hydrating a dried film of lipid and glycophorin,
resulting in asymmetric vesicles in which more than 80%of the carbohydrate groups
are on the vesicle exterior. This is presumably due to the small size of the recon-
stituted vesicle, which limits the fraction of the bulky carbohydrate-containing
groups that can pack into the interior volume.
   Alternative reconstitution techniques involving protein insertion into preformed
vesicles have achieved some success in obtaining asymmetric incorporation. One of
these asymmetric insertion techniques utilizes the detergent octylglucoside. It is pos-
sible to form vesicles in the presence of relatively high detergent concentrations (ap-
proximately 20 mM) which are sufficient to solubilize the spike protein of Semliki
Forest virus [16]. The spike protein consists of a hydrophilic spike and a smaller
hydrophobic anchor portion of the molecule. The anchor portion is solubilized by a
coat of detergent, and this half of the molecule can insert into the preformed bilayer
on dialysis.
   In summary, a large variety of sophisticated and well-defined model membrane
systems are available. The simplest model systems consist of aqueous dispersions of
lipid which can be converted to LUV forms by a variety of techniques. The sub-
sequent incorporation of protein, with well-defined lipid to protein ratios and asym-
metric transmembrane protein orientations, is becoming more feasible. Problems re-
main, however, both in removing the last traces of detergent in reconstituted systems
and in generating the lipid asymmetry observed in biological membrane system.

4. Physical properties o lipids

4.1. Gel-liquid-crystalline phase behavior

As indicated previously, membrane lipids can exist in a frozen gel state or fluid liq-
uid-crystalline state, depending on the temperature [17], as illustrated in Fig. 7.
Transitions between the gel and liquid-crystalline phases can be monitored by a va-
riety of techniques, including nuclear magnetic resonance (NMR), electron spin reso-
nance (ESR), and fluorescence. Differential scanning calorimetry (DSC), which mea-

            kyr chain          Polar head group                     kylchams
                  I             1          Lioid

                Crystalline slale                                        a
                                                         Lquid~cryslalline l e
                Sltd Lo'                                 lluid L.

                 Main lransilion    -,
                                                                                 Y =
                                                                                 f B


                B /-li
                i p
                    1                         + 5 mol% choleslerol

                                                  + 12.5 ma% c k l e r o l
                                                        + 20 moW chdeslerd
                                                        + 32 mot96 chdestwd
                                                        + 50 mot96 cholesletol
                         290         3M           350
                                     m wr r
                                    T p aue(Q
Fig. 7. The phospholipid gel-liquid-crystalline phase transition and the effect of cholesterol. (A) Phospho-
lipids, when fully hydrated, can exist in the gel, crystalline form (Lp) or in the fluid, liquid-crystalline state
(LJ. bilayers of gel-state phosphatidylcholine, the molecules can be packed such that the acyl chains are
tilted with respect to the bilayer normal (Lp state. Raising the temperature converts the crystalline state
into the liquid-crystalline phase transition as detected by DSC. T represents temperature. For dipalmit-
oylphosphatidylcholine (DPPC) the onset of the main transition occurs at approximately 41"C. The
pretransition represents a small endothermic reorganization in the packing of the gel-state lipid molecules
prior to melting. (C) Influence of cholesterol. The enthalpy of the phase transition (represented by the area
under the endotherm) is dramatically reduced. At greater than 30 mol% cholesterol, the lipid phase transi-
tion is effectively eliminated.

sures the heat absorbed (or released) by a sample as it undergoes an endothermic (or
exothermic) phase transition, is particularly useful. A representative DSC scan of
dipalmitoylphosphatidylcholine,which exhibits a gel to liquid-crystalline transition
temperature (T,) of 41"C, is illustrated in Fig. 7. Three parameters of interest in such
traces are the area under the transition peak, which is proportional to the enthalpy of
the transition; the width of the transition, which gives a measure of the 'cooperativity'
of the transition; as well as the transition temperature T, itself. The enthalpy of the

transition reflects the energy required to melt the acyl chains, whereas cooperativity
reflects the number of molecules that undergo a transition simultaneously.
   Before describing the calorimetric behavior of various phospholipid systems, we
emphasize two general points. First, gel-state lipids always assume an overall bilayer
organization, presumably because the interactions between the crystalline acyl chains
are maximized. Thus, the nonbilayer hexagonal (H,,) or other phases discussed in the
following section are not available to gel-state systems. Second, species of naturally
occurring lipids exhibit broad noncooperative transitions due to the heterogeneity in
the acyl chain composition. Thus, sharp gel-liquid-crystal transitions, indicating
highly cooperative behavior, are observed only for aqueous dispersions of molecu-
larly well-defined species of lipid. These can presently be obtained only by synthetic
   The calorimetric behavior of a variety of synthetic phospholipids is given in Table
IV. There are three points of interest. First, for the representative phospholipid spe-
cies, phosphatidylcholine, there is an increase in Tc by approximately 20°C as each
two-carbon unit is added and a corresponding increase in enthalpy (2-3 kcal/mol).
Second, inclusion of a cis double bond at C-9 results in a remarkable decrease in T,,
which is further lowered as the degree of unsaturation is increased. It is interesting to
note that inclusion of only one cis-unsaturated fatty acid at the C-1 or C-2 position
of the glycerol backbone is sufficient to lower Tc from 41"C for dipalmitoyl phospha-
tidylcholine to -5°C for the palmitoyl-oleoyl species, a major molecular subspecies
of phosphatidylcholine in biological membranes. A final point is that the T, and

Temperature (T,) and enthalpy (AH) of the gel-liquid-crystallinephase transition of
phospholipids (in excess water)

Lipid species                   T, ? 2°C             AH ? 1 kcal/mol

12:0/12:0            PC          -1                    3
14:0/14:0            PC          23                    6
16:0/16:0                        41                    8
16:1cd9/16:1cA9                 -36                    9
18:0/18:0                         54
18:1cd9/l8: 1cA9                -20
16:0/16:0                         63
16:0/16:0                         55
16:0/16:0                         41
16:0/16:O                         67

"The code denotes the number of carbons per acyl chain and the number of double bonds. A gives the
position of the double bond. PC, phosphatidylcholine; PE, phosphatidylethanolamine; PS, phosphatidyl-
serine, PG, phosphatidylglycerol; PA, phosphatidic acid; c, cis.

enthalpy are also sensitive to the head-group constituent. For example, molecular
species of phosphatidylethanolamine commonly exhibit T, values 20°C higher than
corresponding species of phosphatidylcholine. The data of Table IV have some pre-
dictive value in that approximate values of T, can be estimated for other molecular
species of lipids.
   The calorimetric behavior of individual lipid species cannot be directly related to
the behavior of the complex lipid mixtures found in biological systems; therefore,
considerable attention has been devoted to the properties of mixtures of pure lipid
species. Two general features have emerged. First, when all component lipids are
liquid crystalline (that is, T T,-), the lipid systems exhibit characteristics consistent
with complete mixing of the various lipids. Second, at temperatures below the T, of
one of the constituents, separation of the component with the highest melting temper-
ature into crystalline domains (lateral phase separation) can occur under certain con-
ditions. For example, equimolar mixtures of two saturated phosphatidylcholines dif-
fering by four carbon units or more (AT, > 20°C) can exhibit lateral phase separation
(indicated by calorimetric and freeze-fracture studies).
   Further studies of the calorimetric behavior of lipid systems have emphasized the
remarkable physical properties of cholesterol [14]. This lipid has the ability to inhibit
the crystallization of lipids to form gel-state systems, as illustrated for dipalmitoyl-
phosphatidylcholine in Fig. 7C. The enthalpy of the transition is progressively re-
duced as the cholesterol content is increased. For phosphatidylcholine: cholesterol
molar ratios of 2: 1, no transition is observable.
   Gel-liquid crystalline transitions profoundly influence the motional properties of
lipids and therefore are readily detected by NMR techniques. In the liquid crystalline
phase, lipids can rotate rapidly about their long axis and diffuse rapidly in the plane
of the bilayer. In the gel phase, such motions are inhibited. 'H NMR is of particular
utility for characterizing the motion in the hydrocarbon region of liquid crystalline
bilayers. This motion is characterized by an order parameter S, derived from *H
NMR spectra, where S = 1 indicates a fully ordered system and S = 0 indicates iso-
tropic (completely disordered) motion where the 'H nucleus is able to assume all
possible orientations with respect to the magnetic field within         s. Order profiles
can be generated employing phospholipids labelled specifically in the acyl chain re-
gion or, more conveniently, by employing phospholipids containing perdeuterated
fatty acids, which allows the complete order profile to be generated in a single experi-
ment [ 181. Hydrocarbon regions of bilayer systems exhibit a characteristic order pro-
file with a 'plateau' region near the headgroup, after which the order decreases rap-
idly towards the center of the bilayer [19]. Hydrocarbon order can be modulated by
a variety of factors such as cholesterol or increased acyl chain saturation, both of
which lead to larger order parameters.
   The relation between the gel-liquid crystalline properties of lipids and the roles of
lipids in biological membranes remains obscure. The observation that individual lip-
id components can adopt gel or liquid-crystalline arrangements has led to the sug-

gestion that segregation of particular lipids into a local gel-state environment may
occur within a biological membrane. This segregation could affect protein function
by restricting protein mobility within the bilayer matrix or could provide packing
defects resulting in permeability changes. There are two major difficulties encoun-
tered with these concepts, however. First, while certain procaryotic systems can ex-
hibit characteristics consistent with the presence of gel-state lipids at temperatures
which allow growth, such observations are by no means universal. In eucaryotic
membranes, for example, there is no evidence for the presence of gel-state lipid com-
ponents at physiological temperatures. The second difficulty concerns the way in
which lateral segregation of lipid into crystalline domains might be regulated. Clearly,
an organism cannot regulate fluidity by regulating temperature; thus, physiological
factors are required which can induce isothermal modulation of the local lipid com-
position. The presence of factors capable of segregating lipids into local crystalline
domains in a biological membrane has not been unambiguously demonstrated.
   The theme that membranes do not require the presence of gel-state lipids is easily
developed for eucaryotic membrane systems, such as the well-characterized erythro-
cyte membrane. Of the erythrocyte membrane lipids, only sphingomyelin exhibits a
Tc close to physiological temperatures, with the attendant possibility of forming local
crystalline domains. However, this possibility is seriously compromised by the pres-
ence in the membrane of equimolar levels of cholesterol, which would be expected to
inhibit such formation, in agreement with the observation that no reversible phase
transition is observable in the intact erythrocyte (ghost) membrane by calorimetric or
other techniques. In other membranes which contain little or no cholesterol, such as
the membranes of various subcellular organelles, the absence of gel-state domains is
indicated by the absence of relatively saturated lipid species, such as sphingomyelin,
as well as by the increased unsaturation of other lipids present.
   In summary, available evidence indicates that membranes require a fluid bilayer
matrix for function and that modulation of local fluidity and function by formation
of crystalline domains is unlikely to be a general phenomenon. The requirement for
a liquid crystalline lipid matrix is more likely related to the consequent ability of
lipids and proteins to diffuse rapidly in the plane of the membrane. Liquid crystalline
lipids exhibit lateral diffusion rates (D,) low8
                                            of      cm2/s or larger, whereas membrane
proteins have D, values of         cm2/s or smaller. This relates the average distance R
a molecule can diffuse in a time At via the relation 22= 4 D 4 . Thus, a lipid in a cell
of 10 pm diameter would be able to diffuse the length of the cell within 25 s.

4.2. Lipid polymorphism

In addition to an ability to adopt a gel or liquid-crystallinebilayer organization, lip-
ids can also adopt entirely different liquid-crystalline structures on hydration [20].
The major structures assumed are illustrated in Fig. 8. These structures have three
general features. First, the predominant structures assumed by isolated species of

Fig. 8. "P-NMR and freeze-fracture characteristics of phospholipids in various phases. The bilayer spec-
trum was obtained from aqueous dispersions of egg yolk phosphatidylcholine and the hexagonal (H,,)
phase spectrum from phosphatidylcholine). The "P-NMR spectrum representing isotropic motion was
obtained from a mixture of 70 mol% soy phosphatidylethanolamine and 30 mol% egg yolk phosphatidyl-
choline. The spectra were recorded at 30°C in the presence of proton decoupling. The freeze-fracture mi-
crographs represent typical fracture faces obtained. The bilayer configuration (total erythrocyte lipids)
gives rise to a smooth fracture face, whereas the hexagonal (HI,)configuration is characterized by ridges
displaying a periodicity of 6-15 nm. Two common conformations that give rise to isotropic motion are
represented in the bottom micrograph: ( I ) bilayer vesicles (less than 200 nm diameter) ofegg phosphatidyl-
choline prepared by extrusion techniques and (2) large lipid structures containing lipidic particles (egg
phosphatidylethanolamine containing 20 mol% egg phosphatidylserine at pH 4.0).

membrane lipids on hydration in excess aqueous buffer are the familiar bilayer or-
ganization and the hexagonal HI, structure. Lipids which form micellar structures,
such as lysophosphatidylcholine, are minority components of membranes. Second,
the HII phase, which consists of a hydrocarbon matrix penetrated by hexagonally
packed aqueous cylinders with diameters of about 20 A,is not compatible with main-
tenance of a permeability barrier between external and internal compartments. This
immediately raises questions concerning the functional role of lipids in membranes
which preferentially adopt this structure in isolation. Finally, in contrast with the
situation for gel-state (crystalline) lipids, it now appears that all biological mem-
branes contain an appreciable fraction (up to 40 mol%) of lipid species which prefer
the HI, arrangement, as well as lipids which prefer bilayer structure.
   The ability of lipids to adopt different structures on hydration is commonly re-

ferred to as lipid polymorphism. Three techniques which have been extensively em-
ployed to monitor lipid polymorphism are X-ray diffraction, 31P-and *H-NMR, and
freeze-fracture procedures. X-ray diffraction is the classical technique, allowing the
detailed nature of the phase structure to be elucidated. The use of "P-NMR for iden-
tification of polymorphic phase characteristics of phospholipids relies on the differ-
ent motional averaging mechanisms available to phospholipids in different struc-
tures and provides a convenient and reliable diagnostic technique. Finally, freeze-
fracture electron microscopy allows visualization of local structure which need not be
arranged in a regular lattice, yielding information not available from X-ray or NMR
   The "P-NMR and freeze-fracture characteristics of bilayer and H,, phase phos-
pholipid systems are illustrated in Fig. 8. Bilayer systems exhibit broad, asymmetric
"P-NMR spectra with a low-field shoulder and high-field peak separated by about
40 ppm, whereas H,, phase systems exhibit spectra with reversed asymmetry which
are narrower by a factor of two. The difference between bilayer and H,, phase "P-
NMR spectra arises from the ability of HI, phase phospholipids to diffuse laterally
around the aqueous channels. Freeze-fracture techniques show flat, featureless frac-
ture planes for bilayer systems, whereas HI, phase structures give rise to a regular
corrugated pattern as the fracture plane cleaves between the hexagonally packed cyl-
   The polymorphic phase preferences of a large variety of synthetic and naturally
occurring phospholipids have been investigated [20], and the results obtained for eu-
Phase preferences of membrane lipids from eucaryotes

Bilayer                                            Hexagonal HI,

PS                                                 PS (pH < 3)
PA                                                 PA (+Ca2+)
                                                   PA (pH < 3)
CL                                                 CL (+Ca2+

                                                   Fatty acids

PC, phosphatidylcholine; SPM, sphingomyelin; PS, phosphatidylserine; PG, phosphatidylglycerol; PI,
phosphatidylinositol; PA, phosphatidic acid; CL, cardiolipin; PE, phosphatidylethanolamine.
"Cholesterol and long-chain unsaturated fatty acids can induce the hexagonal (H,,)  phase in some lipid

caryotic lipid species are summarized in Table V. It is immediately apparent that a
significant proportion of membrane lipids adopt or promote HI, phase structure un-
der appropriate conditions. Phosphatidylethanolamine,which commonly comprises
up to 35% of membrane phospholipids, is perhaps the most striking example, and
particular effort has been devoted to understanding the factors which result in a pre-
dilection for the H,, arrangement. As illustrated in Fig. 9, phosphatidylethanolamine
isolated from erythrocytes can adopt both the bilayer and HI, arrangements, depend-
ing on the temperature. The HI, structure is formed above a characteristic bilayer to
hexagonal (HJtransition temperature T H of about 10°C. Similar or lower values of
T B H have been observed for phosphatidylethanolamine isolated from endoplasmic
reticulum and the inner mitochondria1membrane. Lower TB, values are observed for
more unsaturated species. This dependence of TBH acyl chain unsaturation has
been characterized more definitively, employing synthetic species of phosphatidyl-
ethanolamine, as summarized in Table VI. This table illustrates that a minimal
degree of unsaturation of the acyl chains is required for HI, structure to be adopted
and that increased unsaturation progressively favors the HI, arrangement. Biological
membranes contain mixtures of lipids which individually prefer bilayer or H,, struc-
tures; therefore, the properties of mixed systems are of considerable interest. Studies

Fig. 9.26.4 MHz "P-NMR spectra of an aqueous dispersion of human erythrocytephosphatidylethanola-
mine dispersed in 25 mM Tris-acetic acid @H 7.0) and 2 mM EDTA. These spectra were obtained employ-
ing proton decoupling. See Fig. 8 for explanation of spectra.

Fig. 10. Phase behavior of phosphatidylcholine (PC) and phosphatidylethanolamine(PE) mixtures and the
effects ofcholesterol. (1) 36.4 MHz "P-NMR spectra of aqueous dispersions of mixtures of soy PE and egg
PC. The amount present is expressed as a percentage of the total phospholipid. (2) 36.4 MHz "P-NMR
spectra obtained at 30°C from equimolar mixtures of dioleoyl-PE with dioleoyl-PC in the presence of (a)
0 mol%; (b) I5 mol%; (c) 30 mol% and (d) 50 mol% cholesterol.

The temperature (TEH) the bilayer-hexagonal HI, phase transition for some phosphatidylethanolamines

Phosphatidylethanolamine                           TEH(OC)

                                                    60 to 63
                                                   -15 to -25
                                                   -15 to -30

on model systems show that mixtures of an HI,phase lipid (for example, phosphati-
dylethanolamine) with a bilayer phospholipid (such as phosphatidylcholine) result in
a progressive stabilization of net bilayer structure for the whole mixture as the per-
centage of bilayer lipid increases, as illustrated in Fig. 10. This is a general feature of
mixtures of bilayer and HI, lipids. Depending on the acyl chain composition, tempe-
rature, and head group size and charge, complete bilayer stabilization can be
achieved by the addition of 10 to 50 mol% of the bilayer species. These systems
appear to retain the ideal mixing behavior characteristic of liquid-crystallinesystems.
For example, in phosphatidylethanolamine-phosphatidylcholinemixtures contain-

ing intermediate amounts of the bilayer-stabilizing species, situations can arise where
HI1 phase and bilayer phase components coexist in the same sample. ’H-NMR
studies of ’H-labeled varieties of these lipids indicate a homogeneous lipid composi-
tion, with no preference of the HI,-preferring phosphatidylethanolamine species for
the HI, component or of phosphatidylcholine for the bilayer component [22].
   There are two other features of these mixed systems which are of particular inter-
est. The first concerns cholesterol, which has the remarkable ability to induce H,,
phase structure for phosphatidylethanolamine-containingsystems where bilayer
structure has been stabilized by phosphatidylcholine (Fig. 10). This effect of choles-
terol is also observed in other mixed-lipid systems. The second point concerns the
narrow 3IP-NMR peak occasionally observed in the mixed-lipid systems of Fig. 10.
Such a spectral feature arises from phospholipids, which experience isotropic mo-
tional averaging over all possible orientations. Such a resonance cannot arise from
phospholipids in HI1 or large (diameter 2200 nm) bilayer structures, where the mo-
tion is restricted. Freeze-fracture studies suggest that this isotropic peak corresponds
to a novel particulate feature observed on the fracture face of these systems, as illus-
trated in Fig. 1 1. These ‘lipidic particles’ are a general feature of mixtures of bilayer-
and HI,-preferring lipids, and there is a growing consensus that they correspond to
interbilayer structures such as inverted micellar or cylindrical interlamellar attach-
ment sites [23]. These structures appear to represent intermediaries between the bi-
layer and HI, phases and may be of particular importance, as such nonbilayer struc-
tures can be localized to a particular region of the membrane. Such structures may
have functional utility, since their formation would not result in a large-scale disrup-
tion of the bilayer permeability barrier that necessarily accompanies generation of
macroscopic HII phase lipid structure.

4.3. Factors which modulate lipid polymorphism

The functional roles of nonbilayer lipid structures in membranes have been inves-
tigated by characterizing the influence of divalent cations, ionic strength, pH, and
membrane protein on lipid polymorphism. These factors can strongly influence the
structural preferences of appropriate lipid systems. In the case of pure lipid systems,
for example, reduction of the pH results in HI, phase structure for (unsaturated)
phosphatidylserine and phosphatidic acid systems, and the addition of Ca” to car-
diolipin triggers bilayer-HI, transitions (Table V). Similar observations extend to
mixed-lipid systems, where the addition of Ca2+to bilayer systems containing phos-
phatidylethanolamine and various acidic phospholipids can also trigger HI1phase
   Phosphatidylserine-phosphatidylethanolaminesystems are perhaps the best char-
acterized in this regard, and certain features of the mechanisms involved deserve
emphasis. First, in some binary phospholipid mixtures containing phosphatidylse-
rine, Ca” can segregate the phosphatidylserine component into a crystalline (gel-

Fig. 11. A schematic representation of the proposed fracture plane around an inverted micelle formed at
the contact point between two bilayers. The corresponding ‘lipidic particles’ observed by freeze-fracture
electron microscopy are shown in the micrograph of a mixture of soy phosphatidylinositol(7.5 rnol%)with
soy phosphatidylethanolamine. Each particle is approximately 10 nm in diameter. The scale represents 100

phase) structure with a characteristic morphology described as ‘cochleate’ (as ob-
served by freeze-fracture). In the case of phosphatidylserine-phosphatidylethanol-
amine systems, the bilayer-stabilizing influence of phosphatidylserine is thus re-
moved, allowing the phosphatidylethanolamine to adopt the H,, organization it
favors in isolation. When 30 mol% or more cholesterol is present, however, Ca2+-
dependent generation of HI, structure proceeds by a different mechanism which does
not involve lateral segregation phenomena-rather, all lipid components, including
phosphatidylserine, adopt the HI, organization. The potential biological relevance of
these observations is illustrated by Fig. 12, where it is shown that Ca2+can trigger HI,
formation in a mixture of lipids isolated from human erythrocytes, with a composi-
tion corresponding to that of the erythrocyte inner monolayer (which contains pre-
dominantly phosphatidylethanolamine and phosphatidylserine).

I     I      I     1      I     I      I

     40            0           -40
Fig. 12. Effect ofcalcium on ‘inner monolayer’ phospholipids: 81.0 MHz ”P-NMR spectra at 37°C arising
from an aqueous dispersion of reconstituted inner monolayer lipid isolated from human erythrocyte mem-
brane. The lipid composition is PE:PS:PC:SPM (in the ratios 0.5:0.25:0.13:0.12) and contains equimolar
cholesterol with respect to total phospholipid. The ratio Ca2’ /PS refers to the molar ratio of Ca” to PS.

4.4. The physical basis of lipid polymorphism

The ability of lipids to adopt different macroscopic structures on hydration has
stimulated studies aimed at understanding the physical properties of lipids which dic-
tate these preferences. These studies have given substantial support to a simplistic
hypothesis that a generalized shape property of lipids determines the phase structure
adopted [30]. This concept is illustrated in Fig. 13, where bilayer phase lipids are
proposed to exhibit cylindrical geometry compatible with that organization, while
H,, phase lipids have a cone shape where the acyl chains subtend a larger cross-sec-
tional area than the polar head group region. Detergent-type lipids which form micel-
lar structures are suggested to have reversed geometry corresponding to an inverted
cone shape. It should be noted that ‘shape’ is an inclusive term reflecting the effects
of the size of polar and apolar regions, head group hydration and charge, hydrogen-
bonding processes, and effects of counterions, among other possibilities. The cone

            LIPID                         PHASE                 MOLECULAR



                                           MICELLAR             INVERTED C O N E

  PnOSPHAlIDYLG1YCEdOL                                             . .. .
                                                                  . . . -.

     CARDIOLIPIN- C o 2 +
 ' H o s P n A r i D I c ACID c a 2
             IpH< 3.01
                                      H E X A G O N A L IHIII       CONE

Fig. 13. Polymorphic phases and corresponding dynamic molecular shapes of lipids.

shape of unsaturated phosphatidylethanolamines, for example, can be ascribed to a
smaller, less-hydrated head group (in comparison with phosphatidylcholine). Alter-
natively, the increased predilection of more unsaturated species of phosphatidyleth-
anolamine for the HI*arrangement (Table VI) may be attributed to the increased
cross-sectional area of the unsaturated (compared with saturated) acyl chains. A
striking observation supporting the shape concept is that lipid mixtures containing
detergents (inverted cone shape) and unsaturated phosphatidylethanolamines (cone
shape) can adopt a bilayer structure, which may be attributed to shape complementa-
rity [21].
   More rigorous and quantititive analyses of the molecular basis of lipid polymor-
phism are available. Chief amongst these is the intrinsic radius of curvature theory
developed by Gruner and colleagues [24]. Within the context of this theory, in the
absence of other constraints a lipid monolayer will curl into a cylinder with an intrin-
sic radius of curvature R,, a parameter which can be measured by X-ray diffraction
studies on HI, phase systems. Lipids such as unsaturated phosphatidylethanola-
mines, which avidly prefer HI, structure, exhibit small values of R,. A difficulty with
measurement of R, values is that they can only be measured in the H,, phase. Lipids
or lipid mixtures which have large R, values are often inhibited from adopting H,,
structures by packing constraints associated with the need to fill the interstitial re-
gions between the HI, phase cylinders. They thus remain in a bilayer organization. In
order to measure R,, such constraints are ‘relaxed’ by addition of hydrocarbons such
as long chain alkanes which are thought to fill these interstitial regions and allow HI,
phase structures with equilibrium curvature to be adopted.
   In summary, studies on model systems show that lipids found in biological mem-
branes can exist in a variety of structures in addition to the bilayer phase. These
structural preferences can be modulated by many biologically relevant variables, sup-
porting the possibility that nonbilayer lipid structures play roles in membrane-
mediated phenomena requiring local departures from bilayer organization. As indi-
cated later in this chapter, membrane fusion is a most important example.

5. Lipids and the permeability properties of membranes

The ability of lipids to provide a bilayer permeability barrier between external and
internal environments constitutes one of their most important functions in a biologi-
cal membrane. The nature and selectivity of this barrier to various molecules and
ions of biological interest (water, uncharged water-soluble non-electrolytes, and ion-
ized solutes of varying hydrophobicity) have been extensively investigated. A succinct
review of these data is difficult to achieve, due in part to the different model systems
employed and the discrepancies among experiments. One major problem concerns
the use of the black lipid membrane or LUV systems where residual solvents such as
decane, detergent, ether, or ethanol may influence the permeability properties of the
ion of interest. Here we present a summary of salient general principles of membrane
permeability in relation to properties of component lipids such as fluidity, polar head
group charge, and phase structure. A synopsis of the permeability coefficients ob-
served for different solutes for a variety of membranes is presented in Table VII.

5.I. Theoretical considerations

In order to appreciate the meaning of the permeability coefficient parameter for a
given lipid system, some understanding of the underlying theory is required. A basic
phenomenological treatment of diffusion begins with Fick’s law, which states that the
diffusion rate of a given substance (number of molecules per unit time, dnldt) through
a membrane is directly proportional to the area (A) of the membrane and the differ-
ence in the concentration AC(t) of the material across the membrane. Thus, dnl
dt-AAC(t), which may be rewritten as dnldt = -PAAC(t), where P, which has the
units of length over time (for example, c d s ) , is the permeability coefficient and t is
time. If we consider the special case of a LUV of radius R containing an initial con-
centration of solute C,(O),where the initial external concentration of this solute is

Permeability coefficients (cmls) for some common polar solutes across model and biological membranes
(at 20°C unless otherwise indicated)

Membrane                                  Na'                     K'             c1-

Egg phosphatidylcholine (4°C)             4 . 2 x 10-l4           -
Soy phosphatidylcholine"                   4.0 x 10-13            -
Equimolar soy phosphatidyl-
  choline and cholesterol
  (40°C)                                        0
                                          ~ 2 . x 10-l3       -
Dilinoleoylphosphatidylserine"                  o
                                          ~ 7 . x 10-l3       -
Human red blood cell
  phosphatidylserine"                     <7.0 x 10-13        ~

Human red blood cell                        1.0 x             2.4 x              1.4x 10-4
Squid axon (resting)                        1.5 x lo-'        5.6 x lo-'         1.0 x lo-*
Squid axon (excited)                        5.0 x             1.7 x              1.0 x

'The data were obtained from large unilamellar vesicles prepared by extrusion through filters according to
Hope et al. (1 986).

zero, it is straightforward to show that d C ( t ) = C1(O)exp(-3Pt/ R ) . Under conditions
where the external volume is much greater than the internal trapped volume,
AC(t) = Cl(t) (where Cl(t) is the internal concentration at time t ) ; thus,
C , ( t ) = CI(O)exp(- 3PtIR). For a 100 nm diameter LUV it may therefore be calcu-
lated that the time required for release of one-half of the entrapped material (t%) is
0.1 s for P = lo-' c d s , whereas for P = lo-''' cm/s, tY2 = 2.3 h.
   It should be emphasized that the preceding example, while illustrative, neglects
several important factors which can strongly influence the net flux of molecules
through membranes. These include the effects associated with the 'unstirred' aqueous
layer (more than 20 nm thick) that extends from the lipid-water interface, in which
solute molecules are not mixed to the same extent as in the bulk solution. Such un-
stirred layers can effectively reduce the solute concentration difference AC across the
membrane itself, giving rise to a smaller measured value of P . For charged molecules,
the efflux can be strongly limited by generation of a membrane potential, as will be
discussed later. Finally, the permeability of various solutes through membranes is
strongly temperature dependent, with activation energies E, in the range of 8-20
kcal/mol. A measure of the influence of temperature is given by the observation that
an activation energy of 12 kcal/mol will increase the permeability coefficient by a
factor of two for every 10°C increase in temperature.

5.2. Permeability of water and non-electrolytes

Liquid-crystalline lipid bilayers are remarkably permeable to water, which exhibits a
permeability coefficient in the range of       to     cmis [25]. Membrane systems

enclosing high concentrations of a relatively impermeable solute will swell when
placed in an aqueous medium containing little or no solute, due to a net influx of
water to achieve osmotic balance. Conversely, the reverse condition will lead to
shrinkage. As a result, the relative permeability of different membrane systems to
water can be monitored by measuring swelling rates (employing light scattering tech-
niques, for example) when osmotic gradients are applied. Results obtained from such
studies indicate that increased unsaturation of the fatty acids of the membrane causes
increases in water permeability. Similarly, the inclusion of cholesterol reduces water
permeability, leading to the general conclusion that factors contributing to increased
order in the hydrocarbon region decrease water permeability.
   The diffusion properties of non-electrolytes (uncharged polar solutes) appear to
depend on the properties of the lipid matrix in much the same manner as does the
diffusion of water. i n general, the permeability coefficients observed are at least two
orders of magnitude smaller. For example, the permeability coefficient of urea across
egg phosphatidylcholinebilayers is approximately 4x           c d s at 25°C [26]. Further-
more, for a given homologous series of compounds, the permeability increases as the
solubility in a hydrocarbon environment increases, indicating that the rate-limiting
step in diffusion is the initial partitioning of the molecule into the lipid bilayer. With
regard to the influence of lipid composition on the permeability of non-electrolytes,
the order in the acyl chain region has the same qualitative effects as in the case of
water. Thus, decreased unsaturation of lipids or increased cholesterol content results
in lower permeability coefficients. Gel-phase systems are particularly impermeable.
However, in systems exhibiting lateral phase separation of gel and liquid-crystalline
domains, the permeability can be higher than for liquid-crystalline systems. This in-
creased permeability can be attributed to packing defects at the crystalline-liquid-
crystalline hydrophobic interface.

5.3.Permeability of ions

Lipid bilayers are remarkably impermeable to most small ions (Table VIi). Permea-
bility coefficients of less than lo-'* c d s are commonly observed, and they can be as
small as                           .
                c d s for Na' and K For the example of a 100 nm diameter LUV, this
would correspond to a half-life for release of entrapped Na' of approximately 3.6
years. in contrast, lipid bilayers appear to be much more permeable to H' or OH-
ions, which have been reported to have permeability coefficients in the range of
c d s [25]. The C1- anion also exhibits anomalous permeability behavior, with per-
meability coefficients up to 300 times greater than those observed for Na' in similar
   Measures of the permeability of membranes to small ions are complicated, since
for free permeation to proceed, a counterflow of other ions of equivalent charge is
required; otherwise, a membrane potential is established which is equal and opposite
to the chemical potential of the diffusing species. As an example, for the 100 nm
diameter LUV which has a well-buffered interior pH of 4.0 and an exterior pH of 7.0
in a Na' buffer, the relatively permeable H' ions can diffuse out, but Na' ions cannot
move in. Thus, a membrane potential (Ay)is established (interior negative), where
    Ayl  = -59 log -= -177 mV
and the subsequent efAux of protons is coupled to the much slower influx of Na'
ions. Assuming a membrane thickness of 4 nm and interior dielectric constant of 2,
the capacitance of the vesicle membrane can be calculated as C = 0.5 pF/cm2; thus,
from the capacitance relation Q = CV(where Q is the charge and Vis the transmem-
brane voltage), the number of protons that diffuse out to set up Ayl can be calculated
to be about 150. Subsequent H' emux will occur only as Na' ions permeate in.
   The relation between the physical properties of lipids and the permeability proper-
ties of membranes to small ions is not understood in detail. Difficulties in under-
standing this relationship arise from the different model systems employed, the vari-
ous impurities present, and complexities due to ion counterflow and related mem-
brane potential effects. Vesicles prepared by techniques involving detergents or
organic solvents contain residual detergent or solvent which can strongly influence
the permeabilities observed, and the presence of n-decane or other long chain alkanes
in black lipid membrane systems may also influence permeability. In general, how-
ever, the permeability of a given ion appears to be related to the order in the hydro-
carbon region, where increased order leads to a decrease in permeability.
   The charge on the phospholipid polar head group can also strongly influence per-
meability by virtue of the resulting surface potential @. For example, approximately
30%of the lipid of the inner monolayer of the erythrocyte membrane is the negatively
charged lipid phosphatidylserine. If we assume an area per lipid molecule of 0.6 nm2,
the resultant surface charge density CJ is 8 pC/cm2 (where C is coulombs). The result-
ing surface potential @ can be calculated from the Gouy-Chapman theory [26] for a
150 mM monovalent salt buffer according to the relation @ = 0.052xsinh-'(a/4.5).
This gives a negative surface potential of @ = -69 mV. This potential will repel
anions from and attract cations to the lipid-water interface. For example, the H'
concentration at the inner monolayer interface will be increased by the Boltzmann
factor exp (e@/kT)= 14.5in comparison with the bulk solution, resulting in a signifi-
cantly lower pH at the membrane interface and correspondingly higher H' emux

6. Lipid-protein interactions

Any complete understanding of biological membrane systems necessitates a detailed
understanding of the nature and influence of lipid-protein interactions. Such interac-
tions can be divided into two classes. The first concerns proteins with hydrophobic

segments which penetrate into or through the lipid bilayer (intrinsic, or integral, pro-
teins), whereas the second concerns water soluble proteins which interact electrostati-
cally with negatively charged groups at the lipid-water interface (extrinsic, or periph-
eral, proteins). The effects of intrinsic and extrinsic proteins on membrane lipid fluid-
ity (for example, the gel or liquid-crystalline nature of associated lipids) or lipid
polymorphism will provide the primary focus of this section. However, it should be
noted that studies of lipid-protein interactions [27] have generated a large and often
confusing literature which has not yet led to a generally accepted understanding. We
emphasize here only those points which we believe provide the most important in-

6.1. Extrinsic proteins

The interaction of extrinsic proteins with lipids has been studied using a variety of
proteins, including polylysine, cytochrome c, the A , basic protein from myelin, and
spectrin from the red blood cell. In order for these basic (positively charged) mole-
cules to interact extensively with lipid systems, the presence of acidic (negatively
charged) lipids is required, consistent with an electrostatic protein-membrane asso-
ciation. Two general points can be made. First, while it is possible that such surface
interactions may induce a time-averaged enrichment of the negatively charged lipid
in the region of the protein, there is presently no unambiguous evidence to suggest
that such clustering can induce a local fluidity decrease via formation of crystalline
domains. Indeed, in model membrane systems containing acidic phospholipids, such
extrinsic proteins as cytochrome c, the A , basic protein, and spectrin induce a de-
creased T, and enthalpy of the lipid gel-liquid-crystalline transition, indicating an
increased disorder in the acyl chain region. This effect has been related to an ability
of such proteins to partially penetrate the hydrophobic region, as indicated by in-
creases in permeability and monolayer surface pressure on binding. The second point
is that there is evidence of competition between divalent cations and extrinsic
proteins for binding to membranes. Thus, spectrin can shield the effects of Ca2' on
the gel-liquid phase transition properties of systems containing negatively charged
   Studies on the influence of extrinsic proteins on the polymorphic properties of
lipids [28] also yielded results consistent with a competition between the protein and
divalent cations. For example, polylysine, which is highly positively charged, can to
some extent destabilize the bilayer structure of cardiolipin-phosphatidylethanola-
mine systems and strongly protects against the ability of Ca2' to induce complete HI,
organization in the pure lipid system. A particularly interesting observation is that
cytochrome c can induce nonbilayer structures in cardiolipin-containing systems.
This observation may be related to an apparent ability of cytochrome c to translocate
rapidly across bilayers that contain cardiolipin, possibly including the inner mito-
chondrial membrane.

6.2. Intrinsic proteins

Intrinsic or integral membrane proteins cannot be solubilized without detergent and
contain one or more hydrophobic sequences which span the lipid bilayer one or more
times in a-helical structures. Studies on the interactions of lipids with such proteins
have resulted in a particularly large literature. This work has mainly focused on the
specificity of such lipid-protein interactions and on the physical state of the lipid. In
particular, it has been shown that lipids residing at the lipid-protein interface of in-
trinsic proteins experience a different environment than do bulk bilayer lipids. It has
been speculated that such boundary lipids may be specific to a given protein and
provide environments that are appropriate to, and possibly regulate, function. These
theories were supported by early ESR studies of spin-labeled lipids in reconstituted
systems which demonstrated that such lipids, when in the vicinity of integral proteins,
exhibited increased order parameters (that is, restricted motion of the lipid) in the
acyl chain region. Other studies indicating the importance of the physical state of
boundary lipids demonstrated that gel-state boundary lipids inhibited the function of
the sarcoplasmic reticulum Ca2' ATPase and other membrane-bound enzymes in re-
constituted systems.
   More recent work is pointing to a rather different picture, however. First, with the
exception of a possible requirement for one or two molecules of a particular lipid,
lipid-protein interactions appear relatively nonspecific, in that a large variety of dif-
ferent (liquid-crystalline)lipids can usually support protein activity. The sarcoplas-
mic reticulum ATPase, for example, has excellent activity when reconstituted with a
variety of phospholipids as well as detergents. Similar observations have been made
for many other integral proteins, including cytochrome oxidase. A second point is
that, in general, a long-lived boundary layer of lipid does not appear to exist at the
lipid-protein interface. For example, whereas ESR spin-label studies indicate long-
lived boundary components, 'H-NMR studies on analogous systems containing 'H-
labeled lipids do not reveal such components. This apparent discrepancy has been
reconciled, since ESR and NMR report on phenomena occurring during different
time scales. Boundary-bulk lipid exchange rates in the region 10-6-10-8 s would ap-
pear slow on the ESR time scale but fast on the NMR time scale. These observations,
together with NMR and calorimetric results indicating that integral proteins can
have disordering effects on adjacent lipids, suggest that lipids in the region of intrinsic
protein exchange rapidly (exchange time         s) and do not have gel-statecharacteris-
tics. This does not mean that the lipid composition in contact with the protein is
necessarily the same as the bulk composition, as effects such as electrostatic lipid-
protein interactions may enhance the local concentration of a particular lipid species
on a time-averaged basis. Furthermore, such generalizationsmay not hold for partic-
ular situations. The purple membrane fragments of Halobacterium halobium, which
contain bacteriorhodopsin, for example, exhibit a unique lipid composition distinct
from the rest of the membrane.

   The influence of intrinsic proteins on lipid polymorphism has been investigated by
De Kruijff and coworkers [28]. Interesting features concern, first, the hydrophobic
peptide antibiotic gramicidin, which spans the membrane as a dimer and which has
a very strong bilayer destabilizing capacity and even induces HI, phase structure in
phosphatidylcholine systems. On the other hand, glycophorin, the major asialogly-
coprotein from the erythrocyte, stabilizes the bilayer structure for unsaturated phos-
phatidylethanolamines.These studies have been extended to signal peptides (Chapter
18), which show an ability to induce H,, phase structure [29] leading to the intriguing
possiblity that nonbilayer structures may play a role in protein insertion into, and
translocation across, membranes.
   In summary, our understanding of lipid-protein interactions in biological mem-
branes remains relatively unsophisticated. It may be that some fraction of lipid diver-
sity satisfies relatively nonspecific requirements and provides an appropriate solvent
for the optimal function of integral proteins. Alternatively, specific functions of lipids
may be more related to other membrane properties, such as permeability, than to
protein function per se. In addition, many fundamental questions have not yet been
adequately addressed, including the role of various lipids in sealing proteins within
the bilayer matrix and in providing an interface appropriate for membrane protein-
substrate interactions.

7. Lipids and membrane fusion

Membrane fusion is one of the most ubiquitous membrane-mediated events, occur-
ring in processes of fertilization, cell division, exo- and endocytosis, infection by
membrane-bound viruses, and intracellular membrane transport, to name but a few.
There are strong experimental and theoretical indications that the lipid components
of membranes are directly involved in such fusion processes. For example, model
membrane systems such as LUVs can be induced to fuse in the absence of any protein
factors. In addition, it is topologically impossible for two membrane-bound systems
to fuse together to achieve mixing of internal compartments without a local tran-
sitory departure from the normal lipid bilayer structure at the fusion interface. We
shall discuss the possible nature of the fusion intermediates, as indicated by studies
on model membrane systems. The fusion intermediates are subsequently related to
fusion behavior observed in biological membrane systems.

7.1. Fusion of model systems

For fusion events to proceed in vivo the presence of Ca” is often required. As a
result, numerous studies have been concerned with the induction of Ca2+-stimulated
fusion between vesicle systems and analysis of the lipid factors involved. We shall

discuss in turn the modulation of gel-liquid-crystallineproperties of lipids and the
modulation of the polymorphic properties of lipids in relation to membrane fusion.
   It has been recognized for some time that model membrane SUV systems will un-
dergo fusion when incubated at temperatures in the region of their gel-liquid-crystal-
line transition temperature T,. Continued recycling of sonicated dipalmitoylphos-
phatidylcholine vesicles through T, = 41”C, for example, results in fusion and
formation of larger systems. Isothermal induction of crystalline structure by the
addition of Ca2’ to phosphatidylserine systems results in fusion to form the large
crystalline cochleate structures noted previously. Given the involvement of Ca2+in
biological fusion events, the latter observation suggests that Ca2+may induce lateral
segregation of negatively charged phospholipids, such as phosphatidylserine, in vivo,
which may act as local crystalline nucleation points for fusion. However, phosphati-
dylserine is not always present in membranes which undergo fusion, nor is Ca2+able
to induce crystalline cochleate-type structures for other species of (unsaturated)
negatively charged phospholipids. Furthermore, in more complex lipid mixtures con-
taining phosphatidylethanolamine and cholesterol, for example, there are strong in-
dications that Ca” is not able to induce segregation of unsaturated phosphatidyl-
serines [22). Finally, the concentration of Ca” required to induce crystalline phos-
phatidylserine-Ca2+    complexes is 2 mM or larger, a concentration much higher than
could occur in the cell cytoplasm, for example.
   The hypothesis that membrane fusion proceeds by taking advantage of the poly-
morphic capabilities of component lipids is more viable, but not proven. Three im-
portant observations have been made which support this hypothesis [21]. First, it has
been shown that lipid-soluble fusogens (such as glycerolmonooleate, which induces
cell fusion in vitro), induce HI, phase structures in model and biological membranes,
which is consistent with a role of nonbilayer structure during fusion. Second, MLV
systems composed of lipid mixtures such as phosphatidylethanolamines and phos-
phatidylserines form HI, structures on the addition of Ca” . SUV or LUV systems
with this lipid composition first fuse to form larger lamellar systems exhibiting lipidic
particle structures (as shown in Figs. 1 I and 14), before assuming the HI, arrange-
ment. Finally, a variety of factors which engender HI, organization, such as pH varia-
tion or increased temperatures, can induce fusion of vesicle systems with appropriate
lipid compositions.
   These observations have led to a general hypothesis that factors which tend to
induce nonbilayer (HI, phase) structure will also induce fusion between membrane-
bound systems. There are many attractive features to this hypothesis. In particular,
lipids which adopt HI,organization hydrate poorly in comparison with bilayer lipids
and thus allow the close apposition of membranes required for fusion. In addition,
the ability of such lipids to adopt inverted structures, such as inverted micelles or
inverted cylinders, clearly provides an attractive intermediate structure for fusion.
Furthermore, all membranes appear to contain lipids that can adopt nonbilayer
structures, and a large number of biologically relevant variables can modulate the

Fig. 14. (A) Freeze-fracture micrographs of vesicles composed of soy phosphatidylethanolaminescontain-
ing 20 mol% soy phosphatidylserine. (B) The vesiscle undergo fusion to larger bilayer structures contain-
ing lipidic particles following the addition of 2 mM Ca” at room temperature. Each bar represents 100

structural preferences of these lipids. These facts support the proposition that fusion
proceeds via a nonbilayer intermediate, as shown in Fig. 15. More quantitative sup-
port for the proposal that intermediates in bilayer-nonbilayer phase transitions also
provide intermediate structures for membrane fusion comes from the elegant thermo-
dynamic analyses and cryo-electron microscopy studies performed by Siege1 and co-
workers [23]. Specifically, these authors provide convincing evidence that bilayer-
nonbilayer transitions proceed through transient ‘interlamellar attachment sites’,
which are lipid cylinders formed between adjacent bilayers. Such structures, which
correspond to the lipidic particle structure observed by freeze-fracture electron mi-
croscopy (Fig. 14), are also observed during membrane fusion processes.

7.2. Fusion of biological membranes

Extension of the preceding observations on fusion of model systems to fusion proces-
ses in vivo is difficult to show directly. However, work on several experimental sys-
tems has provided circumstantial evidence in support of the hypothesis that fusion

        (a)                   (b)                  (d                    (4
Fig. 15. Proposed mechanism of membrane fusion proceeding via an inverted cylinder or inverted micellar
intermediate. The process whereby the membranes come into close apposition (a)-(b) is possibly protein
mediated, whereas the fusion event itself (b)-(c) is proposed to involve formation of an inverted lipid inter-

processes rely on the polymorphic capabilities of lipids. One system studied was the
fusion process involved in the exocytotic events occurring during release of the con-
tents of secretory vesicles such as the chromaffin granules of the adrenal medulla.
Such exocytosis is dependent on the influx of Ca”, which stimulates fusion between
the granule and the cytosolic side of the plasma membrane. By analogy to the eryth-
rocyte membrane, the inner (cytosolic) monolayer of the chromaffin cell is likely
composed primarily of phosphatidylserine and phosphatidylethanolamine, whereas
the outer (also cytosolic) monolayer of the secretory granule membrane is enriched in
aminophospholipids. Studies have shown that chromaffin granules will undergo
Ca2’-stimulated fusion with SUVs of inner monolayer lipid composition. Such fusion
appears to depend on the ability of Ca” to promote nonbilayer structures. In another
system, myoblast cells (which fuse to form the multinucleated muscle fibers) have
been studied [30]. Such fusion, which is also Ca2+dependent, may rely on the trans-
membrane distribution of phosphatidylethanolamine and phosphatidylserine, which
appear to reside mainly in the outer monolayer of the myoblast plasma membrane.
  Yet another system concerns the tight junction network formed by epithelial and
endothelial cells to separate apical (membrane facing the lumen) and basolateral (sur-
face opposite the lumen) domains. Such networks may correspond to a situation of

arrested fusion. Freeze-fracture work suggests that the striated patterns characteris-
tic of tight junction assemblies may correspond to long, inverted lipid cylinders simi-
lar to those comprising the HI, phase structure [31]. Similar states of arrested fusion
may correspond to the contact sites between the inner and outer membranes of mito-
chondria and E. coli.

8. Model membranes and drug delivery

The preceding sections have dealt primarily with the use of lipids in various model
membrane systems to gain insight into the physical properties and relative functional
roles of individual lipid components in biological membranes. However, these model
membrane systems have important potential uses in their own right, as carriers of
biologically active agents such as drugs, enzymes, and DNA vectors for clinical appli-
cation [32]. Natural membrane lipid components such as phosphatidylcholine are
remarkably nontoxic and nonimmunogenic and can therefore provide benign car-
riers for more toxic or labile agents encapsulated within lipid vesicles. An important
aim, which has not yet been realized, is to target liposomal systems containing drugs
such as anticancer agents to specific tissues via antibodies attached to the vesicle sur-
face, as indicated in Fig. 16.
   The many difficulties involved in drug delivery via liposomal systems may be sum-
marized as follows: First, vesicle systems must be employed which exhibit an ade-
quate trapped volume to entrap sufficient drug, and a mode of preparation must be
used which allows a high trapping efficiency. Several such procedures exist, including
the reversed-phase evaporation protocol and the extrusion protocol outlined previ-
ously (Table IV), which allow maximum trapping efficiencies in the range of 30-50%
of available drug. In addition, new procedures have become available that rely on
transmembrane pH gradients across LUV membranes and that allow the rapid, effi-
cient encapsulation of lipophilic, cationic drugs such as the anticancer drug doxoru-
bicin. These procedures allow encapsulation efficiencies approaching 100% and ex-
tremely high interior drug concentrations of 300 mM or higher. The second difficulty
concerns the phenomenon of serum-induced leakage of the liposomes due to interac-
tion with serum components such as lipoproteins. This problem can be significantly
alleviated by inclusion of lipids that are more saturated and/or cholesterol in the car-
rier vesicle. A third difficulty for liposomal delivery systems involves uptake of the
liposomes by the fixed and free macrophages of the reticuloendothelial system, which
are primarily localized to the liver (Kupffer cells) and spleen. This problem can be
circumvented by inclusion of certain gangliosides in the liposome. However, other
significant problems remain. For instance, although several procedures exist for cou-
pling antibodies to vesicles, it is unlikely that such targeted systems will be able to
cross the endothelial barrier to gain access to extravascular tissue.
   Despite these problems, the attractive nature of vesicle-mediated drug delivery has

Fig. 16. The delivery of biologically active materials encapsulated in membrane vesicles. Tissue-specific
antibodies, indicated by the Y-shaped symbol, are covalently attached to the surface of the vesicle and
enable the targeting of entrapped material.

engendered increasing interest and effort which have already resulted in protocols of
potential clinical importance. These advances are largely based on the finding that
liposomal encapsulation of anticancer agents (such as doxorubicin) or antifungal
agents (such as amphotericin B) can reduce the toxicity associated with the drug
while maintaining or even increasing efficacy. Other applications take advantage of
the natural targeting of liposomes to fixed and free macrophages of the reticuloen-
dothelial system with two different aims. The first involves parasites which reside in
the macrophages and which are difficult to eliminate by conventional means. How-
ever, encapsulation of an appropriate drug into a vesicle carrier, which is subse-
quently taken up by the macrophages, can result in elimination of parasites such as
Leishmania. An advantage of this method of treatment is that the dose levels needed
are much lower than otherwise required. The second involves incorporation of mac-
rophage- activating factors into the vesicles which, when endocytosed, result in mac-
rophage activation. Such activated macrophages appear to be remarkably effective
for recognizing and destroying diseased tissue, including transformed cells. As sum-
marized elsewhere [32], extensive clinical trials are being conducted on these and
other liposomal formulations.
9. Future directions

The physical properties of membrane lipids are increasingly well-understood. The
relation between these physical properties and the functional roles of lipids remain
relatively obscure, however. General roles of lipids in maintaining the membrane bi-
layer permeability barrier are clear. Similarly, the ability of certain classes of lipids to
adopt nonbilayer structures are very probably of utility in inter-membrane interac-
tions such as fusion, which require local departures from the bilayer organization.
These abilities could be satisfied by a relatively limited subset of the lipids actually
present in membranes. Important additional functions which require clarification in-
clude the detailed roles of lipids in establishing and modulating membrane permea-
bility and membrane protein function as well as their roles in transbilayer signalling

  1. Gorter, E. and Grendel, F. ( I 925) On bimolecular layers of lipids on the chromocytes of the blood. J.
      Exp. Med. 41,43943.
 2. Singer, S.J. and Nicholson, G.L. (1972) The fluid mosaic model of the structure of cell membranes.
      Science 175, 720-3 1.
 3. Gennis, R.B. (1989) Biomembranes: Molecular Structure and Function. New York: Springer.
 4. Van Deenen, L.L.M. and De Gier, J. (1974) Lipids of the red cell membrane. In: The Red Blood Cell
      (D. Surgenor, Ed.) pp. 147-213. New York: Academic Press.
 5. Op den Kamp, J.A.F. (1979) Lipid asymmetry in membranes. Annu. Rev. Biochem. 48,47-71.
 6. Hope, M.J., Redelmeier, T.E., Wong, K.F., Rodrigueza, W. and Cullis, P.R. (1989) Phospholipid
      asymmetry in large unilamellar vesicles induced by transmembrane pH gradients. Biochemistry 38,
 7. Devaux, P.F. (1991) Static and dynamic lipid asymmetry in cell membranes. Biochemistry 30, 1163-
      1 173.
 8. Tanaka, Y.and Schroit, A.J. (1983) Insertion of fluorescent phosphatidylserine into the plasma mem-
      brane of red blood cells: recognition by autologous macrophages. J. Biol. Chem. 258, 1133543.
 9. Kates, M. (1986) Techniques in Lipidology: Isolation, Analysis, and Identification of Lipids. Amster-
      dam: North Holland.
10. Hope, M.J., Bally, M.B., Mayer, L.D., Janoff, AS. and Cullis, P.R. (1986) Generation of multila-
      mellar and unilamellar phospholipid vesicles. Chem. Phys. Lipids 40, 89--108.
1 I . Bangham, A.D., Standish, M.M. and Watkins, J.D. (1965) Diffusion of univalent ions across the la-
      mellae of swollen phospholipids. J. Mol. Biol. 13, 238-51.
12. Szoka, F. and Papahadjopoulos, D. (1980) Comparative properties and methods of preparation of
      lipid vesicles (liposomes). Annu. Rev. Bioeng. 9,467-508.
13. Coronado, R. (1986) Recent advances in planar phospholipid bilayer techniques for monitoring ion
      channels. Annu. Rev. Biophys. Biophys. Chem. 15,259-277.
14. Demel, R.A. and De Kruijff, B. (1976) The function of sterols in membranes. Biochim. Biophys. Acta
      457, 109-32.
15. Madden, T.D. (1986) Current conczpts in membrane protein reconstitution. Chem. Phys. Lipids 40,

16. Helenius, A., Sawas, M. and Simons, K. (1981) Asymmetric and symmetric membrane reconstitution
    by detergent elimination. Eur. J. Biochem. 116, 27-31.
17. Silvius, J.R. (1982)Thermotropic phase transitions of pure lipids in model membranes and their modi-
    fication by membrane proteins. In: Lipid-Protein Interactions (P.C. Jost and O.H. Griffith, Eds.) Vol.
    2. New York: Wiley.
18. Lafleur, M., Fine, B., Sternin, E., Cullis, P.R. and Bloom, M. (1989) Smoothed orientational order
    profile of lipid bilayers by 2H NMR. Biophys. J. 56, 1037-1041.
19. Seelig, J. and Seelig, A. (1980) Lipid conformation in model membranes and biological membranes. Q.
    Rev. Biophys. 13, 19-61.
20. Cullis, P.R. De Kruijff, B. (1979) Lipid polymorphism and the functional roles of lipids in biologi-
    cal membranes. Biochim. Biophys. Acta 559, 399420.
21. Seddon, J.M. (1990) Structure of the inverted hexagonal (H,,) phase and non-lamellar phase transi-
    tions of lipids. Biochim. Biophys. Acta 1031, 1-69.
22. Tilcock, C.P.S. (1986) Lipid polymorphism. Chem. Phys. Lipids 40, 109-126.
23. Siegel, D.P., Bums, J.L., Chestnut, M.H. and Talmon, Y. (1989) Intermediates in membrane fusion
    and bilayer-nonbilayer phase transitions imaged by time resolved cryo-transmission electron micro-
    scopy. Biophys. J. 56, 161-169.
24. Gruner, S.M. (1985) Intrinsic curvature hypothesis for biomembrane lipid composition: a role for
    non-bilayer lipids. Proc. Natl. Acad. Sci. U.S.A. 82, 3665-3669.
25. Deamer, D.W. and Bramhall, J. (1986) Permeability of lipid bilayers to water and ionic solutes. Chem.
    Phys. Lipids 40, 167-1 88.
26. Killian, J.A. and De Kruijff, B. (1986) The influence of proteins and peptides on the phase properties
    of lipids. Chem. Phys. Lipids 40, 259-284.
27. McLaughlin, S. (1977) Electrostatic potentials at membrane-solution interfaces. In: Current Topics in
    Membranes and Transport, Vol. 9, pp. 71-144. New York: Academic Press.
28. Jost, P.C. and Griffith, O.H. (Eds.) (1982) Lipid-Protein Interactions, Vols. 1 and 2. New York: Wi-
29. Killian, J.A., De Jong, A.M.Ph., Bijvelt, J., Verkleij, A.J. and De Kruijff, B. (1990) Induction of non-
    bilayer lipid structures by functional signal peptides. EMBO J. 9, 815-819.
30. Sessions, A. and Horwitz, A.F. (1981) Myoblast aminophospholipid asymmetry differs from that of
    fibroblasts. FEBS Lett. 134, 75-79.
31. Kachar, B. and Reese, T.S. (1982) Evidence for the lipidic nature of tight junction strands. Nature 296,
32. Ostro, M.J. and Cullis, P.R. (1989) Use of liposomes as injectable drug delivery systems. Am. J. Hosp.
    Pharm. 46, 15761587.
This Page Intentionally Left Blank
D.E. Vane and J. Vance (Eds.) Biochemisfry of Lipids, Lipoproteins and Membranes
8 1991 Elsevier Science Publishers B.V. All rights reserved.                              43

                                                                                   CHAPTER 2

                                    Lipid metabolism in procaryotes
                       SUZANNE JACKOWSKI'-2,JOHN E. CRONAN, JR.3 and
                                                CHARLES 0. ROCK'-2

   'Deparfmento Biochemistry, St. Jude Children's Research Hospital, Memphis, T N 38105,
 U.S.A.,'Department of Biochemistry, University qf Tennessee College of Medicine, Memphis,
 TN 38163, U.S.A . and 'Departments of Microbiology and Biochemistry, University of Illinois,
                                                                  Urbana, IL 61801, U.S.A.

1. The study o bacterial lipid metabolism

There are several major advantages of using bacteria to study the regulation of phos-
pholipid biosynthesis and the role of phospholipids in membrane function. The main
advantage bacteria offer over other systems is the use of the genetic approach to
manipulate the experimental system. This approach has allowed bacterial physiolo-
gists to vary the components of a metabolic network in the cell in order to test criti-
cally hypotheses about the role of these components in vivo. Other distinct advan-
tages are that (a) the experimenter has complete control over the conditions of cell
growth, (b) large quantities of precisely grown cells for biochemical analysis are rea-
dily obtained, (c) most bacteria have a simple lipid composition and lack intracellular
membrane systems, and (d) there is a wealth of general information on bacterial me-
tabolism and biochemistry. The biological organism most thoroughly understood at
the molecular level is the gram-negative bacterium Escherichia coli, and this genera-
lity carries over to lipid metabolism. Most of our knowledge of the molecular aspects
of procaryotic lipid metabolism is based on studies of E, coli; hence, research on this
bacterium will dominate the discussions in this chapter. However, the diversity of life
styles, lipid structures, and metabolic pathways represented by bacteria is so large
that E. coli physiology should not be considered typical of the procaryotic kingdom.

2. Historical introduction

The modern era of E. coli phospholipid enzymology began in the early 1960s when
Vagelos and his colleagues discovered that the intermediates of fatty acid biosynthe-
sis were bound to a heat-stable cofactor termed acyl carrier protein (ACP). The reali-
zation that the reactions in fatty acid biosynthesis could be separated and studied

individually precipitated a flurry of activity, primarily by the laboratories of Bloch,
Vagelos, and Wakil, and within a few years the structures of all the intermediates in
fatty acid biosynthesis had been elucidated [ 11. These experiments had a great deal of
impact on the lipid metabolism field as a whole because the fatty acid synthases of
higher organisms are multifunctional protein complexes, and the individual reactions
could not be isolated. In the late 1960s, the enzymatic steps leading to the major
phospholipid classes were established by the classical demonstration, primarily by
Kennedy and co-workers, of the enzymes in crude cell extracts that catalyze these
reactions [2]. Later radiolabeling studies established the rapid turnover of the antici-
pated intermediates in the pathway such as phosphatidic acid, CDP-diacyglycerol,
and phosphatidylserine. Armed with the knowledge of the biochemical pathways and
the intermediates, selection schemes were devised to obtain mutants in specific path-
way enzymes, thus ushering in the most recent work on the regulation of the phos-
pholipid metabolic network that forms the focus of the present chapter.

3. An overview of phospholipid metabolism in E. coli

In E. coli, phospholipids are synthesized exclusively for use in the biogenesis of mem-
branes, and there does not appear to be any significant alternative fate for these lip-
ids. The steps in the biosynthetic pathway leading to the major phospholipid classes
of E. coli have been established, and the biochemistry of these enzymes is covered in
most basic biochemistry texts. The enzymes are isolated as individual protein species,
and to date there is no convincing evidence for the existence of multienzyme complex-
es in the lipid biosynthetic pathway. The exception of this rule is the high molecular
weight fatty acid synthases of phylogenetically advanced bacteria (see the following
discussion). The enzymes of fatty acid biosynthesis are located in the cytoplasmic
compartment, and the enzymes that metabolize phospholipids are bound to the inner
face of the cytoplasmic membrane. Each mole of phospholipid requires 32 mol of
ATP for its biosynthesis from acetyl-CoA and sn-glycerol-3-phosphate, and since ap-
proximately 10% of the dry weight of the cell is phospholipid, a significant amount of
energy expended in the biogenesis of a new cell is used in the production of phospho-
lipid. The advantage to the cell of maintaining fine control over the biosynthesis of
phospholipid is evident from these numbers, and the more recent work on lipid bio-
genesis has focused on uncovering these control mechanisms. The thrust of this chap-
ter will also be on the regulation of this branch of intermediary metabolism.

4. Genetic analysis of lipid metabolism

E. coli has a single chromosome composed of sufficient DNA to encode about 3000
proteins. Two general types of mutants can be obtained. Auxotrophic mutants are

strains that require a growth supplement which the organism isolated from nature
(called the wild-type strain) does not require. The auxotrophic lipid metabolic
mutants are those that require the addition of fatty acids or other lipid precursors (for
example, sn-glycerol-3-phosphate) to the growth medium. The reason for the nutri-
ent supplement is generally that the mutant organism has lost the function of a key
enzyme required in the biosynthesis of the compound; however, the reason for the
requirement can be more complex (for example, theplsB and acpS mutants, Table I).
   Conditional lethal mutations represent a second class of mutants. Many com-
pounds, such as phospholipids, are not readily taken up from the growth medium by
bacteria. Therefore, mutants unable to synthesize a phospholipid required for mem-
brane function would be nonviable (dead) and thus could not be isolated and studied
using the approach just described. However, the isolation of conditionally lethal mu-
tants allows the study of such pathways. Conditionally lethal mutants function nor-
mally under one set of conditions but become defective when shifted to a second set
of conditions. Although several types of conditionally lethal mutants have been used
in bacterial genetics, only temperature-sensitive mutations have been used in the
study of lipid metabolism. Temperature-sensitive mutants are strains that grow at a
low temperature (for instance, 30"C), but not at 42°C (wild-type E. coli grows at
844°C). The mutants owe this restricted growth temperature range to a mutation in
the gene encoding a required protein such that the mutant protein denatures at an
abnormally low temperature. Therefore, the protein is functional in cells incubated at
30"C, but at 42"C, the protein is nonfunctional and growth ceases. The mutational
alteration in protein structure is usually a single amino acid change, and the increased
thermal lability of the protein can often be demonstrated in vitro. Such a demonstra-
tion is strong evidence that the mutation is within the gene encoding the protein.
   Three general approaches have been used to isolate the E. coli mutants listed in
Table I. In all three approaches, the wild-type strain is treated with a strongly muta-
genic chemical to increase the rate of mutation, after which mutants are sought. The
first technique is to isolate auxotrophic mutants that require either fatty acids or
sn-glycerol-3-phosphate by use of standard bacterial genetic selection methods. A
second type of selection is to incorporate large quantities of highly radioactively la-
beled lipid precursors into the mutagenized cells at the nonpermissive growth
temperature (usually 42°C) and then store the labeled cells. During storage, those
cells competent in lipid synthesis are irradiated by the disintegration of the incor-
porated radioactive precursor, and they die, whereas those cells mutationally defec-
tive in lipid synthesis survive due to their lack of incorporation of the radioactive
precursor. The third method is to screen colonies of mutagenized cells either by
performing a given lipid synthetic enzyme assay on isolated bacterial colonies (Fig.
 1) or by screening such colonies for the inability to synthesize a given lipid. The use
of such brute-force colony autoradiography schemes has proven extremely valuable
in isolating phospholipid mutants [ 3 ] .
    The chromosome of E. coli is circular and is divided into 100 minutes (map units)

Table I
Genes in E. coli lipid biosynthesis

Gene        Proteinifunction                    Map position   Mutant biochemical
            affected                            (min)          or growth phenotype

aas         Acyl-ACP synthetase                 61             Accumulation of lysophosphatidyl
                                                               ethanolamine and acyl-
acpP        Acyl carrier protein                24             Unknown
acps        Acyl carrier protein synthase       43             Accumulation of apo-ACP
cdh         CDP-diacylglycerol hydrolase        89             Defective enzyme activity
cdsA        CDP-diacylglycerol synthase          4             pH sensitive growth
CdSS        CDP-diacylglycerol synthase         69             Suppresses pH sensitivity of cds
rfa         Cyclopropane fatty acid synthase    36             Cells lack cyclopropane fatty acids
CIS         Cardiolipin synthase                21             Deficient in cardiolipin
dgkA        Diacylglycerol kinase               92             Cells osmotically sensitive
dgk R       Diacylglycerol kinase               94             Overproduction of dgkA protein
fahA        3-H ydrox ydecano yl-ACP            22             Unsaturated fatty acid auxotroph
            deh ydrase
fabAup      3-H ydroxydecanoyl- ACP             22             Overproduction of saturated fatty
            dehydrase                                          acids
fab B       3-Ketoacyl-ACP synthase I           50             Unsaturated fatty acid auxotroph
fabD        Malonyl-CoA:ACP transacylase        24             Temperature-sensitive growth
fabE        Acetyl-CoA carboxylase              71             Temperature-sensitive growth
fubF        3-Ketoacyl-ACP synthase I1          24             Altered thermal regulation deficient
                                                               in cis-vaccenate
fadR        Positive regulation offabA and      26              Growth on decanoate
            negative regulation offud regulon
@A          Unknown                             69             Utilization of trans-unsaturated
                                                               fatty acids
gPsA        sn-Glycerol-3-phosphate             81             Glycerol-3-phosphate auxotroph
lpxA        UDP-N-acetylglucosamine              4             Temperature-sensitive defect in lipid
            acetyltransferase                                  A biosynthesis
lpxB        Lipid A disaccharide synthase        4             Accumulation of lipid X
mdoA        Glucosyl transferase                23             Lacks membrane-derived
mdoB        Phosphoglycerol transferase I       99             Lacks membrane-derived
PgPA        Phosphatidylglycerol phosphate 9                   Defective enzyme activity
PgPB        Phosphatidylglycerol phosphate 28                  Defective enzyme activity
PgsA        Phosphatidylglycerol phosphate 42                  Defective enzyme activity
pldA        Phospholipase A, (detergent    86                  Defective enzyme activity

Table I (continued)
Genes in E. co/i lipid biosynthesis

Gene        Proteinlfunction                   Map position   Mutant biochemical
            affected                           (min)          or growth phenotype

pldB        Lysophospholipase L,               86             Defective enzyme activity
plsB        sn-Glycerol-3-phosphate            92             Glycerol-3-phosphate auxotroph
plsC        1-Acylglycerolphosphate            65             Temperature-sensitive growth
plsX        Unknown                            24             Required for growth of plsB
            Phosphatidylserine decarboxylase   95             Accumulate PtdSer and lack PtdEtn;
                                                              Cells require Mgz+
pnA         Phosphatidylserine synthase        56             Defective enzyme activity
pssR        Phosphatidylserine synthase        85             Overproduction of pssA protein
tesB        Thioesterase I1                    10             Defective enzyme activity

as determined by the time-of-entry in interrupted-conjugation experiments. The en-
tire chromosome is also linked by cotransduction data using bacteriophage PI allow-
ing the construction of a detailed genetic linkage map [4]. Each minute of the map
corresponds to about 45 kilobases of DNA. There is also a physical restriction en-
zyme map of the E. coli chromosome generated by the isolation and ordering of hun-
dreds of genomic clones in bacteriophage 1 to produce a set of overlapping clones
that span the entire chromosome [5]. The correspondence between the physical and
genetic maps is excellent, and often the 1 phage carrying a particular gene can be
identified if the map position is known. The locations of the known genetic loci of the
enzymes involved in phospholipid metabolism are shown in Fig. 2, with the 100/0
point of the map located at the top center of the figure. The general impression is that
the genes of phospholipid metabolism are scattered throughout the circular map and
thus far only two examples of operon structure (cotranscribed genes) have been
found, both in fatty acid synthesis. Several other genes of phospholipid metabolism
occupy adjacent positions on the chromosomal DNA but are not cotranscribed.
These enzymes are indispensable for cell proliferation; therefore, these enzymes are
not inducible, but rather are expressed at the same relative level regardless of the
growth condition. This means that the regulation of the phospholipid biosynthetic
pathway is exerted at the level of individual enzyme activities and not at the level of
gene expression. In addition to the genes listed in Table I, there are many other loci
that have been extremely useful in studying phospholipid metabolism. A listing of
these mutations and their uses can be found in [6].
   It should be noted that mutants are defined by two characteristics: the phenotype
of a mutant is its outward manifestation, whereas the genotype is the genetic altera-
tion which causes the phenotype. Examples are the fubA and fa6B mutations. Both

Fig. 1, Isolation of mutants by colony autoradiography. Mutagenized colonies growing on the surface of
an agar Petri plate are blotted onto a sheet of filter paper. Much of the colony adheres to the paper. The
paper is then immersed in a solution which lyses the colonies (the membranes and proteins of the lysed cells
remain bound to the paper) and is next immersed in a second solution containing the precursors and buffer
necessary for the enzyme reaction. A water-soluble substrate of the reaction is included in a radioactively
labeled form. The immobilized lysed cells are allowed to incorporate the radioactive precursor for a short
time; then the filter is washed repeatedly with a solution in which lipids are insoluble but in which water-
soluble compounds remain in solution. The washed paper is stained with a specific protein stain, dried, and
exposed to a sheet ofX-ray film. When the resulting autoradiogram (B) is compared with the stained filter
paper (A), colonies that lack enzyme activity (arrow) can be identified. The mutant colony is then isolated
from the original agar plate.

the JubA and f k b B mutants (Table I) have the same phenotype, a requirement for
unsaturated fatty acids for growth. However, the two mutants have different geno-
types, since their phenotype is due to lesions in two different genes (Fig. 2) which
encode different fatty acid biosynthetic enzymes.

5. Membrane systems of E. coli

The primary metabolic fate for phospholipids is the formation of the two membrane
systems of E. coli (Fig. 3). Like other gram-negative bacteria, E. coli has an inner
cytoplasmic membrane that contains the enzymes of phospholipid biosynthesis, elec-
tron transport, metabolite and ion transport, and other metabolic processes. Between
the inner and outer membranes there is an osmotically active compartment called the
periplusmic space. Membrane-derived oligosaccharides, peptidoglycan, and binding
proteins involved with metabolite transport are found in this compartment. The

  pssR             minutes


              E . coli K




Fig. 2. Location of the known genes of lipid metabolism on the E. colichromosome.

                                                            Lipid A

                                                       I+    Phospholipids

                 (             Cytoplasm         && c Proteins

Fig. 3. Membrane systems of E. coli. The major structures indicated are the outer membrane, the periplasm
and the inner (cytoplasmic) membrane. MDO, membrane-derived oligosaccharides; KDO, 3-deoxy-~-
niunno-octulosonic acid. Reprinted with permission from [lo].

outer membrane is considerably different from the inner membrane. Pores exist in
this structure that allow the passage of molecules having a molecular weight less than
600, whereas the inner membrane is impermeable to solutes unless specific transport
systems are present. The outer layer of the outer membrane is composed primarily of
lipopolysaccharides rather than phospholipid. The outer membrane is rich in struc-
tural lipoproteins and proteins involved in the transport of high molecular weight
compounds. Some of these proteins also function as receptors for bacteriophages
that infect E. coli.
   The marked difference in the composition of the inner and outer membrane sys-

              10       20     30 40 50
                  Fraction Number
Fig. 4. Separation of the inner (cytoplasmic) membrane from the outer membrane by sucrose density gra-
dient centrifugation. The two membrane systems are separated by isopycnic centrifugation on linear suc-
rose gradients (30-55%, wlw). The distribution of [3H]glycerol-labeledphospholipids in the gradient is
shown in the figure. CM, cytoplasmic membrane; OM, outer membrane; and M, an intermediate fraction
composed of a mixture of the two membrane systems.

tems of E. coli has been established due to the development of a reliable method for
the separation of these two structures [7]. If the associations between the two mem-
brane systems are broken, the two structures can be clearly resolved using density-
gradient centrifugation (Fig. 4). The presence of lipopolysaccharides and the higher
protein:lipid ratio in the outer membrane results in the banding of the outer mem-
brane at a higher density (1.22 g/ml) than the inner membrane (1.16 g/ml). Virtually
everything we know about the subcellular distribution of phospholipids and phos-
pholipid enzymes in E. coli is derived from the analysis of such gradients.

6. Lipid biosynthetic pathways in E. coli

6.1. Initiation o fatty acid biosynthesis

The precursors for fatty acid biosynthesis are derived from the acetyl-CoA pool. Ma-
lonyl-CoA is required for all the elongation steps and is formed by the carboxylation
of acetyl-CoA by acetyl-CoA carboxylase (Fig. 5). Acetyl-CoA carboxylase is com-
posed of three individual proteins: biotin carboxylase, biotin carboxyl carrier pro-
tein, and transcarboxylase. There appear to be two pathways for the initiation of
fatty acid biosynthesis (Fig. 6). In both cases, malonyl-ACP is a substrate for the
condensing enzyme and is formed by malonyl transacylase which catalyzes the inter-
conversion of malonyl-CoA and malonyl-ACP. Acetoacetyl-ACP synthase (3-ketoa-
cyl-ACP synthase 111) catalyzes the condensation of acetyl-CoA and malonyl-ACP.
This enzyme does not utilize acetyl-ACP and specifically catalyzes the first condensa-
   ATP + H C O j


             BlOT IN CARBOXYLASE



Fig. 5. The acetyl-CoA carboxylase system in E. coli is composed of three separate proteins that catalyze

the two step process. First, biotin carboxylase carboxylates the N" position of the biotin ring attached to
biotin carboxyl carrier protein (BCCP). Second, the transcarboxylase component transfers the C 0 2moiety
from biotin carboxyl carrier protein to acetyl-CoA to form malonyl-CoA.

                                   CH3 -C-SCOA
                                                                     CO2 +&ASH

                ACPSH       CoASH

  P        P       U        -        P       P
                                 0-C-CHI-C-SACP                           P      P
                                                                     CH3 -C-CHI-C-SACP

                        @         W q! f B A
                                 W Ay fKB S
                ACPSH       CoASH

       P                           CH3-C-SACP
                                                                     CO2 +ACPSH

Fig. 6. The initiation phase of fatty acid biosynthesis. The initiation of new acyl chains is accomplished by
the action of four enzymes: ( I ) malonyl-CoA:ACP transacylase, (2) acetyl-CoA:ACP transacylase and (3)
acetoacetyl-ACP synthase or, alternatively (4) 3-ketoacyl-ACP synthase.

tion reaction. A second pathway begins with the transacylation of acetyl-CoA to ace-
tyl-ACP by acetyl transacylase followed by the condensation of acetyl-ACP and ma-
lonyl-ACP by 3-ketoacyl-ACP synthase I. The reasons for the existence of two path-
ways for initiation of fatty acid biosynthesis and their relative importance is an area
of active investigation.

6.2. Elongation o ucyl chains

Four enzymes participate in each cycle of chain elongation; the general reaction
scheme is shown in Fig. 7. First, 3-ketoacyl-ACP synthase I or I1 add an additional

                         ?                            ? ?
     CH3-(CHz)x -CH   CH-C-S-ACP         CH3-(CHz ),-C-CHz -C-S-ACP

Fig. 7. The elongation cycle of fatty acid biosynthesis. The elongation of a growing acyl chain is accom-
plished by the action of four enzymes: ( I ) 3-ketoacyl-ACP synthase, (2) 3-ketoacyl-ACP reductase. (3) 3-
hydroxyacyl-ACP dehydrase, and (4) truns-2-acyl-ACP reductase (enoyl reductase).

two-carbon unit from malonyl-ACP. The resulting ketoester is reduced by an
NADPH-dependent 3-ketoacyl-ACP reductase, and a water molecule is then re-
moved by the 3-hydroxyacyl-ACP dehydrase. The last step is catalyzed by enoyl-
ACP reductase to form a saturated acyl-ACP, which in turn can serve as the substrate
for another condensation reaction. NADPH is probably the preferred cofactor for
the enoyl-ACP reductase, but there may be two enzymes involved - one specific for
NADPH and the other specific for NADH. An important point to remember about
the elongation phase of fatty acid biosynthesis is that the condensing enzymes cata-
lyze the only irreversible steps in the elongation cycle.

6.3. Product diversification

Three major fatty acids are produced by the E. coli fatty acid synthase system, name-
ly, palmitic, 16:0, palmitoleic, 16:1(49), and cis-vaccenic, 18: l(dI I), acids. A specific
dehydrase enzyme, 3-hydroxydecanoyl-ACP dehydrase, first described by Bloch and
coworkers [8], catalyzes a key reaction at the point where the biosynthesis of sat-
urated fatty acids diverges from unsaturated fatty acids (Fig. 8). This dehydrase is a
distinctly different enzyme from the 3-hydroxyacyl-ACP dehydrase that participates
in the elongation cycle in that the 3-hydroxydecanoyl-ACP dehydrase is also capable
of isomerizing trans-2- to cis-3-decenoyl-ACP. Genetic studies have also shown that
the two condensing enzymes, I and 11, are responsible for different aspects of the
elongation reactions in the unsaturated branch of the pathway. 3-Ketoacyl-ACP syn-
thase I is absolutely required for one of the elongation reactions in this branch. and
3-ketoacyl-ACP synthase I1 is responsible for the elongation of palmitoleate to cis-



  10:1A2 trans-;-decenoyl-ACP
                                             10:1A3 cis-3-decenoyCAC


Fig. 8. Product diversification in fatty acid biosynthesis. Three main fatty acids are produced by the E. coli
fatty acid synthase system. The ratio of these fatty acids is controlled by the activity of three enzymes: (1)
3-hydroxydecanoyl-ACP dehydrase (encoded by thefnhA gene) is a specific dehydrase that introduces the
double bond into the acyl chain, (2) 3-ketoacyl-ACP synthase I (encoded by thefabB gene) catalyzes an
essential step in the unsaturated fatty acid elongation pathway, ( 3 ) both 3-ketoacyl-ACP synthases I and
I1 can elongate saturated fatty acids, and (4) 3-ketoacyl-ACP synthase 11 (encoded by the fubF gene) is
responsible for the elongation of 16:1(49) to 18:1(4ll),

                    0 -
                RI -C -S ACP
                                             R2-C   -S -ACP         0
HO - CH                                                  Rz-C-O-CH
                                HzC-O-P03'                     HzC -0 -P03'
                    ACPSH                       ACPSH

Fig. 9. Fatty acid transfer to the membrane. A fatty acid is transferred from the acyl-ACP pool to the
 I -position of glycerol-3-phosphate and, subsequently, a different fatty acid is tranferred from acyl-ACP
and added to the 2-position of I-acylglycerolphosphate to form the first membrane phospholipid in the
pathway. The substrate specificity of the sn-glycerol-3-phosphate acyltransferase (encoded by the plsB
gene) determines the preponderance of palmitic acid in the l-position, and similarly, the substrate specific-
ity of the 1-acyl-sn-glycerol-3-phosphate acyltransferase (encoded by the plsC gene) governs the preferen-
tial incorporation of unsaturated fatty acids into the 2-position.

vaccenate. Control of product distribution is one of the most important adaptive
responses in bacterial physiology and the regulation of the pathway shown in Fig. 8
will be covered in detail in later sections.

6.4. Transfer to the memhrune

The transformation of the water-soluble acyl-ACP end products of fatty acid biosyn-
thesis into a membrane phospholipid is accomplished by the glycerolphosphate
acylation system (Fig. 9). The first enzyme (glycerolphosphate acyltransferase) trans-
fers an acyl moiety from acyl-ACP (or acyl-CoA) to the 1-position of glycerolphos-
phate. The second enzyme (1-acylglycerolphosphateacyltransferase) esterifies the 2-
position of the glycerol backbone with a second acyl chain. Like most other phospho-
lipids in nature, bacterial phospholipids have an asymmetric distribution of fatty
acids between the 1- and 2-positions of the glycerolphosphate backbone (Fig. 9), and
the acyl chain specificity of the glycerolphosphate acyltransferase system is consid-
ered to account for this important aspect of membrane phospholipid structure. How-
ever, some bacteria do not have diacylphospholipids. Archaebacteria contain glyce-
rolipids containing phytanoyl ether groups at the 1- and 2-positions of glycerol, and
Clostridia contain an abundant amount of alk-1’-enyl (Chapter 8) groups at the 1-
position of their glycerolphosphatides.The details of the biosynthesis of these more
unusual phospholipid structures remain to be elucidated.

6.5. Diversijication of polar head groups

Phosphatidic acid formed by the glycerolphosphate acyltransferase is utilized by
phosphatidate cytidylyltransferase(Fig. 10) along with CTP to form CDP-diacylgly-
cerol, the key intermediate in the formation of the diverse phospholipid species found
in bacterial systems. The diversity of polar head groups in the procaryotic kingdom
defies adequate description in this short space; the reader is referred to Goldfine’s
review [9] for a more comprehensive treatment of bacterial phospholipid structures.
   Phospholipid biosynthesis has been most carefully studied in E. coli [lo]. This or-
ganism possesses one of the simplest phospholipid compositions, consisting prima-
rily of phosphatidylethanolamine (75%), phosphatidylglycerol(15-20%), and cardio-
lipin (5-10%). The two more abundant phospholipids are synthesized from CDP-
diacylglycerol (Fig. 10). Phosphatidylserine synthase exchanges L-serine for CMP to
form phosphatidylserine, which is subsequently decarboxylated by phosphatidylse-
rine decarboxylase to form the most abundant phospholipid, phosphatidylethanola-
mine. Similarly, phosphatidylglycerol arises from the exchange of CMP for sn-glyce-
rol-3-phosphate to form phosphatidylglycerolphosphate, which is subsequently de-
phosphorylated to form phosphatidylglycerol. Cardiolipin is synthesized by the con-
densation of two molecules of phosphatidylglycerol.

6.6. Central role of acyl carrier protein

A quick examination of Figs. 6-10 shows that ACP is a critical cofactor in four of the
five phases of membrane phospholipid biogenesis; consequently, this protein has re-
ceived considerable experimental attention [11,12]. ACP is one of the most abundant
proteins in E. coli (0.25% of the total soluble protein). It has been purified to homoge-
neity and its complete amino acid sequence and solution structure have been deter-
mined by high resolution nuclear magnetic resonance spectroscopy (Fig. 11). ACP,
whose molecular weight is 8860, has a number of characteristic structural features. It


                                        Phorphrtlac Add

                                                                        @ H-
                                                                         - *
                                                                         c         CH- CH*

                       0                                                    0
                       I                                                    I

                                                                        @- CH p-          FH- Clip- OH


Fig. 10. Synthesis of phospholipid polar head groups. The three phospholipid species found in E. coli are
synthesized by a series of reactions utilizing six enzymes: (1 ) phosphatidate cytidylyltransferase, (2) phos-
phatidylserine synthase, (3) phosphatidyl serine decarboxylase, (4) phosphatidylglycerolphosphate
synthase, (5)phosphatidylglycerolphosphate phosphatase, and ( 6 ) cardiolipin synthase. PG, phosphatidyl-

Fig. 11. Backbone structure of E. coli acyl carrier protein determined by high-resolution nuclear magnetic
resonance spectroscopy. The right view is produced by a rotation of 90" about the X-axis of the protein.
The structure is dominated by three a-helices which are designated by the single, double or triple bars in
the two views of the protein. The prosthetic group, which is the linkage point for a fatty acid chain, is
attached to serine-36 at the bottom of the second helix. The acyl chain extends up along the second helix
and makes contact with residues isoleucine-54 and alanine-59. Reprinted with permission from [I 31.

has a preponderance of acidic residues, resulting in an isoelectric point of pH 4.1, and
a high content of a-helical secondary structure. Hydrodynamic studies have shown
that ACP is an asymmetrically shaped protein with a major axis:minor axis ratio of
1.4. The intermediate of fatty acid biosynthesis is bound to the protein as a thioester
attached to the terminal sulfhydryl of the 4'-phosphopantetheine prosthetic group.
The prosthetic group sulfhydryl is the only thiol group in E. coli ACP. The prosthetic
group is in turn attached to the protein via a phosphodiester linkage to Ser-36 of the
protein. A /?-turn is commonly found between structural domains of proteins and
consists of a group of four amino acids that constitute an approximate 180" reversal
in the direction of the amino acid backbone. Ser-36 is located in such a /?-turn struc-
ture located between the second and third a-helical segments of ACP (Fig. 11). Se-
quence data are available for ACP in fungi, plants, and animals; all of these ACPs
have a helical secondary structure and a primary sequence that is highly conserved in
the region of the protein on either side of the attachment site for the phosphopan-
tetheine prosthetic group.
   The protein portion of ACP is metabolically stable in E. coli, but its phosphopan-
tetheine prosthetic group is metabolically active and undergoes considerable turn-
over in vivo. The prosthetic group turnover cycle is mediated by the action of two
enzymes: [ACPIsynthase, which is responsible for the transfer of the phosphopante-
theine portion from CoA to Ser-36 of apo-ACP, and [ACPIphosphodiesterase, which
initiates the turnover cycle by cleaving the prosthetic group from ACP (Fig. 12). Each
round of prosthetic group turnover results in the expenditure of two molecules of
ATP that are used in the conversion of phosphopantetheine to CoA.
   The measurement of the rate of prosthetic group turnover in vivo presents an inte-
resting biochemical challenge. A conventional pulse-chase experiment cannot be
used to study ACP turnover because the CoA pool is both large and metabolically

 4'- phosphopantetheine


A              c         u
    3:5'-ADP       CoA

Prosthetic Group Turnover
Fig. 12. Turnover of the phosphopantetheine prosthetic group ofacyl carrier protein. The hydrolysis of the
prosthetic group is accomplished by the enzyme ACP-phosphodiesterase, and the phosphopantetheine is
added to the apo-protein by the enzyme ACP-synthase.

stable and thus cannot be effectively chased. Two approaches have been used to cir-
cumvent this difficulty. In the first, a mutant unable to synthesize the pantothenate
portion of phosphopantetheine is supplemented with exogenous [3H]pantothenateto
uniformly label the CoA and ACP pools. Next, the CoA pool is severely depleted by
depriving the mutant of the pantothenate supplement required for the production of
CoA. After the intracellular CoA supply is virtually exhausted, the cells are resus-
pended in media containing ['4C]pantothenate, thereby initiating growth and CoA
synthesis. The rate of prosthetic group turnover is determined by measuring the
change in the tritium:carbon-14 ratio in the CoA pool to estimate the rate of [panto-
thenate- 'HIACP turnover [14].
   A second approach is to exploit the effect of heavy isotopes on the solution struc-
ture of ACP (the constitutional isotope effect) to differentiate between newly synthe-
sized and preexisting ACP in a pulse-chase experiment. The incorporation of deute-
rium into nonexchangeable positions (for example, -CH2-) of the ACP amino acid
backbone results in the destabilization of the protein and increased sensitivity to pH-
induced hydrodynamic expansion. Therefore, deuterio-ACP can be separated from
the normal protio-ACP by conformationally sensitive gel electrophoresis [ 151. This
technique separates ACP species according to their stability to pH-induced denatura-
tion. After the ACP prosthetic group is prelabeled with tritiated prosthetic group
precursor (/3-alanine) during growth on deuterium oxide medium, the cells are chased
with ''C-labeled P-alanine without deuterium oxide. The rate of prosthetic group
turnover is determined by measuring the loss of tritium and its replacement with I4C
in preexisting deuterio-ACP. At low intracellular CoA levels, ACP prosthetic group
turnover was four times faster than the rate of new ACP synthesis, but at higher CoA
concentrations characteristic of logarithmic growth, turnover was an order of magni-
tude lower, amounting to 25% of the ACP pool per generation. Despite continued
interest in this topic, the role of ACP prosthetic group turnover in bacterial lipid
metabolism remains an enigma.
   Although ACP plays an indispensable role in fatty acid biosynthesis, the size of the

ACP pool can be severely depleted without significantly affecting phospholipid pro-
duction [ 161. This point was investigated by the analysis of [3H]leucine-labeledcell
extracts by conformationally sensitive gel electrophoresis, since the removal of the
prosthetic group to form apo-ACP also destabilizes the protein to pH-induced de-
naturation. Normally a significant pool of inactive apo-ACP does not exist in vivo,
but in E. coli strains containing the acpS mutation (abnormal ACP synthase func-
tion), apo-ACP accumulates and becomes the major form of ACP in vivo. Although
the acpS mutant contains much less ACP than wild-type strains, it still possesses a
normal 1ipid:protein ratio, indicating that ACP concentration per se is not a factor in
determining the overall rate of phospholipid biosynthesis.
   The gene encoding ACP (called acpP) has recently been isolated. Isolation of this
gene was unusually difficult and it now seems clear that this was due to cell death
caused by overproduction of ACP. Construction of a synthetic gene encoding ACP
showed that production of increased levels of ACP caused growth inhibition and
death of E. coli. The earlier attempts to isolate the ACP gene utilized gene libraries
present in high copy number plasmid cloning vectors and thus the resulting overpro-
duction would have eliminated cells carrying ACP plasmids from the banks. The rea-
son for the toxicity of ACP overproduction is an area of current investigation.
   The acpP gene has been shown by genetic and molecular genetic techniques to be
located beween thefabD and fabF genes encoding malonyl-transacylase and 3-ketoa-
cyl-ACP synthase 11, respectively. Another gene that lies betweenfabD and acpP has
been detected by sequence analysis which contains a consensus N A D N A D P binding
domain. This gene seems likely to encode an enzyme catalysing a reductive step in
fatty acid synthesis (e.g., reaction 2 or 4 of Fig. 7). This gene is cotranscribed with
ACP and thus provides one of the first examples of operon structure in lipid metabo-

7. Lipopolysaccharide biosynthesis

The outer leaflet of the outer membrane of E. coli contains minor amounts of phos-
pholipid and is instead composed mostly of lipopolysaccharide (Fig. 3). These lipo-
polysaccharides consist of three regions of contrasting chemical and biological
properties. The outermost region consists of the 0-specific polysaccharide and forms
the basis for the serological differences between closely related bacteria. The O-anti-
gen region is linked to a core polysaccharide region, which is common to groups of
bacteria. This region is in turn attached via a 2-keto-3-deoxyoctonate trisaccharide to
the lipid component termed lipid A. Lipid A anchors the lipopolysaccharide to the
outer membrane and also functions as an endotoxin and a mitogen during bacterial
  The only major E. coli fatty acid that is not a component of the phospholipids is
3-hydroxymyristic acid. Rather, this fatty acid is attached by both ester and amide

linkages to the saccharide residue of the lipid IV, portion of the outer membrane
lipopolysaccharide (Fig. 13). The available evidence suggests that 3-hydroxymyristic
acid is derived from the central fatty acid biosynthetic machinery. The steps involved
in the transfer of 3-hydroxymyristic acid to the saccharide residue are shown in Fig.
 13 [ 171. However, the mechanism that determines whether the 3-hydroxymyristoyl-
ACP is channeled to lipopolysaccharide biosynthesis rather than elongation to pal-
mitic acid and hence to phospholipid is unknown.
   Some confusion has been created in this area, since the originally published struc-
ture of the lipopolysaccharide was in error with regard to the placement of the 3-
hydroxymyristoyl moieties. The genetic approach provided the key to unraveling the
chemical structure of the lipopolysaccharide [ 181. Using the colony autoradiography
approach, strains defective in phosphatidylglycerolphosphate synthase were isolated
that contained less than 5% of the wild-type enzymatic activity. Interestingly, these
mutants (pgsA) were not temperature-sensitive for growth and contained normal lev-
els of phosphatidylglycerol. In order to obtain mutants with reduced phosphatidyl-
glycerol content, second-step mutants were generated by starting with a parent har-
boring a defective pgsA gene. This approach yielded a temperature-sensitive orga-
nism that contained less than 5% of the normal amount of phosphatidylglycerol after
3 h of incubation at 42°C. This second mutation was designated pgsB; it is far re-
moved from pgsA on the bacterial chromosome. It was then discovered that two
molecules related to the lipid A biosynthetic pathway accumulate in strains harbor-
ing both the pgsA and pgsB mutations. It was found that the pgsB gene is the same
as the lpxB gene that catalyzes the formation of tetra-acyldisaccharide phosphate
(Fig. 13). The structure of these simple intermediates suggested a lipid A structure
different from the prevailing model, thus providing the impetus to reevaluate lipo-
polysaccharide structure and biosynthesis [ 171. The currently accepted lipid A struc-
ture is synthesized by the condensation of UDP-2,3-diacylglucosamine with 2,3-dia-
cylglucosamine- l-phosphate to form a @l+6) tetraacyldisaccharide-l-phosphate,
which is subsequently phosphorylated at the C-4' position. The acyltransferase reac-
tions that attach 3-hydroxymyristic acid to the glucosamine residue have been char-
acterized and are specific for 3-hydroxymyristoyl-ACP [ 191. 3-Hydroxymyristic acid
is taken up by the bacteria and can be used as a carbon source via the P-oxidation
pathway; however, exogenous 3-hydroxymyristate is not incorporated into lipid A
because it is not converted to 3-hydroxymyristoyl-ACP.

8. Degradative pathways

8.1. Phospholipases

There are 10 reported enzymatic activities that degrade phospholipids, intermediates
in the phospholipid biosynthetic pathway, or triacylglycerol (Table 11). The best char-


                                                              (Lipid X)

Fig. 13. Biosynthesis of glucosamine-based phospholipids in E. coli. Uridine diphospho-N-acetylglucosa-
mine (UDP-GlcNAc) acyltransferase is the first committed enzyme. The lpxA and IpxB genes code for the
enzymes indicated in the pathway. Lipid IV, is modified further by the addition of two 3-deoxy-D-munno-
octulosonic acid moieties. the 0-acylation of hydroxymyristate. and the attachment of additional core
sugars and 0-antigen to yield the mature lipopolysaccharide. (Reprinted with permission from Raetz and
Dowhan, 1990, J . Bid. Chem. 265, 1235.)

Table I1
Lipid degradative enzymes in E. coli

Enzyme                                 Location             Substrates

Phospholipase A,                       Outer membrane       Phosphatidylethanolamine,
                                                            cardiolipin and lyso derivatives
Phospholipase A                        Cytoplasm            Phosphatidylglycerol
Lysophospholipase L,                   Inner membrane       Lyso-phosphatid ylethanolamine
Lysophospholipase                      Cytoplasm            Lyso-phosphatidylethanolamine,
                                                            l yso-phosphatidylglycerol
Phospholipase C                        Unknown              Phosphatidylethanolamine
Phospholipase D                        Cytoplasm            Cardiolipin
Phospholipase D                        Cytoplasm            Phosphatidylserine
Lipase                                 Membrane             Triacy Iglycerol
CDP-diacylglycerol hydrolase           Inner membrane       CDP-diacylglycerol
Phosphatidic acid phosphatase          Membrane             Phosphatidic acid
Thioesterase I                         Cytoplasm            Acyl-CoA
Thioesterase 11                        Cytoplasm            Acyl-CoA

acterized of these is the detergent-resistant phospholipase A, of the outer membrane.
This enzyme is unusually resistant to inactivation by heat and ionic detergents and
requires calcium for maximal activity. The phospholipase has been purified to homo-
geneity and exists as a single subunit with a molecular weight of 28 kDa. Hydrolysis
of fatty acids from the 1-position of phospholipids is the most rapid reaction, but the
enzyme will also hydrolyze 2-position fatty acids, as well as both isomeric forms of
lysophosphatides. A detergent-sensitive phospholipase A has also been described.
This enzyme differs from the detergent-resistant protein in that it is located in the
soluble fraction of the cell, is inactivated by heat and ionic detergents, and in contrast
with the broad substrate specificity of the outer membrane phospholipase, has a high
degree of specificity for phosphatidylglycerol. The cytoplasmic phospholipase A also
requires calcium for activity.
   There are also inner membrane and cytoplasmic lysophospholipases. The best char-
acterized is the inner membrane lysophospholipase L, which has been purified and
cloned. This enzyme hydrolyzes 2-acylglycerophosphoethanolamineefficiently, but
is barely active on the I -acyl isomer. This lysophospholipase also catalyzes the trans-
fer of fatty acids from 2-acylglycerophosphoethanolamine phosphatidylglycerol to
form acyl-phosphatidylglycerol.
   The physiological role of these degradative enzymes remains unknown. Mutants
lacking the detergent-resistant phospholipase ($A), lysophospholipase L, (PldB),or
both enzymes do not have any obvious defects in growth, phospholipid composition,
or turnover. Moreover, strains that overproduce the detergent-resistant enzyme
(constructed by molecular cloning) also grow normally. It has been established that
the detergent-resistant phospholipase is responsible for the release of fatty acids from
phospholipids that occurs during infection with T4 and A. phages. However, phos-
pholipid hydrolysis is not essential for the life cycle of these bacteriophages. One
possible function for the phospholipases is that they are actually biosynthetic pro-
teins that act as hydrolases in the absence of suitable acceptor molecules in the assay
systems employed. Examples of such enzymes are the phospholipase D and CDP-
diacylglycerol hydrolase activities that are associated with phosphatidylserine syn-
thase. Phosphatidylserine synthase appears to function via a phosphatidyl-enzyme
intermediate, and in the absence of a suitable acceptor such as serine or CMP, the
phosphatidyl- enzyme complex can be hydrolyzed by water; thus, the enzyme exhibits
either phospholipase D or CDP-diacylglycerol hydrolase activity. Recently, CDP-
diacylglycerol hydrolase of the inner membrane (an enzyme different from phospha-
tidylserine synthase) has been shown to be a cytidylyl donor to inorganic phosphate
and other phosphomonoester acceptors, which suggests that this enzyme is a bio-
synthetic cytidylyltransferase, although the identity of the acceptor molecule in vivo
has not been determined. Finally, some of these enzyme activities may reflect a broad
substrate specificity of a single enzyme rather than the presence of several distinct
protein species. For example, the observed lipase activity that cleaves the l-position
fatty acids from triacylglycerols (a lipid usually not found in E. coli) is probably due
to the presence of the detergent-resistant phospholipase A, acting on triacylglycerol
as an alternate substrate. In conclusion, much remains to be learned about the role
of phospholipid degradative enzymes in the lipid metabolism of bacteria.

8.2. Thioesteruses

E. coli contains two thioesterases that catalyze the hydrolysis of acyl-CoA molecules
(Rhodopseudomoms sphaeroides has two similar enzymes). Thioesterase I is a small
(molecular weight of 22 kDa) serine esterase that hydrolyzes only long-chain acyl-
CoAs. Thioesterase I1 is a much larger protein (molecular weight of 122 kDa) that is
insensitive to serine esterase inhibitors and cleaves a broad range of acyl-CoA chain
lengths. Both enzymes are much more active on acyl-CoA molecules than on acyl-
ACPs because the native solution structure of ACP protects the thioester bond from
attack by these enzymes. No function for either enzyme is known. Mutants (called
tesB) completely lacking thioesterase I1 have recently been isolated and the mutation
has no effect on growth or fatty acid composition. The gene has been isolated and
massive overproduction of thioesterase I1 also has no detectable effect on lipid me-
tabolism. The sequence of the tesB gene indicates that thioesterase I1 represents a
class of thioesterase distinct from the serine esterase-type thioesterases described in
eucaryotic cells. Acyl-CoAS are found in E. coli only when the /I-oxidation pathway
is operative. These thioesterases could play a role in the synthesis of various fatty-
acid-containing molecules in the cell other than phospholipids, as described in the

next section. It is possible that these thioesterases actually are acyltransferases that in
the absence of the unknown physiological acceptor molecules transfer the acyl chain
to water. A thioesterase that specifically hydrolyzes acyl-ACP has not been identi-

8.3. Fatty acid oxidation in bacteria

E. coli has an inducible system for the uptake and oxidation of fatty acids as a carbon
source for growth. The genes comprising the regulon for fatty acid degradation are
scattered throughout the bacterial chromosome, and the expression of these genes is
controlled by a single genetic locus CfadR). An important point to keep in mind is that
acyl-CoAs serve as the substrates for the enzymes of /?-oxidation, whereas the fatty
acid biosynthetic enzymes utilize acyl-ACPs. The biochemistry of the enzymes of
fatty acid oxidation in E. coli is covered in detail in Chapter 3.

9. Phospholipid turnover

9.1. The diacylglycerol cycle

Early observations on phospholipid metabolism showed that the polar head group of
phosphatidylglycerol was lost in a pulse-chase experiment, whereas that of phospha-
tidylethanolamine was quite stable. At first it was thought that the phosphatidylgly-
cerol was being degraded. Upon the discovery that E. coli contains cardiolipin, it was
realized that some of the phosphatidylglycerol ‘turnover’ was actually the conversion
of phosphatidylglycerol to cardiolipin catalyzed by cardiolipin synthase. However,
cardiolipin synthesis did not account for all the loss of ”P-labeled phosphatidylglyce-
rol observed in pulse-chase experiments nor did it explain why the nonacylated glyce-
rol of phosphatidylglycerol was labeled (and chased) more rapidly than the acylated
glycerol moiety. A nonlipid phosphate-containing compound derived from the head
group of phosphatidylglycerol was sought, and a family of molecules called mem-
brane-derived oligosaccharides (MDO) was discovered [20]. These molecules are
composed of sn-glycerol-1-phosphate (derived from phosphatidylglycerol), glucose,
and (usually) succinate moieties, have molecular weights of 4000-5000 Da and are
found in the periplasm of gram-negative bacteria. The periplasm, the space between
the cytoplasmic and outer membranes of these organisms (Fig. 3), is an osmotically
sensitive compartment. The synthesis of the MDO compounds is regulated by the
osmotic pressure of the growth medium (decreased osmotic pressure gives an in-
creased rate of MDO synthesis); thus, MDO compounds seem to be involved in
osmotic regulation.
   The discovery of the MDO compounds provided a function for the well-studied,
but enigmatic, enzyme diacylglycerol kinase. In the synthesis of MDO (Fig. 14), the

sn-glycerol-1-phosphate polar group of phosphatidylglycerolis transferred to the oli-
gosaccharide, with 1,2-diacylglycerolas the other product. Diacylglycerol kinase will
phosphorylate the diacylglycerol to phosphatidic acid, which can reenter the phos-
pholipid biosynthetic pathway (Fig. 10) to complete the diacylglycerolcycle (Fig. 14).
In the overall reaction only the sn-glycerol- 1 -phosphate portion of the phosphatidyl-
glycerol molecule is consumed; the lipid portion of the molecule is recycled back into
phospholipid. It is clear that MDO synthesis is responsible for most of the metabolic
instability of the polar group of phosphatidylglycerol, since its turnover is greatly
decreased if MDO synthesis is blocked at the level of oligosaccharide synthesis (by
lack of UDP-glucose). Moreover, the rate of accumulation of diacylglycerol in
strains lacking diacylglycerol kinase (dgk) depends on both the presence of the oligo-
saccharide acceptor and the osmotic pressure of the growth medium. It should be
noted that some species of MDO contain phosphoethanolamine. Although direct
proof is lacking, it is likely that the ethanolamine moiety is derived from phosphati-
dylethanolamine, as this is the only known source of ethanolamine.


                                                   8-H ~ - C H - C H ~ -
                                                     C                     OH

Fig. 14. The 1,2-diacylglycerolcycle. First, the sn-glycerol-1-phosphate moiety is removed from phosphati-
dylglycerol for the biosynthesis of membrane-derived oligosaccharides (MDO). The resulting 1,2-diacyl-
glycerol is then phosphorylated by 1,2-diacylglycerolkinase to phosphatidic acid, which is subsequently
reutilized in the synthesis of membrane phospholipids (see Fig. 10).

9.2. The 2-acylglycerolphosphoethanolamine cycle

2-Acylglycerolphosphoethanolamineacyltransferase is another inner membrane en-
zyme that participates in a metabolic cycle. This acyltransferase esterifies the 1-posi-
tion of 2-acyl-lysophospholipids utilizing acyl-ACP as the acyl donor (Fig. 15). Un-
like the sn-glycerol-3-phosphate acyltransferase, 2-acylglycerolphosphoethanola-
mine acyltransferase does not utilize acyl-CoA thioesters. 2-Acylglycerolphospho-
ethanolamine has been identified as a minor membrane lipid in E. coli and is ap-
parently the substrate for this enzyme in vivo. There is a small amount of fatty acid

        -C                 -OH
             ATP, Mg       +'                    AMP   +   PPi

              ACP - SH                                  ,,
                                                 ACP-S- 0

                                    &-o]    0

     -  apo   -
                   Cys - NH2

                                                            - NH - C    + V

Fig. 15. Relationship between turnover at the I-position of phosphatidylethanolamine and the acylation of
Brdun's lipoprotein in E. coli. The turnover cycle is initiated by the transfer of the I-position fatty acid of
phosphatidylethanolamineprimarily to the amino terminus of the lipoprotein. The resulting 2-acyl-glyce-
rolphosphoethanolamine (GPE) is reacylated by 2-acyl-GPE acyltransferase using acyl-ACP as the acyl
donor. Free fatty acids are ligated to ACP by acyl-ACP synthetase to generate the acyl-ACP. The fatty acid
activation and acyltransferase steps are carried out by a single inner membrane enzyme possessing a bound
ACP subunit.

turnover at the 1-position of phosphatidylethanolamine, and one metabolic fate of
the 1-position fatty acids appears to be acylation of the amino terminus of Braun’s
lipoprotein (Fig. 15) 1211 (Section 12.6).
   Normally, acyl-ACP intermediates are not observed in the 2-acylglycerol phos-
phoethanolamine acyltransferase assay system (Fig. 15); however, after the addition
of chaotropic salts (such as LiCl), acyl-ACP accumulates. This partial reaction of the
acyltransferase (acyl-ACP synthetase) is most active with saturated fatty acids as sub-
strates, although the enzyme ligates a broad range of fatty acids to ACP. This enzy-
matic activity is of considerable practical significance since acyl-ACP synthetase is
used to prepare native acyl-ACPs for use as substrates in the study of other fatty acid
and phospholipid biosynthetic enzymes. Biochemistry and genetics have demonstrat-
ed that 2-acylglycerolphosphoethanolamineacyltransferase and acyl-ACP synthe-
tase activities are catalyzed by the same protein.

10. Inhibitors of lipid metabolism

10.1. 3-Decynoyl-N-acetylcysteamine

The dehydrase enzyme that catalyzes both the dehydration and isomerization reac-
tions of fatty acid biosynthesis (Fig. 8) is specifically and irreversibly inhibited by the
acetylenic substrate analog 3-decynoyl-N-acetylcysteamine (decynoyl-NAC; Fig. 16)
[22]. The allenic inhibitor, 2,3-decadienoyl-NAC, inhibits dehydrase activity even
more effectively. 3-Decynoyl-NAC concentrations of 10 to 50 pM completely inhibit
bacterial growth, but growth inhibition is relieved by addition of unsaturated fatty
acids to the medium. Saturated fatty acid synthesis continues at its normal pace in the
presence of 3-decynoyl-NAC and supplies necessary precursors for lipopolysaccha-
ride production. This inhibitor has been useful in modifying membrane lipid compo-
sition and in the selection of mutants in the production of thefabA gene product.
Although the dehydrase is specific for dissociated fatty acid synthase systems charac-
teristic of procaryotes, a drawback to the use of dehydrase inhibitors as antimicrobial
drugs is that they would be ineffective in physiological environments where unsatu-
rated fatty acids are available to rescue the microbes.

10.2. Cerulenin

Cerulenin, (2R)(3S)-2,3-epoxy-4-oxo-7,10-dodecadienolyamide 16), is a fungal
product that is an irreversible inhibitor of 3-ketoacyl-ACP synthase I and I1 activities
and is extremely effective in blocking the growth of a broad spectrum of bacteria [23].
Cerulenin is a noncompetitive, irreversible inhibitor. Cerulenin blocks 3-ketoacyl-
ACP synthase activity by covalent modification of the synthase active site and inhibi-
tion correlates with the binding of 1 mole of cerulenin per mole of enzyme. 3-Ketoa-
                                              0                           0
                                              II                          1
CH3- CHZ- CHZ- CHz- CHZ-CHZ- C E C CHZ- C          - S - CHZ- CHZ - NH - C - C H 3
                      3-Decynoy CN- Acetylcysteamine

                                                   :: /”\         0


                                          ,c =c,
                                     HO              CH 3

Fig. 16. Structure of inhibitors of fatty acid biosynthesis. 3-Decenoyl-N-acetylcystamine a selective in-
hibitor of 3-hydroxydecanoyl-ACP dehydrase and blocks unsaturated fatty acid synthesis in procaryotes,
but not the multifunctional eucaryotic fatty acid synthase systems. Cerulenin is a fungal antibiotic that
inactivates the 3-ketoacyl-ACP synthase reaction in both Type I and Type I1 systems. Thiolactomycin is
a fungal antibiotic that blocks fatty acid production at the 3-ketoacyl-ACP synthase step of Type 11, but
not Type I, fatty acid synthase systems. (Reprinted with permission from Jackowski, Fatty Acid Biosyn-
thesis. In Sutcliffe and Georgopapadakou, Eds., Emerging Tcrrgets,forAntibacterial and Antifungal Chem-
otherapy.New York: Chapman and Hall, in press.)

cyl-ACP synthases I and I1 contain both a fatty acyl and a malonyl-ACP binding site.
Incubation of 3-ketoacyl-ACP synthases with acyl-ACP protects the enzymes from
cerulenin inhibition. These data strongly support the concept that cerulenin binds to
the fatty acyl site of the condensing enzyme. Acetoacetyl-ACP synthase (synthase 111)
is not inhibited by cerulenin, indicating that this condensing enzyme lacks the fatty
acyl binding site. Although cerulenin has proven to be a versatile biochemical tool
[23], it is not a suitable antibiotic because it is also a potent inhibitor of the condens-
ing enzyme reaction catalyzed by the multifunctional mammalian fatty acid synthase.
This observation is not surprising since the multifunctional synthases have a fatty
acyl binding site analogous to the site on prokaryotic 3-ketoacyl-ACP synthases I
and 11.

10.3. Thiolactomycin

Thiolactomycin, (4S)(2E,5E)-2,4,6-trimethyl-3-hydroxy-2,5,7-octatriene-4-thiolide
(Fig. 16), is a unique antibiotic structure that inhibits dissociated, but not multifunc-
tional fatty acid synthases. The antibiotic is not toxic to mice and affords significant
protection against urinary tract and intraperitoneal bacterial infections. An analysis
of the individual enzymes of the procaryotic fatty acid synthase system shows that
the 3-ketoacyl- ACP synthase activity and the acetyl-CoA:ACP transacylase activity

are the only activities inhibited by thiolactomycin. The observations that malonyl-
ACP protects the synthases from thiolactomycin inhibition and that they are com-
petitively inhibited with respect to malonyl-ACP are consistent with thiolactomycin
interacting with the malonyl-ACP site on the condensing enzymes rather than the
acyl-ACP site. All three condensing enzymes of E. coli are inhibited by thiolacto-

I I . Regulation of fatty acid composition in E. coli

11.1. Role oj3-hydroxydecanoyl-ACP dehydrase

The reaction in which the double bond is introduced into the growing fatty acid chain
is catalyzed by 3-hydroxydecanoyl-ACP dehydrase (Fig. 8). This enzyme, a homodi-
mer of an 18 kDa molecular weight subunit, catalyzes the specific dehydration of
3-hydroxydecanoyl-ACP to a mixture of trans-2-decenoyl-ACP and cis-3-decenoyl-
ACP. The double bond of the trans-2 intermediate is reduced to decanoyl-ACP by
enoyl-ACP reductase, whereas the cis-3 double bond is retained through subsequent
elongation steps and becomes the double bond of the unsaturated fatty acids of E.
coli, namely, palmitoleic acid and cis-vaccenic acid. The dehydration reaction pro-
ceeds first by formation of the trans-2-decenoyl-ACP as an enzyme-bound interme-
diate. A portion of this intermediate can dissociate from the enzyme and be converted
to saturated fatty acids by enoyl-ACP reductase, whereas the remaining trans-2 iso-
mer (still enzyme-bound) is isomerized to cis-3-decenoyl-ACP by the dehydrase. The
phenotype of mutants lacking this dehydrase demonstrates that the isomerase reac-
tion is the activity required for unsaturated fatty acid synthesis.
   Mutants lacking the dehydrase (calledfahA) were the first mutants isolated in the
fatty acid biosynthetic pathway. These mutants require unsaturated fatty acids for
growth but synthesize saturated fatty acids normally. In vitro, the mutant enzymes
are unable to catalyze the formation of trans-2-decenoyl-ACP or cis-3-decenoyl-
ACP. This finding, together with continued synthesis of saturated fatty acids ob-
served in vivo, indicated that trans-2-decenoyl-ACP can be synthesized by a dehy-
drase other than 3-hydroxydecanoyl-ACP dehydrase. However, this second dehy-
drase is unable to catalyze the isomerization to cis-3-decenoyl-ACP. This second
dehydrase is presumably the enzyme that catalyzes the dehydration of the 3-hydroxy-
acyl-ACPs of shorter and longer chain length (Fig, 7).
   ThefabA gene has been isolated and sequenced and the transcriptional regulation
of the enzyme has been studied. Surprisingly, thefubA gene is transcribed from two
promoters, a weak constitutive promoter and a strong promoter requiring the pres-
ence of FadR protein for function. Indeed, FadR protein has been found to bind to
a DNA segment that overlaps the major promoter. Thus, FadR seems to regulate
both the fabA gene and the /3-oxidation regulon (see Chapter 3) acting as both a

positive regulator of fatty acid synthesis and as a negative regulator of fatty acid
degradation (Fig. 17).

11.2. Role of 3-ketoacyl-ACP synthase I

Another class of fatty acid mutants VabB) has the same phenotype as the fubA
strains. Both mutants synthesize saturated fatty acids normally, but require for
growth the addition of unsaturated fatty acid to the medium. However, fabB strains
contain normal levels of 3-hydroxydecanoyl-ACP dehydrase, and genetic analysis
shows that the fabA and fubB mutations are in two different genes (Fig. 2; Table I).
When first isolated, thefabB mutants seemed an enigma, since the same set of fatty
acid biosynthetic enzymes (except 3-hydroxydecanoyl-ACP dehydrase) was thought
to function in both unsaturated and saturated fatty acid synthesis. Hence, a mutation
resulting in the loss of one of these enzymes was expected to block the synthesis of
saturated as well as unsaturated fatty acids.
   The puzzle was resolved by enzymological studies showing that two distinct en-
zymes catalyze the 3-ketoacyl-ACP synthase reaction in E. coli [25].Therefore, the


Fig. 17. Regulation by FadR protein. In the upper left panel (Repression) FadR protein (circles) is synthe-
sized and binds to a specific DNA sequence within the promoters of the genes encoding the proteins of the
fatty acid degradation (oxidation) pathway thus preventing transcription of these genes by RNA polymer-
ase (ovals). Fatty acids added to the culture medium (induction, lower left panel) enter the cells and bind
to FadR protein resulting in disassociation of the FadR-DNA complex, allowing RNA polymerase to
transcribe thep-oxidation genes and produce the required proteins. In the right panel FadR protein binds
to a DNA sequence within the promoter of thefabA gene thus facilitating binding of RNA polymerase to
an otherwise inactive promoter (positive control). Addition of fatty acids results in disassociation of the
FadR-DNA complex and inactivation of the promoter.

loss of one enzyme could yield a fabB mutant, since a second enzyme was available
for the elongation steps required in saturated fatty acid synthesis. Indeed, it was soon
shown that fab3 mutants lacked synthase I and that the fabB gene encodes the
synthase I enzyme. Therefore, it is clear that synthase I catalyzes a key, but uni-
dentified, reaction in unsaturated fatty acid synthesis in which synthase I1 is unable
to function. Both synthases I and I1 are capable of elongating saturated fatty acids.

11.3. Role of 3-ketoacyl-ACP synthase 1

The discovery of the role of 3-ketoacyl-ACP synthase I1 in unsaturated fatty acid
synthesis was an outcome of the investigation of a phenomenon called thermal con-
trol of fatty acid synthesis. E. coli, in common with most (if not all) other organisms,
synthesizes phospholipids with a greater proportion of unsaturated fatty acids when
grown at low, rather than high, temperatures. This regulatory system is thought to be
designed to ameliorate the effects of temperature change on the physical state of the
membrane phospholipids. As discussed in Chapter 1, the proportion of fluid (disor-
dered) lipid to nonfluid (ordered) lipid in cell membranes plays a major role in mem-
brane function. Increased incorporation of unsaturated fatty acids decreases the melt-
ing temperature of the membrane phospholipids, whereas increased incorporation of
saturated fatty acids has the opposite effect. The thermal regulatory system can thus
adapt the membrane lipid for optimal functioning at new growth temperatures.
   A key finding in unraveling the mechanism of thermal regulation in E. coli was that
protein synthesis was not required for the synthesis of new phospholipids with a fatty
acid composition characteristic of the new temperature. This result indicated that
thermal regulation was exerted by a protein synthesized at all growth temperatures
but active only at low temperatures. A clue to the identify of this protein was the
observation that the unsaturated fatty acid synthesized in greater quantity at low
temperature was cis-vaccenic acid; the amount of palmitoleic acid did not vary with
temperature. This result suggested that thermal regulation involved the conversion of
palmitoleic acid to cis-vaccenic acid. This suggestion was buttressed by the isolation
of a mutant (called fabF) defective both in the conversion of palmitoleic acid to cis-
vaccenic acid and in thermal regulation. The fabF strain had only traces of cis-vac-
cenic acid in its phospholipids, and its fatty acid composition (chiefly palmitic and
palmitoleic acids) did not change with growth temperature.
   The search for the enzyme missing in the fabF strains paralleled that described
from thefabB mutant and resulted in the demonstration that fabF strains are defec-
tive in the gene encoding 3-ketoacyl-ACP synthase 11. Moreover, analysis of strains
carrying mutations in both fabB (a temperature-sensitive mutant was used) andfabF
showed that these double mutants were defective in the production of long chain fatty
acids arguing against the existence of an analogous third synthase in E. coli.
   The analysis of the fabF mutants suggests that 3-ketoacyl-ACP synthase I1 func-
tions much better than synthase I in the elongation of palmitoleoyl-ACP to the 3-keto

precursor of cis-vaccenoyl-ACP and that the difference is accentuated at lower tem-
peratures (Fig. 18). Since mutants containing either synthase I or synthase I1 synthe-
size saturated fatty acids normally, either enzyme can catalyze all the elongation steps
required in saturated fatty acid synthesis. These predictions were borne out by in
vitro studies on the purified enzymes. A simple model emerged in which temperature
alters the activity of synthase 11, which in turn regulates the fatty acid composition by
producing more cis-vaccenoyl-ACP for incorporation into phospholipid [26].
   Though all the available data were consistent with this model, one dilemma re-
mained. It could be argued either that the lack of synthase I1 resulted in the loss of
thermal regulation or, conversely, that the lack of cis-vaccenoyl-ACP, the product of
elongation of palmitoleoyl-ACP by synthase 11, caused the lack of the temperature
regulation (thermal regulation would, in this case, be exerted at a later step). Al-
though indirect evidence indicated that the first argument was correct, direct proof
was needed. This was provided by an experiment in which normal amounts of cis-
vaccenic acid were synthesized in the absence of 3-ketoacyl-ACP synthase I1 using a
cloned synthase I gene [26]. Although synthase I elongates palmitoleoyl-ACP more
poorly than does synthase 11, the synthase I reaction proceeds at a measurable rate

            16:O palmitoyl-ACP

       16:1A9 palmitoleoyl-ACP

                                            +   sn-glycerol-3-@   -    Phosphatidic Acid

    18:lAll cis-vaccenoyl-ACP    -I                                             t
                                                                       Membrane Bilayer

            genotype             relative levels        growth        ratio
                                 of synthases         temperature   18:1116:1
                                   I        II            Ca

            Wild-type              1        1             30          0.7
                                                          37          0.4
                                                          42          0.3
            fab6 clone            10       1              30          2.0
                                                          37          1.2
                                                          42          0.8
            fabF                   1       0              30          0.03
                                                          37          0.03
                                                          42          0.02
            fabFlfab6 clone       10       0              30          0.7
                                                          37          0.6
                                                          42          0.5

Fig. 18. Thermal regulation of fatty acid biosynthesis. 3-Ketoacyl-ACP synthase I1 is primarily responsible
for the temperature control of E. coti fatty acid composition by being more active in conversion of palmi-
toleate to cis-vaccenate at lower temperatures than at higher temperatures.

with this substrate. Thus, the investigators reasoned that if synthase I could be over-
produced by molecular cloning into a multicopy plasmid, appreciable cis-vaccenic
acid synthesis should occur. The results of this experiment (Fig. 18) show that
the presence of a plasmid carrying the fabB gene results in the synthesis of a normal
level of cis-vaccenic acid in the absence of synthase 11. More importantly, the fatty
acid composition of these strains is not altered by growth temperature. This exper-
iment therefore demonstrates that 3-ketoacyl-ACP synthase 11 is the only protein
responsible for thermal regulation of membrane lipid composition in E. coli (Fig.

11.4. Factors affecting fatty acid chain-length distribution

Palmitate, palmitoleate, and cis-vaccenate comprise the bulk of the fatty acids found
in E, coli membranes. One likely candidate for the site of chain-length regulation is
at the level of the reactions of 3-ketoacyl-ACP synthases I and 11. Substrate specific-
ity studies on these enzymes in vitro support the view that the reason membrane
phospholipids are devoid of chain lengths longer than 18 carbons is due in part to the
reduced activity of these enzymes on 18-carbon substrates. The fatty acid composi-
tion of mutants lacking one or the other of the condensing enzymes has also given
some clues to their substrate specificity in vivo (see preceding discussion). The fabB
mutants lack all unsaturated fatty acids; therefore, it is not possible to assign an es-
sential role for synthase I in controlling chain length. Synthase I1 mutants (fabF) are
defective in the elongation of palmitoleate to cis-vaccenate; therefore, synthase I1
plays a critical role in determining the amount of 18-carbon fatty acids in the mem-
brane. Although these data suggest that the condensing enzymes play a significant
role in determining chain length, additional physiological experiments indicate that
the activity of the glycerolphosphate acyltransferase is also important. When phos-
pholipid biosynthesis is arrested at the acyltransferase step (by glycerol starvation of
a plsB mutant), the fatty acids that accumulate in the absence of their utilization by
the acyltransferase have abnormally long chain lengths (e.g., 20 and 22 carbons)
compared with the normal distribution of fatty acids synthesized in the presence of
acyltransferase activity. These data suggest that competition between the rate of
elongation and utilization of the acyl-ACPs by the acyltransferase may be a signifi-
cant determinant of fatty acid chain length in E. coli.

11.5. Synthesis of cyclopropanefatty acids

Cyclopropane fatty acids are a third major type of fatty acid found in E. coli and
many other bacteria. These acids are formed by a postsynthetic modification of the
unsaturated fatty acids of membrane phospholipids. In vitro, neither free fatty acids
nor their thioesters are substrates for cyclopropane fatty acid synthase, only phos-
pholipid dispersed in a micelle will suffice. The reaction involves the addition of a

CH, group from S-adenosylmethionine to the double bond to form a cyclopropane
ring. The cis double bond is converted to a cis cyclopropane group.
   This novel reaction has several interesting aspects. For example, cyclopropane
fatty acid synthesis occurs primarily as bacterial cultures enter the stationary growth
phase (that is, as the cultures cease growth due to oxygen or nutrient limitation);
rapidly growing cells accumulate few cyclopropane fatty acids. However, the levels of
cyclopropane fatty acid synthase vary little with growth phase; thus, the timing of
cyclopropane fatty acid synthesis is not due to a change in enzyme production. There
is no good rationale for the timing of synthesis. Recent work shows that cells carrying
a cloned segment of DNA containing the cyclopropane fatty acids synthase gene syn-
thesize the acids throughout log phase. Despite this modification, the cells grow
normally; therefore, the presence of these acids does not antagonize normal growth.
The continuous synthesis of cyclopropane acids in strains carrying a cloned ~$7 gene
indicates that whatever the regulatory process that inhibits cyclopropane fatty acid
synthesis in rapidly growing cells, it can be overcome by increased production of the
enzyme. This result suggests a stoichiometric inhibition, such as a protein-protein
interaction, but no further data are available.
   Another intriguing question is the physiological rationale for the modification,
since the physical and chemical properties of the cyclopropane fatty acids are rather
similar to those of the unsaturated fatty acids from which they are derived. Many
functions have been proposed for cyclopropane fatty acids. However, mutants of E.
coli (called c f . ) have been isolated which completely lack cyclopropane fatty acids
(fewer than 100 molecules per cell) due to disruption of the gene encoding the en-
zyme. These mutants grow and survive various environmental stresses (such as sta-
tionary phase and high and low oxygen tension) as well as strains which accumulate
cyclopropane acids in a normal manner. We are left with an enigma: cyclopropane
fatty acids are widely conserved among bacteria but seem to play no essential role in
cellular metabolism. We can conclude only that these acids play a role in the natural
environment which has not yet been duplicated in the laboratory.

11.6. Importance o the fatty ucid composition of the membrune phospholipids

An important lesson learned from studies of the various mutants of E. coli blocked
in fatty acid synthesis is that the organism tolerates a wide variation in the fatty acid
composition of the membrane phospholipids. Although saturated fatty acids alone
will not support growth, a wide variety of cis unsaturated fatty acids (mono-, di-, or
tri-unsaturated) will support growth of the unsaturated fatty acid auxotrophs, fubA
andfubB. It is clear that the double bond per se plays no chemical role in metabolism.
Indeed, even unsaturated fatty acids with a centrally located trans double bond (a
type of fatty acid not found in E. coli and very rarely found in nature) will suffice.
However, in order to utilize trans fatty acids to support growth, an additional muta-
tion, calledfutA is required. The function of the protein encoded by this gene that

allows E. coli to tolerate trans unsaturated fatty acids in place of cis is a mystery. The
role of the double bond is apparently to decrease the temperature of the phase transi-
tion of the phospholipid in which it resides, since a number of fatty acids lacking
double bonds also support growth. These acids (cis- or truns-cyclopropane, branched,
centrally brominated) do, however, share with double bonds the ability to disrupt the
close packing of phospholipid acyl chains and lower the phase transition (Chapter 1).
This property is purely physical. The presence of a substituent or a double bond in
the middle of the hydrocarbon chain sterically disrupts strong hydrophobic interac-
tions with other acyl chains.
   However, there are limits to the fatty acid compositions which allow growth. The
finding that fubA and fubB mutants require an unsaturated or equivalent fatty acid
for growth indicates that a membrane composed of phospholipids containing only
saturated fatty acids is nonfunctional. Indeed, these mutants undergo cell lysis when
deprived of the unsaturated fatty acid supplement. Thus, E. coli requires some fluid
lipid for a functional membrane. Several laboratories have reported that if less than
half of the membrane lipid was in the fluid state, the E. coli membrane became non-
functional. A similar argument can be made for the importance of some nonfluid
lipid for a functional membrane. Mutants (fabD, fubE, or fubB plusfubF) which
block fatty acid synthesis at an early step require both a saturated and an unsaturated
fatty acid for growth. If only an unsaturated fatty acid were added, the cells would
leak internal components and eventually lyse and die.
   The conclusions from these experiments are straightforward. A functional E. coli
membrane requires that the composition of the membrane phospholipids be within
the limits of the phase transition. If all the phospholipids were in either the ordered
state or the disordered state, the membrane would be nonfunctional. However, quite
wide variations in fluidity are tolerated; that is, the cells do not have to maintain a
precise ratio of fluid to nonfluid lipid to have functional membranes. However, there
does seem to be an optimal fluidity at which cell growth is most rapid, and the ther-
mal control regulatory systems seem designated for optimizing the fluidity within the
tolerated range, rather than for extending the range.

12. Regulation of phospholipid biosynthesis in E. coli

12.1. Control of fatty acid positional distribution

The sn-glycerol-3-phosphate acyltransferase step in the pathway for phospholipid
synthesis (Fig. 9) represents the transition point from soluble to membrane-bound
enzymes and intermediates and has received considerable experimental attention.
The acyltransferase is an integral cytoplasmic membrane protein, which makes it in-
trinsically more difficult to work with than the soluble enzymes of fatty acid biosyn-
thesis. Primarily through the use of gene-cloning procedures, the glycerolphosphate

acyltransferase has become one of the best characterized membrane-bound enzymes
of phospholipid biosynthesis. The finding that the acyltransferase could be solubil-
ized from the membrane with the detergent Triton X-100, led to the most important
breakthrough in dealing with this enzyme: the isolation of hybrid plasmids harboring
the glycerolphosphate acyltransferase (pfsB)structural gene. Strains containing this
plasmid overproduce the acyltransferase 1O-fold. Solubilization of the membrane
and three column-chromatographic steps produce a single protein species having a
molecular weight of 83 kDa [28]. Each step of the purification is carried out in deter-
gent-containing buffers, but as is common for membrane enzymes, reconstitution of
the protein with phospholipids is required for enzymatic activity. The complete pri-
mary sequence (806 residues) of the acyltransferase has been determined by a combi-
nation of protein- and DNA-sequencing techniques [29]. The single polypeptide cata-
lyzes the formation of 1-acylglycerolphosphate from either acyl-CoA or acyl-ACP
acyl donors. This lack of substrate specificity is not shared by Rhodopseudomonas
sphaeroides and Clostridium butyricum glycerolphosphate acyltransferases, which
utilize acyl-ACP thioesters exclusively. These data suggest that the utilization of both
acyl-CoA and acyl-ACP by the E. cofi acyltransferase may be a special adaptation of
this organism rather than a general rule for bacterial acyltransferases.
   The 1-acyl-sn-glycerol-3-phosphate    acyltransferase is poorly characterized com-
pared with the glycerolphosphate acyltransferasejust described. A biochemical assay
for 1-acylglycerolphosphateacyltransferase activity in isolated membranes is availa-
ble. This enzyme utilizes acyl-CoA and acyl-ACP as the acyl donor in vitro. Mutants
(pfsc) that are temperature-sensitive in 1-acylglycerolphosphate acyltransferase ac-
cumulate 1-acylglycerolphosphateat the nonpermissive temperature.
   Naturally occurring phospholipids are generally characterized as having a saturat-
ed fatty acid at the l-position and an unsaturated fatty acid at the 2-position of the
glycerol backbone. The substrate specificity of the glycerolphosphate acyltransferase
system is considered the most likely origin of acyl-group asymmetry in bacterial
phospholipids. Accordingly, higher V,,, and lower K, values are found for saturated
rather than unsaturated acyl donors when either the purified or membrane-bound
forms of sn-glycerol-3-phosphate acyltransferase are used as the enzyme source.
These data demonstrate that the acyltransferase has a substrate specificity that is
consistent with the role of this enzyme in controlling the positional placement of fatty
acids on the glycerol backbone. Acyltransferase specificity can also be investigated in
vivo using unsaturated fatty acid auxotrophs. When E. cofi mutants that are unable
to synthesize unsaturated fatty acids (eitherfabA orfubB) are deprived of their exo-
genous unsaturated fatty acid supplement, a significant accumulation of disaturated
molecular species of phospholipid is observed. Restoration of unsaturated fatty acids
to the medium results in the synthesis of molecular species having the typical fatty
acid positional asymmetry. These data demonstrate that the acylation specificity of
the glycerophosphate acyltransferase is not absolute in vivo, but is controlled in part
by the supply of fatty acids. This is another good example of how the genetic ap-

proach was used to dissect the interrelationships between enzymes in a metabolic
pathway that were not directly evident from in vitro studies.

12.2. Regulation of total phospholipid synthesis

The biosynthesis of phospholipid is an energy-intensive process, and there is abun-
dant evidence that phospholipid production is tightly regulated in vivo. There are
two possible sites where regulation of the pathway could occur. Phospholipid pro-
duction could be controlled either at the level of fatty acid supply or at the level of
fatty acid incorporation into phospholipid. This question has been addressed in vivo
by measuring the pool size of long-chain acyl-ACP substrates for the acyltransferase
during balanced growth and after the cessation of phospholipid synthesis. To accom-
plish this goal, the genetic approach was used to construct strains of E. coli contain-
ing both the pis3 acyltransferase mutation and the panD defect in the CoA biosyn-
thetic pathway. The plsB mutation results in a reduced affinity of the glycerolphos-
phate acyltransferase for glycerol-3-phosphate and therefore allows phospholipid
biosynthesis to be turned on and off by the presence or absence of the glycerolphos-
phate supplement in the culture medium. The panD allele renders the cell auxotrophic
for p-alanine, a precursor to the 4'-phosphopantetheine prosthetic group of ACP,
thus allowing the ACP pool to be uniformly and specifically labeled. Cell samples
were labeled with /3-[3H]alanine,the ACP and acyl-ACP were extracted, and the con-
centration of acyl-ACP substrates for the acyltransferase determined by reversed-
phase high-pressure liquid chromatography. If the utilization of acyl-ACP was rate
limiting in phospholipid production, a large pool of acyl-ACP molecules awaiting
acyltransfer would be anticipated. On the other hand, if the supply of acyl-ACP was
rate limiting, the concentration of long-chain acyl-ACP species would be low. In
these experiments, the acyl-ACP concentration was approximately 10% of the total
ACP pool, and the majority of these acyl-ACPs were not of chain lengths that are
substrates for the acyltransferase. Hence, it would appear that the supply of acyl-
ACP limits the rate of phospholipid synthesis in E. coli. When phospholipid bio-
synthesis was inhibited by the removal of glycerolphosphate, the acyl-ACP pool rose
to about 34% of the total ACP, thereby establishing that acyl-ACPs are the acyl
donors for phospholipid synthesis. As a check on this point, the long-chain acyl-CoA
pool was also measured before and after inhibition of phospholipid synthesis. Long-
chain acyl- CoAs were not detected under either of these conditions. These data
strongly point to a step in the initiation phase of fatty acid biosynthesis (Figs. 5 and
6) as the primary rate-controlling step in the pathway of phospholipid synthesis.
More work is needed to establish which enzyme is the pacemaker of phospholipid
synthesis. Control of the pathway at the level of initiation appears to be a logical
hypothesis, since 94% of the ATP used in the biosynthesis of a phospholipid is ex-
pended in the synthesis of the fatty acid components.
   An essential player in the initiation phase is acetyl-CoA carboxylase which catalyz-

es the synthesis of malonyl-CoA from acetyl-CoA, ATP, and bicarbonate (Figs. 5
and 6). The key component of this enzyme is a biotin moiety covalently bound to
biotin carboxyl carrier protein (BCCP) and two enzyme activities, biotin carboxylase
which catalyzes the ATP dependent formation of carboxybiotin on BCCP and car-
boxyltransferase which carboxylates acetyl-CoA (using carboxyl BCCP) to form ma-
lonyl-CoA. The genes encoding BCCP, biotin carboxylase, and one of the two car-
boxyl transferase subunits have been isolated and sequenced. The BCCP and biotin
carboxylase genes are cotranscribed and this transcription is regulated by an un-
known mechanism. The carboxyltransferase subunits are found on the genetic map
at a position distinct from BCCP and from one another. It has also been shown that
the rate of synthesis of the biotin cofactor is regulated by the rate of apo-BCCP syn-

12.3. Coordinution of phospholipid synthesis with cellular metabolism

The exponential phase of bacterial growth is typically referred to as balanced since
the myriad of biochemical pathways are well coordinated. These regulatory mecha-
nisms have proven difficult to isolate experimentally during balanced growth, but
some progress has been made using amino acid starvation to perturb the metabolic
system. Wild-type cells respond to the decreased availability of any aminoacyl-tRNA
species by a dramatic reduction in stable RNA accumulation in addition to several
other metabolic adjustments that down-regulate cellular metabolism. A single site
mutation, relA, abolishes this entire set of adjustments, conferring a phenotype called
relaxed. These regulatory effects are mediated by a family of nucleotides, most nota-
bly guanosine-5'-diphosphate-3'-diphosphate      (ppGpp) that accumulate in rel+ but
not relA strains. Several laboratories have reported that phospholipid production
decreases dramatically following the starvation of rel+but not relA strains [30].Care-
ful measurement of both the ppGpp concentration and the rate of phospholipid bio-
synthesis shows a quantitative correlation between ppGpp concentration and phos-
pholipid accumulation. Although the in vivo evidence strongly supports the role of
ppGpp in the regulation of total phospholipid production, the identity of the enzyme
or enzymes affected by ppGpp remains to be established.

12.4. Regulation of phospholipid head group composition

The biochemical mechanisms that control the ratio of phosphatidylethanolamine,
phosphatidylglycerol, and cardiolipin in the membranes of E. coli are less well under-
stood than the factors that control the phospholipid fatty acid composition. Mutants
in the enzymes of phospholipid biosynthesis can be used to render any of the steps in
Fig. 10 rate limiting; however, these data do not shed much light on the mechanisms
that operate in vivo. Similarly, gene-cloning experiments provide little insight. For
example, a 10-fold overproduction of phosphatidylserine synthase in strains harbor-

ing a hybrid plasmid containing the pss gene does not significantly affect the ratio of
polar head groups in the membrane. It therefore seems that either the phospholipid
head group composition is not regulated by the levels of the synthetic enzymes or the
enzyme levels measured in vitro are not an accurate indication of the activity present
in vivo [lo]. The isolation of mutants that cause overproduction of diacylglycerol
kinase (dgkR) and phosphatidylserine synthase (pssR) indicate that regulatory mech-
anisms do exist [2]. However, both the method whereby the synthesis of these en-
zymes is controlled and the reasons for this control (since overproduction does not
affect phospholipid composition) are unknown. Since these enzymes are generally
integral membrane proteins, their activity could be modified by the phospholipid en-
vironment in vivo, although there is little evidence for this type of mechanism either
in vivo or in vitro. Selection schemes for the isolation of strains with abnormal phos-
pholipid compositions and second-site revertants of strains possessing lesions in
phospholipid enzymes may provide additional insight into this important problem in
the future.

12.5. E. coli tolerates variations in phospholipid composition

Studies of mutants blocked early in the phospholipid biosynthetic pathway (for ex-
ample, plsB) show that cell growth and macromolecule synthesis continue for about
a cell generation following severe inhibition of phospholipid synthesis. Conse-
quently, the cell membranes become unusually dense due to continued membrane
protein synthesis in the absence of lipid synthesis. Thus, other metabolic processes
are not tightly coupled to membrane phospholipid synthesis.
   Analysis of mutants blocked later in the phospholipid pathway shows that the
membrane lipid composition of E. coli (although closely maintained in normally
growing cells) can be significantly altered without abrupt effect on cell growth and
membrane function (Table 111). For example, cls strains have less than 10% of the
normal level of cardiolipin but grow normally. However, it should be noted that car-
diolipin is a major membrane component only in cells grown into stationary phase.
Mutants blocked in the other late steps of phospholipid synthesis (pss, psd, or pgsA)
do show severe disturbances in growth. However, these problems do not arise until
the synthesis of a given lipid has been inhibited for one to two generations. At this
time, cell division is often inhibited, and disruption of the barrier properties of the
membrane also becomes evident. These experiments show that a precisely aligned
phospholipid composition is not required for growth and membrane function. Func-
tional membranes can also contain large amounts of phospholipids which are only
trace components of wild-type strains in E. coli, such as phosphatidylserine, a major
component of psd strains and phosphatidic acid, a major component of cds strains
(Table 111). Indeed, by fusion of artificial liposomes with Salmonella typhimurium
cells, such abnormal lipids as cholesterol and phosphatidylcholine have been intro-
duced into bacterial membranes without affecting bacterial physiology.

Table I11
E. Cali mutants for the modification of membrane lipid composition

Gene               Growth                            Membrane lipid modification
defect             phenotype

aas                None                              Accumulation of lysophosphatidylethanolamine
                                                     and acylphosphatidylglycerol;incorporation of
                                                     phosphatidylcholine to 5%
cdsA               pH-sensitive                      Reduced phosphatidylethanolamine; phosphatidic
                                                     acid accumulation
cfa                None                              Accumulation of cyclopropane fatty acids
CIS                None                              Lack of cardiolipin
dgkA               Osmotic sensitivity               Accumulation of diacylglycerol and other neutral
fabA or fabB       Unsaturated fatty acid            Accumulation of disaturated phospholipid molec-
                   auxotroph                         ular species
                   None                              Lack of 18:l fatty acids
                   None                              Accumulation of lipid X
                   Ts (42°C)'                        Lack of phosphatidylglycerol
                   Glycerol auxotroph                Lipid:protein ratio reduced 40%
                   Ts (42°C)                         Accumulation of acylglycerolphosphate
                   Ts (42°C)                         Accumulation of phosphatidylserine; lack of phos-
                   Ts (42°C)                         Reduced phosphatidylethanolamine; increased
                                                     phosphatid ylglycerol plus cardiolipin
                   Cytidine auxotroph                Accumulation of phosphatidic acid; 1ipid:protein
                                                     ratio increases threefold

'Ts, temperature-sensitive

   It is not clear why cells synthesize a diversity of phospholipid species when a single
species would suffice to form a lipid bilayer (Chapter 1). Some evidence suggests that
the differing physical properties of bilayers composed of various lipid mixtures could
be important, but definitive evidence is lacking. Recent reports suggest two alterna-
tive possibilities [lo]. First, the reconstitution of dnaA gene product function in the
initiation of DNA synthesis in vitro requires the presence of either phosphatidyl-
glycerol or cardiolipin. Second, the accumulation of unprocessed precursors of outer
membrane proteins in the cytoplasm of mutants deficient in phosphatidylglycerol
suggests that translocation of proteins across the plasma membrane requires the
presence of acidic phospholipids.

12.6. Outer membrane lipoprotein

Lipids are also used in bacteria in the synthesis of outer membrane lipoproteins. The
most well known of these is Braun's lipoprotein. This lipoprotein is one of the most

abundant proteins in the cell, and the mature form possesses an amino terminal cys-
teine that is modified with both a diacylglycerol and an amide-linked fatty acid (Fig.
19). The lipid modification of this protein is presumably important to its proper at-
tachment to the membrane. An important structural point is that the diacylglycerol
is linked via a thioether to the C-1 carbon of sn-glycerol and not to the C-3 position.
This finding indicates that the precursor is not a diacylglycerol moiety of a phospho-
lipid (linkage to C-3 would be expected in this case). There is strong evidence that the
source of the sn-1-glycerol is the nonacylated polar head group of phosphatidylglyce-
rol. The ester-linked fatty acids are subsequently added to the protein, followed by
the amide-linked fatty acid. The exact source of all these fatty acids has not been
determined, but the data strongly argue that they are derived from the phospholipid
pool and not from the fatty acid biosynthetic pathway. Likewise, the nature of the
transacylase enzymes responsible for the placement of these fatty acids on the lipo-
protein remains to be elucidated. Therefore, phospholipids are used as intermediates
in the biosynthesis of membrane proteins in addition to fulfilling their structural role.

13. Lipid metabolism in bacteria other than E. coli

It must be emphasized that the lipid metabolism of E. coli differs greatly from that of
some other bacteria. Although many bacteria follow the E. coli paradigm, others
ignore it. What is most striking about procaryotic lipids is their incredible diversity.

13.1. Bacteria lacking unsuturuted fatty ucids

Many bacteria, such as the very successful Bacillus genus, possess only very low levels
of unsaturated fatty acids under most growth conditions. Instead of unsaturated fatty
acids, the major fatty acids imparting membrane fluidity are terminally branched
chain fatty acids, which have physical properties similar to those of unsaturated fatty
acids. Indeed, it has been shown that E, coii can use such acids as unsaturated fatty

                   c- s

               a      CH2
                      I II
                    - CH - C - NH - Ser - Ser-

Fig. 19. Location of acyl chains on the amino terminus of Braun’s lipoprotein of E. coli.

acid substitutes. The terminally branched chain acids are made by substituting isobu-
tyryl-CoA or 2-methylvaleryl-CoA for acetyl-CoA as the primer of fatty acid bio-
synthesis. The branched chain acyl-CoA primer is transacylated to ACP and is used
as the acceptor of malonyl groups in processive cycles of fatty acid synthesis.

13.2. Bacteria containing phosphatidylcholine

Most bacteria lack phosphatidylcholine. However, a few bacteria possess this lipid
(for example, Rhodopseudomonas sphaeroides), and these tend to be rather highly
specialized or highly evolved bacteria such as photosynthetic or nitrogen-fixing bac-
teria. Bacterial phosphatidylcholine is synthesized by three successive methylations
of phosphatidylethanolamine (Chapter 7). These organisms seem to lack the ability
to incorporate choline directly into phospholipid.

13.3. Bacteria synthesizing unsaturated fatty acids by an aerobic pathway

The pathway used by plants and animals to synthesize monounsaturated fatty acids
involves formation of a double bond in an oxygen-requiring step (Chapter 5 ) . Al-
though most bacteria such as E, coli use the anaerobic pathway, some obligately aer-
obic bacteria (for instance, Bacilli) synthesize unsaturated fatty acids by an oxygen-
requiring 'reaction which resembles that of higher cells. In the Bacilli, significant
amounts of unsaturated fatty acid are synthesized only at low growth temperatures,
a situation reminiscent of the thermal control of E. coli. However, it should be noted
that new protein synthesis seems to be required from commencement of unsaturated
fatty acid synthesis upon shift to low temperature; thus, synthesis of some new
protein(s) at the lower temperature is probably required. However, it has not been
possible to study the desaturation reaction in vitro, so a detailed analysis of this sys-
tem is not yet available. The oxygen-dependent is also used by various Bacilli to syn-
thesize diunsaturated fatty acids. Polyunsaturated fatty acids are abundant compo-
nents of all eucaryotic membranes, but few procaryotes synthesize these acids. How-
ever, the diunsaturated fatty acids of Bacilli have very different double-bond posi-
tions from those commonly found in eucaryotes (for example, 5,lO-hexadecadienoic
acid in Bacillus licheniformis).

13.4. Bacteria with a multifunctional fatty acid synthase

The fatty acid synthase complex of Mycobacterium smegmutis represents an excep-
tion to the general rule that the enzymes of bacterial fatty acid synthesis do not form
multifunctional complexes [l]. It was discovered that this bacterium has a fatty acid
synthase complex composed of six identical subunits, each having a molecular weight
of 290 kDa. Each of these subunits is a multifunctional polypeptide that contains all
six of the reaction centers required for saturated fatty acid synthesis, similar to the

liver enzyme described in Chapter 4. The fatty acid synthase of M. smegmatis also
differs from the E. coli system in that the products of the synthase are acyl-CoAs
having chain lengths ranging from 16 to 24 carbons. Another unusual feature is that
the fatty acid synthase system is markedly stimulated by methylated polysaccharides
that are polymeric forms of either 6-0-methylglucose or 3-0-methylmannose. These
polysaccharide structures have hydrophobic domains that bind long-chain acyl-
CoAs and stimulate fatty acid production by relieving the synthase system from feed-
back inhibition by acyl-CoA. Interestingly, it appears that the diffusion of the acyl-
CoA from the enzyme surface is the rate-limiting step for the synthesis of fatty acids
in M . smegmatis. Corynebacterium diphtheriae has a fatty acid synthase similar to
that in M. smegmatis, although the aggregate molecular weight is somewhat larger
(2.5 x       Da). Brevibacterium ammoniagenes has an unusual multienzyme complex
(molecular weight of 1.23 x       Da) that synthesizesboth saturated and unsaturated
fatty acids. The unsaturated acids are produced in the absence of oxygen and, there-
fore, appear to be synthesized by a modification of the anaerobic pathway used by E.
coli. B. ammoniagenes are members of a highly developed group of bacteria that is
thought to be the progenitor of fungi; thus, the finding that the organization of their
fatty acid synthase resembles that of the fungi is not surprising.

13.5. Bacteria with intracytoplasmic membranes

Some specialized bacteria elaborate intracytoplasmic membrane systems that harbor
specific metabolic processes in response to changes in the environment. Rhodopseu-
domonas sphaeroides is an example of the type of system used to study the production
and differentiation of intracytoplasmic membranes. When this organism is grown
phototropically, the cytoplasmic membrane invaginates and differentiates into an in-
tracytoplasmic membrane that contains the reaction centers required for photo-
synthetic growth. The quantity of intracytoplasmic membrane produced is inversely
related to the intensity of the incident light. The phospholipid components of the
intracytoplasmic membrane are acquired discontinuously during the cell cycle,
resulting in cyclic alterations in the composition of the intracytoplasmic membranes
in synchronously dividing cell populations. The enzymes responsible for the biosyn-
thesis of intracytoplasmic membrane phospholipids are located in the cytoplasmic
membrane compartment and the phospholipids are translocated to the intracytoplas-
mic membrane.

13.6. Other bacterial oddities

One bacterium is known (Bacteroides)which synthesizes sphingolipids.Although not
yet studied in detail, the biosynthetic pathway seems very similar to that used in
mammals (Chapter 11). A number of bacterial species (for example, Clostridium)
synthesize 1-alk-1'-enyl lipids (plasmalogens) and in some cases further modify the

ether group. The synthetic pathways of these lipids are unknown but bacterial plas-
malogens are clearly synthesized by a pathway different from that in mammals
(Chapter 8). The bacterial plasmalogens are made by strictly anaerobic bacteria.
Since the mammalian pathway requires oxygen (Chapter 8), a markedly different
pathway must be used by these bacteria. Several bacteria synthesize methyl-branched
fatty acids with the methyl group located in the center (rather than at the end) of the
acyl chain. These acids are synthesized by a reaction that resembles cyclopropane
fatty acid synthesis in that the donor of the -CH, group is S-adenosylmethionine,and
the substrate is the fatty acid residue of an intact phospholipid molecule.

13.7. Lipids o nonbacterial (but related) organisms

It has recently been realized that there is a group of what were considered to be bac-
teria but which actually form a group of organisms distinct from the common bacte-
ria (eubacteria) and eucaryotes. These organisms, called the Archaebacteria, have a
very unusual lipid composition in that the building block of their lipids is the six-
carbon unit, mevalonic acid, rather than the two-carbon unit, acetic acid. The result
is that phytanyl chains are bound to the glycerol moieties of the complex lipids by
ether linkages; thus, these lipids differ in several basic features from those of the com-
mon bacterial and eucaryotic cells.

14. Future directions
Although our knowledge of the individual enzymatic steps in the phospholipid bio-
synthetic pathway and their regulation has increased dramatically over the last 20
years, there are still a number of fundamental questions that are unanswered. The
identity of the enzyme or enzyme system that functions as the pacemaker of phospho-
lipid biosynthesis is not known. Similarly, the mechanisms that operate to produce
the observed distribution of phospholipid polar head groups has not been uncovered,
and the details of how phospholipid and macromolecular biosynthesis are coordinat-
ed during normal growth and nutritional stress remain to be established. The regula-
tory point that controls the divergence of 20% of the acyl-ACP to lipid A biosynthesis
rather than to phospholipid production is a mystery. Further experiments to measure
the intracellular level of intermediates in the phospholipid biosynthetic pathway in
vivo along with an in vitro biochemical analysis of isolated regulatory enzymes will
be needed to complete our understanding of the rate-determining steps in the biosyn-
thesis of bacterial phospholipids. The genetic approach will continue to be an impor-
tant tool, and recent advances in genetic-cloning procedures have begun to have a
profound effect on our understanding of both the biochemistry and physiology of E.
coli. The selection of regulatory mutants holds the key to testing our hypotheses
about the regulation of the pathway in vivo. Designing selection schemes to isolate
regulatory mutants will not be straightforward and represents one of the most chal-
lenging aspects of our future work.

 I. Bloch, K. and Vance, D.E. (1977) Control mechanisms in the synthesis of saturated fatty acids. Annu.
    Rev. Biochem. 46,263-298.
 2. Raetz, C.H.R. (1978) Enzymology, genetics and regulation of membrane phospholipid synthesis in
    Escherichiu coli. Microbiol. Rev. 42, 6111659.
 3. Raetz, C.H.R. ( I 982) Genetic control of phospholipid bilayer assembly. In: Comprehensive Biochem-
    istry (Hawthorne, J.N. and Ansell, B., Eds.) Vol. 4. pp. 435477, Elsevier, Amsterdam.
 4. Bachmann, B.J. (1990) Linkage map of Escherichia coli K-12, edn. 8. Microbiol. Rev. 54. 130-197.
 5. Kohara, Y., Akiyama, K. and Isono, K. (1987) The physical map of the whole E. coli chromosome.
    Application of a new strategy for rapid analysis and sorting of a large genomic library. Cell 50, 495-
 6. Clark, D.P. and Cronan, J.E., Jr. (1981) Bacterial mutants for the study of lipid metabolism. Methods
    Enzymol. 72,693-707.
 7. Osborn, M.J. and Munson. R. (1974) Separation of the inner (cytoplasmic) and outer membrane of
    gram-negative bacteria. Methods Enzymol. 3 I,642 -652.
 8. Bloch, K. (1971)P-Hydroxydecanoyl thioester dehydrase. In: The Enzymes (Boyer, P., Ed.) pp, 441-
    464, Academic Press, New York, NY.
 9. Goldfine, H. (1982) Lipids of prokaryotes - structure and distribution. In: Current Topics in Mem-
    branes and Transport (Razin, S. and Rottem, S.,    Eds.), Vol. 17, pp. 1 4 3 , Academic Press, New York,
10. Raetz, C.R.H. and Dowhan, W.      (1990) Biosynthesis and function of phospholipids in Escherichiu coli.
    J. Biol. Chem. 265, 123551238,
11. Prescott, D.J. and Vagelos, P.R. (1972) Acyl carrier protein. Adv. Enzymol. 36, 269-31 1.
12. Vanden Boom, T. and Cronan, J.E., Jr. (1989) Genetics and regulation of bacterial lipid metabolism.
    Annu. Rev. Microbiol. 43. 31 7-343.
13. Kim, Y. and Prestegard, J.H. (1991) Refinement of the NMR structures for acyl carrier protein with
    scalar coupling data. Proteins 6, 377-385.
14. Powell, G.L., Elovson, J. and Vagelos, P.R. (1969) Acyl carrier protein. XU. Synthesis and turnover
    of the prosthetic group of acyl carrier protein in vivo. J. Biol. Chem. 244, 5616-5624.
15. Jackowski, S. and Rock, C.O. (1983) Ratio of active to inactive forms of acyl carrier protein in Esche-
    richia coli. J. Biol. Chem. 259, 15186-15191.
16. Jackowski. S.and Rock. C.O. (1984) Turnover of the 4'-phosphopantetheine prosthetic group of acyl
    carrier protein. J. Biol. Chem. 260, 1891-1895.
17. Raetz, C.R.H. (1990) Biochemistry of endotoxins. Annu. Rev. Biochem. 59, 129-170.
18. Nishijima, M. and Raetz, C.R.H. (1979) Membrane lipid biogenesis in Escherichiu coli. Identification
    of genetic loci for phosphatidylglycerophosphate synthetase and construction of mutants lacking
    phosphatidylglycerol. J. Biol. Chem. 254, 7837--7844.
19. Anderson, M.S. and Raetz, C.R.H. (1987) Biosynthesis of lipid A precursors in Eschrrichia co/i. A
    cytoplasmic acyltransferase that converts UDP-N-acetylglucosamine to UDP-3-O-(R-3-hydroxymyr-
    istoy1)-N-acetylglucosamine. J. Biol. Chem. 262, 5159-5169.
20. Van Golde, L.M.G.. Schulman, H. and Kennedy, E.P. (1973) Metabolism of membrane phospholipids
    and its relation to a novel class of oligosaccharides in Escherichiu coli. Proc. Natl. Acad. Sci. U.S.A.
    70, 1369- 1372.
21. Jackowski, S.and Rock, C.O. (1986) Transfer of fatty acids from the I-position of phosphatidyletha-

      nolamine to the major outer membrane lipoprotein of Escherichiu roli. J. Biol. Chem. 261. 11328-
22.   Helmkarnp, G.M., Rando, R.R., Brock, D.J.H. and Bloch, K. (1968) P-Hydroxydecanoyl thioester
      dehydrase. Specificity of substrates and acetylenic inhibitors. J. Biol. Chem. 243, 3229-323 1.
23.   Omura, S. (1976) The antibiotic cerulenin, a novel tool for biochemistry as an inhibitor of fatty acid
      synthesis. Bacteriol. Rev. 40, 681-697.
24.   Nishida, I., Kawaguchi, A. and Yamada, M. (1986) Effect of thiolactomycin on the individual en-
      zymes of the fatty acid synthase system in Escherichiu coli. J. Biochem. (Toyko) 99, 1447-1454.
25.   D’Agnolo, G., Rosenfeld, I.S. and Vagelos, P.R. (1975) Multiple forms ofg-ketoacyl-acyl carrier pro-
      tein synthetase in Escherichiu coli. J. Biol. Chem. 250, 5289-5294.
26.   DeMendoza, D. and Cronan, J.E., Jr. (1983) Thermal regulation of membrane lipid fluidity in bacte-
      ria. Trends Biochem. Sci. 8,49-52.
27.   DeMendoza, D., Ulrich, A.K. and Cronan, J.E., Jr. [ 1983) Thermal regulation of membrane fluidity
      in Escherichh coli. Effects of overproduction of P-ketoacyl-acyl carrier protein synthase I.J. Biol.
      Chem. 258,2098-2101.
28.   Lightner, V.A., Larson, T.J., Tailleur, P., Kantot, G.D.. Raetz, C.H.R., Bell, R.M. and Modrich, P.
      (1980) Membrane phospholipid synthesis in Escherichiu coli. Cloning of a structural gene (plsB) of the
      sn-glycerol-3-phosphate acyltransferase. J. Biol. Chem. 255, 9413-9420.
29.   Larson, T.J., Lightner, V.A., Green, P.R., Modrich, P., and Bell, R.M. (1980) Membrane phospholip-
      id synthesis in Escharichiu roli. Identification of the sn-glycerol-3-phosphate acyltransferase polypep-
      tide as theplsB gene product. J. Biol. Chem. 255,9421-9426.
30.   Rock, C.O. and Cronan, J.E., Jr. (1982) Regulation of bacterial membrane lipid synthesis. In: Current
      Topics in Membranes and Transport, (Razin, S . and Rottem, S., Eds.) Vol. 17, pp. 207-227, Academic
      Press, New York, NY.
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemistry of Lipids, Lipoproteins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                    87

                                                                                         CHAPTER 3

                                                           Oxidation of fatty acids
                                                                                    HORST SCHULZ

                  City College of CUNX Department of Chemistry, New York, N Y 10031. U.S.A.

1. The pathway o p-oxidation: a historical account

Fatty acids are a major source of energy in animals. The study of their biological
degradation began in 1904 when Knoop [l] performed the classical experiments
which led him to formulate the theory of!-oxidation. In his experiments Knoop used
fatty acids which contained a phenyl residue in place of the terminal methyl group.
The phenyl residue served as a reporter group because it was not metabolized, but
instead was excreted in the urine. When Knoop fed phenyl-substituted fatty acids
with an odd number of carbon atoms, like phenylpropionic acid (C,H,-CH,-CH,-
COOH) or phenylvaleric acid (C6H,-CH2-CH2-CH2          -CH,-COOH), to dogs, he isola-
ted from their urine hippuric acid (C,H,-CO-NH-CH2-COOH), the conjugate of ben-
zoic acid and glycine. In contrast, phenyl-substituted fatty acids with an even number
of carbon atoms, such as phenylbutyric acid (C,H,-CH2-CH,-CH,-COOH), were de-
graded to phenylacetic acid (C,H,-CH,-COOH) and excreted as phenylaceturic acid
(C,H,-CH2-CO-NH-CH,-COOH). These observations led Knoop to propose that
the oxidation of fatty acids begins at carbon atom 3, thep-carbon, and that the result-
ing p-keto acids are cleaved between the a-carbon and p-carbon to yield fatty acids
shortened by two carbon atoms. Knoop’s experiments, and later ones performed by
Dakin, prompted the idea that fatty acids are degraded in a stepwise manner by suc-
cessive !-oxidation. In the years following Knoop’s initial study, Dakin [2] per-
formed similar experiments with phenylpropionic acid. Besides hippuric acid he was
able to isolate the glycine conjugates of the following !-oxidation intermediates:
phenylacrylic acid (C,H,-CH=CH-COOH), p-phenyl-p-hydroxypropionic acid
(C,H5-CHOH-CH2-COOH),and benzoylacetic acid (C6H5-CO-CH2-COOH). the              At
same time Embden and coworkers demonstrated that in perfused livers unsubstituted
fatty acids are degraded by /?-oxidation to ketone bodies. Thus, by 1910 the basic
information necessary for formulating the pathway of /.?-oxidation was available.
   After a 30-year period of little progress, Munoz and Leloir in 1943, and Lehninger
in 1944, demonstrated the oxidation of fatty acids in cell-free preparations from liver.
Their work set the stage for the complete elucidation ofp-oxidation. Detailed investi-

gations with cell-free systems, especially the studies of Lehninger, demonstrated the
need for energy to ‘spark’ the oxidation of fatty acids. ATP was shown to meet this
requirement and to be essential for the activation of fatty acids. Activated fatty acids
were shown by Wakil and Mahler, as well as by Kornberg and Pricer, to be thioesters
formed from fatty acids and coenzyme A. This advance was made possible by earlier
studies of Lipmann and coworkers who isolated and characterized coenzyme A, and
Lynen [3] and coworkers who proved the structure of ‘active acetate’ to be acetyl-
CoA. Acetyl-CoA was found to be identical with the two-carbon fragment removed
from fatty acids during their degradation. The subcellular location of the /?-oxidation
system was finally established by Kennedy and Lehninger, who showed that mito-
chondria were the cellular components most active in fatty acid oxidation. The mito-
chondrial location of this pathway agreed with the observed coupling of fatty acid
oxidation to the citric acid cycle and to oxidative phosphorylation. The most direct
evidence for the proposed ,!%oxidationcycle, as shown in Fig. 1, emerged from enzy-
mological studies carried out in the fifties primarily in the laboratories of Green in
Wisconsin, Lynen in Munich, and Ochoa in New York. Their studies were greatly

                                                              Cell membrone


              FABP?                                           cytosoi


                                                              Outer mttochondrial

      R-CH-CH2-COSCoA                 R - C - CHr-COSCoA
                       2 ;
                      Y /
                      NAD+        NADH t Ht

Fig. 1. Pathway of B-oxidation in mitochondria. AS, acyl-CoA synthetase; CPT carnitine palmitoyltrans-
ferase; T, carnitine:acylcarnitine translocase; AD, acyl-CoA dehydrogenase; EH, enoyl-CoA hydratase;
HAD, L-3-hydroxyacyl-CoAdehydrogenase; KT, 3-ketoacyl-CoA thiolase; FABP, fatty acid binding pro-

facilitated by newly developed methods of protein purification and by the use of spec-
trophotometric enzyme assays with chemically synthesized intermediates of P-oxida-
tion as substrates.

2. Uptake and activation of fatty acids in animal cells
Fatty acids are transported between organs either as unesterified fatty acids com-
plexed to serum albumin or in the form of triacylglycerols associated with lipopro-
teins. Triacylglycerols are hydrolyzed outside of cells by lipoprotein lipase to yield
free fatty acids (Chapter 15). The mechanism by which free fatty acids enter cells
remains poorly understood despite a number of studies performed with isolated cells
from heart, liver, and adipose tissue [4]. Kinetic evidence has been obtained for both
a saturable and nonsaturable uptake of fatty acids. The saturable uptake which pre-
dominates at low concentrations of free fatty acids is presumed to be carrier-
mediated and possibly involves a 40 kDa plasma membrane protein. The nonsatur-
able uptake, which is significant only at higher concentrations of free fatty acids, has
been attributed to nonspecific diffusion across the membrane.
   Once long-chain fatty acids have crossed the plasma membrane, they either diffuse
or are transported to mitochondria, peroxisomes and the endoplasmic reticulum
where they are activated by conversion to their CoA thioesters. Whether this transfer
of fatty acids between membranes is a facilitated process or occurs by simple diffu-
sion remains a matter of speculation. The identification of low-molecular-weight ( 1 4
15 kDa) fatty acid binding proteins (FABPs) in the cytosol of various animal tissues
has led to the suggestion that these proteins may function as carriers of fatty acids in
the cytosolic compartment [5]. FABPs may also be involved in the cellular uptake of
fatty acids, their intracellular storage, or the delivery of fatty acids to sites of their
utilization. However, little evidence in support of these proposed functions has been
presented. In contrast, a wealth of structural information about four distinct, but
homologous, FABPs has been obtained. They are liver FABP which is found in liver
and intestine, intestinal and adipocyte FABPs, which are present in intestine and adi-
pose tissue, respectively, and heart FABP, which is abundantly present in heart and
red skeletal muscle, but is also found, although at lower concentrations, in other tis-
sues including lung, kidney, and brain [5].
   The primary structures of most well-characterized FABPs have been determined
by protein and/ or cDNA sequencing. Sequence comparisons revealed high degrees
of homologies (80-95%) for the same type of FABP from different animals and mod-
erate homologies (25-35%) for different types of FABPs from one animal [5].An
exception is the relatively high degree of homology (6&70%) observed between heart
and adipocyte FABPs. The tertiary structure of the rat intestinal FABP has been
solved to 2.5 A resolution. The protein consists of 10 anti-parallel P-strands which
form a clam shell-like structure. A single fatty acid binds to the interior of the pro-
tein with its carboxylate group facing an arginine residue. FABPs, other than the in-

testinal form, possibly bind more than one mole of fatty acid per mole of protein.
Although FABPs, especially liver FABP, have been reported to bind ligands other
than fatty acids, e.g. fatty acyl-CoA, heme, bilirubin, sterols, and steroids, serious
questions about the apparently broad ligand specificities of FABPs persist. A case in
point is the reported tight binding of fatty acyl-CoA to liver FABP which may be due
to contaminating amounts of a 10 kDa acyl-CoA binding protein which binds fatty
acyl-CoAs but not fatty acids and which shows no homology with any FABP.
   The metabolism of fatty acids requires their prior activation by conversion to fatty
acyl-CoA thioesters. The activating enzymes are ATP-dependent acyl-CoA syntheta-
ses, which catalyze the formation of acyl-CoA by the following two-step mechanism [6]:

                                                 0       0
                                                 II      II
                                                                   +   PPi

                               eCH3-C-SCoA       + AMP

   The evidence for this mechanism was primarily derived from a study of acetyl-CoA
synthetase. Although the postulated intermediate acetyl-AMP does not accumulate
in solution, and therefore only exists bound to the enzyme, the indirect evidence for
this intermediate is very convincing. Synthetic acetyl-AMP is a donor of acetate in
the reaction with CoASH to form acetyl-CoA and is a donor of AMP in the pyro-
phosphate-dependent formation of ATP. All exchange reactions agree with this two-
step mechanism, which is also supported by the observed incorporation of one oxy-
gen atom of acetate into AMP. Other fatty acids are believed to be activated by a
similar mechanism, even though less evidence to support this hypothesis has been
obtained. The activation of fatty acids is catalyzed by a group of acyl-CoA syntheta-
ses which differ in their subcellular locations and in their specificities with respect to
the chain length of their fatty acid substrates. The chain-length specificities are the
basis for classifying these enzymes as short-chain acyl-CoA synthetase, medium-
chain acyl-CoA synthetase, long-chain acyl-CoA synthetase and very-long-chain
acyl-CoA synthetase.
   A short-chain specific acetyl-CoA synthetase has been isolated in highly purified
form from beef heart mitochondria. The enzyme is most active with acetate as a sub-
strate, but exhibits some activity towards propionate. Acetyl-CoA synthetase has
been found in mitochondria of heart, skeletal muscle, kidney, adipose tissue and in-
testine, but not in liver mitochondria. A cytosolic acetyl-CoA synthetase was identi-
fied in liver, intestine, adipose tissue and mammary gland, all of which have high
lipogenic activities. It is possible that the cytosolic enzyme synthesizes acetyl-CoA for

lipogenesis, whereas the mitochondrial acetyl-CoA synthetase activates acetate
headed for oxidation. Evidence for the presence of a distinct propionyl-CoA synthe-
tase in liver mitochondria has been obtained. This enzyme is active with acetate, pro-
pionate, and butyrate, but the K,,, value for propionate is much lower than the K,
values for the other two substrates.
   Medium-chain acyl-CoA synthetases are present in mitochondria of various mam-
malian organs. The partially purified enzyme from beef heart mitochondria acts on
fatty acids with 3-7 carbon atoms, but is most active with butyrate. In contrast, a
partially purified enzyme from bovine liver mitochondria activates fatty acids with
4 12 carbon atoms with octanoate being the best substrate. This enzyme also acti-
vates branched-chain, unsaturated, and hydroxy-substituted medium-chain carbox-
ylic acids and, more surprisingly, acts on aromatic carboxylic acids like benzoic acid
and phenylacetic acid.
   Long-chain acyl-CoA synthetase is a membrane-bound enzyme associated with
the endoplasmic reticulum, peroxisomes, and the outer mitochondrial membrane.
The rat liver enzymes purified from mitochondria and the endoplasmic reticulum are
identical as judged by several molecular and catalytic properties [7]. Immunological
evidence indicates that long-chain acyl-CoA synthetases associated with the three
subcellular organelles are structurally identical or very similar. The enzyme acts effic-
iently on saturated fatty acids containing 10-18 carbon atoms and on unsaturated
fatty acids containing 16-20 carbon atoms. The multiple subcellular locations of this
enzyme may reflect its functions in both lipid synthesis and fatty acid oxidation. If so,
the mitochondrial and peroxisomal enzymes would synthesize fatty acyl-CoA thioes-
ters for oxidation, whereas the enzyme of the endoplasmic reticulum would provide
substrates for lipid (for example, glycerolipid) synthesis. Evidence has been obtained
for the existence of a distinct very-long-chain acyl-CoA synthetase, which, in contrast
to long-chain acyl-CoA synthetase, is more active with lignoceric acid and arachidon-
ic acid than with palmitic acid.
   In addition to ATP-dependent acyl-CoA synthetases, a group of GTP-dependent
acyl-CoA synthetases has been described. The best known of these is succinyl-Coh
synthetase, which cleaves GTP to GDP plus phosphate and functions in the tricar-
boxylic acid cycle. Although a mitochondrial GTP-dependent acyl-CoA synthetase
activity has been described, the existence of a distinct enzyme with such activity has
been questioned.

3. B-Oxidation in mitochondria

3.1. Mitochondria1 uptake of fatty acids

Since the inner mitochondrial membrane is impermeable to CoA and its derivatives,
fatty acyl-CoA thioesters formed at the outer mitochondrial membrane cannot di-

rectly enter the mitochondrial matrix where the enzymes of /I-oxidation are located.
Instead, the acyl residues of acyl-CoA thioesters are carried across the inner mito-
chondrial membrane by L-carnitine. This carnitine-dependent translocation of fatty
acids across the inner mitochondrial membrane is schematically shown in Fig. 1 [8].
The reversible transfer of fatty acyl residues from CoA to carnitine is catalyzed by
carnitine palmitoyltransferase I (CPT I), which is an enzyme of the outer mitochon-
drial membrane. The resultant acylcarnitines cross the inner mitochondrial mem-
brane via the carnitine:acylcarnitine translocase [9]. This carrier protein facilitates a
slow unidirectional flux of carnitine either in or out of the mitochondrial matrix in
addition to catalyzing a much faster mole to mole exchange of acylcarnitine for car-
nitine, carnitine for carnitine, and acylcarnitine for acylcarnitine. The unidirectional
flux of carnitine may be an important mechanism by which mitochondria of various
organs acquire carnitine, which is synthesized in the liver. The fast exchange, espe-
cially of acylcarnitine for carnitine, is believed to be essential for the translocation of
long chain fatty acids from the cytosol into mitochondria. In the mitochondrial
matrix, carnitine palmitoyltransferase I1 (CPT II), an enzyme of the inner mito-
chondrial membrane, catalyzes the transfer of acyl residues from carnitine to CoA to
form acyl-CoA thioesters which then enter the /I-oxidation spiral. CPT I1 from mi-
tochondria of bovine heart and rat liver has been purified to near homogeneity. The
purified enzymes have molecular weights of approximately 70 kDa and catalyze the
reversible transfer of acyl residues with 8-16 carbon atoms between CoA and car-
nitine. The cDNAs of rat and human CPT I1 have recently been cloned and se-
quenced. The predicted amino acid sequences of the corresponding 74 kDa-proteins
show a better than 80% homology with each other. In contrast, CPT I, which is in-
hibited by malonyl-CoA, has not yet been isolated without substantial or total loss of
activity and consequently no definite structural information about this protein is
   In addition to CPT I and CPT 11, mitochondria contain a carnitine acetyltransfer-
ase which has been purified to homogeneity. The bovine heart enzyme has an esti-
mated molecular weight of 60 kDa and is composed of a single polypeptide chain. It
catalyzes the transfer of acyl groups with 2-10 carbon atoms. The function of this
enzyme has not been conclusively established. Perhaps, the enzyme regenerates free
CoA in the mitochondrial matrix by transferring acetyl groups and other short-chain
or medium-chain acyl residues from CoA to carnitine. The resulting acylcarnitines
can leave mitochondria via the carnitine:acylcarnitine translocase and can be utilized
by the same or other tissues or can be excreted in urine. Although carnitine acetyl-
transferase and medium-chain carnitine acyltransferase activities have been detected
in peroxisomes and microsomes, their functions remain to be elucidated.
   Short-chain and medium-chain fatty acids with less than 10 carbon atoms can en-
ter mitochondria as free acids independent of carnitine. They are activated in the
mitochondrial matrix where short-chain and medium-chain acyl-CoA synthetases
are located.

3.2. Enzymes of /?-oxidation in mitochondria

The first reaction of /?-oxidation is the dehydrogenation of acyl-CoA to 2-trans-en-
oyl-CoA according to the following equation:
This reaction is catalyzed by acyl-CoA dehydrogenases which have tightly, but non-
covalently, bound flavin adenine nucleotides (FAD) as prosthetic groups. Three dif-
ferent acyl-CoA dehydrogenases function in the /?-oxidation of fatty acids [101. They
are: short-chain acyl-CoA dehydrogenase, medium-chain acyl-CoA dehydrogenase,
and long-chain acyl-CoA dehydrogenase. The same set of acyl-CoA dehydrogenases
is present in liver and heart mitochondria. Isolation and purification of these en-
zymes has permitted detailed studies of their molecular and kinetic properties. All
three dehydrogenases have similar molecular weights between 170 and 190 kDa and
are composed of four identical subunits. Their cDNAs have been cloned and se-
quenced. High degrees of homology (close to 90%) have been observed for the same
enzyme from human and rat and significant homologies (30-35%) are apparent when
different enzymes from one source are compared. The tertiary structure of medium-
chain acyl-CoA dehydrogenases at 3 A resolution confirms the homotetrameric
structure of the enzyme with one FAD bound per subunit in an extended conforma-
tion. The three dehydrogenases differ in their specificities for substrates of various
chain lengths. Short-chain acyl-CoA dehydrogenase or butyryl-CoA dehydrogenase
acts only on hexanoyl-CoA and shorter-chain substrates. Medium-chain acyl-CoA
dehydrogenase is most active with medium-chain substrates from hexanoyl-CoA to
dodecanoyl-CoA, whereas long-chain acyl-CoA acts on octanoyl-CoA and longer-
chain substrates. Kinetic measurements with all three dehydrogenases yielded K,,
values of 1-10 p M for good substrates. The dehydrogenation of acyl-CoA thioesters
proceeds by removal of a proton from the a-carbon of the substrate followed by
hydride transfer from the /?-carbon to the FAD cofactor of the enzyme to yield 2-
trans-enoyl-CoA and enzyme-bound FADH,. Reoxidation of FADH, of the dehy-
drogenase occurs by single electron transfer to the FAD prosthetic group of a second
flavoprotein named electron-transferring flavoprotein (ETF), which donates elec-
trons to an iron-sulfur flavoprotein named ETFxbiquinone oxidoreductase. The
latter enzyme, a component of the inner mitochondria1 membrane, feeds electrons
into the mitochondrial electron transport chain via ubiquinone. The flow of electrons
from acyl-CoA to oxygen is schematically shown in the following flow chart:
    R-CH,-CH,-CO-SCoA+FAD (Acyl-CoA dehydrogenase) -+ FAD (ETF)                    +
    FeS (ETFxbiquinone oxidoreductase) + ubiquinone +++-+oxygen
ETF is a soluble matrix protein with a molecular weight close to 60 kDa. It is com-
posed of two non-identical subunits of similar molecular weights. Only one of the
subunits contains an FAD prosthetic group.

  In addition to the three aycl-CoA dehydrogenases involved in fatty acid oxidation,
two acyl-CoA dehydrogenases specific for metabolites of branched-chain amino
acids have been isolated and purified. They are isovaleryl-CoA dehydrogenase and
2-methyl-branched chain acyl-CoA dehydrogenase.
  The second step in p-oxidation is the hydration of 2-trans-enoyl-CoA to L-3-hydrox-
yacyl-CoA catalyzed by enoyl-CoA hydratase as shown in the following equation:

     R-CH=CH-CO-SCoA      + HZO +R-CH(OH)-CH,-CO-SCoA
The reaction is reversible and at equilibrium the ratio of L-3-hydroxyacyl-CoA to
2-enoyl-CoA is close to 3. Enoyl-CoA hydratase also catalyzes the hydration of 2-
cis-enoyl-CoA to D-3-hydroxyacyl-CoA.
   Two enoyl-CoA hydratases have been identified in heart mitochondria [1I]. The
better characterized of the two is enoyl-CoA hydratase, or crotonase, which has been
purified to homogeneity and crystallized. Its molecular weight is 165 kDa and it is
composed of six identical subunits. The cDNA of rat liver crotonase has been cloned
and sequenced. A long-chain enoyl-CoA hydratase, which appears to be at least part-
ially membrane associated, has been separated from crotonase. The pig heart long-
chain enoyl-CoA hydratase does not act on crotonyl-CoA (CH,-CH=CH-CO-
SCoA), but is active with longer-chain enoyl-CoA thioesters. In contrast, pig heart
crotonase is most active with crotonyl-CoA and exhibits a decreasing activity with
increasing chain lengths of its substrates. The activity of crotonase with 2-hexadece-
noyl-CoA is only one-fortieth of the activity observed with crotonyl-CoA as a sub-
strate. Kinetic measurement with both hydratases yielded K,,, values between 10 yM
and 30 pM for virtually all substrates. Since most of the long-chain enoyl-CoA hy-
dratase activity of pig heart muscle is due to the long-chain enzyme, the suggestion
has been made that crotonase and long-chain enoyl-CoA hydratase cooperate to as-
sure high rates of hydration of all enoyl-CoA intermediates in fatty acid oxidation.
However, the definite metabolic function of long-chain enoyl-CoA hydratase re-
mains to be established.
   The third reaction in the p-oxidation spiral is the reversible dehydrogenation of
L-3-hydroxyacyl-CoAto 3-ketoacyl-CoA catalyzed by L-3-hydroxyacyl-CoAdehy-
drogenase as shown in the following equation:

     R-CH(OH)-CH,-CO-SCoA      + NAD'+R-CO-CH, -CO-SCoA + NADH + H'
L-3-Hydroxyacyl-CoAdehydrogenase is a soluble matrix enzyme, which has a molec-
ular weight of approximately 65 kDa and is composed of two identical subunits [4].
The pig heart enzyme has been sequenced and its conformation at 2.8 8, resolution is
indicative of a bilobal protein with an NAD' binding site at the amino-terminal do-
main. The enzyme is specific for NAD' as a coenzyme, but acts on L-3-hydroxyacyl-
CoAs of various chain lengths. Although the K,,, values for all L-3-hydroxyacyl-CoAs

studied so far are below 10 p M , the dehydrogenase exhibits little activity (only 6%)
with a long-chain substrate like L-3-hydroxyhexadecanoyl-CoA as compared to the
medium-chain substrate L-3-hydroxydecanoyl-CoA. Thus, a recently described
membrane-bound L-3-hydroxyacyl-CoAdehydrogenase with a preference for long-
chain substrates may complement the soluble dehydrogenase to assure high rates of
dehydrogenation over the whole spectrum of /?-oxidation intermediates.
  The last reaction in the /?-oxidation spiral is the cleavage of 3-ketoacyl-CoA cata-
lyzed by thiolase as shown in the following equation:

     R-CO-CH,-CO-SCoA       + COASH+R-CO-SCOA + CH,-CO-SCoA

The products of the reaction are acetyl-CoA and an acyl-CoA shortened by two car-
bon atoms. The equilibrium of the reaction is far to the side of the thiolytic cleavage
products. All thiolases that have been studied in detail contain an essential sulfhydryl
group which participates directly in the carbon-carbon bond cleavage as outlined in
the following equations where E-SH represents thiolase:

    E-SH + R-CO-CH,-CO-SCoA          *R-CO-S-E + CH,-CO-SCoA
           R-CO-S-E + COASH         *R-CO-SCoA+E-SH
According to this mechanism, 3-ketoacyl-CoA binds to the enzyme and is cleaved
between its 01 and/? carbon atoms. An acyl residue, which is two carbons shorter than
the substrate, remains covalently linked to the enzyme via a thioester bond, while
acetyl-CoA is released from the enzyme. Finally, the acyl residue is transferred from
the sulfhydryl group of the enzyme to CoASH to yield acyl-CoA.
   Several types of thiolases have been identified, some of which exist in multiple
forms [ 121. Mitochondria contain two classes of thiolases: acetoacetyl-CoA thiolase
or acetyl-CoA acetyltransferase, which is specific for acetoacetyl-CoA as a substrate,
and 3-ketoacyl-CoA thiolase or acetyl-CoA acyltransferase, which acts on 3-ketoa-
cyl-CoA thioesters of various chain lengths. The latter enzyme is essential for /?-oxi-
dation, whereas acetoacetyl-CoA thiolase possibly functions only in ketone body
synthesis and degradation. Both mitochondrial thiolases have been purified to homo-
geneity. The cDNA of rat liver 3-ketoacyl-CoA thiolase has been cloned and se-
quenced. A comparison of amino acid sequences proves the two mitochondrial
thiolases to be different, but homologous enzymes. Both types of thiolases are com-
posed of four identical subunits with molecular weights close to 42 kDa. Kinetic
measurements with 3-ketoacyl-CoA thiolase from pig heart have yielded K, values
that decrease from 17 pM to 2 pM as the acyl chain length of the substrate increases
from 4 to 8 carbon atoms. This enzyme acts equally well on all substrates tested ex-
cept for acetoacetyl-CoA which is cleaved at half the maximal rate observed with
longer chain substrates. Some tissues, such as liver, also contain a cytosolic acetoa-
cetyl-CoA thiolase which functions in cholesterol biosynthesis.

3.3. P-Oxidation o unsaturated und odd-chainfatty acids

Unsaturated fatty acids, which usually contain cis double bonds, also are degraded
by /?-oxidation. However, their double bonds must either be removed or moved dur-
ing the degradation process [ 131. All double bonds found in unsaturated and polyun-
saturated fatty acids can be classified either as double bonds extending from odd-
numbered carbon atoms, like the 9-cis double bond present in oleic acid, linoleic acid
and many other polyunsaturated fatty acids, or as double bonds extending from
even-numbered carbon atoms like the 12-cis double bond of linoleic acid. Since both
classes of double bonds are present in linoleic acid, its degradation exemplifies the
breakdown of all other unsaturated fatty acids. A summary of the oxidation of lino-
leic acid is presented in Fig. 2. Linoleic acid, after conversion to its CoA thioester (I),
undergoes three cycles of /?-oxidation to yield 3-cis,6-cis-dodecadienoyl-CoA(11)
which is isomerized to 2-trans,6-cis-dodecadienoyl-CoA         (111) by A3-cis-A2-trans-en-
oyl-CoA isomerase, an auxiliary enzyme of /?-oxidation. This enzyme can also iso-
merize 3-trans-enoyl-CoA thioesters to their 2-trans isomers, although at lower rates.
2-trans,6-cis-Dodecadienoyl-CoA       (111) is a substrate of P-oxidation and can pass
once through the cycle to yield 4-cis-decenoyl-CoA (IV) which is dehydrogenated to
2-trans,4-cis-decadienoyl-CoA by acyl-CoA dehydrogenase. Only medium-chain
dehydrogenase is capable of catalyzing this reaction, even though long-chain acyl-
CoA dehydrogenase will act on decanoyl-CoA. 2-trans,4-cis-Decadienoyl-CoA              (V)
cannot be directlyp-oxidized, but is first reduced by NADPH in a reaction catalyzed
by 2,Cdienoyl-CoA reductase. The product of this reduction, 3-trans-decenoyl-CoA
(VI), is isomerized by A3-cis-A2-trans-enoyl-C isomerase to 2-trans-decenoyl-CoA
(VII), which can be completely degraded by passing four times through the /?-oxida-
tion spiral. The conclusion is that the degradation of unsaturated fatty acids in mito-
chondria requires A3-cis-A2-trans-enoyl-Co isomerase and 2,4-dienoyl-CoA reduc-
tase as auxiliary enzymes in addition to the enzymes of the /?-oxidation spiral.
   Although the mitochondrial pathway of linoleate degradation (Fig. 2) is well estab-
lished, a number of textbooks continue to present a wrong degradation scheme. Ac-
cording to the incorrect scheme, 4-cis-enoyl-CoA intermediates derived from unsat-
urated fatty acids, for example compound IV shown in Fig. 2, pass once more
through a complete cycle of /?-oxidation to yield 2-cis-enoyl-CoAs which are hydrat-
ed by enoyl-CoA hydratase to D-3-hydroxyacyl-CoAs. The latter compounds are
then converted to their L-isomers by 3-hydroxyacyl-CoA epimerase before being
completely degraded by /?-oxidation. However, the absence of 3-hydroxyacyl-CoA
epimerase from mitochondria [ 131 and the inability of the mitochondrial /?-oxidation
system to act on 2-trans,4-cis-decadienoyl-CoA[ 121 disprove the epimerase-depend-
ent pathway of unsaturated fatty acid p-oxidation.
   A3-cis-d2-trans-Enoy1  -CoA isomerase has been isolated and purified from hog liver
as well as rat liver, The rat liver enzyme is a dimeric protein with a molecular weight
of 60 kDa. In addition to catalyzing the isomerization of CoA derivatives of 3-cis-

 (I)                                                      SCoA

     i                                             0

                                I Enoyl - C o A   isomerase
                                to    -

                           , 2        5CoA
               7    6
                            One cycle of p-oxidation

(              5    4

                           1Acyl- C ~ dehydrogenase

                          2 . A - Dienoyl-CoA reductase

                           Enoyl- C a A isomerase

                           Four cycles of p-oxidation


Fig. 2. B-Oxidation of linoleoyl-CoA.

enoic acids and 3-trans-enoic acids with six to sixteen carbon atoms to the corre-
sponding 2-rrans-enoyl-CoAs, the enzyme catalyzes the conversion of 3-acetylenic
acyl-CoA to 2,3-dienoyl-CoA,
   2,4-Dienoyl-CoA reductases from bovine and rat liver have been purified to homo-
geneity. They are homotetramers with native molecular weights of 124 kDa. The re-
ductase has a specific requirement for NADPH. NADH neither substitutes for
NADPH nor inhibits the enzyme. K,,, values for NADPH and 2-trans,Qcis-decadie-
noyl-CoA are 94 p M and 3 p M , respectively.
   The oxidation of fatty acids with an odd number of carbons proceeds by /?-oxida-
tion and yields, in addition to acetyl-CoA, one mole of propionyl-CoA per mole of
fatty acid. Propionyl-CoA, which is further metabolized to succinate, is also formed
during the degradation of amino acids such as methionine, valine, and isoleucine.
Propionyl-CoA is carboxylated by biotin-containing propionyl-CoA carboxylase to

     CH,-CH,-CO-SCoA       + HCO3- + ATP + -0OC-CH-CO-SCOA + ADP + Pi
The D-isomer is isomerized to the L-isomer by methylmalonyl-CoA racemase. In the
final step of this pathway, L-methylmalonyl-CoA is isomerized to succinyl-CoA,
which is an intermediate of the tricarboxylic acid cycle:

     -OOC-CH-CO-SCOA        + -OOC-CH~-CH,-CO-SCOA
This reaction is catalyzed by methylmalonyl-CoA mutase, one of the few enzymes
requiring cobalamin as a cofactor. All reactions of propionyl-CoA catabolism occur
in mitochondria.

3.4. Regulation o rnitochondriul/3-oxidation

The rate of fatty acid oxidation is a function of the plasma concentration of unesteri-
fied fatty acids. Unesterified or free fatty acids are released from adipose tissue into
the circulatory system which carries them to other tissues or organs. The breakdown
of triacylglycerols (lipolysis) in adipose tissue is regulated by hormones like gluca-
gon, which stimulates, and insulin, which inhibits this process. The stimulatory effect
of glucagon is due to its activation of adenylate cyclase and the resulting increase in
the concentration of cellular cyclic AMP, which activates protein kinase A. One of
the substrates of protein kinase A in adipose tissue is the hormone-sensitive lipase,
which is activated by phosphorylation and inactivated by dephosphorylation. Thus,
when the concentration of glucose is low, as in the fasting animal, a high [glucagon]/
[insulin] ratio results in an increased plasma concentration of unesterified free fatty
acids. These fatty acids will enter cells, where they can be either degraded to acetyl-
CoA or incorporated into other lipids. The utilization of fatty acids for either oxida-
tion or lipid synthesis depends on the nutritional state of the animal, more specifically
on the availability of carbohydrates. Because of the close relationship between lipid
metabolism, carbohydrate metabolism, and ketogenesis, the regulation of fatty acid
oxidation in liver differs from the regulation of /&oxidation in tissues like heart and
skeletal muscle, which have an overwhelming catabolic function. For this reason, the
regulation of fatty acid oxidation in liver and heart will be discussed separately.
   The direction of fatty acid metabolism in liver depends on the nutritional state of
the animal. In the fed animal the liver breaks down carbohydrates to synthesize fatty
acids, while in the fasted animal fatty acid oxidation, ketogenesis, and gluconeogene-
sis are the more active processes. Clearly, there exists a reciprocal relationship be-
tween fatty acid synthesis and fatty acid oxidation. Although it is well established

that lipid and carbohydrate metabolism are under hormonal control, it has been
more difficult to identify the specific sites at which fatty acid synthesis and oxidation
are regulated and to elucidate the regulatory mechanisms. McGarry and Foster [14]
have proposed that the concentration of malonyl-CoA, the first committed inter-
mediate in fatty acid biosynthesis, determines the rate of fatty acid oxidation. The
essential features of their hypothesis are presented in Fig. 3. In the fed animal, where
glucose is actively converted to fatty acids, the concentration of malonyl-CoA is
elevated. Malonyl-CoA at micromolar concentrations is a reversible and effective
inhibitor of CPT I. As a result of the inhibition of CPT I by malonyl-CoA, fatty acyl
residues are not transferred from acyl-CoA to carnitine and therefore are not translo-
cated into mitochondria. Consequently, B-oxidation is depressed. When the animal
changes from the fed to the fasted state, hepatic metabolism shifts from glucose
breakdown to gluconeogenesis with a resulting decrease in fatty acid synthesis. The
concentration of malonyl-CoA decreases, and the inhibition of CPT I is relieved.
Furthermore, starvation also causes an increase in the total CPT I activity as well as
a decrease in the sensitivity of CPT I towsrd malonyl-CoA. Altogether, during
starvation acylcarnitines are more rapidly formed and translocated into mitochon-
dria thereby stimulating B-oxidation and ketogenesis.
   It appears that the cellular concentration of malonyl-CoA is directly related to the
activity of acetyl-CoA carboxylase, which is hormonally regulated. The short-term
regulation of acetyl-CoA carboxylase involves the phosphorylation and dephospho-
rylation of the enzyme (see Chapter 4). Phosphorylated acetyl-CoA carboxylase is
the less active form of the enzyme. In the fasted animal, a high [glucagon]/[insulin]
ratio causes an increase in the cellular concentration of CAMP, which is responsible

                      Fatty a c i d synthesis               Fatty a c i d oxidation

                          Glucose                           COpyKetone        bodies

                          Pyruvate                          Acetyl-CoA

                       Acetyl -CoA
                                                                 f    /?-oxidation
                                                            Fatty ocyl -CoA

 AACyP                                                            f   CPTII
inactive                                                  Fatty acyl carnitine

                       Malonyi-CoA-       --    - - 8- - - --c        CPTI

                                                          Fatty acyl- CoA

                        f a t t y acids
                                                            Fatty acids


Fig. 3. Proposed regulation of fatty acid oxidation in liver. @, Stimulation;          0,
                                                                                        inhibition; 0, enzymes
subject to regulation. ACC, acetyl-CoA carboxylase.

for the phosphorylation and inactivation of acetyl-CoA carboxylase. As a conse-
quence, the concentration of malonyl-CoA and the rate of fatty acid synthesis de-
crease, while the rate of /?-oxidation increases. A decrease of the [glucagon]/[insulin]
ratio reverses these effects. Thus, both fatty acid synthesis and fatty acid oxidation
are regulated by the ratio of [glucagon]/[insulin].
  Whether or not fatty acid oxidation in non-hepatic tissues is controlled by malo-
nyl-CoA via the regulation of CPT I remains to be established. However, in heart,
and possibly in other tissues, the rate of fatty acid oxidation is tuned to the cellular
energy demand in addition to being dependent on the concentration of plasma free
fatty acids. At sufficiently high concentrations (>0.6 mM) of free fatty acids the rate
of fatty acid oxidation is only a function of the cellular energy demand. Studies with
perfused hearts and isolated heart mitochondria have shown that a decrease in the
energy demand results in increases in the concentrations of acetyl-CoA and NADH
and in decreases of those of CoASH and NAD'. The resultant increases in the ratios
of [acetyl-CoA]/[CoASH]and [NADH]/[NAD'] in the mitochondrial matrix may be
the cause for the reduced rate of /?-oxidation. Experiments with isolated heart mito-
chondria support this view and moreover demonstrate that the ratio of [acetyl-CoA]/
[CoASH], and not the [NADH]/[NAD'] ratio, controls the rate of /?-oxidation. Al-
though the site of regulation has not been determined beyond doubt, it is likely that
the [acetyl-CoA]/[CoASH] ratio determines the activity of 3-ketoacyl-CoA thiolase
and thereby controls the flux of fatty acids through the B-oxidation spiral.

3.5. Inhibitors of mitochondria1/?-oxidation

The study of hypoglycin, which causes Jamaican vomiting sickness in humans [ 151,
stimulated an interest in inhibitors of fatty acid oxidation. Ingestion of the amino
acid hypoglycin, which is present in the arillus of the unripe ackee fruit, induces se-
vere hypoglycemia presumably as a result of inhibiting fatty acid oxidation. In ani-
mals, hypoglycin is metabolized by deamination and oxidative decarboxylation to
methylenecyclopropylacetyl-CoA (Table I), which inactivates several acyl-CoA de-
hydrogenases and thereby inhibits /?-oxidation. Efforts to develop other inhibitors of
fatty acid oxidation resulted in the design of several compounds that inactivate either
CPT I or thiolases [16].
   Several inhibitors of CPT I are long-chain fatty acids with reactive substitutents, as
for example 2-tetradecylglycidic acid and 2-bromopalmitic acids (Table I). These
compounds are converted intracellularly to their CoA thioesters whereupon they
bind to CPT I to form covalent adducts or tight complexes and inactivate the enzyme.
As a consequence, the uptake and oxidation of long-chain fatty acids by mitochon-
dria are inhibited. However, the oxidation of medium-chain or short-chain fatty
acids, which enter mitochondria directly without the assistance of carnitine, is unaf-
fected by inhibitors of CPT I.
   Another group of /?-oxidation inhibitors inactivates thiolases. Compounds like 4-

Inhibitors of mitochondria1/?-oxidation. Shown are also the active metabolites derived from the inhibitors
and the inhibited enzymes.
lnhibi tor                     Inhibitory metobolite            Inhibited enzyme

                               CH~=C-CH-CHZ-COSCOA              A C ~ -COA
                                                                       I     dehydrogenases
2-Tetradecylglycidic acid      CH3-(CH2)?3-C -COSCoA            CPT I
2-Bromopalrnitic acid          CH3-(CH2),3-CH-COSCoA            CPT I

4-Pentenoic acid
                               CH2 CH-C-CH2-COSCoA              Thiolases

                                             0 Br
                                           II I
2-Brornooctanoic acid          CH3- (CH2I4-C-CH -COSCaA         Thiolases

                               Br   0
                               I    II
4-Brornocrotonic acid          CH2-C-CH2-COSCoA

pentenoic acid, 2-bromooctanoic acid, and 4-bromocrotonic acid enter the mito-
chondrial matrix, where they are activated and metabolized by /?-oxidation to their
 3-keto derivatives (Table I). The resultant 3-ketoacyl-CoA compounds, with either a
4-double bond or bromine residues at carbon atoms 2 or 4, are highly reactive and
bind covalently to the active site of thiolases, thereby inactivating them and inhibiting

4. p-Oxidation in peroxisomes

Peroxisomes and glyoxysomes, collectively referred to as microbodies [ 171, are sub-
cellular organelles capable of respiration. They do not have an energy-coupled elec-
tron transport system like mitochondria, but instead contain flavine oxidases, which
catalyze substrate-dependent reductions of oxygen to H202.Since catalase is present
in these organelles, H20, is rapidly reduced to water. Thus, peroxisomes and glyoxy-
somes are organelles with a primitive respiratory chain where energy released during
the reduction of oxygen is lost as heat. Glyoxysomes are peroxisomes that contain the
enzymes of the glyoxylate pathway in addition to flavine oxidases and catalase. Per-
oxisomes or glyoxysomes are found in all major groups of eukaryotic organisms in-
cluding yeasts, protozoa, plants and animals.
   The presence of an active /%oxidation system in microbodies similar to the mito-
chondrial system was first detected in glyoxysomes of germinating seeds. When rat
liver peroxisomes were shown to contain a /?-oxidation system [I 81, originally
thought to be limited to mitochondria, interest in this pathway was greatly stimu-
lated. Studies of peroxisomal p-oxidation were aided by the use of certain drugs, like

clofibrate and di(-2-ethylhexyl)phthalate, which cause peroxisomes to proliferate and
in addition induce the synthesis of the enzymes of peroxisomal /%oxidation. Within
a short period of time the pathway had been elucidated (Fig. 4) and the rat liver
enzymes had been purified and characterized [ 193.
   The mechanism by which fatty acids are taken up by peroxisomes has not been
fully elucidated. Either free fatty acids enter peroxisomes and are converted to their
CoA thioesters by peroxisomal acyl-CoA synthetase, or fatty acids cross the peroxi-
soma1 membrane as their CoA thioesters. Carnitine does not stimulate fatty acid oxi-
dation in peroxisomes The first step in peroxisomal p-oxidation (Fig. 4) is the dehy-
drogenation of acyl-CoA to 2-trans-enoyl-CoA catalyzed by acyl-CoA oxidase. This
enzyme, in contrast to the mitochondria1 dehydrogenases, transfers two hydrogens
from the substrate via its cofactor flavine adenine dinucleotide to oxygen, which is
thereby reduced to H,Oz. The rat liver acyl-CoA oxidase is a homodimer with a mo-
lecular weight close to 150 kDa. This enzyme is inactive with butyryl-CoA and hexa-
noyl-CoA as substrates, but dehydrogenates all longer chain substrates with similar
maximal velocities. The K,,, values for long-chain substrates are approximately 10
pM. The next two reactions in b-oxidation, the hydration of 2-enoyl-CoA to L-3-
hydroxyacyl-CoA and the NAD+-dependent dehydrogenation of L-3-hydroxyacyl-
CoA to 3-ketoacyl-CoA, are catalyzed in rat liver peroxisomes by a trifunctional po-
lypeptide which harbors both enoyl-CoA hydratase and L-3-hydroxyacyl-CoAdehy-
drogenase in addition to d3-cis-d2-truns-enoyl-CoA    isomerase. This trifunctional en-


                                      Acyl-CoA oxidase


                                       Trifunctional enzyme

      It      I1

           CoASH                  3-Ketoacyl-CoA   thiolase

      0                    0
   A-C-SCOA        + CH3-C-SCOA

Fig. 4. Pathway of j3-oxidation in peroxisomes.

zyme consists of a single polypeptide chain with a molecular weight close to 80 kDa.
The enoyl-CoA hydratase associated with the trifunctional enzyme is most active
with crotonyl-CoA and exhibits decreasing activities with increasing chain lengths of
the substrates. The K, value for crotonyl-CoA is 83 pM,while those for all other
substrates, are approximately 10 pM. The NAD+-specificL-3-hydroxyacyl-CoAde-
hydrogenase of the trifunctional enzyme is almost equally active with substrates of
various chain lengths. However, the K, values decrease with increasing chain length
from 42 pM for 3-hydroxybutyryl-CoA to 1 pM for 3-hydroxydecanoyl-CoA. The
last reaction ofg-oxidation, the CoA-dependent cleavage of 3-ketoacyl-CoA, is cata-
lyzed by 3-ketoacyl-CoA thiolase. The peroxisomal enzyme is a homodimer with a
molecular weight close to 80 kDa. This thiolase exhibits little activity toward acetoa-
cetyl-CoA, but is highly active with all longer chain substrates. The K, values for all
substrates are in the low micromolar range.
   Cloning and sequencing of the genes and cDNAs of the rat peroxisomal /I-oxida-
tion enzymes revealed the presence of two partially different mRNAs for acyl-CoA
oxidase and the presence of two different 3-ketoacyl-CoA thiolase genes, one of
which is constitutively expressed, whereas the other is highly expressed in response to
peroxisomal proliferators [20]. Sequencing of cDNAs definitely established that the
mitochondria1 and peroxisomal p-oxidation enzymes are products of different genes.
   Unsaturated fatty acids are degraded in peroxisomes mostly by the reductase-de-
pendent pathway outlined in Fig. 2. However, a small fraction (2%) of 2,4-dienoyl-
CoA intermediates may pass once more through the /I-oxidation spiral due to inter-
mediate channeling on the trifunctional enzyme [ 131. The resultant 2-cis-enoyl-CoAs
are then hydrated to D-3-hydroxyacyl-CoAsby enoyl-CoA hydratase of the trifunc-
tional enzyme. The further metabolism of D-3-hydroxy intermediates requires epi-
merization to their L-isomers. In rat liver peroxisomes, epimerization occurs by a
two-step dehydratiodhydration reaction sequence which is catalyzed by a novel D-3-
hydroxyacyl-CoA dehydratase and by enoyl-CoA hydratase of the trifunctional en-
zyme (Fig. 5). Interestingly, all peroxisomes with /I-oxidation systems also seem to
have the capacity to epimerize 3-hydroxyacyl-CoAs in contrast to mitochondria
which do not contain this activity. In peroxisomes of yeast, fungi, and plants the
3-hydroxyacyl-CoA epimerase activity is, at least in part, associated with the trifunc-

R+SC~A                  R-SC~A
                  EH          II
  OH 0                        0
Fig. 5. Conversion of 2-cis-enoyl-CoA to L-3-hydroxyacyl-CoA in peroxisornes. EH, enoyl-CoA
hydratase; D-HAD, D-3-hydroxyacyl-CoAdehydratase.

tional enzyme which, however, is devoid of A3-cis-A2-trans-enoyl-CoA      isomerase ac-
   The products of peroxisomal /.?-oxidation in animals are chain-shortened acyl-
CoAs, acetyl-CoA, and NADH. These compounds may exit from peroxisomes via
pores that have been observed in isolated rat liver peroxisomes and that appear to
permit the influx of substrates and efflux of products of /.?-oxidation. Acyl-CoAs,
including acetyl-CoA, may also be converted to acylcarnitines by a unique peroxiso-
ma1 carnitine octanoyltransferase before leaving peroxisomes. This reaction, as well
as the observed hydrolysis of acetyl-CoA, would regenerate CoASH in peroxisomes.
   Although rat liver peroxisomes are capable of chain-shortening long-chain fatty
acids, their main function seems to be the partial /?-oxidation of very-long-chain fatty
acids, prostaglandins, dicarboxylic acids, xenobiotic compounds, like phenyl fatty
acids, and hydroxylated 5-/.?-cholestanoicacids, formed during the conversion of
cholesterol to cholic acid.

5. Fatty acid oxidation in E. coli
 The presence of an active fatty acid oxidation system in E. coli was demonstrated
 with cells grown on long-chain fatty acids as the sole carbon source [21]. Under such
 growth conditions, the enzymes of fatty acid oxidation are highly induced. Isolation
 of fatty acid oxidation mutants facilitated the mapping of fatty acid degradation 0
 genes, which are located at five different locations on the E. coli chromosome, but
 together form a regulon. Expression of thefud regulon is repressed by the fudR gene
 product, the repressor protein, in the absence of fatty acids. In the presence of fatty
 acids with ten or more carbon atoms the fad regulon is coordinately induced, most
 likely as a result of fatty acyl-CoAs binding to the 29 kDa repressor protein. Expres-
 sion of the enzymes of fatty acid oxidation is also repressed by glucose in the growth
 medium in a manner that resembles catabolite repression involving the catabolite
 gene activator protein (CAP) and cyclic AMP.
    The uptake of fatty acids with more than 10 carbon atoms by E. coli requires a
 functional fadL gene product, which is a 43 kDa protein of the outer membrane and
 is believed to function as a fatty acid permease [22]. The uptake of long-chain fatty
 acids is closely coupled to their activation by acyl-CoA synthetase. This conclusion
 is based on the observation that an E. coli mutant constitutive for the enzymes of
 fatty acid oxidation, but with a defective acyl-CoA synthetase, was unable to take up
 and metabolize fatty acids. Medium-chain fatty acids (fatty acids with 6-10 carbon
 atoms) can enter E. coli cells either via the long-chain uptake system or by a nonsatu-
 rable process characteristic of simple diffusion.
    E. coli seems to contain only one acyl-CoA synthetase (Fig. 6), the product of the
fudD gene, which can activate both medium-chain and long-chain fatty acids [22].
 This ATP-dependent (AMP-forming) acyl-CoA synthetase has been purified to ho-

mogeneity. The molecular weights of the native enzyme and its subunit were estimat-
ed to be 130 kDa and 47 kDa respectively. This synthetase activates fatty acids with
6-1 8 carbon atoms, but is most active with dodecanoic acid.
   Acyl-CoA dehydrogenases and the electron-transferring flavoprotein (ETF) have
not been purified or characterized but their genes vudE,EG)seem to be located in the
5-min-region of the E. coli chromosome unlinked to otherfad genes. Purification of
the enzymes ofp-oxidation from E. coli led to the isolation of a homogeneous protein
which exhibited enoyl-CoA hydratase (crotonase), L-3-hydroxyacyl-CoA dehy-
drogenase, 3-ketoacyl-CoA thiolase, ~l~-cis-d~-tvuns-enoyl-Co  A isomerase, and 3-hy-
droxyacyl-CoA epimerase activities (Fig. 6 ) . This multienzyme complex of fatty acid
oxidation contains all enzymes of the P-oxidation spiral with the exception of acyl-
CoA dehydrogenase [23]. The chain length specificities of 3-ketoacyl-CoA thiolase,
L-3-hydroxyacyl-CoA dehydrogenase and enoyl-CoA hydratase of the E. coli com-
plex are similar to those of the mammalian enzymes. All three enzymes act on sub-
strates of various chain lengths. However, enoyl-CoA hydratase is most active with
short-chain substrates, whereas 3-ketoacyl-CoA thiolase and L-3-hydroxyacyl-CoA

                                          Acyl-CoA   synthetase ( f a d D )

               FAOHz                    LI

                                        Cis -4'-trons-A2_Enoyl-CoA        isomerase

         A     S     C     o     A   *-         -       -         -
                                        - - - - - - - - - - - -- - - -   - - - - - - - -- -   I
                                                                                              ?               SCoA


 C           (LIOH
                         -I             Enoyl-CoA    hydratase

                                        3-Hydroxyacyl -CoA epimerase                              (D)OH   0
          4SCoA                                                     -
                                     *- - _ - _ - _ - - - _ - - --------- - - - _ - I?            a           SCoA

                                        3-Hydroxyacyl -COP. dehydrogenase
 P        H++NADH                                                 ( f a d B)

                                        P                                              P
                                        3-Keioacyl -CoA ihlolose ( f a d A )

Fig. 6 . Pathway of fatty acid oxidation and organization of theb-oxidation enzymes in E. coli. The 78 kDa
and 42 kDa subunits are maked a and b, respectively.

dehydrogenase exhibit their optimal activities with medium-chain substrates. The
complex has an estimated molecular weight of 260 kDa and is composed of two types
of subunits with molecular weights of 7 8 kDa and 42 kDa. The quaternary structure
of the complex is a&, where a andp denote the 78 kDa and 42 kDa subunits, respec-
tively. Phospholipids characteristic of the E. coli membrane are associated with the
purified complex. They constitute approximately 4% of the protein mass. Immu-
nological studies suggest that the total activities of 3-ketoacyl-CoA thiolase, L-3-hy-
droxyacyl-CoA dehydrogenase, and crotonase present in E. coli extracts are as-
sociated with the complex. However, a long-chain enoyl-CoA hydratase is appar-
ently not part of the complex. The subunit locations of five component enzymes were
determined by chemical modifications [24]. 3-Ketoacyl-CoA thiolase appears to be
the only enzymatic activity associated with the 42 kDa 8-subunit, whereas enoyl-CoA
hydratase, L-3-hydroxyacyl-CoAdehydrogenase, A3-cis-A2-truns-enoyl-CoA          isomer-
ase, and 3-hydroxyacyl-CoA epimerase are located on the 78 kDa a-subunit. The
complex is coded for by thefadBA operon which has been cloned and sequenced.The
operon contains two structural genes: the fudB gene, which codes for the large a-
subunit with a calculated molecular weight of 79,593 Da, and the fudA gene, which
codes for the smallp-subunit with a calculated molecular weight of 40,889 Da. Kinet-
ic studies performed with the fatty acid oxidation complex are indicative of the chan-
neling ofp-oxidation intermediates,like ~-3-hydroxyacyl-CoA,      from the active site of
enoyl-CoA hydratase to that of L-3-hydroxyacyl-CoAdehydrogenase without equili-
brating with the bulk medium.
   Even though E. coli does not synthesize polyunsaturated fatty acids, it can degrade
unsaturated fatty acids by the reductase-dependent pathway outlined in Fig. 2. The
auxiliary enzymes are A3-cis-d2-trans-enoyl-CoA     isomerase which is a component en-
zyme of the fatty acid oxidation complex, and 2,4-dienoyl-CoAreductase, which is a
monomeric 70 kDa flavoprotein coded for by the fudH gene that is unlinked to other
fad genes.

6. Inherited diseases of fatty acid oxidation

Disorders of fatty acid oxidation in humans have been recognized with increasing
frequency since the early 1970s. A number of these disorders, which compromise the
functions of liver, muscle, and other organs have been traced to deficiencies of p-
oxidation enzymes or low carnitine levels.
   In myopathic carnitine deficiency the concentration of carnitine is normal in plas-
ma, but low in muscle possibly due to impaired carnitine uptake by muscle [25]. This
deficiency results in muscle weakness and abnormal accumulation of lipids in muscle,
but generally is not life-threatening.The most frequently observed enzyme deficiency
related to fatty acid oxidation in muscle is CPT deficiency [25]. Patients with low
levels (545% of normal) of CPT have recurrent muscle weakness and myoglobinuria

often precipitated by prolonged exercise, fasting, or both. CPT deficiency is paral-
leled by impaired fatty acid oxidation and an increased dependence on carbohydrate
   More recently, disorders caused by deficiencies of short-chain, medium-chain and
long-chain acyl-CoA dehydrogenases have been described [26]. All of these defects
are inherited in an autosomal recessive manner. Medium-chain acyl-CoA dehydro-
genase (MCAD) deficiency is the most common disorder of fatty acid oxidation. It is
characterized by non-ketotic hypoglycemia induced by fasting. The carnitine levels of
afflicted patients are low, presumably due to the excretion of medium-chain acylcar-
nitines in urine. Other excreted fatty acid metabolites are glycine conjugates of me-
dium-chain fatty acids and medium-chain dicarboxylic acids as well as dicarboxylic
acids themselves. Between episodes, patients with MCAD deficiency appear quite
normal. Therapy is designed to prevent fasting and includes L-carnitine supplementa-
   Other known disorders of /?-oxidation are due to deficiencies of electron-trans-
ferring flavoprotein (ETF), ETF dehydrogenase, L-3-hydroxyacyl-CoA dehydrogen-
ase, and 2,4-dienoyl-CoA reductase. Deficiencies of ETF and ETF dehydrogenase
give rise to a spectrum of disorders, including multiple acyl-CoA dehydrogenation
defects, in which the dehydrogenation of numerous substrates is impaired. A defect
of L-3-hydroxyacyl-CoA dehydrogenase is characterized by urinary excretion of me-
dium-chain 3-hydroxy fatty acids and 3-hydroxy dicarboxylic acids. However, this
disorder is poorly understood at the molecular level due to the insufficient charac-
terization of L-3-hydroxyacyl-CoA dehydrogenases. Finally, a deficiency of 2,4-
dienoyl-CoA reductase, which was detected and characterized very recently, is the
one disorder that affects only the @-oxidation of unsaturated fatty acids [27]. This
disorder is characterized by the urinary excretion of 2-trans,4-cis-decadienoylcar-
nitine derived from the corresponding CoA derivative, which is a metabolite of li-
noleic acid and would be expected to accumulate in .the absence of 2,4-dienoyl-CoA
   A general conclusion of these studies is that an impairment of p-oxidation makes
fatty acids available for microsomal o-oxidation by which fatty acids are oxidized at
their terminal (w) methyl group or at their penultimate (w-1) carbon atom. Molecular
oxygen is required for this oxidation and the hydroxylated fatty acids are further
oxidized to dicarboxylic acids. Long-chain dicarboxylic acids can be chain-shortened
by /%oxidation in peroxisomes to medium-chain dicarboxylic acids which are excret-
ed in urine.
   Several disorders associated with an impairment of peroxisomal /?-oxidation have
been described [28]. Of these, Zellweger syndrome and neonatal adrenoleukodys-
trophy are characterized by the absence or low levels of peroxisomes due to a defec-
tive biogenesis of this organelle. As a result of this deficiency, compounds that are
normally metabolized in peroxisomes accumulate in plasma, for example very long-
chain fatty acids, dicarboxylic acid, hydroxylated 5-/?-cholestanoicacid and also phy-

tanic acid. Infants with Zellweger syndrome rarely survive longer than a few months
due to hypotonia, seizures and frequently cardiac defects. In addition to disorders of
peroxisome biogenesis, defects of each of the three enzymes of the peroxisomal p-
oxidation spiral have been reported. All of these patients were hypotonic, developed
seizures, and failed to make psychomotor gains.
   The importance of a-oxidation in humans has been established as a result of study-
ing Refsum’s disease, a rare and inherited neurological disorder. Patients afflicted
with this disease accumulate large amounts of phytanic acid (Fig. 7), which is derived
from phytol, a component of chlorophyll. Because of a methyl substituent at its p-
carbon, phytanic acid cannot bep-oxidized, but it can undergo a-oxidation to prista-
nic acid. This minor pathway of fatty acid oxidation involves the hydroxylation at the
a-carbon followed by decarboxylation. Pristanic acid, in contrast to phytanic acid,
can be degraded by p-oxidation. A deficiency of a-oxidation prevents the metabolism
of phytanic acid and results in its accumulation in various body compartments.

7. Future directions

Although fatty acid oxidation has been extensively studied, important questions
about this basic biological process remain unanswered. Moreover, the recognition
that animal cells contain two P-oxidation systems and reports of new inherited disor-
ders of fatty acid oxidation have expanded the list of unresolved problems. Most
important are efforts to elucidate the regulation of the pathway in order to under-
stand the dynamics of the process. Progress toward this goal requires the complete
structural and functional characterization of the carnitine-dependent uptake system
by which fatty acids enter liver mitochondria. Since less is known about the regula-
tion of fatty acid oxidation in extrahepatic tissues, more needs to be done. Interest in
the regulation of P-oxidation has also prompted efforts to elucidate the organization
of ,&oxidation enzymes in the mitochondrial matrix. Topics that have been ad-
dressed, but have not been fully answered, are the contribution of peroxisomes to
fatty acid oxidation in animals and the cooperation of the mitochondrial and peroxi-

 CH3    CH3
              CH3       CH3
                              COOH Phytanic acid

       C o 2 d a-Oxidation

 CH3    CH3       CH3   CH3
                                   Pristanic acid

Fig. 7. Metabolism of phytol.

soma1 /?-oxidation systems in the metabolism of compounds that are degraded by
p-oxidation. The characterization of the peroxisomal p-oxidation system, more so
than the mitochondrial one, has been, and continues to be aided by molecular
biological approaches. Although studies of inherited disorders of fatty acid oxidation
have provided some important answers, they have generated even more questions.
For example, questions have emerged about the number of/?-oxidation enzymes that
catalyze the same reaction, but differ in their chain length specificities. Even today,
not all steps of the lengthy pathway of /?-oxidation, from a long-chain fatty acid to
acetyl-CoA, have been experimentally verified. Once this has been done with the
powerful and sensitive analytical techniques available today, for example high per-
formance liquid chromatography and gas chromatography in combination with
mass spectrometry, the pathway may need to be modified. Clearly, this well estab-
lished area of biochemistry is alive with opportunities for further research.

  I . Knoop, F. (1904) Der Abbau aromatischer Fettsauren im Tierkorper. Kuttruff, Freiburg.
 2. Dakin, H. (1909) The mode of oxidation in the animal organism of phenyl derivatives of fatty acids.
      Part IV. Further studies on the fate of phenylpropionic acid and some of its derivatives. J. Biol. Chem.
      6, 203-219.
  3. Lynen, F. (1952-1953) Acetyl coenzyme A and the fatty acid cycle. Harvey Lectures 48, 210-244.
 4. Schulz, H. (1991) Beta oxidation of fatty acids. Biochim. Biophys. Acta 1081, 109-120.
  5. Veerkamp, J.H., Peeters, R.A. and Maatman, R.G.H.J. (1991) Structural and functional features of
      different types of cytoplasmic fatty acid-binding proteins. Biochim. Biophys. Acta 1081, 1-24.
 6. Groot, P.H.E., Scholte, H.R. and Hulsmann, W.C. (1976) Fatty acid activation: specificity, localiza-
      tion, and function. In: Advances in Lipid Research (Paoletti, R. and Krichevsky, D., Eds.), Vol. 14, pp.
      75-126, Academic Press, New York.
 7. Tanaka, T., Hosaka, K., Hoshimaru, M. and Numa. S . (1979) Purification and properties of long-
      chain acyl coenzyme A synthetase from rat liver. Eur. J. Biochem. 98. 165-172.
  8. Bieber, L.L. (1988) Carnitine. Annu. Rev. Biochem. 57, 261-283.
 9. Pande, S.V. (1975) A mitochondrial carnitine acylcarnitine translocase system. Proc. Natl. Acad. Sci.
      U.S.A. 72.883-887.
10. Ikeda, Y. and Tanaka, K . (1990) Purification and characterization of five acyl-CoA dehydrogenases
      from rat liver mitochondria. In: Fatty Acid Oxidation. Clinical, Biochemical and Molecular Aspects
      (Tanaka, K . and Coates, P.M., Eds.), pp. 37-54, Liss, New York, NY.
I I. Fong, J.C. and Schulz, H. (1981) Short-chain and long-chain enoyl-CoA hydratases from pig heart
      muscle. Methods Enzymol. 71, 390-398.
12. Middleton. B. (1975) 3-Ketoacyl-CoA thiolases of mammalian tissues. Methods Enzymol. 35, 128-
13. Schulz, H. and Kunau, W.-H. (1987) Beta-oxidation of unsaturated fatty acids: a revised pathway.
      Trends Biochem. Sci. 12,403- 406.
14. McGdrry, J.D. and Foster, D.W. (1980) Regulation of hepatic fatty acid oxidation and ketone body
      production. Annu. Rev. Biochem. 49, 395420.
15. Tanaka, K. and Ikeda, Y.(1990) Hypoglycin and Jamaican vomiting sickness. In: Fatty Acid Oxida-
      tion. Clinical, Biochemical and Molecular Aspects (Tanaka, K. and Coates, P.M., Eds.) pp. 167-184,
      Liss, New York, NY.

16. Schulz, H. (1987) Inhibitors of fatty acid oxidation. Life Sci. 40, 1443-1449.
17. De Duve C. (1983) Microbodies in the living cell. Sci. Am. 248,7484.
18. Lazarow, P.B. and De Duve, C. (1976) A fatty acyl-CoA oxidizing system in rat liver peroxisomes;
    enhancement by clofibrate, a hypolipidemic drug. Proc. Natl. Acad. Sci. U.S.A. 73,2043-2046.
19. Hashimoto, T. (1990) Purification, properties, and biosynthesis of peroxisomal B-oxidation enzymes.
    In: Fatty Acid Oxidation. Clinical, Biochemical and Molecular Aspects (Tanaka, K. and Coates, P.M.,
    Eds.), pp. 137-152, Liss, New York, NY.
20. Osumi, T. (1990) Molecular cloning and sequencing of the peroxisomalB-oxidation enzymes. In: Fatty
    Acid Oxidation. Clinical, Biochemical and Molecular Aspects (Tanaka, K. and Coates, P.M., Eds.),
    pp. 681496, Liss. New York, NY.
21. Overath, P., Raufuss, E.M.,   Stoffel, W. and Ecker, W. (1967) The induction of the enzymes of fatty
    acid degradation in Escherichiu coli. Biochem. Biophys. Res. Commun. 29,28-33.
22. Nunn, W.D. (1986) A molecular view of fatty acid catabolism in Escherichiu coli. Microbiol. Rev. 50,
23. Binstock, J.F., Pramanik, A. and Schulz, H. (1977) Isolation of a multienzyme complex of fatty acid
    oxidation from Escherichiu coli. Proc. Natl. Acad. Sci. U.S.A. 74,492495.
24. Yang, S . and Schulz, H. (1983) The large subunit of the fatty acid oxidation complex from Escherichiu
    coli is a multifunctional polypeptide. J. Biol. Chem. 258, 9780- 9785.
25. Di Mauro, S., Trevisan, C . and Hays, A. (1980) Disorders of lipid metabolism in muscle. Muscle
    Nerves 3, 369- 388.
26. Roe, C.R. and Coates, P.M. (1989) Acyl-CoA dehydrogenase deficiencies. In: The Metabolic Basis of
    Inherited Disease (Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D., Eds.) Vol. 1, 6th edn., pp.
    889-914, McGraw-Hill, New York, NY.
27. Roe, C.R., Millington, D.S., Norwood, D.L., Kodo, N., Sprecher, H., Mohammed, B.S., Nada, M.,
    Schulz, H. and McVie, R. (1990) 2,4-Dienoyl coenzyme a reductase deficiency. A possible new disor-
    der of fatty acid oxidation. J. Clin. Invest. 119, 1703- 1707.
28. Lazarow, P.B. and Moser, H.W. (1989) Disorders of peroxisome biogenesis. In: The Metabolic Basis
    of Inherited Disease (Scriver, C.R., Beaudet, A L., Sly, W.S. and Valle, D., Eds.) Vol 11, 6th edn., pp.
    1479-1510, McGraw-Hill, New York, NY.
D.E. V a n e and J. Vance (Eds.) Biochemistry of Lipids, Lipoproteins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                 111

                                                                                       CHAPTER 4

                                 Fatty acid synthesis in eucaryotes
                                                                             ALAN G. GOODRIDGE

                     Depurtment of Biochemistry, University of Iowa, Iowa City, IA 52242, U.S.A.

1. Introduction

Long-chain fatty acids serve two primary functions in animals. As parts of phospho-
lipids and other complex lipids, fatty acids are critical structural components of cellu-
lar membranes. As parts of triacylglycerols, fatty acids represent stored energy. The
latter function is the primary concern of this chapter. Homeothermic animals main-
tain a constant body temperature and, in adults, a constant body weight. To do so,
birds and mammals balance the amount of energy consumed in their diets with the
amount lost as heat. Nature has solved the problem of erratic or cyclic food availabil-
ity by evolving a complex, highly regulated system which ensures that energy can be
stored when food is abundant; the stored energy can be utilized when food is scarce.
   Triacylglycerol contains about twice as many calories per gram as protein or car-
bohydrate and is the major form in which energy is stored. In addition, the energy is
stored without the concomitant deposition of large quantities of water. In an average
70-kg human male, triacylglycerols constitute 85% of the total 166,000 kcal stored in
body tissues. Carbohydrate, the other easily mobilized form of energy, constitutes
less than 1000 kcal. By contrast, the diets of many animals contain a large amount of
carbohydrate. Thus, energy storage involves conversion of carbohydrates to fatty
   The rate of de novo synthesis of long-chain fatty acids is rapid in well-fed animals,
especially when the diet has little or no fat, and slow in starved animals. This regula-
tion is important because glucose, a substrate for lipogenesis, is required as an energy
source for the brain and for erythrocytes, even during starvation. Thus, inhibition of
the conversion of glucose to fatty acids during starvation preserves glucose for those
tissues that require it. Regulation of fatty acid synthesis by diet is also important for
another reason. The last two enzymes of fatty acid synthesis involve conversion of
acetyl-CoA to long-chain fatty acids. The product of the reverse pathway, oxidation
of fatty acids, is acetyl-CoA. If flux through the terminal steps in fatty acid synthesis
were not inhibited during starvation, energy would be wasted in the futile cycling of
acetyl-CoA to fatty acids and back to acetyl-CoA again.

   The cells of most tissues synthesize fatty acids at low rates. The liver, however, has
a large capacity to synthesize fatty acids. Triacylglycerols are synthesized in the liver
and transported to adipose tissue for storage. A few species, including rodents, con-
vert dietary carbohydrate to triacylglycerol in both liver and adipose tissue. Other
tissues also synthesize large amounts of long-chain fatty acids but for specialized pur-
poses. For example, the lactating mammary gland converts carbohydrate into tria-
cylglycerol used to nourish the newborn. Fatty acids synthesized by sebaceous glands
are secreted as ester waxes and triacylglycerols, secretions used to condition skin,
hair, and feathers and to lubricate external surfaces. Regulation of fatty acid synthe-
sis in mammary and sebaceous glands is different from that in liver or adipose tissue.
This chapter focuses on regulation of fatty acid synthesis in the latter organs.

2. Signals in blood that mediate the effects of diet

Incorporation of ['4C]glucose or [I4C]acetateinto long-chain fatty acids is inhibited
by starvation, diabetes, and diets high in fat. Treatment of diabetic animals with insu-
lin stimulates fatty acid synthesis, as does feeding starved animals, especially if the
diet contains a high proportion of carbohydrate. These early findings suggest that
insulin mediates the effects of diet on lipogenesis. Two other facts strengthen this
postulate. Glucose stimulates secretion of insulin from the B cells of the islets of Lan-
gerhans, and insulin stimulates the metabolism of glucose by liver and adipose tissue.
However, the failure of insulin, in vivo or in vitro, to restore fatty acid synthesis in
tissues from starved rats suggests the involvement of other factors.
   Glucagon, another pancreatic hormone, is a second important factor in the regula-
tion of fatty acid synthesis. If glucagon is added to incubations of isolated tissues,
fatty acid synthesis is inhibited. In vivo, glucagon blocks the stimulation of hepatic
fatty acid synthesis caused by the refeeding of starved rats. Thus, glucagon may me-
diate a major part of the inhibition of fatty acid synthesis caused by starvation. The
inverse relationship between glucose concentration in the blood and secretion of glu-
cagon is consistent with this hypothesis. Glucagon may be more important than insu-
lin in the regulation of fatty acid synthesis. In perfused livers or isolated liver cells, the
effects of glucagon are usually greater in magnitude than those of insulin. Further-
more, the glucagon concentration in blood is elevated in diabetic animals and de-
creased when diabetic animals are treated with insulin. This is due to the inability of
glucose to inhibit glucagon secretion in the absence of insulin. Therefore, the low rate
of fatty acid synthesis in diabetic animals may be attributable to a combination of
inhibition by glucagon and lack of stimulation by insulin rather than simply to a lack
of stimulation by insulin.
   Glucagon and insulin are polypeptide hormones that interact with specific, but
different receptors on the outer surface of the plasma membrane. Several intracellu-
lar compounds have been postulated to mediate the actions of insulin, but the actual

intracellular pathway remains largely unknown. The intracellular signalling pathway
for the action of glucagon is better known. Binding of glucagon to its receptor acti-
vates adenylate cyclase, an enzyme bound to the cytoplasmic face of the plasma
membrane. The resulting increase in the intracellular concentration of cAMP acti-
vates the catalytic subunit of protein kinase A. The rapid inhibition of fatty acid
synthesis caused by glucagon is due primarily to the phosphorylation by this protein
kinase of enzymes that are rate-controlling in the pathway or of enzymes that regu-
late the production of important regulatory intermediates. Exogenous cAMP mimics
the inhibitory effect of glucagon on fatty acid synthesis in isolated liver preparations.
   The concentrations of unesterified fatty acids in plasma and long-chain fatty acyl-
CoAs in liver are elevated by starvation. In addition, diets high in fat cause inhibition
of fatty acid synthesis and elevation of the hepatic concentration of long-chain fatty
acyl-CoA. Thus, unesterified fatty acids in plasma also may regulate fatty acid syn-
thesis. The inhibition of fatty acid synthesis caused by unesterified fatty acids in isola-
ted hepatocytes and in perfused livers provides further evidence for this hypothesis.
The effects of glucagon and insulin may be amplified via this mechanism because
glucagon stimulates, and insulin inhibits, the release of fatty acids from adipose tis-
sue. (Fig. 1 demonstrates the effects of insulin, glucagon, and unesterified fatty acids
on fatty acid synthesis from I4C-labeled acetate in isolated hepatocytes.)
   Insulin, glucagon, and unesterified fatty acids are important regulators of the path-
way of fatty acid synthesis but not the only agents that do so. Fatty acid synthesis in
liver is increased in the hyperthyroid state and decreased in the hypothyroid state.

Fig. 1. The effects of insulin, glucagon, and free fatty acids on incorporation of [I -I4C]acetate into total
fatty acids in isolated hepatocytes from the newly hatched chick [I].

Furthermore, thyroid hormones stimulate fatty acid synthesis in liver cells in culture.
The active thyroid hormone is triiodothyronine; its circulating levels rise when
starved animals are fed and fall when fed animals are starved. The changes are small
compared to the changes in fatty acid synthesis but are consistent with a role for
triiodothyronine in the regulation of fatty acid synthesis during cycles of starvation
and feeding. Limited space prevents the cataloging of every compound that affects
the rate of fatty acid synthesis in adipose tissue or liver. Insulin, thyroid hormone,
glucagon, and unesterified fatty acids act directly on isolated hepatocytes or adipo-
cytes and have physiologically relevant effects. We shall concentrate on the mecha-
nisms by which these agents exert their effects on fatty acid synthesis.

3. Which enzymes regulate fatty acid synthesis?

Having described extracellular regulators, we now turn to identification of those en-
zymes with the potential to regulate flux through the pathway. More than 25 enzymes
are involved in the conversion of glucose to long-chain fatty acids. These include 14
enzymes or multienzyme complexes in the pathway per se, 4 6 that catalyze potential
futile cycles, 3 that produce essential cofactors, 2 or more transporter proteins, and
several enzymes that synthesize or degrade regulatory intermediates. Regulation of a
metabolic pathway through alteration of the catalytic activity of an enzyme in that
pathway must be exerted at a reaction for which concentrations of the reactants are
far from thermodynamic equilibrium. If the concentrations of the reactants were
close to equilibrium, the reverse reaction would proceed at almost the same rate as
the forward reaction, despite net flux through the overall pathway in the forward
direction. A change in the catalytic activity of such an enzyme would have little effect
on unidirectional flux through the pathway because both forward and reverse reac-
tions would be activated to the same extent. An enzyme is classified as regulatory if
its mass action ratio (product of the concentrations of the products divided by the
product of the concentrations of the substrates) is displaced 50-fold or more from
thermodynamic equilibrium. (The reader is referred to [2] for a more detailed analy-
sis). Enzymes that catalyze several essentially irreversible reactions involved in the
conversion of glucose to fatty acids meet this criterion.
   After determining which enzymes have the potential to regulate carbon flux, physi-
cal and kinetic properties of purified regulatory enzymes are examined to determine
the potential for regulation of catalytic efficiency in intact cells. In addition, enzyme
concentrations are analyzed to assess regulation of the number of enzyme molecules.
For the section of the pathway for fatty acid synthesis that converts citrate to long-
chain fatty acids, all three enzymes, ATPcitrate lyase, acetyl-CoA carboxylase, and
fatty acid synthase, are candidates for regulation because the reactants for all three
reactions are far from equilibrium in intact cells. ATP:citrate lyase and fatty acid
synthase do not exhibit physical or kinetic properties consistent with an ability to

regulate catalytic efficiency in intact cells, whereas acetyl-CoA carboxylase does.
Analysis of the physical and kinetic properties of this carboxylase is the subject of a
subsequent section of this chapter.
  Another way to regulate flux through a metabolic pathway is to control delivery of
substrates to the pathway. Therefore, before discussing acetyl-CoA carboxylase, we
shall review briefly regulation of the delivery of citrate, NADPH, ATP, and CoA to
the terminal segment of the pathway for fatty acid synthesis.

4. Regulation of substrate supply

4.1. Production ofpyruvate from glucose

Most of the carbon destined for fatty acid synthesis flows through the pyruvate pool.
In liver, glucose is synthesized from pyruvate via gluconeogenesis, and pyruvate is
produced from glucose via glycolysis. Most of the enzymes in these pathways are
freely reversible and ‘near equilibrium,’ functioning equally well in either direction.
Complete reversibility at every step would make regulation impossible, and energy
would be wasted in futile cycles. Different and essentially irreversible reactions cata-
lyze the interconversion of glucose and glucose-6-phosphate, fructose-6-phosphate
and fructose- 1,6-bisphosphate, and pyruvate and phosphoenolpyruvate (Fig. 2).
Regulation occurs at each of these steps.
   When animals are fed high-carbohydrate diets and (or) when the ratio of insulin to
glucagon in the blood is high, the activities of glucokinase, phosphofructokinase I,
and pyruvate kinase are increased, and the activities of glucose-6-phosphatase, fruc-
tose- 1, 6-bisphosphatase, and phosphoenolpyruvate carboxykinase are decreased.
These changes in activities are due to a combination of changes in enzyme concentra-
tion and changes in catalytic efficiency. Catalytic efficiency is regulated by a combi-
nation of covalent modification (phosphorylation) and allosteric mechanisms ([3-61
contain more detailed discussions of these mechanisms). These regulatory adjust-
ments cause net pyruvate production from glucose in the livers of well-fed animals
and net glucose production from pyruvate in livers of starved animals.

4.2. Production of citrate from pyruvate

Metabolism of pyruvate is key to the disposition of carbon from carbohydrate or
protein precursors (Fig. 2). Inhibition of pyruvate kinase during active gluconeogen-
esis prevents phosphoenolpyruvate from being reconverted to pyruvate in the cyto-
sol, thus inhibiting futile cycling. Other mechanisms direct incoming pyruvate toward
gluconeogenesis, oxidation, or lipogenesis depending on the fuel and hormonal mi-
lieu. Conditions favoring glucose synthesis (such as starvation) cause the inhibition

 aspartate   //   rnalate
                                    citrate -9citrate
                                    j                          11   1
                                                               12   1
                              Mitochondrion Cytoplasm

of pyruvate dehydrogenase in the mitochondrion. As a consequence, pyruvate flux is
directed toward oxaloacetate and thence to glucose via the pyruvate carboxylase
reaction. Both covalent modification and allosteric regulation by substrates, prod-
ucts, and other mitochondrial metabolites are involved in the complex regulation of
pyruvate dehydrogenase [6]. The net result of stimulation of this enzyme activity is
rapid formation of citrate in the mitochondria when environmental conditions favor
lipogenesis. Citrate is transported from the mitochondrion to the cytosol, where fatty
acid synthesis occurs. Citrate leaves the mitochondrion on the tricarboxylate anion
carrier in exchange for another organic anion, probably malate. The rate of efflux of
citrate is a function of the citrate concentration gradient, the malate concentration
gradient, and the activity of the tricarboxylate anion carrier. Fatty acyl-CoA, a prod-
uct of the fatty acid synthesis pathway, may inhibit this carrier. The amount of citrate
in the cytosol is an important determinant of the rate of fatty acid synthesis, both as
substrate for the pathway and as a positive regulator of acetyl-CoA carboxylase.
Once in the cytosol, citrate is cleaved to acetyl-CoA and oxaloacetate in a reaction

catalyzed by ATP:citrate lyase. The four carbons of oxaloacetate can be recycled to
the mitochondrion, either as malate or as pyruvate and COz.

4.3. Production of NADPH

Fatty acid synthesis utilizes two molecules of NADPH for each molecule of acetate
incorporated into long-chain fatty acids. In liver, glucose-6-phosphate dehydrogen-
ase and 6-phosphogluconate dehydrogenase (Fig. 2) probably furnish about half of
the NADPH used in fatty acid synthesis, with the other half coming from malic en-
zyme [L-ma1ate:NADPH' oxidoreductase (decarboxylating)]. Isocitrate dehydrogen-
ase [threo-d,-isocitrate:N ADP' oxidoreductase (decarboxylating)], appears to utilize
NADPH rather than produce it. The activities of the two dehydrogenases of the pen-
tose phosphate pathway and of malic enzyme correlate positively with the rate of
fatty acid synthesis under a wide variety of conditions. Thus, the rate of production
of NADPH could regulate fatty acid synthesis. However, in liver, each of these en-
zymes is usually near equilibrium with respect to its substrates and products. As a
consequence, changes in the activities of these enzymes do not alter the rate of
production of NADPH. Dietary and hormonal conditions that stimulate hepatic
fatty acid synthesis decrease the intracellular concentration of NADPH. Under most
conditions, therefore, the rate of production of NADPH is a function of the rate of
utilization of NADPH, rather than the converse.

5. Regulution of the catalytic efJiciency of acetyl-CoA carboxylase

5.1. A key regulutory reaction

Acetyl-CoA carboxylase catalyzes the carboxylation of acetyl-CoA to malonyl-CoA.
Hydrolysis of ATP provides the energy to drive this essentially irreversible reaction.
Acetyl-CoA carboxylase is considered the key regulatory enzyme in the conversion of
citrate to long-chain fatty acids because (1) the concentrations of its substrates and
products are far from thermodynamic equilibrium, indicating that this reaction is
catalyzed by a regulatory enzyme; (2) the maximum velocity of the enzyme, as meas-
ured in cell extracts under optimal conditions, is usually the slowest of all enzymes in
the pathway; (3) the concentration of the product of the enzyme, malonyl-CoA, in-
creases when flux through the pathway increases (if the catalytic efficiency of fatty
acid synthase were regulated, malonyl-CoA concentration would decrease with in-
creased flux); (4) acetyl-CoA carboxylase catalyzes the first committed step in the
pathway, the most appropriate step at which to regulate a metabolic pathway; and (5)
despite extensive analysis, there is little evidence for physiologically relevant regula-
tion of the catalytic efficiency of either ATP:citrate lyase or fatty acid synthase, the
other two enzymes in this section of the pathway.

5.2. Structure and reaction mechanism

Acetyl-CoA carboxylase has two distinct catalytic sites, each of which carries out one
of the following partial reactions:

      (1) E-Biotin + HC0,- + ATP t)E-biotin-CO, + ADP + Pi
      (2) E-Biotin-CO, + acetyl-CoA + E-biotin + malonyl-CoA
      Net: ATP + HC03- + acetyl-CoA + malonyl-CoA + ADP + Pi

In E. coli these reactions require the participation of three different proteins (Chapter
2). In both yeast and animals, the analogous reactions are catalyzed by multifunc-
tional polypeptides. The subunit molecular weight of animal acetyl-CoA carboxylase
is about 265,000. The smallest form of the native enzyme is a dimer (protomer) that
lacks enzyme activity.

5.3. Regulation by citrate

Conversion from the catalytically inactive protomer to the catalytically active poly-
mer is stimulated by tricarboxylate anions. Concomitant with the increase in enzyme
activity, the protomers assemble into filamentous polymers with molecular weights
of up to 1 x 10’. Independent measurements of catalytic activity and rate of polymer-
ization, using stopped-flow techniques, indicate that activation by citrate occurs
much more rapidly than polymerization. Thus, polymerization is not a prerequisite
for increased catalytic efficiency. Activation by citrate is unusual in that it increases
the rate of reaction of bound substrate, V,,,, rather than affecting the apparent affin-
ity, K,, of the enzyme for substrates. Both reactions 1 and 2 in Section 5.2 are stimu-
lated by tricarboxylate anions. Thus, these effectors appear to induce a conforma-
tional change that causes the biotin prosthetic group to be reoriented with respect to
the biotin carboxylase and carboxyl transferase active sites, facilitating efficient ca-
talysis [7].
   The ability of citrate to activate the pace-setting enzyme in the terminal segment of
the pathway for fatty acid synthesis suggests a teleologically satisfying mechanism for
regulation of flux from glucose to fatty acids in intact cells. Under conditions favor-
ing fatty acid synthesis (for example, animals fed a high-carbohydrate diet), citrate
production in the mitochondria is high. EHux of this citrate into the cytosol also
should be high, leading to an increased concentration of citrate in that compartment.
This would supply substrate for fatty acid synthesis via ATP:citrate lyase and stimu-
late the activity of acetyl-CoA carboxylase.
   In isolated avian hepatocytes, the rate of fatty acid synthesis correlates positively
with the concentration of citrate. In some instances, therefore, citrate may be a phys-
iological regulator of acetyl-CoA carboxylase. Several earlier studies with intact rats
and perfused rat liver did not find a correlation, suggesting that the relative contribu-

tion of citrate to regulation of this enzyme may vary greatly under different condi-
tions and in different animals. The varying requirements for citrate were at least par-
tially resolved with the introduction of a method for rapidly isolating acetyl-CoA
carboxylase under conditions that prevented proteolysis and phosphorylation [8].
The degree of dependence of rat liver acetyl-CoA carboxylase activity on citrate var-
ied as a function of the degree of phosphorylation of the enzyme; the enzyme purified
by ‘rapid freezing’ procedures had a high specific activity in the absence of citrate and
lower phosphate content. Enzyme purified with previous ‘standard’ procedures had
a low specific activity without citrate, was substantially activated by citrate and had
a much higher phosphate content. Dephosphorylation of the latter enzyme resulted
in a preparation that resembled the enzyme purified from ‘freeze-clamped’ liver.
These phenomena probably have physiological significance because the physical and
kinetic properties of the enzyme purified from starved liver by the freeze-clamp pro-
cedure were similar to those of enzyme purified from liver of fed rats by the old meth-
ods [9]. Similarly, treatment of normally fed rats with insulin rapidly stimulated the
specific activity of acetyl-CoA carboxylase purified from liver by the freeze-clamping
procedure; the resulting enzyme had a decreased phosphate content and increased
citrate-independent activity. Treatment of rats with glucagon or epinephrine de-
creased specific activity of purified acetyl-CoA carboxylase, increased phosphate
content and decreased citrate-independent activity. Thus, citrate’s primary role may
be when acetyl-CoA carboxylase is in its more phosphorylated state.
   Despite an increased understanding of the allosteric regulation of acetyl-CoA car-
boxylase, the physiological role of citrate in the regulation of acetyl-CoA carboxylase
activity in intact cells and the molecular basis for the effect of citrate on enzyme activ-
ity remain unclear. Kim et al. [ 101suggest that a citrate-metal ion complex binds near
the amino terminus of the enzyme and inhibits an interaction between the amino
terminus and a part of the enzyme near residue 1200, a serine phosphorylated by
protein kinase A. According to this hypothesis, binding of citrate displaces the amino
terminal segment, increasing access of substrates to the active site.

5.4. Regulation by long-chain furty acyl-CoA

Diets containing a high concentration of fat inhibit fatty acid synthesis, suggesting
that long-chain fatty acids and (or) long-chain acyl-CoA derivatives may regulate a
key step in the pathway. Their ability to regulate acetyl-CoA carboxylase activity was
tested with the earliest preparations of partially purified acetyl-CoA carboxylase.
Unesterified fatty acids had no effect on enzyme activity when used at physiological
concentrations. Fatty acyl-CoA derivatives, however, were potent inhibitors of the
enzyme, effective at low concentrations. Unfortunately, fatty acyl-CoAs inhibited the
activity of almost every enzyme that was tested. In most instances, the inhibition was
irreversible and occurred at concentrations higher than the critical micellar concen-
tration of the acyl-CoA. Thus, it was concluded that all the inhibitory effects of long-

chain fatty acyl-CoA derivatives were caused by their detergent properties and were
not of physiological significance. More detailed studies of the actions of fatty acyl-
CoAs on acetyl-CoA carboxylase have led to quite a different conclusion. Inhibition
of the activity of acetyl-CoA carboxylase by fatty acyl-CoA is competitive with ci-
trate, is reversed by citrate or albumin, and has an apparent K, of about 0.2 pM, well
below the critical micellar concentration of fatty acyl-CoAs. When several different
enzyme activities were challenged with submicromolar concentrations of long-chain
acyl-CoA, only acetyl-CoA carboxylase showed significant inhibition. Furthermore,
binding of 1 mol of palmitoyl-CoA to 1 mol of acetyl-CoA carboxylase completely
inhibited enzyme activity. Thus, long-chain fatty acyl-CoAs are specific inhibitors of
acetyl-CoA carboxylase. Subsequent studies have shown that glucokinase and citrate
synthase also are inhibited by specific interactions with long-chain acyl-CoA.
   The question of physiological relevance remains. The total concentration of long-
chain fatty acyl-CoA in liver is inversely correlated with the rate of fatty acid synthe-
sis [ l l ] (Table I). However, virtually all the long-chain fatty acyl-CoA in a cell is
bound to protein. The concentration and affinity constants for cellular proteins that
bind fatty acyl-CoA are largely unknown. With the exception of the enzymes that
metabolize fatty acyl-CoA and fatty acid binding protein, an abundant cytosolic pro-
tein which binds fatty acids and fatty acyl-CoAs (see Chapter 3), the identities and
properties of binding proteins are unknown. Changes in the concentrations of specif-
ic binding proteins would alter the free concentrations of fatty acyl-CoAs. Further
complicating the analysis is a lack of understanding of which species of long-chain
fatty acyl-CoA are present under different conditions. The abilities of different spe-
cies of fatty acyl-CoA to inhibit acetyl-CoA carboxylase are variable. In summary,

The correlation between rate of fatty acid synthesis and concentration of long-chain fatty acyl-CoA in liver
of rats

Treatment                                  Fatty acid synthesis"        Long-chain fatty acyl-CoAb

Control                                    1.o                           83
Starved 3 days                             0.2                          131
Starved 3 days, high-fat diet 3 days       0.2                          144
Starved 3 days, high-carbohydrate diet     2.4                           12
3 days
Diabetic                                   0.1                          105
Diabetic treated with insulin              0.8                           13

Source: Greenbaum, Gumaa, and McLean [I 11.
"Fatty acid synthesis was estimated by measuring the incorporation of [U-'4C]glucoseinto total fatty acids
in liver slices (mmol/g livedh).
bFatty acyl-CoA was measured in the insoluble fraction of perchloric acid extracts of freeze-clamped liver
(nmol/g liver).

indirect evidence suggests that long-chain fatty acyl-CoA may participate in the regu-
latory process in intact cells; formal proof is lacking.

5.5. Regulation by covalent modiJication

The first indication that acetyl-CoA carboxylase might be regulated b y phosphoryla-
tion was the finding of 2 mol of covalently bound phosphate per mole of subunit in
highly purified enzyme from rat liver. Acetyl-CoA carboxylase that has been purified
by rapid procedures that minimize protease and phosphatase activity has at least 6
mol of alkali-labile phosphate per mol of enzyme subunit. Phosphorylation of par-
tially purified acetyl-CoA carboxylase is accompanied by loss of enzyme activity,
dephosphorylation by an increase in activity [I21 (Fig. 3). Purified acetyl-CoA car-
boxylase can be phosphorylated by purified protein kinase A; loss of enzyme activity
is concomitant with phosphorylation. Addition of purified protein phosphatase
reverses both the phosphorylation and the inhibition of enzyme activity. Phosphory-
lation by protein kinase A does not change the apparent K , values for substrates, but
does decrease V,,, by about 50%. The more phosphorylated enzyme is more sensitive
to inhibition by long-chain fatty acyl-CoA. These results establish the potential for
regulation of acetyl-CoA carboxylase by a phosphorylation/dephosphorylation
mechanism. The allosteric regulators, citrate and long-chain fatty acyl-CoA, may act
synergistically with covalent modification in regulation of the activity of this enzyme.

                    10          20                         10           20

Fig. 3. Kinetics of inactivation and phosphorylation of acetyl-CoA carboxylase activity (left panel) and of
activation and dephosphorylation of the enzyme (right panel) in partially purified enzyme preparations. In
the left panel the reactions contained [y-”P]ATP and endogenous protein kinase. In the right panel, ATP
was removed after the phosphorylation was completed and fresh liver extract added as a source of protein
phosphatase. At different time points, the reaction was terminated and acetyl-CoA carboxylase precipitat-
ed by a specific antiserum. The immunoprecipates were assayed for [”PI radioactivity. These curves have
been replotted from the data presented in Carlson and Kim [12]. to which the reader should refer for more

   Glucagon stimulates phosphorylation of acetyl-CoA carboxylase in rat hepato-
cytes and adipocytes and inactivates acetyl-CoA carboxylase activity. The two tryp-
tic peptides of acetyl-CoA carboxylase that are phosphorylated in intact glucagon-
treated cells are the same as those phosphorylated by protein kinase A in vitro. One
of these peptides has three phosphorylation sites, however, and the one phosphory-
lated in glucagon-treated cells is not the one that is phosphorylated by protein kinase
A in vitro. The site phosphorylated in intact cells corresponds to the site phosphory-
lated by an AMP-activated protein kinase [14]. Thus, in the case of acetyl-CoA car-
boxylase, the effect of CAMP may be indirect. In other words, protein kinase A phos-
phorylates and activates a kinase kinase that, in turn, phosphorylates the AMP-acti-
vated kinase which, in turn, phosphorylates and inactivates acetyl-CoA carboxylase.
The kinase kinase that phosphorylates the AMP-activated protein kinase also is acti-
vated by long-chain fatty acyl-CoA. Thus, the inhibition of acetyl-CoA carboxylase
caused by this effector may be amplified by virtue of its direct allosteric effect on
acetyl-CoA carboxylase plus an indirect effect that results in increased phosphory-
lation of that enzyme. These relationships are described diagrammatically in Fig. 4.
   The complete nucleotide and deduced amino acid sequences have been determined
for acetyl-CoA carboxylases from rat and chicken. Rat liver acetyl-CoA carboxylase

               CARBOHYDRATE AND
                                          -                                                 \
FAlTY ACID                            I                                                         I

                                                   AMP-ACTIVATED                            I
                                                   PROTEIN KINASE
                  FAlTY ACYL-CoA          I          (LOW AMIVlty)           I             /
                             I        I
                             \     4                                     /         '   e
                                 L '
                                                                     /       - / M O

Fig. 4. Model for the roles of cAMP and fatty acyl-CoA in the regulation of acetyl-CoA carboxylase.
Acetyl-CoA carboxylase (ACC) is phosphorylated and converted to a less active form by AMP-activated
protein kinase. The latter enzyme is phosphorylated and activated by a kinase kinase. Arrows with dotted
lines indicate either positive (+) or negative (-) allosteric effects of fatty acyl-CoA on the specified enzyme.
Extent of allosteric inhibition is indicated by the number of (-) symbols. The kinase kinase that phosphor-
ylates AMP-activated protein kinase is activated indirectly by cAMP as indicated by multiple arrows; the
question mark indicates that the CAMP-activation of this step has not been demonstrated definitively.

is coded by 7035 nucleotides, which encode 2345 amino acids. Peptides which contain
phosphorylation sites corresponding to the effects discussed above and some less
characterized ones have been isolated and their amino acid sequences determined.
Seven potential phosphorylation sites have been identified in the rat enzyme and lo-
calized in the primary structure [lo]. Six of the seven serines are clustered in the
amino terminal end of the protein at positions 23, 25, 29, 76, 77, and 95; the seventh
is at position 1200. Loss of the amino terminal end of acetyl-CoA carboxylase by
limited proteolysis leads to activation and loss of citrate dependence suggesting that
the amino terminal phosphorylation sites play important roles in regulation of en-
zyme activity. Phosphorylation of the serine at position 1200 by protein kinase A
decreases enzyme activity and decreases activation by citrate.
   A novel mechanism for achieving tissue-specific regulation of acetyl-CoA carboxy-
lase activity was discovered when the nucleotide sequences of different cDNA clones
from a rat mammary gland library were compared. Two types of cDNAs were char-
acterized; one type lacked and one contained 24 bases that code for 8 additional
amino acids located 4 residues upstream of serine 1200. Enzyme synthesized in vitro
from the longer mRNA was not phosphorylated by protein kinase A at serine 1200,
but enzyme synthesized from the shorter mRNA was phosphorylated at this site.
Both mRNAs were detected in liver, adipose tissue and mammary gland, but in dif-
ferent relative amounts. Thus, acetyl-CoA carboxylase in some tissues may be more
susceptible to regulation by mechanisms mediated by protein kinase A than enzyme
in other tissues.
   Expression of other isozymic forms of acetyl-CoA carboxylase also may be in-
volved in achieving tissue-specific regulation of enzyme activity. Two isoforms of the
enzyme have been discovered in rat organs. One isoform has a subunit molecular
weight of 280 kDa, is expressed in cardiac and skeletal muscle, and, in those organs,
is unaffected by starvation and refeeding. The other isoform has a subunit molecular
weight of 265 kDa, is expressed in liver, adipose tissue and mammary gland, and is
regulated by starvation and refeeding. The larger form is also expressed in liver and
regulated by starvation and feeding. Partially purified forms of the two species differ
in their sensitivity to activation by citrate and affinity of acetyl-CoA. The physiologi-
cal significance of these isoforms and their relationship to the multiple mRNAs de-
scribed above, if any, is unknown.
   The activity of acetyl-CoA carboxylase in rat adipocytes and hepatocytes is acti-
vated by insulin within minutes of hormone addition. The mechanism for this effect
is unclear. Since glucagon inhibits acetyl-CoA carboxylase activity by promoting
phosphorylation of the enzyme, it was postulated initially that insulin might activate
via a dephosphorylation mechanism, However, in intact adipocytes insulin stimulates
the phosphorylation of acetyl-CoA carboxylase within tryptic peptides distinct from
those phosphorylated by protein kinase A or the AMP-activated protein kinase. The
phosphopeptide that accounts for most of the increase in total phosphorylation of
the enzyme is identical to that containing the sites phosphorylated by casein kinase-2

in vitro. Casein kinase-2 activity is increased by insulin in several cell-types. Thus,
changes in the activity of casein kinase-2 or a kinase with a similar substrate speci-
ficity may cause the increased phosphorylation of acetyl-CoA carboxylase in re-
sponse to insulin. But, phosphorylation of acetyl-CoA carboxylase by casein kinase-2
has no effect on the kinetic properties of the enzyme. Phosphorylation by casein
kinase-2 accelerates the rate of dephosphorylation of acetyl-CoA carboxylase at sites
that are sensitive to protein kinase A. It follows that insulin-dependent phosphoryla-
tion might activate acetyl-CoA carboxylase by reversing an inhibitory phosphoryla-
tion. In intact adipocytes, however, the phosphorylation of tryptic peptides contain-
ing sites sensitive to protein kinase A is unchanged 15 min after insulin administra-
tion, even though activation of acetyl-CoA carboxylase is evident by this time. The
significance of the insulin-dependent phosphorylation of acetyl-CoA carboxylase
remains uncertain.
   There is additional evidence that changes in phosphorylation state do not mediate
the activation of acetyl-CoA carboxylase caused by insulin in rat adipocytes. First,
when extracts of insulin-treated cells are incubated with the purified catalytic subunit
of protein phosphatase-2A, the insulin-sensitive sites are dephosphorylated without
reversing the effects of insulin on enzyme activity. Second, the effect of insulin on
acetyl-CoA carboxylase activity, but not its effect on phosphorylation, is lost on puri-
fication of the enzyme by avidin-Sepharose chromatography. Insulin’s effect on en-
zyme activity also is lost if crude extracts are incubated at high ionic strength fol-
lowed rapidly by gel filtration to separate the enzyme from low molecular weight
effectors. The activation of acetyl-CoA carboxylase by insulin may be mediated by a
low molecular weight effector which dissociates from the enzyme at high ionic
   Two low molecular weight activators of acetyl-CoA carboxylase are released from
rat liver membranes in response to insulin. These activators have an inositol phos-
phate-glycan structure (Chapter 7) and are produced by an insulin-sensitive hydro-
lysis of a glycosyl-phosphatidylinositol precursor in the plasma membrane. Hydro-
lysis is catalyzed by a specific phospholipase C [131. These phospho-oligosaccharides
also modulate the activity of other insulin-sensitive enzymes and may function as
second messengers for some actions of insulin.
   Allosteric mechanisms and phosphorylation-dephosphorylation mechanisms
probably play complementary roles in regulating the catalytic activity of acetyl-CoA
carboxylase. Glucagon, for example, stimulates phosphorylation of the hepatic en-
zyme at an inhibitory site. Glucagon also stimulates the hormone-sensitive lipase in
adipose tissue causing an increase in the rate of release of unesterified fatty acids to
the plasma and, consequently, an increase in unesterified fatty acid concentration in
plasma and other tissues including liver. In liver, the concentration of long-chain
fatty-acyl CoA varies in equal proportion with that of unesterified long-chain fatty
acids. The increased phosphorylation of acetyl-CoA carboxylase caused by glucagon
increases sensitivity of the enzyme to inhibition by fatty acyl-CoA. As described

previously, glucagon also inhibits the flux of carbon from glucose to pyruvate. De-
creased production of extramitochondrial citrate is a consequence of the decreased
production of pyruvate via glycolysis. Increased phosphorylation of acetyl-CoA car-
boxylase at its inhibitory site decreases sensitivity of this enzyme to its activator,
citrate. The combined result of the three actions (direct inhibition of activity due to
phosphorylation, elevation of the long-chain fatty acyl-CoA level, and diminution of
the citrate level) results in an amplification of the inhibitory response that neither
allosteric mechanisms nor covalent modification alone could have achieved. Regula-
tion of fatty acid synthesis via phosphorylation of acetyl-CoA carboxylase was
reviewed recently by Hardie [ 141.
   Thus far, we have discussed regulation by allosteric factors and covalent modifica-
tion, mechanisms that alter the catalytic efficiency of enzyme molecules. This type of
regulation accounts for the minute-to-minute changes in enzyme activity that occur
under different hormonal and nutritional conditions. The concentration of acetyl-
CoA carboxylase molecules also is regulated by hormones and nutritional status. The
concentration of the lipogenic enzymes are regulated coordinately, so this kind of
regulation will be analyzed after a discussion of the physical and enzymatic proper-
ties of fatty acid synthase.

6. Fatty acid synthase

The synthesis of long-chain fatty acids from acetyl-CoA and malonyl-CoA involves
a sequence of six reactions for each two-carbon addition; the sequence is repeated
several times to produce a long-chain fatty acid (Fig. 4). Although the structural or-
ganization of the process varies greatly in different organisms, the enzymatic mecha-
nisms are very similar. In E. coli and in the plastids of green plants, enzymes cata-
lyzing the individual reactions are discrete monofunctional proteins that can be sep-
arated and analyzed individually (Chapter 2). In yeast, synthesis of fatty acids from
acetyl-CoA and malonyl-CoA is catalyzed by a fatty acid synthase complex that con-
sists of two multifunctional polypeptides, each coded by a different gene. Mam-
malian and avian fatty acid synthases are also multifunctional polypeptides, but all
enzyme activities are localized on a single polypeptide chain encoded by a single gene.
This section will concentrate on the structure and functional organization of this fas-
cinating multifunctional polypeptide.

6.1. Animal fatty acid synthase: the component reactions

The enzymatic reactions for the synthesis of palmitate catalyzed by animal fatty acid
synthase are listed in Fig. 5. In addition to the 31 listed reactions, each cycle requires
the transfer of malonyl-CoA to the 4'-phosphopantetheine residue of the enzyme,
making a total of 37 enzymatic reactions for the synthesis of each molecule of palmi-
I. Acyl transferase

2. Acvf transferase
                          + HS-pan-E         -         0
                                                    CH&--Gpan-E                  + CoA

   -OCCH&-S-CoA               + HS-pan-€            -    -OCCH2C-S-pan-E                      + CoA
3. pKetoacyl synthase
            0                                                      0
            1                                                      II
   (a) CH3C-S-pan-E             + HS-cys--E-                  CH&-S-Cys-E                    + HS-pan-€
            0                       0 0
            I                          II      II
   (b) CH$-S-cyS-E              + -OCCHZC-S-pan-E
                                                              0             0
                                                              1             II
                                                      CH3CCH2-C-S-pan-E                         + HS-cys-E        + COZ
4. pKetoacy1 reductase
       0    0                                                            0
       II       II                                                   H   II
   CH,CCH&-S-pan--E               + NADPH + H+-                   CH&CHpC-S-pan-E                       + NADP+

5 pHydroxyacy1 dehydrase

   CH&CH,C-S-pan-E               -             H
                                                                                 + H20


6. Enoyl reductase
          0                                                                        0
                                  + NADPH + H+           -                         II
                                                                  CH3CH2CH2C-S-pan--E                    f   NADP'

7. p-Ketoacyl synthase
(a) CH,CH2CHzC-S-pan--E                + HS-cys-E             -    CH3CH,CH2C-S-cys-E
                                                                                                          + HS-pan-€
(b) CH3CH2CH~C-S-cys-E
                                       + -OCCH2C!--S-pan-E                       -                 0
                                                                                           + HS-vS-E

                                                                                                       + CO2

8-10, Repeat reactions 4. 5. and 6. forming Hexanoyl-pan-E

11-30. Repeat reaction 3.4.5, and 6 five times, with the molecule growing by 2 carbons with each
       cycle to produce palmitoyl-pan-E

31. Thioesterase
    Palmltoyl-pan-E          + H20            Palmitate + HS-pan-€

Fig. 5 . The component reactions of animal fatty acid synthase. The abbreviations HS-cys and HS-pan
indicate cysteinyl residues and 4'-phosphopantetheine groups, respectively.

tate. In the non-aggregated fatty acid synthase of E. coli, the growing acyl chain is
attached to a small peptide (acyl carrier protein) via a 4'-phosphopantetheine residue.
The eucaryotic equivalent of the bacterial acyl carrier protein is part of the linear
structure of the multifunctional fatty acid synthase polypeptide. In other respects, the
reactions of the animal fatty acid synthases are like those of E. coli (Chapter 2).

6.2. Animal fatty acid synthuse: the subunits are identical

Despite the successful purification and characterization of the component enzymes of
E. coli fatty acid synthase, the animal enzyme proved highly resistant to dissociation.
Because the bacterial pathway for conversion of malonyl-CoA to fatty acids contain-
ed several independent gene products, the animal enzyme was generally assumed to
be a multienzyme complex highly resistant to dissociation rather than a multifunc-
tional polypeptide. On the contrary, however, if protease activity was inhibited dur-
ing purification and analysis, the totally denatured complex had only a single compo-
nent. Based on the nucleotide sequences of cloned DNAs complementary to fatty
acid synthase mRNA, the rat and chicken enzymes contain 2505 and 2512 amino
acids, respectively, and molecular weights of about 273 kDa. The native enzyme has
a molecular weight of about 500 kDa, indicating a dimeric structure.
   The two subunits of avian and mammalian fatty acid synthases are probably iden-
tical. Based on polyacrylamide gel electrophoresis in the presence of sodium dodecyl
sulfate, the two subunits are identical in size. They also have the same shape and
charge. Free N-terminal amino acid residues cannot be detected in the purified en-
zyme, indicating that both subunit peptides have blocked N-termini. Each fatty acid
synthase dimer contains 1 . 4 1 . 8 mol of 4'-phosphopantetheine per mole of dimer.
Each dimer contains two sites each for the enzymatic activities of thioesterase, keto-
reductase, enoyl reductase, and P-ketoacyl synthase. These stoichiometries are based
on measurement of the extent of reaction of inhibitors of specific subreactions of
fatty acid synthase. Peptides containing three of the enzyme's active sites have been
purified from the homogeneous dimeric enzyme after cleavage with proteases. Thio-
esterase can be removed from the native enzyme by mild treatment with several dif-
ferent proteases (Fig. 6). Active thioesterases with molecular weights of about 35 kDa
have been purified from several fatty acid synthases. A single N-terminal amino acid
sequence, at least eight amino acids in length, has been determined for thioesterase
purified from protease-treated, native dimeric fatty acid synthase of chicken liver. A
chymotryptic peptide containing the active serine of thioesterase was purified from
the dimeric fatty acid synthase of goose uropygial gland; it also contained a single
sequence of eight amino acids. Similarly, peptides containing the 4'-phosphopante-
theine prosthetic group of the acyl carrier peptide region and the NADPH binding
site of the enoyl reductase region had single amino acid sequences of 64 and 1 1 amino
acids, respectively. Despite their purification from native enzymes containing both
subunits, each of the foregoing peptides contained only one type of amino acid se-


                  AcCoA    en'ry


                                      Substrate    ACCOA

Fig. 6 . Proposed functional map of the chicken fatty acid synthase. The sketch is based on a detailed
analysis described in [15.16]. The abbreviations for partial activities are AT, acetyl transferase; MT, malo-
nyl transferase; KS, b-ketoacyl synthase; KR, /%ketodcylreductase; DH, dehydrase; ER, enoyl reductase;
TE, thioesterase; ACP, acyl carrier peptide; Ac, acetyl; Mal, malonyl. The wavy lines represent the 4'-
phosphopantetheine prosthetic group. Taken from Tsukamoto et al. [ 161 with permission of the authors.

quence. If the subunits were not identical, some of the four peptides purified from
dimeric enzyme should have contained more than one amino acid sequence. There-
fore, the two subunits probably have identical sequences throughout their entire

6.3. Animal fatty acid synthase: structural organization

Multifunctional proteins are organized into globular domains. The component cata-
lytic activities and regulatory sites are located on different domains. The domains are
connected to one another by polypeptide bridges that are susceptible to proteolytic
attack. Wakil and co-workers analyzed the fragmentation pattern of chicken liver
fatty acid synthase using several different proteases [ 151 and identified three principal
domains (Fig. 6). The locations of the various functional centers of the enzyme were
determined by analyzing enzyme activity of fragments and by localizing labeled, site-
specific reagents to specific fragments [I 61 (Fig. 6). A similar map has been developed
for the fatty acid synthase purified from rabbit mammary gland [17]. (The reader is

referred to [15, 161 for a detailed description of the analysis of chicken liver fatty acid
   The amino acid sequences deduced from the nucleotide sequences of the cloned
cDNAs has permitted tentative assignment of the linear order of the activities along
the peptide chain. Assignments were based on sequence similarities between regions
of the deduced sequence and peptide fragments that reacted with site-specific rea-
gents or were based on sequence similarities to catalytic sites in other enzymes of
similar function. The P-ketoacyl synthase is located near the amino terminus, fol-
lowed by the acylimalonyl-CoA transferase, a relatively long stretch of residues with-
out known function, dehydrase, enoyl reductase, P-ketoacyl reductase, acyl carrier
protein and finally, the thioesterase at the carboxy terminus [18-211. This order
agrees well with that predicted from the protease studies.
   The head-to-tail organization of the subunits (Fig. 6) is based on cross-linking
studies and explains why synthesis of palmitate is blocked when the identical subunits
of fatty acid synthase are dissociated. Upon dissociation, the P-ketoacyl synthase
reaction is disrupted because this reaction requires the participation of the 4'-phos-
phopantetheine prosthetic group of the opposite subunit. The other activities that use
the 4'phosphopantetheine prosthetic group are in the same subunit as the pantetheine
group that they use. Further work is required to substantiate the model described
here, but this hypothesis provides a satisfying synthesis of the known experimental

6.4. Comparison of yeast and animalfatty acid synthases

The structural organization of the yeast enzyme is intermediate between that of E.
coli and animals. The six catalytic sites and the acyl carrier function are present on
two different multifunctional polypeptides, a and P. Elegant genetic studies [22] es-
tablished that the acyl carrier function, ketoacyl synthase activity, and ketoacyl re-
ductase are on the a subunit. The acetyl transferase, malonyl transferase, dehydrase,
and enoyl reductase activities are on the /3 subunit. These assignments were subse-
quently confirmed based on the amino acid sequences deduced from the cloned a and
P genes. The native molecular weight of the yeast enzyme is about 2.3 X lo6. The a-
and p- subunits have molecular weights of about 200 kDa each, suggesting an 0$6
structure for the native enzyme. Electron microsopic analysis of the purifed protein
also supports an a& structure [23].
   In addition to the organizational differences between the yeast and animal en-
zymes, there are several functional differences. The yeast enzyme has separate acetyl
and malonyl transferases, whereas the animal enzyme has a single acyl transferase for
both substrates. The product of the yeast enzyme is palmitoyl-CoA, whereas that of
the animal enzyme is free palmitate. Finally, the yeast enoyl reductase component
requires FMNHz and NADPH as cofactors, but the animal enzyme requires only
  The multifunctional character and the domain structure of animal fatty acid syn-
thases suggests that the animal enzymes have evolved from a set of independent genes
that coded for monofunctional enzymes, in part at least, through a series of gene
fusion events. If so, does the yeast enzyme represent a step in the evolution of the
animal enzyme, or does it represent the result of an independent series of gene fusion
events? According to the protease mapping studies and the deduced amino acid se-
quences, the linear order of the individual enzymatic activities is different for the
animal and yeast enzymes. These differences and the other functional and structural
differences suggest that these two fatty acid synthases have evolved by independent
gene fusion events.

7. Regulation of enzyme concentration

Rapid changes in the flux of carbon from glucose to long-chain fatty acids are initiat-
ed by a combination of changes in delivery of substrates, concentration of allosteric
effectors, and degree of phosphorylation of enzymes, as discussed previously. In con-
trast, increases or decreases in fatty acid synthesis caused by changes in the total
activities (that is, amounts) of the lipogenic enzymes occur over periods of hours or
days rather than the seconds or minutes typical of changes in catalytic efficiency
caused by allosteric or phosphorylation mechanisms. Total activity is defined here as
the maximum activity that can be demonstrated in cell-free extracts under optimal
assay conditions with respect to substrates, cofactors, and effectors. Thus, the total
activities of glucose-6-phosphate dehydrogenase, 6-phosphogluconate dehydrogen-
ase, malic enzyme, ATP-citrate lyase. acetyl-CoA carboxylase, and fatty acid syn-
thase are high in the livers of well-fed animals, especially if the diet is high in carbo-
hydrate and low in fat. These activities are decreased by starvation, high-fatAow-
carbohydrate diets, or diabetes. The slowness of the changes and their manifestation
under optimal assay conditions suggests that they are due to changes in enzyme con-
   Regulation of the catalytic efficiency of a constant quantity of enzyme can be distin-
guished experimentally from regulation of the number of enzyme molecules per cell by
using immunological techniques. Each of the lipogenic enzymes listed was purified to
homogeneity, and the homogeneous enzyme used to raise monospecific antisera. With
a few quantitatively minor exceptions, immunological analyses for each of these lipo-
genic enzymes indicates that dietary and hormonal control of the total activities of all
the lipogenic enzymes involves regulation of enzyme concentration.

7.1. Messenger R N A levels regulate enzyme synthesis rutes

The concentration of an enzyme protein is a function of both its rate of synthesis and
its rate of degradation. Without exception, changes in the concentrations of lipogenic

enzymes have been associated with quantitatively comparable changes in the relative
synthesis rates for those enzymes (see [24] for a more detailed discussion). Synthesis
of a specific enzyme can be regulated by controlling (a) the efficiency with which a
constant quantity of mRNA is translated into protein or (b) the relative abundance
of its mRNA. The analysis of the regulation of expression of specific mRNAs re-
quired the development of techniques and the isolation of reagents for measuring
mRNA concentration. Cloned complementary DNAs (cDNAs) are such reagents
and have been isolated for ATP:citrate lyase, malic enzyme, glucose-6-phosphate de-
hydrogenase, 6-phosphogluconate dehydrogenase, acetyl-CoA carboxylase and fatty
acid synthase. These cloned cDNAs have been used in hybridization assays to deter-
mine the abundance of specific mRNAs in crude mixtures of total cellular RNA. In
these assays, total RNA is extracted from liver or other organs, separated by size by
agarose gel electrophoresis, transferred to nitrocellulose membranes (or similar sup-
ports), and fixed to the membrane with heat or UV irradiation. Cloned cDNAs are
labeled with 32Pand hybridized to the membrane-bound RNAs. After hybridization,
the membranes are washed to remove all but the specifically bound 32P-DNA,      dried,
and exposed to X-ray film. The intensity of the autoradiographic signal for a specific
mRNA is proportional to the amount of that mRNA in the original RNA mixture.
Based on such Northern-blot analyses, hepatic concentrations of the mRNAs for the
lipogenic enzymes were found to correlate with enzyme synthesis rates in starved
versus fed animals (Fig. 7), in untreated and insulin-treated diabetic animals and in
animals treated with thyroid hormone. Thus, synthesis rates of the lipogenic enzymes
are controlled by regulating the concentrations of their respective mRNAs; regula-
tion is pretranslational.

7.2. Transcription is usually the regulated step

Accumulation of specific mRNAs is regulated at the level of gene transcription, proc-
essing of nuclear transcripts, or stability of nuclear or mature mRNA. At present,
there are two methods for evaluating the point at which regulation is exerted. The
first involves a Northern-blot analysis of nuclear precursors to specific mRNA. Ordi-
narily, the concentration of nuclear RNA precursors is very low. To assure detecta-
bility, nuclear RNA is isolated and analyzed. This procedure usually eliminates ma-
ture cytoplasmic RNA, but to be sure that cytoplasmic RNAs are not measured in
the assay, the best probe DNAs are those from unique intronic regions of the gene in
question. If abundance of nuclear intermediates correlates positively with abundance
of the mature cytoplasmic mRNA, transcription is probably the regulated step. If
processing or degradation of the mature mRNA is regulated, concentration of the
nuclear precursors should be little affected. For avian malic enzyme, such data sug-
gest that starvation and feeding regulate transcription (Fig. 8). For rat malic enzyme,
diet, thyroid hormone and insulin may regulate gene expression at both transcriptio-

Fig. 7. The effects of feeding in vivo on the level of malic enzyme mRNA. RNA was extracted from the
liver of 2-day-old chicks that were starved from hatching or fed for 24 h as indicated on the figure. Total
polyadenylated RNA was separated by size by electrophoresis in an agarose gel, blot-transferred to a
nitrocellulose membrane, and hybridized to "P-labeled, single-stranded malic enzyme cDNA. After hy-
bridization, the membranes were washed to remove non-specific DNA and subjected to autoradiography.
Each lane contained 20pg of RNA. Taken from Goodridge et al. [25] with permission of the Journal of

nal and post-transcriptional steps [24]. Other lipogenic genes have not been evaluated
with this technique.
   The transcription 'run-on' assay is a second method to evaluate steps involved in
regulation of mRNA abundance. Elongating RNA polymerase I1 molecules remain
bound to DNA when nuclei are isolated. When the isolated nuclei are incubated in
vitro with the appropriate nucleoside triphosphates, one or more of which is labeled,
nascent RNA chains elongate at about the same rate and for about the same distance
irrespective of the gene or physiological state of the cell of origin. These labeled run-
on transcripts are purified and hybridized to specific cDNAs or genomic DNAs that
have been fixed to membranes. The intensity of the resulting autoradiographic signal

Fig. 8. Nuclear precursors for malic enyzme mRNA in liver from starved and refed chicks. Nuclear RNA
was isolated from 12-14-day chicks starved for 48 h and then either starved for an additional 6 h or refed
for 6 h. Nuclear RNA (25 pg) was separated by size on a 0.5% agarose, 2.2 M formaldehyde gel and blotted
onto a ‘Genescreen’ membrane. Identical strips of membrane were hybridized with two intron probes from
the malic enzyme gene and a cDNA probe for glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
After hybridization the membranes were washed and subjected to autoradiography. ’*P-labeled probes
were prepared by nick-translation. GAPDH RNA was a control for selectivity; the abundance of its
mRNA is unaffected by starvation and refeeding. Taken from Ma et al. 1261with permission of The Jour-
nal of Biological Chemistry.

is proportional to the number of RNA polymerase molecules that were engaged in
transcription along that part of the gene corresponding to the probe DNA. If initia-
tion is the rate limiting step, the number of polymerase molecules engaged in tran-
scription per unit length of probe DNA should be directly proportional to the rate of
initiation of the corresponding gene. For avian malic enzyme and fatty acid synthase,
such experiments have confirmed transcription to be the primary, perhaps exclusive,
regulated step. In other systems, the results are less clear. Evidence for post-tran-
scriptional regulation most often takes the form of a lack of correlation between
transcription as measured by the run-on assay and mRNA abundance as measured
by hybridization analysis. As noted above, the transcription run-on assay is an indi-
rect measure of transcription initiation and is subject to several artifacts. In the
absence of positive evidence for post-transcriptional regulation, results from run-on
assays must be viewed with skepticism unless exceptional care has been taken (1) to

insure that transcription is catalyzed by RNA polymerase 11, (2) to minimize
transcription from the opposite strand, repetitive elements and GC-rich regions
(bona fide parts of the gene or added to probe DNAs during cloning), and (3) to
eliminate the possibility of internal re-initiation.

7.3. Regulation in cells in culture

Nutritional regulation of lipogenesis and its associated enzymes occurs in two or-
gans, liver and adipose tissue. There are complex interactions of liver and adipose
tissue with other organs in the body via the nervous system and numerous hormones
and fuels in the blood. These interactions make it difficult to identify extracellular
regulatory molecules and to analyze the intracellular molecular mechanisms that reg-
ulate fatty acid synthesis. Two useful model systems have been developed.

7.3.1. 3T3-Ll cells - a pre-adipocyte cell line
The 3T3 cell line is derived from mouse embryo fibroblasts that survived senescence
in culture. Foci of cells containing fat droplets appear when 3T3 cells are held at
confluence for more than a week. The 3T3-LI subline was derived from cells in one
of these foci of adipocyte-like cells. In the 3T3-LI cell line, almost all cells are con-
verted into adipocytes if the cells are held at confluence for a sufficiently long time.
Treatment of confluent 3T3-LI cells with dexamethasone (a synthetic glucocorticoid
hormone) and isobutylmethyl xanthine (an inhibitor of phosphodiesterase) or with
insulin-like growth factor I causes a rapid and synchronous differentiation of the
pre-adipocytes into adipocytes. Based on both morphological and biochemical crite-
ria, the differentiated 3T3-Ll cells are remarkably similar to normal adipocytes. In
the course of differentiation, many proteins, including the lipogenic enzymes, accum-
ulate to levels characteristic of mouse adipose tissue. The increases in the levels of the
lipogenic enzymes that occur when 3T3-Ll pre-adipocytes are converted to adi-
pocytes are due to increased rates of synthesis of these enzymes in the differentiating
cells. For those mRNAs that have been measured during differentiation, increases in
enzyme synthesis are accompanied by similar increases in abundance of the corre-
sponding mRNA. For some adipocyte-specificproteins, regulation is primarily tran-
scriptional. For the lipogenic enzymes, however, the relative roles of transcriptional
and post-transcriptional mechanisms have not been clarified.
   The process of conversion of fibroblast-like pre-adipocytes into adipocytes prob-
ably mimics the terminal differentiation of adipocytes that occurs in vivo. At the
molecular level, however, little is known about the intracellular events that initiate
and maintain this differentiation process. Mechanisms involved in the commitment
of cells to a particular differentiation path are unlikely to be the same as those that
modulate metabolic function in terminally differentiated cells. 3T3-Ll adipocytes do
not exhibit all of the regulatory properties exhibited by intact adipose tissue. For
example, the activities of the lipogenic enzymes in mature adipocytes of intact rats are

stimulated by insulin and thyroid hormone and inhibited by treatments that increase
the intracellular concentration of CAMP. In mature 3T3-Ll adipocytes, this type of
regulation is either lacking entirely or small in magnitude compared with the in vivo
phenomena. The basis for these differences between 3T3-LI adipocytes and intact
adipose tissue is unknown.

7.3.2. Hepatocytes in maintenance culture
Cell lines derived from minimum-deviation hepatomas are not good models for regu-
lation of fatty acid synthesis in the liver because neither the tumors in vivo nor the
derived cell lines have the capacity to modulate lipogenesis or the activities of the
lipogenic enzymes in a manner that is typical of normal liver. However, maintenance
cultures of hepatocytes from both rats and chickens have been very useful for analy-
sis of the regulation of the concentration of the lipogenic enzymes. In avian hepato-
cytes maintained in a chemically defined medium, insulin plus triiodothyronine
causes 33- and 8-fold increases in the activities of malic enzyme and fatty acid
synthase, respectively (Table 11). The addition of glucagon (or CAMP) blocks the
increases caused by insulin and triiodothyronine. Enzyme activity is regulated by
controlling enzyme concentration; enzyme concentration is regulated by controlling
enzyme synthesis (Table 11); enzyme synthesis is regulated by controlling mRNA
concentration; and mRNA concentration is regulated primarily by controlling the
rate of transcription (Fig. 9).
   The DNA sequence elements that specify which gene will respond with increased
transcription initiation to which intracellular signal are found at the 5' ends of genes,
usually 5' to the start site for transcription. Analyses designed to identify and charac-

Effects of triiodothyronine and glucagon on the activities, synthesis rates, and mRNA abundances of malic
enzyme and fatty acid synthase in avian hepatocytes in maintenance culture

Measurement                     Control            Triiodothyronine         Triiodothyronine
                                                                            plus glucagon

Malic enzyme
  Activity                        3                100
  Relative Synthesis              1                100
  mRNA abundance                  3                100
Fatty acid synthase
  Activity                       12                100                      21
  Relative synthesis              6                100                      16
  mRNA abundance                  8                100                      35

Sources: Fischer and Goodridge [27]; Goodridge and Adelman [28].
The results are expressed as a percentage of the values in cells incubated with triiodothyronine. All incuba-
tions contained insulin.

Fig. 9. The effect of triiodothyronine and cAMP on transcription of the malic enzyme gene. Chick embryo
hepatocytes were isolated and maintained in medium containing insulin [28]. After 42 h of incubation,
triiodothyronine (T3) was added to the medium; 24 h later, dibutyryl cAMP (Bt2 CAMP) was added, and
the cells were harvested at 0.5, I , and 3 h. Nuclei were isolated and nascent RNA chains extended in vitro
with [32P]UTP.   The labeled transcripts were isolated and hybridized to the indicated genomic or comple-
mentary DNAs which had been fixed to 'Genescreen' membranes. After hybridization, the membranes
were washed and subjected to autoradiography. M.E.-2.6, M.E.-4.8-5'; and M.E.-4.8-3' are unique frag-
ments of genomic DNA from the malic enzyme gene. M13mp18Rf and pUC19 are vector DNAs included
as negative controls for specificity of the hybridization. P-Actin and GAD (glyceraldehyde-3-phosphate
dehydrogenase) are cDNA probes, and are controls for selectivity; neither triiodothyronine nor cAMP
affect their transcription rates. C, control cells supplemented with insulin alone for the entire incubation
period. Taken from [29] with permission of The Journal of Biological Chemistry.

terize those sequences and the cellular proteins that bind thereto are just beginning.
The rat genes for acetyl-CoA carboxylase and malic enzyme and the avian genes for
fatty acid synthase and malic enzyme have been isolated and their 5' ends partially
characterized. The 5'-flanking DNA of these genes has been ligated to the structural
gene of a bacterial enzyme, chloramphenicol acetyltransferase and transferred into
cells in culture. For malic enzyme and fatty acid synthase, addition of triiodothyro-
nine to the culture medium causes an increase in chloramphenicol acetyltransferase
activity. Thus, the 5' ends of these genes contain cis-acting DNA elements that desig-
nate these genes to respond to triiodothyronine.
   Hormone response elements have not been detected in the 5' flanking DNA of the
acetyl-CoA carboxylase gene. What has been discovered, however, is heterogeneity
at the 5' ends of the mRNA for acetyl-CoA carboxylase. Two different promoters

have been mapped, each specific for certain cell-types. Alternative RNA processing
accounts for a total of five kinds of acetyl-CoA carboxylase mRNA [30]. The regula-
tory roles of the different mRNAs, each of which encodes the same acetyl-CoA car-
boxylase protein, are not known.

8. Future directions

There are four stages in the analysis of the regulation of a metabolic pathway. They
are ( 1 ) identification of regulatory enzymes; (2) analysis of the physical, kinetic, and
regulatory properties of those enzymes; ( 3 ) development of a hypothesis to explain
the observed regulation of the pathway based on the physical, kinetic, and regulatory
properties; and (4) testing and modification of that hypothesis as necessary. For eu-
caryotic fatty acid synthesis, stages 1-3 are rather complete. Thus, the hypotheses
outlined in this chapter must now be tested and refined. In addition, there is a fifth
stage, analysis of the mechanisms by which a!terations in the structure of regulatory
enzymes result in functional regulation. The new technologies of genetic engineering
will play crucial roles in the analyses at stages 4 and 5 and in analyzing aspects of the
regulation of enzyme activity that remain poorly understood (stage 2). Actively stud-
ied areas will be (1) regulation of gene expression, (2) regulation of the catalytic ef-
ficiency of acetyl-CoA carboxylase, and ( 3 ) structure-function relationships in the
multifunctional fatty acid synthase polypeptide.
   The molecular mechanisms by which gene expression is regulated is an area of
intense research. The genes for the lipogenic enzymes are not unique, but features of
their expression make them especially interesting. Regulation of these genes by triio-
dothyronine, insulin, and CAMPmake them part of a small set of identified proteins
regulated by these agents. Other features of interest include coordinate regulation of
expression and tissue-specific expression and regulation. Several research groups
have already begun analyses designed to identify cis-acting sequence elements and to
identify, purify and characterize trans-acting factors that bind to cis-acting DNA ele-
ments and regulate transcription. Such studies should make important contributions
to our understanding of the structure and function of the lipogenic genes, and of
hormonal and nutritional regulation of gene expression.
   The physiological significance of phosphorylation and allosteric effectors in the
regulation of acetyl-CoA carboxylase activity will continue to receive considerable
attention. So too will the molecular mechanisms whereby phosphorylation and allo-
steric effectors regulate catalytic efficiency of the enzyme. The entire amino acid se-
quences of both rat and chicken acetyl-CoA carboxylase have been determined and
published. Site-directed mutagenesis coupled with over-expression in bacterial or an-
imal cells will permit direct analysis of the importance of various residues to the reg-
ulation of enzyme activity by allosteric and covalent modification mechanisms. Us-
ing homologous recombination, it is possible to mutate selectively essentially any

endogenous gene of cells in culture or intact mice [31]. Re-expression of mutant ace-
tyl-CoA carboxylases altered by site-directed mutagenesis should permit analysis of
both allosteric and covalent modification mechanisms under physiological condi-
tions in intact cells and animals.
   Genomic clones for yeast fatty acid synthase and genomic and cDNA clones for
avian and mammalian fatty acid synthases have been isolated and characterized.
Analyses of the relationships between enzymatic function and domain structures of
fatty acid synthase are now possible.

 1. Goodridge, A.G. (1973) Regulation of fatty acid synthesis in isolated hepatocytes prepared from the
    livers of neonatal chicks. J. Biol. Chem. 248, 1924-1931.
 2. Rolleston, F.S. (1972) A theoretical background to the use of measured concentrations of intermedi-
    ates in study of the control of intermediary metabolism. Curr. Top. Cell. Regul. 5,47775.
 3. El-Maghrabi, M.R., Claus, T.H., McGrane, M.M. and Pilkis, S.J. (1982) Influence of phosphoryla-
    tion on the interaction of effectors with liver pyruvate kinase. J. Biol. Chem. 257,23340.
 4. Spence, J.T. (1983) Levels of translatable mRNA coding for rat liver glucokinase. J. Biol. Chem. 258,
 5. Hers, H.-G. and Van Schaftingen, E. (1982) Fructose 2,6-bisphosphate 2 years after its discovery.
    Biochem. J. 206, 1-12.
 6. Denton, R.M. and Halestrap, A.P. (1979) Regulation of pyruvate metabolism in mammalian tissues.
    Essays Biochem. 15, 37-77.
 7. Lane, M.D., Moss J. and Polakis, S.E. (1974) Acetyl coenzyme A carboxylase. Curr. Top. Cell. Regul.
    8, 139-95.
 8. Thampy, K.G. and Wakil, S.J. (1989) Regulation of acetyl-coenzyme A carboxylase. 11. Effect of fast-
    ing and refeeding on the activity, phosphate content, and aggregation state of the enzyme. J. Biol.
    Chem. 263,6454-58.
 9. Thampy, K.G. and Wakil, S.J. (1989) Regulation of acetyl-coenzyme A carboxylase. I. Purification
    and properties of two forms of acetyl-coenzyme A carboxylase from rat liver. J. Biol. Chem. 263,
10. Kim, K.-H., Lopez-Casillas, F., Bai, D.H., Luo, X. and Pape, M.E. (1989) Role of reversible phos-
    phorylation of acetyl-CoA carboxylase in long-chain fatty acid synthesis. FASEB J. 3, 225&56.
11. Greenbaum, A.L., Gumaa, K.A. and McLean, P. (1971) The distribution of hepatic metabolites and
    the control of the pathways of carbohydrate metabolism in animals of different dietary and hormonal
    status. Arch. Biochem. Biophys. 143. 617-63.
12. Carlson, C.A. and Kim, K.-H. (1974) Regulation of hepatic acetyl coenzyme A carboxylase by phos-
    phorylation and dephosphorylation. Arch. Biochem. Biophys. 164,478-89.
13. Saltiel, A.R., Doble, A., Jacobs, S. and Cuatrecasas, P. (1983) Putative mediators of insulin action
    regulate hepatic acetyl CoA carboxylase activity. Biochem. Biophys. Res. Commun. 110, 789-795.
14. Hardie, D.G. (1989) Regulation of fatty acid synthesis via phosphorylation of acetyl-CoA carboxy-
    lase. Prog. Lipid Res. 28, 117-146.
15. Mattick, J.S., Tsukamoto, Y., Nickless J. and Wakil, S.J. (1983) The architecture of the animal fatty
    acid synthetase. I. Proteolytic dissection and peptide mapping. J. Biol. Chem. 258, 15291-99.
16. Tsukamoto, Y.,   Wong, H., Mattick, J.S., and Wakil, S.J. (1983) The architecture of the animal fatty
    acid synthetase complex IV. Mapping of active centers and model for the mechanism of action. J. Biol.
    Chem. 258, 15312-22.

17. McCarthy, A.D., Goldring, D. and Hardie, D.G. (1983) Evidence that the multifunctional polypep-
    tides of vertebrate and fungal fatty acid synthases have arisen by independent gene fusion events.
    FEBS Lett. 162, 300-304.
18. McCarthy, A.D., Aitken, A., Hardie, D.G., Santikarn, S., and Williams, D.H. (1983) Amino acid
    sequence around the active serine in the acyl transferase domain of rabbit mammary fatty acid syn-
    thase. FEBS Lett. 160, 296300.
19. Schweizer, M., Takabayashi, K., Laux, T., Beck, K.-F., and Schreglmann, R. (1989) Rat mammary
    gland fatty acid synthase: localization of the constituent domains and two functional polyadenylatiord
    termination signals in the cDNA. Nucleic Acids Res. 17, 567-586.
20. Amy, C.M., Witkowski, A., Naggert, J., Williams, B., Randhawa, Z. and Smith, S . (1989) Molecular
    cloning and sequencing of cDNAs encoding the entire rat fatty acid synthase. Proc. Natl. Acad. Sci.
    U.S.A. 86,3114-18.
21. Chang, S.-I. and Hammes, G.G. (1989) Homology analysis of the protein sequences of fatty acid syn-
    thases from chicken liver, rat mammary gland and yeast. Proc. Natl. Acad. Sci. U.S.A. 86, 8373-76.
22. Knobling, A. and Schweizer, E. (1975) Temperature-sensitive mutants of the yeast fatty-acid-synthase
    complex. Eur. J. Biochem. 59,415-21.
23. Wakil, S.J., Stoops, J.K. and Joshi, V.C. (1983) Fatty acid synthesis and its regulation. Annu. Rev.
    Biochem. 52,537-79.
24. Goodridge, A.G. (1987) Dietary regulation of gene expression. Enzymes involved in carbohydrate and
    lipid metabolism. Annu. Rev. Nutr. 7, 157-85.
25. Goodridge, A.G., Crish, J.F., Hillgartner, F.B. and Wilson, S.B. (1989) Nutritional and hormonal
    regulation of the gene for avian malic enzyme. J. Nutr. 119, 299-308.
26. Ma, X.-J., Salati, L.M., Ash, S.E., Mitchell, D.A., Klautky, S.A., Fantozzi, D.A. and Goodridge,
    A.G. (1990) Nutritional regulation and tissue-specific expression of the malic enzyme gene in the
    chicken. Transcriptional control and chromatin structure. J. Biol. Chem. 265, 1 8 4 3 5 4 .
27. Fischer, P.W.F. and Goodridge, A.G. (1978) Coordinate regulation of acetyl coenzyme A carboxylase
    and fatty acid synthetase in liver cells of the developing chick in vivo and in culture. Arch. Biochem.
    Biophys. 190, 33244.
28. Goodridge, A.G. and Adelman, T.G. (1976) Regulation of malic enzyme synthesis by insulin, triio-
    dothyronine, and glucagon in liver cells in culture. J. Biol. Chem. 251, 3027-32.
29. Salati, L.M., Ma, X.-J., McCormick, C.C., Stapleton, S.R. and Goodridge, A.G. (1991) Triiodothyro-
    nine stimulates and cyclic AMP inhibits transcription of the gene for malic enzyme in chick embryo
    hepatocytes in culture. J. Biol. Chem. 266,4010-16.
30. Luo, X., Park, K., Lopez-Casillas, F. and Kim, K.-H. (1989) Structural features of the acetyl-CoA
    carboxylase gene: mechanisms for the generation of mRNAs with 5 end heterogeneity. Proc. Natl.
    Acad. Sci. U.S.A. 86, 404246.
31. Capecchi, M.R. (1989) Altering the genome by homologous recombination. Science 244, 1288-92.
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemistry oflipids, Lipoproteins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                    141

                                                                                          CHAPTER 5

    Fatty acid desaturation and chain elongation
                                   in eucaryotes
                                                                                   HAROLD W. COOK

        Department of Pediatrics and Biochemistry and the Atlantic Research Centre for Mental
                                     Retardation, Dalhousie University, Halifax, N .S.,Canada

1. Introduction

Fatty acyl chains, esterified to the complex lipids of biological membranes, are major
components of these membranes, both quantitatively and qualitatively. Fatty acyl
chains account for more than half the mass of most major phospholipids and are
primarily responsible for the apolar nature of the membrane bilayer. Depending on
their chain length and degree of unsaturation, they contribute to fluidity and other
physical and chemical properties of the membrane [1,2]. As major membrane compo-
nents, acyl chains influence a variety of membrane functions, such as ion channels
and transport, endocytosis and exocytosis, and the activities of membrane-associated
enzymes. Furthermore, polyunsaturated fatty acids derived from essential fatty acids
also serve as precursors of biologically active molecules, such as prostaglandins,
thromboxanes, leukotrienes, and a variety of hydroxy and hydroperoxy fatty acids
(see Chapter 10).
   Why is the variety of acyl chains in the lipids of biological membranes of eucaryot-
ic organisms necessary? How are these chains derived and modified, and what regu-
lates metabolism of acyl chains prior to their esterification to membrane lipids?
   Three major forces hold lipid molecules within the membrane: (1) electrostatic in-
teractions between polar groups of lipids and oppositely charged groups in adjacent
proteins; (2) hydrogen bonding between oxygen and nitrogen atoms in lipids and
adjacent proteins; and (3) London-Van der Waals dispersion forces between CH,
pairs in hydrocarbon tails of adjacent lipid molecules. Of these three, only the Lon-
don-Van der Waals forces are involved along the acyl chains and may be the major
force holding membrane lipid molecules together [3]. Since there is a symmetrical
electron distribution along the linear acyl chain, no net charge results, but there are
gravitational forces between adjacent acyl chains. These London-Van der Waals for-
ces are relatively weak; nevertheless, they are additive and proportional to the num-

ber of overlapping methylene groups and the distance between them. For a long acyl
chain the total bonding strength is greater than the electrostatic and hydrogen bond-
ing of polar head groups.
  Thus, length is a crucial parameter in acyl chain contribution to membrane struc-
ture and stability. The weak apolar interactions determine how the membrane will
interact with its environment. For example, restricted solubility of lipids in the sur-
rounding aqueous milieu is governed by the tendency of acyl chains to remain in
association with one another rather than to associate with the aqueous environment.
Similarly, response to temperature fluctuations is modulated by the extent to which
thermal influences cause acyl chains to dissociate from one another and assume a
more randomized structure. Accordingly, solubility is decreased and melting point
increased as fatty acyl chain length is increased.
  Where a more loosely packed membrane structure is advantageous, the rigidity of
lengthy saturated acyl chains can be countered by acyl chains with double bonds.
Introducing a double bond of cis geometric configuration results in a bending of the
chain with a change of approximately 30" from the linearity of the saturated chain
(Fig. 1). As London-Van der Waals forces vary inversely as the sixth power of the
distance between the acyl chains, factors that increase that distance markedly de-
crease chain-chain interaction. Double bonds also are non-rotating and restrict acyl
chain movement. Further, some charge concentration around the double bond in-
creases polarity in the acyl chain.

                                MELTING       SPATIAL
Fig         ABBREVIATION         POINT       WIDTH {nm)
                                                                 POSSIBLE CONFIGURATIONS

Stearic           18:O            700           0.25

 Oleic          c-l8:l(n-9)       16O           0.72

Elaidic         1-18:l(n-9)        430          0.31

Linoleic      c,c.l8:2(n.6)       -50            1.13

Fig. 1 . Some physical characteristics of fatty acids. See Table I for nomenclature of the fatty acids. Single
bonds have a length of 0.154 nm and an angle of approximately 1 1 1"; double bonds have a length of 0.133
nm and an angle of approximately 123".

  The extent to which double bonds cause bends or curved shapes (potentially, an
acyl chain, with six double bonds, could assume a U-shape or be nearly circular)
within the biological membrane is not well understood. Physical and biochemical
data show that unsaturated acids decrease membrane rigidity. Accordingly, within
membrane phospholipids, acyl chain length and the number and position of double
bonds markedly influence fluidity, permeability and stability of biological mem-

2. Historical background

The vital contribution of lipid molecules to the hydrophobic character of membranes
was recognized late in the nineteenth century. However, the nutritional importance
of specific lipid molecules was first revealed through the pioneering work of Burr and
Burr in 1929 [4]. They fed rats a fat-free diet and observed retarded growth, scaly
skin, tail necrosis and eventual death. This disorder was reversed by feeding a specific
fatty acid fraction. Linoleic acid was recognized as the active agent and the term
‘essential fatty acid’ was coined.
   During the following two decades, progress toward understanding the metabolism
of unsaturated fatty acids was limited by the analytical techniques available. How-
ever, the number of double bonds in fatty acids could be determined by iodination
and progress was made towards describing the process of fatty acid desaturation.
   Two factors emerged during the 1950s that lead to our current understanding of
fatty acid metabolism. The advent of chromatographic techniques (gas-liquid and
thin-layer chromatography were prime contributors) and the greater availability of
appropriate substrates and precursors labeled with isotopes, such as I4C and ’H,
greatly enhanced capabilities for studying single species of fatty acids and monitoring
their conversions by analyzing relatively small amounts of sample.
   During the last four decades, considerable detail has been revealed about metabol-
ic pathways of elongation and desaturation of saturated, monounsaturated and sev-
eral families of polyunsaturated fatty acids. In the 1950s and 1960s, in vivo evalua-
tion of fatty acid metabolism was supplemented by in vitro assays of specific enzy-
matic steps [ 5 ] . Activity of49 desaturase was measured in yeast, liver microsomes and
plants [5-71. It was determined that polyunsaturated fatty acid formation in animal
tissues involved 46 and A5 desaturation whereas in plants, further desaturation oc-
curred at the 412 and 415 positions [7-141. Chain elongation of long chain fatty acids
was found in mitochondria and in microsomes [9-161. During this period a relation-
ship between essential fatty acids and prostaglandins was elucidated. Through subse-
quent work, we now understand primary and alternate pathways of fatty acid desatu-
ration and chain elongation, competitive interactions of fatty acids, and factors that
regulate long chain fatty acid metabolism.

3. Chain elongation of long chain fatty acids

De novo synthesis of fatty acids by the soluble, cytosolic enzymes of the acetyl-CoA
carboxylase and fatty acid synthase complexes (Chapter 4) produces mainly palmi-
tate, with minor amounts of stearate. Quantitatively, these chain lengths are major
components of many membrane lipids and qualitatively appear to be related to the
optimum width of the membrane lipid bilayer. On the other hand, in membranes
many major acyl chains are longer than 16 carbons (Table I). For example, in the
myelin surrounding axonal processes of neuronal cells, fatty acyl chains of 18 car-
bons or greater make up more than 60% of the total, and in sphingolipidsin particu-
lar, acyl chains of 24 carbons are prominent.
   Many eucaryotic cells have the capacity for 2-carbon chain elongation (Fig. 2),
both of endogenously synthesized acids and of exogenous, dietary acids. In liver,
brain and other tissues there are two primary systems for elongation, one in the endo-
plasmic reticulum, and the other in mitochondria. Recent evidence indicates that liv-
er peroxisomes also contain an acetyl-CoA dependent elongation system that is en-
hanced after treatment of animals with peroxisomal proliferators; the significance
and characteristics of the activity remain to be elucidated.

1. Condensation

a. Microsomes
                            9          a
                                                                  P             + H-S-C~A
                                                              t cot
 b. Mitochondria
    R-LS-COA                   P
                       + CH,-C-S-C~A              e
                                                             ?             + H-S-C~A

2. Reduction ( p-keto acyl-CoA reductase )
                   B            + NAD(P)H + H+*                        ?
                                                  + NAo(P)+
3. Dehydration ( 0-hydroxy acyl-CoA dehydrase )

                        B                        8
                                   *R-CH-CH-C-S-COA              + Hi0

4. Reduction ( 2-trans enoyl4oA reductase )

                   ?            + NAO(P)H + H                      B
                                                   b R-CH2-CH2-C-S-CoA
                                                   + NAO(P)+
Fig. 2. Reactions in 2-carbon chain elongation of long chain fatty acids.

Nomenclature and bond positions of major long-chain fatty acids

Common name                                     Systematic namea        Abbreviations    Bond positions

Palrnitic acid                                 hexadecanoic acid           16:O
Palmitoleic acid                             9-hexadecenoic acid           16:l(n-7)        A9
                                             6-hexadecenoic acid           16:1(n- 10)      A6

Stearic acid                                     octadecanoic acid          18:O
Oleic acid                                     9-octadecenoic acid          18:1(n-9)         A9
Vaccenic acid                                 11-octadecenoic acid          18:l(n-7)         All
Petroselenic acid                              6-octadecenoic acid          18:l(n-12)        A6
Elaidic acid                                 t-9-octadecenoic acid        1-18: l(n-9)      t-A9
Linoleic acid                           9,12-octadecadienoic acid           l8:2(n-6)         A9,12
Linoelaidic acid                  t. t -9-1 2-octadecadienoic acid      t,t-l8:2(n-6)    1, t-A9,12
or-Linolenicacid                  9,12,15-octadecatrienoic acid             18:3(n-3)         A9,12, 15
y-Linolenic acid                   6,9,12-octadecatrienoic acid             18:3(n-6)         A6,9,12
Stearidonic acid              6,9,12,15-octadecatetraenoic    acid          18:4(n-3)         A6,9,12,15

Arachidic acid                                       eicosanoic acid       20:o
Gadoleic acid                                      9-eicosenoic acid       20:l(n-ll)       A9
Gondoic acid                                      1 1-eicosenoic acid      20:l(n-9)        All
Dihomo-y-linolenic                     8, I I , ICeicostrienoic acid       20:3(n-6)        A8,I1,14
Mead acid                               5,8,11 -eicosatrienoic acid        203(n-9)         A5,8,11
Arachidonic acid                 5,8,1 I , 14-eicosatetraenoic acid        20:4(n-6)        A5,8,11,14
Tirnnodonic acid             5,8,11, I 4,17-eicosapentaenoic acid          20:5(n-3)        A5,8, I 1,14,17

Behenic acid                                  docosanoic acid             22:o
Cetoleic acid                              11-docosenoic acid             22:l(n-ll)       All
Erucic acid                                13-docosenoic acid             22:l(n-9)        A13
Adrenic acid                  7,10,13,16-docosatetraenoicacid             22:4(n-6)        A7,10,13,16
Docosapentaenoic            4,7,10,13,16-docosapentaenoicacid             22:5(n-6)        A4,7,10,13,16
Docosapentaenoic                                        acid
                           7,10,13,16,19-docosapentaenoic                 22:5(n-3)        A7,10,13,I6, I9
Clupanodonic acid                                      acid
                          4,7,10,13,16,19-docosahexaenoic                 22:6(n-3)

Lignoceric acid                              tetracosanoic acid           24:O
Nervonic acid                              15-tetracosenoic acid          24:l(n-9)        A1 5

Cerotic acid                                 hexacosanoic acid            26:O
Ximenic acid                               17-hexacosanoic acid           26: I (n-9)      A1 7

"The full designation for double bonds would be all-cis, for example, all-cis-9,12,15-octadecatrienoicacid.
Unless otherwise indicated as t for a trans double bond, all bonds are of the cis geometric configuration.

3.1. The microsomal elongation system

The more active fatty acyl chain elongation system is associated with the endopIasmic
reticulum. The 2-carbon condensing unit is malonyl-CoA, very limited activity is seen
with acetyl-CoA. Avidin, a protein that binds biotin and inhibits biotin-dependent
carboxylases, does not alter microsomal elongation activities in vitro, indicating that
a microsomal acetyl-CoA carboxylase is not involved. CoA derivatives appear to be
the active form of the malonyl group and of fatty acyl acceptors. Fatty acyl-CoAs can
be formed from free fatty acids by fatty acyl-CoA ligase in the presence of ATP, MgZ'
and CoA. Evidence suggests that microsomal elongation of polyunsaturated acids
might occur in the absence of CoA; however, acyl-carrier protein is not involved.
Microsomal elongation is active with both saturated and unsaturated fatty acids, the
latter having higher activity; y-linolenate is the most effective of the unsaturated sub-
strates. Either NADPH or NADH can be utilized as electron donors with NADPH
being more active.
   In a manner analogous to fatty acid synthase, four component reactions occur in
the 2-carbon elongation process (Fig. 2). The condensation of fatty acyl-CoA and
malonyl-CoA to form thep-ketoacyl CoA derivative is rate-limiting and can be meas-
ured in vitro in the absence of NADPH. The rate of condensation is equivalent to the
overall rate of elongation and is dependent on the chain length and number and posi-
tion of double bonds in the primer. At least two condensation enzymes, one for sat-
urated and one for unsaturated primers, are indicated from differences in rates with
various labeled substrates and response to inhibitors. When microsomes are prela-
beled with fatty acid, acyl intermediates do not readily exchange, suggesting channel-
ing of the metabolites.
   The condensing enzymes funnel thep-keto acyl-CoAs to a common set of enzymes
for completion of elongation. The second reaction in elongation, catalyzed by a re-
ductase which utilizes NADPH in the formation of p-hydroxy acyl-CoA (reaction 2,
Fig. l), cannot be readily assayed as a single reaction since fully elongated product is
formed when NADPH is present. The third reaction, catalyzed by a dehydrase, can
be monitored by measuring 2-trans-enoyl-CoA formation from p-hydroxyacyl-CoA
in the absence of reduced pyridine nucleotide. This dehydrase has been solubilized
with detergent and purified nearly 100-fold.The final reaction, catalyzed by 2-trans-
enoyl-CoA reductase in the presence of NADPH, can be measured separately. In
contrast to the condensation enzyme, the activities of the reductases and dehydrase
are not influenced by diet or substrate modifications.
   Cytochrome b, participation in transfer of electrons from pyridine nucleotides to
the reductases has been proposed based on increased reoxidation of microsomal cy-
tochrome b, in the presence of NADPH, ATP and malonyl-CoA and on marked
reduction of malonyl-CoA incorporation in the presence of antibodies to cytochrome
b5. Developmental profiles of microsomal cytochrome reductases closely parallel
those of fatty acid elongation in both liver and brain.

   Relationships among component enzymes of elongation within endoplasmic reti-
culum membranes are not clear. It has been argued that each reaction is catalyzed by
a discrete enzyme, rather than by a multifunctional complex, since CoA derivatives
of all intermediates have been isolated. There are two or more condensation enzymes
and partially purified /?-hydroxy acyl-CoA dehydrase neither contains nor requires
the other activities. Others suggest a covalent linkage of the acyl-CoA to a multifunc-
tional enzyme after condensation.
   Most studies have utilized liver enzymes because of high activity in this tissue; how-
ever, in preweanling rats, brain activity generally exceeds that in liver. Studies with
brain, using various radioactively labeled fatty acyl precursors, such as palmitate,
stearate, arachidate and behenate, suggest the existence of different enzymes reacting
with dissimilar acyl chain lengths. Differences in pH optima, relative rates of sub-
strate utilization and activity profiles during development have been documented. In
Quaking mice (a mutant with defective myelination), chain elongation of 20:O to 22:O
and 24:O was reduced by about 70%, whereas elongation of 16:O and 18:O was unal-
tered relative to control mice. Treatment of microsomes with phospholipase A, in-
hibits condensation activity; the interaction between phospholipids and condensa-
tion enzyme for 20:O-CoA elongation seems different from that for 16:O-CoA or 20:4-
CoA. Existence of multiple condensing enzymes also is supported by studies with
diabetic rats where inhibited 16:O elongation is restored to normal after insulin treat-
ment but inhibition of 18:3(n-6)elongation is not reversed. In studies of elongation
of saturated, dienoic and trienoic substrates by rat liver microsomes, fasting de-
pressed all three activities similarly; in contrast, the increase in response to refeeding
was much greater with saturated substrates than with unsaturated fatty acids. This
suggestion of more than one elongation system in liver was further supported by
competitive substrate experiments.

3.2. The mitochondria1 elongation system

Although less active than the microsomal system, mitochondria1 chain elongation
has been extensively investigated, particularly in liver and brain. In contrast to the
microsomal system, the 2-carbon donor in mitochondria is acetyl-CoA (Fig. 2, reac-
tion 1b). Saturated and monoenoic fatty acyl-CoA derivatives support higher activity
than polyunsaturates; generally, monoenes are more active than saturates, particu-
larly in brain. While there is no uniform agreement about nucleotide requirements,
maximal mitochondria1elongation in liver, brain, kidney, and adipose tissue seems to
require both NADPH and NADH; heart, aorta and muscle, however, may require
only NADH. The effectivenessof mixtures of NADPH and NADH can be explained
if NADH is optimal in the first reduction and NADPH in the second.
   During the early 1970s, mechanism of mitochondrial chain elongation were eluci-
dated [16]. Although /?-oxidation (Chapter 3) and chain elongation have the same
subcellular location, reversal of /?-oxidation is not thermodynamically feasible; the

FAD-dependent acyl-CoA dehydrogenase of B-oxidation is substituted by a more
thermodynamically favorable enzyme, enoyl-CoA reductase, to produce overall neg-
ative free energy for the sequence.
   Enoyl-CoA reductase isolated from liver mitochondria using centrifugation and
chromatographic techniques was distinct from the microsomal reductase, based on
pH optima and specificities for saturated and unsaturated acyl-CoA derivatives. Kin-
etic studies suggested that enoyl-CoA reductase was rate limiting in mitochondrial
chain elongation.

3.3. Functions of the two elongation systems

Microsomal chain elongation appears to be the major source of acyl chains greater
than 16 carbons during growth and maturation, when required long chain acids may
not be supplied adequately in the diet. For example, 18- to 24-carbon saturates and
monoenes and 20- and 22-carbon polyunsaturates are required for normal brain
myelination, regardless of dietary fluctuations during development.
   The function of the mitochondrial elongation system is less clear. In view of the
relatively low activity toward 16- and 18-carbon acyl chains, a primary role for the
mitochondrial system in the formation of long acyl chains for membrane synthesis is
questionable. It has been proposed that mitochondrial elongation in liver may serve
a transhydrogenase function, moving electron equivalents from NADPH-generating
substances to the respiratory chain. During cellular anoxia, chain elongation might
conserve reducing equivalents or acetate units through the formation of acyl chains.
Considering the low capacity for chain elongation in mitochondria relative to other
reactions involving NADH or NADPH generation and utilization, a significant role
in nucleotide balance must be viewed cautiously.

4. Formation of monounsaturated fatty acids by oxidative desaturation

The spectrum of fatty acyl chains required to meet the physical and biochemical re-
quirements of lipid storage and of membrane synthesis and maintenance cannot be
supplied by diet, de novo synthesis and chain elongation alone. Unsaturated fatty
acids must be supplied by oxidative desaturation of saturated chains and of dietary
acids of plant and animal origin.

4.1. Nomenclature to describe double bonds

Before discussing desaturation enzymes, abbreviations used to describe the number
and position of double bonds in acyl chains (Table I) will be outlined using oleic acid
as an example.
(1) To indicate that oleic acid is an 18-carbon fatty acid with one double bond, the

    shorthand 18:1 will be used. The number before the colon denotes the number of
    carbon atoms and the number following refers to the number of double bonds.
    This form is adequate only where the position or nature of the double bond is not
    known, for example, in chromatographic analysis of an acyl chain mixture where
    positional isomers are not separated.
(2) To assign the position of an individual double bond or the specificity of an en-
    zyme inserting it, the delta (A) nomenclature will be used. This describes a bond
    position relative to the carboxyl carbon of the acyl chain. For oleic acid, the
    double bond is in the A9 position, between carbons 9 and 10, and would be intro-
    duced into the 18-carbon chain by a A9 desaturase enzyme.
(3) To designate an individual fatty acid within a ‘family’ of structurally related
    acids, the (n- ) nomenclature, promoted by the IUPAC-IUB Commission on No-
    menclature, will be used. Here, the position of the first double bond from the
    methyl end is described. Thus, 18: I (n-9) indicates that the double bond closest to
    the methyl end (the only double bond in this example) is in the A9 position, since
    n=18 and (n-9) is 9 carbons from both the carboxyl and methyl ends. This con-
    vention is particularly useful in designating groups of fatty acids derived from the
    same parent compound and in which metabolic reactions do not occur on the
    methyl side of an existing double bond.
(4) To indicate the geometric configuration of a double bond, the designation will
    be preceeded by c- for cis or t- for trans. Thus, c-18: l(n-9) distinguishes oleic acid
    from its trans isomer, elaidic acid. Generally, double bond configuration is cis.
Among other conventions, the w-designation is still widely used to designate the posi-
tion of a double bond from the methyl end (w-carbon) and is similar to the (n- )

4.2. Characteristics of the monoene-forming desaturation enzymes

Monounsaturated fatty acids are formed in mammalian systems by direct oxidative
desaturation (a removal of two hydrogens) of a preformed long chain saturated fatty
acid. The oxygenase type of enzyme is associated with the endoplasmic reticulum
(microsomes) and can be isolated as microsomes from liver, mammary gland, brain,
testes and adipose tissue. The A9 desaturase usually is the predominant, if not exclu-
sive, desaturation enzyme for saturated acids in these tissues.
   The A9 desaturase acts on fatty acyl-CoA; with crude microsomal preparations in
vitro, free fatty acids can be used if ATP, Mg2+and CoA are supplied to supplement
activation by fatty acyl-CoA ligase, which is usually not rate limiting. For most tis-
sues, 14- to 18-carbon saturated fatty acyl chains are good substrates, with stearoyl-
CoA being most active. Reduced pyridine nucleotide is required, and generally
NADH is more active than NADPH. The A9 desaturase has an absolute requirement
for molecular oxygen-which acts as an electron acceptor for two pairs of hydrogens,

one from NADH and the other from the fatty acyl-CoA-and is highly sensitive to
inhibition by cyanide.
   Most assays of A9 desaturase use stearoyl-CoA (or stearic acid and cofactors) la-
beled with 14C in the carboxyl carbon or 3H in the acyl chain. Following in vitro
incubation, lipids are extracted and methyl esters are formed by trans-esterscation.
Saturated substrate and monounsaturated products can be separated by gas liquid
chromatography, high performance liquid chromatography on reversed phase
columns, or argentation thin-layer chromatography. By the latter technique, silica
gel impregnated with silver ions effectively separates acyl chains according to their
degree of unsaturation.
   The A9 desaturation system consists of three major proteins: (1) NADH-cyto-
chrome b5 reductase, (2) cytochrome b5, and (3) a terminal desaturase component or
cyanide-sensitive protein (Fig. 3). Under most circumstances, electron transport
greatly exceeds the activity of the rate-limiting desaturase component. Most of the
neutral lipid and some phospholipids can be extracted from microsomes without sig-
nificant loss of A9 desaturase activity; a phospholipid fraction, including phosphati-
dylcholine, is required for desaturation.
   The firm association of the A9 desaturase system with the endoplasmic reticulum
retarded early characterization of the complex. Loss of activity during solubilization
was largely overcome through the use of carefully controlled ratios of detergents to
protein and combinations of mild extraction solvents. Activity in a purified complex
was achieved by recombination of the three major protein components and egg leci-
thin or synthetic phosphatidylcholine.

      NADH : Cytochrome    Cytochrome            Desaturase
        bg Reductase            b3        (Cyanide Sensitive Factor)
Fig. 3. Diagrammatic representation of the A9 desaturase complex, including the electron transport pro-

   The desaturase component, highly sensitive to cyanide, has a molecular weight of
53 kDa, one atom of non-heme iron per molecule as the prosthetic group, and 62%
non-polar amino acid residues. The desaturase component is largely within the mi-
crosomal membrane, with the active center exposed to the cytosol. Interaction of the
desaturase with specific reagents suggests that arginyl residues play a role at the bind-
ing site for the negatively charged CoA moiety of the substrate, and tyrosyl residues
may be involved in chelation of the iron prosthetic group. Stearoyl-CoA desaturase,
purified 150-fold, appears homogeneous as judged by SDS-polyacrylamidegel elec-
trophoresis. Based on this, the level of A9 desaturase in rat liver at maximal induction
could account for 0 . 5 4 8 % of microsomal protein.
   NADH-cytochrome bSreductase (a flavoprotein of molecular weight 43 kDa) and
cytochrome b, (a heme-containing protein of molecular weight 16.7 kDa) are more
readily solubilized than the desaturase. Cytochrome b, has a catalytic hydrophilic
region of 85 residues (including the NH,-terminal) and the protein terminates in a
hydrophobic COOH-terminal tail of approximately 40 amino acids. The latter at-
taches the protein to the membrane.
   The mechanism of hydrogen removal from the saturated acyl chain is not fully
understood. Stereochemical studies have shown that only the D-hydrogens at posi-
tions 9 and 10 are removed to give a cis double bond. This appears to occur by con-
certed removal of the hydrogens but the mechanism of oxygen involvement in the
enzyme:substrate complex is unknown. Attempts to demonstrate a hydroxyacyl in-
termediate have been negative and hydroxyacyl-CoAs are not readily desaturated. It
remains uncertain whether one or two cytochrome b, molecules per complex are re-
quired to transfer two electrons from NADH to molecular oxygen.

4.3. Modification of A9 desaturase activities in vitro

To understand mechanisms and regulation of the A9 desaturase system, it is impor-
tant to have probes that alter its activity. As cyanide completely inhibits A9 desatura-
tion in rat liver by acting on the terminal desaturase component, the latter is fre-
quently referred to as the cyanide-sensitive factor. Cyanide inhibition appears to be
related to accessibility of the non-heme iron in the desaturase. Some A9 desaturases
(for example in yeast) are not inhibited by cyanide. Iron chelators interacting with the
non-heme iron in the desaturase component, are inhibitory in vitro. Less readily ex-
plained is the reduction of A9 desaturase activity of liver microsomes with copper
deficiency and increased 18:1/18:0 ratio in triacylglycerols of adipose tissue following
copper supplementation.
   Cyclopropenoid fatty acids found in stercula and cotton seeds are potent inhibitors
of the A9 desaturase. Sterculoyl- and malvaloyl-CoA (18- and 16-carbon derivatives,
respectively, with cyclopropene rings in the A9 position) specifically inhibit A9 desatu-
ration in vitro. When hens are fed meal containing these fatty acids, decreased 18:I/
18:O ratios in their egg yolks are found. Cyclopropene acids have been used in vitro

to alter A9 and 46 activities of microsomes from developing brain differentially, thus
distinguishing the relative contributions of these two enzymes in perinatal brain de-
   Acceptors for the products of desaturation generally stimulate the A9 desaturase
system. Both glycerol-3-phosphate and lysolecithin remove monoenoic products as
they are produced and probably stimulate overall activity by reducing feedback inhi-
bition by the unsaturated fatty acid.

4.4. Dietary and hormonal regulation of A9 desaturase

A remarkable feature of A9 desaturase is the extreme response to dietary deprivation
and alterations. When rats are not fed for 12-72 h, liver A9 desaturase activity de-
clines markedly to levels less than 5% of control values, as stored energy reserves are
mobilized from adipose triacylglycerols (Fig. 4). When the rats are refed, A9 desatu-



0)   200-

4    100-
ri     0-
cc    40-
5                        RAT BRAIN




            10 DAY OLD

Fig. 4. Effects of fasting for 48 h and subsequent refeeding of a normal chow diet for 24 h on the in vitro
A9 desaturase activities of brain and liver from 10-day-old and adult rats. Adapted from Cook and Spence
[31]. Values are nmol/min/mg protein.

rase activity increases dramatically to levels of greater than 2-fold above normal. The
restoration has been termed ‘super-induction’, as levels of enzyme activity can rise
more than 50-fold above the fasted state, particularly when the rats are refed a fat-
free diet enriched in carbohydrate or protein. With protein synthesis inhibitors and
immunological techniques, it has been shown that synthesis of the desaturase compo-
nent is altered quantitatively. Such responses of the liver enzyme to dietary intake
probably explain the so-called circadian changes in A9 desaturase activity, where ac-
tivities can fluctuate 4-fold over a 24-h period; highest liver activity (around mid-
night) corresponds to maximal food intake in the nocturnal rat.
   In contrast to the liver enzyme, brain A9 desaturase activity is little altered by die-
tary restrictions. This ensures a constant level of activity during crucial stages of
brain development. Brain A9 desaturase activity is greatest during the perinatal and
suckling period in rats and is generally higher than in liver. However, when rats are
weaned, liver activity rapidly increases in response to solid food intake to achieve
adult levels 100- to 200-fold greater than in neonates; in contrast, brain A9 desaturase
activity slowly declines. In addition to the role that diets low in fat play in increasing
liver A9 desaturase activity, there is evidence that dietary polyunsaturated acids, par-
ticularly linoleic acid, selectively inhibit monoene formation by an unknown mecha-
   Advantage has been taken of dietary induction of A9 desaturase to facilitate isola-
tion of poly(A+) RNA and preparation of cDNA for the enzyme [ 171. Translation-
ally active levels of mRNA are increased 40- to 60-fold following refeeding of fasted
animals. A9 desaturase is translated on cytoplasmic polysomes with post-translation-
a1 binding of iron and insertion into the endoplasmic reticulum. Desaturase peptides
synthesized using plasmid expression vectors can be reconstituted with cytochrome
b, and reductase to give A9 desaturase activity. The gene encoding stearoyl-CoA de-
saturase, expressed with a 63-fold increase during induced differentiation of preadi-
pocytes, also has been characterized [18]. Adipocyte A 9 desaturase has 92% identity
with the rat liver enzyme but is distinct from a second form of A9 desaturase found
in brain, kidney, adipose tissue and lung but not in liver. A structural gene for A9
desaturase of yeast also has been isolated and characterized; liver A9 desaturase can
effectively substitute for the missing activity in yeast mutants [19].
   Hormonal regulation of A 9 desaturase is not completely defined, despite close as-
sociation between the activities of lipogenic enzymes and carbohydrate metabolism
[20]. Insulin appears to be an inducer of A9 desaturase in vivo. Rats made diabetic by
streptozotocin-induced destruction of /?-cells of the pancreas have depressed A9 de-
saturase activity in liver, mammary gland and adipose tissue and insulin restores the
activity if administered in vivo; if added to an in vitro assay, insulin is without effect
on desaturase activity. The in vivo insulin effect is blocked by actinomycin D or puro-
mycin, indicating synthesis of new protein. Significant changes in the content of cyto-
chrome b, and the reductase do not occur. Time course studies with transcription and
translation inhibitors indicate that the mRNA for the desaturase is rather stable.

Other hormones and effectors, such as glucagon and cyclic AMP, do not appear to
alter A9 desaturase activity, whereas epinephrine and thyroxine may enhance
monoene formation. It remains to be demonstrated whether insulin and other hor-
mones merely alter the flux of carbohydrate metabolites or if, as with de novo fatty
acid synthesis, the desaturase is covalently modified.

4.5. Formation of monounsaturated fatty acids in plants

In plants, desaturation occurs both in the cystosol and chloroplasts. Most plants uti-
lize stearoyl-ACP as substrate, with free fatty acid or fatty acyl-CoA being inactive.
Euglena is an interesting case. When grown in the dark, its chloroplasts become non-
functional and 18:O-CoA is the required form of the saturated substrate, whereas cells
grown in light require 18:O-ACP. Somewhat analogous to animal systems, desatura-
tion in plants requires molecular oxygen, NADPH, a flavoprotein reductase (ferre-
doxin: NADPH reductase) and ferredoxin as an intermediate electron receptor. The
desaturase component is inhibited by cyanide but not by carbon monoxide.

5. Formation of polyunsaturated fatty acids

5.1. Characteristics and restrictions in animal systems

All eucaryotic organisms contain polyenoic fatty acyl chains in the complex lipids of
their membranes, and most mammalian tissues can modify acyl chain composition
by introducing more than one double bond.
(1) The first double bond introduced into a saturated acyl chain is generally in the A9
    position so that substrates for further desaturation contain either a A9 double
    bond or one derived from the A9 position by chain elongation. An exception is
    the relatively large amount of (n-10) monounsaturated fatty acyl chains in neona-
    tal rat brain (16: l(n-10) and 18:l(n-10) comprise up to 35% of the monoene frac-
    tion). In vitro evidence indicates that these acids are formed by a A6 desaturation
    enzyme distinct from the A9 desaturase; however, the qualitative significance of
    this unusual isomer composition during brain development is unclear.
(2) Like A9 desaturation that inserts the first double bond, further desaturation is an
    oxidative process requiring molecular oxygen, reduced pyridine nucleotide and
    an electron transfer system consisting of a cytochrome and related reductase en-
(3) Animal systems cannot introduce double bonds beyond the A9 position. Thus,
    second and subsequent double bonds are always inserted between an existing
    bond and the carboxyl end of the acyl chain, never on the methyl side of an exist-
    ing double bond (Fig. 5). Plants, on the other hand, generally introduce second
    and third double bonds between the existing double bond and the terminal me-

                  ,             INSECTS
                                    I             ,   LOWERPLANTS

Fig. 5. Positions of fatty acyl chain desaturation by enzymes of animals, plants, insects and lower plants
such as Euglena.

    thy1 group. The ability displayed by diatoms and Euglena to desaturate on either
    side of an existing bond apparently is true of insects also [21]; at least 15 insect
    species form 18:2(n-6) by de novo synthesis and also 20:4(n-6). Consequently, in
    animals double bonds are inserted at the 49,46, A5 and A4 positions, in plants at
    the A9,A12 and A15 positions and in insects in the A5,A6, A9,412 and A15 posi-
    tions. Well-established evidence confirms A9, A6 and A5 desaturases in a variety
    of animal tissues. Whereas A4 desaturase has not been directly characterized, the
    abundance of long chain polyenoic acids containing 44 double bonds in tissues
    such as brain, retina and testes, coupled with the other general restrictions on
    polyene formation, argue strongly for a distinct A4 desaturase. Comparative
    studies with tissue explants, primary cultures and neoplastic cells in culture sug-
    gest that A4 desaturase is characteristic of differentiated tissue.
    In most organisms, and certainly in higher animals, conjugated double bonds are
    extremely rare. Accordingly, given the limitations of mammalian desaturases,
    chain elongation usually alternates with desaturation to maintain methylene in-
    terruption in polyunsaturated chains (Fig. 6).
    All bonds introduced by oxidative desaturation are in the cis geometric configu-
    ration. When acyl chains containing trans double bonds are introduced to animal
    systems through diet or intestinal bacteria, the trans acyl chains seem to be recog-
    nized as biologically distinct from those containing cis bonds.

S.2. Essential fatty acids-a        contribution of plant systems

Requirements for polyunsaturated acyl chains cannot be met solely by de novo meta-
bolic processes within animal tissues. Animals are absolutely dependent on plants (or
insects) for the two major precursors of the (n-6) and (n-3) fatty acids, linoleic and
linolenic acids. In animal tissues these acyl chains can be converted to fatty acids
containing 3-6 double bonds.


                                                                                                  22:Yn - 3)

24:D           24:lln-9)


Fig. 6 . Major pathways of fatty acid biosynthesis by desaturation and chain elongation in animal tissues.
Note the alternating sequence of desaturation in the horizontal direction and chain elongation in the verti-
cal direction. Reduced intensity of the arrows distinguishes minor pathways. Boldness of type and size for
individual fatty acids reflects, in a general way, relative accumulation in tissues.

   Severe effects observed in experimental animals and humans in the absence of
these dietary acids include a dramatic decrease in weight, dermatosis and increased
skin permeability to water, enlarged kidneys and reduced adrenal and thyroid glands,
cholesterol accumulation and altered fatty acyl composition in many tissues, im-
paired reproduction, and ultimate death. The four (n-6) acids in the sequence from
18:2(n-6) to 20:4(n-6) individually have similar potency in reversing these effects of
deficiency, whereas the activity of 18:3(n-3) alone is much lower. Thus, the term 'es-
sential fatty acid' clearly applies at least to the major (n-6) acids. The essential role of
(n-3) acids is more difficult to define but there is increasing evidence that minimal
amounts of (n-3) acids, probably 0.2-0.5% of total fatty acids, are required for nor-
mal tissue development.
   A function for 18:2(n-6), in addition to its role as precursor of 20:4(n-6), seems
likely [22]. Cats, unable to synthesize 20:4(n-6) from 18:2(n-6),require both 18:2(n-6)
and 20:4(n-6) in their diets. As carnivores, they rely on other animals to make 20:4(n-
6) for them. Moreover, some fatty acids that cannot serve as prostaglandin precur-
sors (for example, columbinic acid which is linoleic acid with an extra trans bond in
the A5 position) prevent signs of essential fatty acid deficiency.

5.3. Families of fatty acids and their metabolism

Relationships among fatty acids in metabolic conversions can be evaluated by con-
sidering groups or families of fatty acids. The predominant fatty acid families are the
(n-6) acids derived from 18:2(n-6), the (n-3) acids derived from 18:3(n-3), the (n-9)
acids derived from 18:l(n-9), and the (n-7) acids derived from 16:l(n-7) (Fig. 6).

5.3.1. The (n-6) family
Arachidonate [20:4(n-6)], an abundant polyenoic acyl chain found in most animal
tissues, can be formed from 18:2(n-6) by the alternating sequence of 46 desaturation,
chain elongation of the 18:3(n-6) intermediate, and A5 desaturation of 20:3(n-6) (Fig.
6). 20:4(n-6) is a component of phospholipids contributing to the structural integrity
of membranes and is the primary precursor of several classes of oxygenated deriva-
tives (for example, prostaglandins) with a variety of biological activities (Chapter 10).
Frequently, 20:4(n-6) is referred to as an essential fatty acid. Indeed, it is absolutely
required in many tissues but an adequate dietary supply of 18:2(n-6)can be converted
to 20:4(n-6) under most circumstances. There are exceptions in specific tissues or
cells. For example, neutrophils require 20:4(n-6) for leukotriene production but can-
not synthesize it from 18:2(n-6).
   In liver and most other tissues of animals in a normal, balanced state, the only
members of the (n-6) family to accumulate in relatively large quantities are 18:2(n-6)
and 20:4(n-6); much lower levels of the intermediates 18:3(n-6) and 20:3(n-6) are de-
tected. Such observations support a rate-limiting role for the A6 desaturase in the
sequence. On the other hand, A5 desaturase activity measured in vitro with 20:3(n-6)
as substrate is approximately equivalent to, or lower than, A6 desaturase activity (Ta-
ble 11) and a possible regulatory function for A5 desaturase in some circumstances
should not be dismissed. When A6 desaturase appears to be limited and potentially
related to a deficiency or disease state, this initial step can be by-passed by providing

Relative rates of desaturation by rat liver and brain

                                                        Liver                    Brain

Desaturation           Substrate           Adult            10-day-old   Adult     10-day-old

A9                     18:O                100                  2        4         12

A6                     18:2(n-6)            30               9           1          8
                       18:3(n-3)            41              17           1         13

A5                     20:3(n-6)            17                  4        1          1

Source: Adapted from Sprecher [12], Cook and Spence [31], and Cook [32].

18:3(n-6), the next fatty acid in the sequence, from enriched sources such as evening
primrose oil.
   Indirect evidence supports the existence of distinct A6 and A5 desaturase enzymes.
The enhancement of desaturase activity observed upon refeeding after fasting can be
suppressed by glucagon or dibutyryl cyclic AMP in the case of A6 desaturase activity
but not for A5 desaturase. Some cultured cell lines have lost 46, but retain A5, desatu-
rase activity. Also convincing is evidence that the A5 desaturase of rat liver micro-
somes can act directly on a phospholipid substrate to form arachidonoyl phosphati-
dylcholine from the eicosatrienoyl phospholipid precursor whereas no such activity
with esterified acyl chains has been demonstrated for A6 desaturase.
   The A9, 46, and A5 desaturases of liver and brain microsomes are stimulated by
cytosolic proteins, bovine serum albumin or catalase. Part of the activation may be
related to the acyl chain binding properties of these proteins regulating availability of
fatty acyl substrates or removing products. Catalase may also influence oxidation-
reduction reactions of the electron transport sequence, particularly for the 46 and A5
desaturases. The physiological significance of direct action by the soluble proteins on
desaturation should be considered cautiously, however, as partially purified desatu-
rase is not stimulated by either albumin or cytosolic protein.
   Using antibodies to cytochrome b5, it has been shown that A6 desaturase activity
is dependent on electron transfer similar to the A9 desaturase. The terminal A6 desa-
turase has been purified as a single polypeptide of 66 kDa with 41% nonpolar amino
acids, contains one atom of nonheme iron, is inhibited by cyanide andp- chlormercu-
ribenzenesulfonate and is predominantly on the cytosolic side of the endoplasmic
reticulum [23]. The A5 desaturase also requires cytochrome b5 but has been purified
only eight-fold.
   The physical relationship of the desaturases and chain elongation enzymes within
the membranes of the endoplasmic reticulum remains to be demonstrated. Indirect
evidence indicates that the sequence of reactions, including esterification of products
to phospholipids, proceeds in a concerted manner without release of free fatty acyl
intermediates [24].
   An alternate pathway for 20:3(n-6) and 20:4(n-6) formation from 18:2(n-6) would
be possible if mammalian tissues had a AS desaturase. Rigorous testing of liver, brain
and mammary tissue in vivo and in vitro with 20:l(n-9) or 20:2(n-6) has shown that
addition of another double bond is in the A5, not the A8 position. Apparently, these
tissues do not have a A8 desaturase. On the other hand, low A8 desaturation activity
has been reported in testes and tumors. Possibly this activity is due to lack of specific-
ity of the A9 desaturase, but expression of a separate A8 desaturase cannot be dis-
   The polyunsaturated acyl chains, 22:4(n-6) and 22:5(n-6), are quantitatively sig-
nificant components of some tissues (Table 111). Involvement of a A4 desaturase in
the formation of 22:5(n-6) has been questioned based on feeding studies with rats, in
which 22:4(n-6) did not give rise to 22:5(n-6) in tissue lipids. On the other hand, A4

Comparison of the relative content of total (n-6) and (n-3) fatty acyl chains in various mammalian tissues
and fluids

Tissue                                            % of total                  Approximate
                                      (n-6)                    (n-3)          (n-6)/(n-3)

Brain                                 12-15                     8-14          1:l
Retina                                10-14                    21-36          1.3
Spermatozoa                            8-12                    32-36          1 :4

Testes                                20-15”                    6-10         4: 1

Liver                                 20-24                      6-10        4:I
Heart                                 3540                       8-14        4: 1
Kidney                                3540                       6-8         5:1
Adipose                                4-22                      1-3         5: 1
Milk                                   24                      0.2-1         5:l

’Testes is distinctive for its high content of 22:5(n-6).

desaturation of dienoic, trienoic and tetraenoic acids of 20-carbons or fewer has been
reported by several workers. Exogenous 22:4(n-6) may be a preferred substrate for
‘retroconversion’.This process of partial degradation involves loss of either a single
two-carbon fragment or a double bond and either two or four carbon atoms by p-
oxidation reactions. In general, retroconversion utilizes fatty acids of 20 carbons or
greater. Since only double bonds in the A4 position are lost, this process could pro-
duce fatty acids with the first double bond in the A5 position. The quantitative signifi-
cance of this process is not well established. Recent evidence indicates a deficiency of
retroconversion of 22:6(n-3) to 20:5(n-3) in fibroblasts from a Zellweger’s patient
compared to control cells. This supports a role for peroxisomes (which have a p-
oxidation system-see Chapter 3) in retroconversion as these mutant cells are defec-
tive in peroxisomal assembly.

5.3.2. The (n-3) family
Generally, the most abundant (n-3) acyl chains are 20:5(n-3) and 22:6(n-3)in animal
tissues such as cerebral cortex, retina, testes and muscle; in retinal rod outer seg-
ments, phospholipids may contain 40-60% 22:6(n-3). Clearly, six double bonds, not
possible in 22 carbon (n-6) acids, could markedly influence membrane fluidity, al-
though physical studies indicate that after the first double bond in an acyl chain,
additional double bonds have a diminished effect.
   Recently there has been major interest in the benefits of increased consumption of
fish and fish oil products abundant in 20:5(n-3)and 22:6(n-3) [25]. Enzymes in phyto-
plankton consumed by fish, or in fish themselves, form these fatty acids from 18:3

(n-3). Consumption of (n-3) fatty acids leads to altered acyl chain composition of
plasma phospholipids, platelets, neutrophils and red cells. Enrichment of (n-3) fatty
acids in peripheral body tissues takes longer and is dependent on the level of dietary
intake. Generally, increases in (n-3) content are at the expense of (n-6) acids, within
a narrow limit of overall unsaturated fatty acid content in biological membranes.
Some populations, such as Greenland Inuit, routinely consume high levels of (n-3)
relative to (n-6) fatty acids. Compared to control populations with lower dietary (n-3)
fatty acids, these Inuit have a lower incidence of ischemic heart disease and longer
bleeding times, related to a reduction in platelet aggregation. Some clinical trials sup-
port a positive effect of (n-3) fatty acids on decreasing platelet aggregation, lowering
blood pressure, reducing circulating triacylglycerols and producing modest changes
in serum cholesterol and lipoproteins. Beyond the cardiovascular system, studies
support reduced severity of arthritis or impaired kidney functions involving abnor-
mal immune and inflammatory responses. Other potential areas of involvement of
(n-3) fatty acids include enchanced insulin sensitivity in the diabetic state, countering
(n-6) fatty acids in chemically induced, transplanted or metastatic tumors, or altered
visual acuity and response to learning tests. Current evidence provides positive indi-
cations but remains inconclusive as to general beneficial effects of (n-3) fatty acids.
An appropriate balance of dietary (n-3) and (n-6) fatty acids must be defined. The
relatively high concentration of (n-3) fatty acids in some body tissues with highly
specialized functions (Table 111) and tenacious retention of (n-3) acids during dietary
deprivation suggest important structural and physical roles in these tissues [26].
   Lack of clearly defined deficiency models, free of complications related to imbal-
ance of (n-6) acids, restricts precise definition of daily requirements for (n-3) fatty
acids. Suggestions that 0.2 - 0.3% of energy as (n-3) fatty acids is adequate for adults
and 0.5% during pregnancy, lactation and infancy must be coupled with indications
that the (n-3) to (n-6) ratio should be between 1:4 and 1:8 to promote normal growth
and development. Reported deficiencies in humans on long term intravenous or gas-
tric tube feeding have been corrected by supplementation with (n-3) fatty acids within
this range.

5.3.3. Competition between the (n-6) and (n-3)families
Competition between fatty acids of the (n-3) and (n-6) families occurs at the level of
desaturation and chain elongation. With the A6 desaturase enzyme, 18:3(n-3) is a
better substrate than 18:2(n-6).Accordingly, an abundance of 18:3(n-3)can effective-
ly decrease formation of 20:4(n-6) from 18:2(n-6). Simultaneous feeding of deute-
rated 18:2(n-6) and 18:3(n-3) to human subjects indicates that conversion of 18:3(n-3)
to 20:5(n-3) and 22:6(n-3) is much greater than conversion of 18:2(n-6) to 20:4 (n-6)
[27]. Competition between (n-6) and (n-3) fatty acids has been demonstrated in a
variety of in vivo and in vitro experimental models and alterations of desaturation
and elongation go beyond relative availability of substrates [24]. Studies with a range
of competing geometric and positional isomers indicate differential interactions of

fatty acid isomers, some as competitors and others as activators and support com-
plexity beyond simple interactions at each step of a common enzyme sequence [28].

5.3.4. The (n-9) family
The prominent acyl chain in this family is 18: 1(n-9). Generally, competition from
18:2(n-6) and 18:3(n-3) for A6 desaturase prevents formation and accumulation of
polyunsaturated (n-9) acids. However, in animals on a diet deficient in essential fatty
acids, competition is removed and 18:l(n-9) is utilized as a substrate for the rate-
limiting 46 desaturase (Fig. 6). Further chain elongation of 18:2(n-9)to 20:2(n-9) and
A5 desaturation results in accumulation of 20:3(n-9). While 20:3(n-9) may partially
substitute for some physical functions of the essential fatty acids within membranes,
it is not a precursor of prostaglandins and cannot alleviate the signs of essential fatty
acid deficiency.
    Since deficiency of essential fatty acids markedly reduces 20:4(n-6) while increasing
20:3(n-9), the ratio of triene to tetraene in tissues and serum has been used as an index
of essential fatty acid deficiency. With capillary and high-efficiency gas chromatog-
raphy (see Fig. 7), isomer separations can be achieved in analyses of tissue and fluid
lipids so that the ratio of 20:3(n-9) to 20:4(n-6) (normally less than 0.2) becomes an
even more precise indication of a deficiency. Even use of this ratio has limitations; for
example, inhibition of 46 desaturase would reduce formation of both 20:3(n-9) and
20:4(n-6) resulting in a deficiency state without a marked alteration of the ratio. The
total amount of (n-6) acids has been proposed as a more accurate reflection of essen-
tial fatty acid deficiency.

5.3.5. The (n-7) family
The primary (n-7) acid in membranes and circulating lipids is 16:l(n-7). As most
analyses do not distinguish specific 18:1 isomers, the contribution of 18:1(n-7) to the
18:l fraction is seldom appreciated. In developing brain, for example, 18:l(n-7)
formed by chain elongation of 16:l(n-7) comprises up to 25% of total 18-carbon
monoene. On the other hand, high levels of polyunsaturated fatty acids derived from
16:l(n-7) are not detected even on a fat-free diet, although increased levels of 16:l-
(n-7) frequently accompany deficiency of essential fatty acids. Potentially, 20:4(n-7),
with only a single carbon shift of the double bonds compared to 20:4(n-6), could be
formed from 16:1(n-7). Accumulation of 20:3(n-7) and 20:4(n-7) has been reported in
the absence of exogenous polyunsaturated fatty acids but these minor ‘endogenous’
polyunsaturated acids are rapidly replaced when (n-3) or (n-6) fatty acids are provid-

5.4. Dietary and hormonal alterations of polyunsaturated acid synthesis

Regulation of desaturase and chain elongation enzyme activities involved in polyun-
saturated fatty acid synthesis is not fully defined, possibly reflecting the complexity

         R E T E N T I O N TIME ( m i n l
Fig. 7. Gas liquid chromatographic pattern of acyl chain distribution in the serum of an infant on total
parenteral nutrition (intravenous feeding) for 5 weeks without appropriate lipid supplementation (upper
panel) and after 1 week on lipid and zinc supplementation (lower panel). Fatty acyl chain methyl esters
from the serum phospholipids were analyzed by gas liquid chromatography at 180°C using a 50 meter
capillary column coated with Silar 9CP.

of the mechanisms involved. Since capacity of the cytochrome system greatly exceeds
the requirements of the desaturases, electron transfer is not considered a point of
control. Responses of the desaturases to dietary alterations observed in different lab-
oratories have sometimes been conflicting, but A6 and A5 desaturases are less dra-
matically influenced by fluctuations in dietary intake than is A9 desaturase (Table 4;
Fig. 4). Generally,A6 and A5 desaturase activities are reduced during fasting, but the
‘superinduction’ seen with the A9 desaturase is not observed upon refeeding. In-
creased A6 desaturase activity induced by protein-enriched diets is blocked by inhib-
itors of protein synthesis. The observations that glucose and fructose increase de-
saturase activities in fasted animals in the short term, but repress activity with
prolonged feeding, may reflect an initial response of the desaturases to insulin. In-
sulin increases the activities of all three desaturases, particularly when they have been
suppressed by induced experimental diabetes. Other hormones induce a differential

Effects of dietary, hormonal, and other manipulations on A9, A6 and A5 desaturation activities i experi-
mental animals

                                                         Effect on desaturation

Treatment                                         A9                A6                A5

  High glucose-short term                         7                 t                 7
               long term                          t                 1
  High protein                                    t                 tt                t
  Fasting                                         11                1                 1
  Re feeding                                      tt                t                 t
 Insulin                                          t                 T                 t
 Glucagon                                         -                 1                 1
 Epinephrine                                      t                 1                 1
 ACTH                                                               5                 5
 CAMP                                             -                 1                 1
 Glucocorticoids                                  -                 1                 1
 Thyroxine                                        7                 1
 Hypothyroidism                                   11                1
  Sterculic acid                                  1                 -
  Cytosolic proteins                              T                 tt                t
  Retinoic acid                                   L                 7
The dash indicates no significant change and blank space indicates an absence of definitive information for
that particular treatment. Adapted primarily from Jeffcoat [ 101 and Brenner [20].

effect on the 49,46, A5 desaturases [20]. For example, glucagon and dibutyryl cyclic
AMP block the response of 46 desaturase to refeeding after starvation but have little
or no effect on A9 desaturase. Epinephrine also suppresses 46 and 45 desaturases,
apparently through B-receptors activating the adenylate cyclase system, but enhances
A9 desaturase. Generally, 46 and 45 desaturases are similarly inhibited by glucocor-
ticoids, other steroids and adrenocorticotropic hormone (ACTH). More research is
required to define the role of cyclic nucleotides and protein kinases as mediators in
the action of hormones on control of polyunsaturated fatty acid biosynthesis.

5.5. Two or more double bonds in plants

Studies by James and coworkers of specificities of plant desaturases that insert sec-
ond and third double bonds into monounsaturated acyl chains, suggested two ope-
rative systems: (1) an enzyme with an active site that interacts with an existing double

bond at a fixed distance (optimally 9 carbons) from the carboxyl end, and (2) an
enzyme with an active center that interacts with an existing double bond at a fixed
distance (optimally 9 carbons) from the methyl end. It has been speculated that the
former accepts thioesters as substrates, whereas the latter may accept an acyl chain
esterified to a complex lipid substrate. 18:l(n-9) would be a substrate for both sys-
tems and is the most actively desaturated monoenoic acid in plants. However, studies
indicate that desaturation of 18:l(n-9) to 18:2(n-6) and of 18:2(n-6) to 18:3(n-3)
occurs primarily on phosphatidylcholine in plant microsomes, not by acyl-CoA de-

6. Unsaturated fatty acids with trans double bonds

 Trans unsaturated fatty acids, the geometric isomers of naturally occurring cis-acids,
are not produced by mammalian enzymes but are formed enzymatically by microor-
ganisms in the gastrointestinal tract of ruminant mammals, and chemically during
commercial partial hydrogenation of fats and oils [13,29]. Trans-acids have been de-
scribed inaccurately as unnatural, foreign or non-physiological; through diets con-
taining beef fat, milk fat, margarines and partially hydrogenated vegetable oils, trans-
acids are ingested, incorporated and modified in animal tissues.
   Early studies suggested selective exclusion of trans-acids from metabolic processes
and lack of incorporation into membrane lipids. Later studies have indicated little
selectivity for absorption, esterification or oxidation of trans isomers compared to cis
isomers. Significant discrimination of specific positional isomers of trans acids (for
example A13 trans acids) may occur. In general, trans acids appear to be recognized
as a distinct class of acyl chains with properties intermediate between saturated and
cis monounsaturated acids. Short-term accumulation of trans-acids in tissues gener-
ally is proportional to dietary levels. Lack of preferential accumulation over the long
term suggests that cis and trans isomers turn over similarly.
   Nonetheless, the influence of trans-unsaturated fatty acids in biological systems
continues to be controversial. Specific interactions of trans-acids with desaturation
and chain elongation enzymes of animal tissues have been reported. Positional iso-
mers of t-18:l (except for the A isomer) are desaturated by the A9 desaturase of rat
liver microsomes, resulting in a series of cis,trans-dienoic isomers that, in some cases,
are desaturated again by the A6 desaturase to unusual polyunsaturated structures.
Chain elongation of trans isomers occurs at a slower rate than A9 desaturation. Thus,
certain monoenoic trans-isomers can both inhibit desaturase activities and form
products with as yet undetermined biological activities.
   Several trans, trans-dienoic fatty acid isomers, including t,t- 18:2(n-6), act as sub-
strates for A6 desaturation in liver and brain, albeit at a much lower rate than for
c,c-18:2(n-6). Dienoic isomers of 18:2 containing trans bonds lack the properties of
essential fatty acids and interfere with normal conversion of 18:2(n-6) to 20:4(n-6).

Inhibition is primarily at the 46 desaturase, although d 5 desaturase activity increases
in some tissues. Dietary supplements containing trans-acids greatly intensify the signs
of mild essential fatty acid deficiency. Since the A desaturase is inhibited, conversion
of available 18:2(n-6) to 20:4(n-6) is reduced as is accumulation of 20:3(n-9) from
conversion of 18:l(n-9). Although trans-dienes are minor components (usually less
than 1%) of hydrogenated vegetable oils, complex interactions and interconversions
are possible.
   Questions about long-term effects of increased trans fatty acid intake still remain
unresolved. Continued investigations of trans-acids as a distinct, and significant class
of fatty acids are necessary to determine their influence on normal tissue develop-
ment and function, development of atherosclerosis, and altered cell metabolism in
tumor tissue.

7. Abnormal patterns of distribution and metabolism of long chain
   saturated and unsaturatedfatty acids
Despite the diversity of enzymes involved in unsaturated fatty acyl chain formation,
documented clinical defects in unsaturated acyl chain metabolism are relatively few.
This indicates that the major desaturation and elongation enzymes are essential in
supporting life-sustaining cellular processes. However, defects or deficiencies result-
ing in abnormal patterns of unsaturated fatty acid distribution have been document-

7.1. Essentialfatty acid deficiency

An inadequate supply of essential fatty acids resulting in the deficiency signs de-
scribed previously for rats (Section 5.2) is very rare in humans. Most normal diets
contain enough 18:2(n-6) and 18:3(n-3), or their metabolic products, to meet tissue
demands and adipose stores provide a protective buffer. However, severe deficiency
states have been observed in humans (especially in premature infants with restricted
adipose stores) on prolonged intravenous feeding or artificial milk formulations
without adequate lipid supplements (Fig. 7). Marked alterations of serum fatty acid
patterns characterized by depletion of (n-6) acids and a major increase in the 20:3-
(n-9) to 20:4(n-6) ratio are accompanied by severe skin rash, loss of hair and irritabil-
ity. These signs are reversed rapidly by supplementing with lipid emulsions contain-
ing esterified 18:2(n-6).
   In experimental animals, dietary deficiency of essential fatty acids is accompanied
by changes in fatty acyl composition of tissue and circulating lipids. Brain tissue
shows exceptional resistance to loss of essential fatty acids but modification of acyl
patterns can be achieved if the diet is started at an early age and continued long

7.2. Zinc deficiency

Many gross signs of zinc deficiency are similar to those observed in essential fatty
acid deficiency. Possible relationships between zinc, essential fatty acids and prosta-
glandins have been proposed, but direct connections at the metabolic level have not
been shown. Some studies, including those with humans, show a positive correlation
between plasma zinc and 20:4(n-6) levels. In chicks and rats, however, zinc deficiency
increases accumulation of 20:4(n-6) and apparently interferes with its normal me-
tabolism, possibly causing prostaglandin deficiency. Direct involvement of zinc in
desaturation and chain elongation has been proposed but not demonstrated. Other
potential sites for zinc involvement include intestinal absorption of fatty acids,
release and mobilization of acyl chains from complex lipids, and conversion of fatty
acids to oxygenated derivatives.
   Acrodermatitis enteropathica, a rare and serious disorder characterized by skin
lesions, gastrointestinal disturbances and retarded growth, is accompanied by a low
serum content of polyunsaturated fatty acids without accumulation of 20:3(n-9). Ad-
ministration of zinc dramatically reverses the symptoms, with an apparent correla-
tion between zinc and 18:2(n-6) levels. Relationships between zinc and essential fatty
acid metabolism in this disorder, and in general, are complex.

7.3. Other clinical disorders

In several human diseases or disorders abnormal patterns of polyunsaturated fatty
acids, attributable to insufficient dietary 18:2(n-6) or to abnormal metabolism of the
essential fatty acids, have been described [30]. Statistical comparisons of total fami-
lies of fatty acids, individual fatty acid components, and individual products of chain
elongation and desaturation steps permit broad groupings. Some disorders, includ-
ing cystic fibrosis, Crohn’s disease, Sjogren-Larsson syndrome, peripheral neurop-
athy, and congenital liver disease have diminished capabilities for desaturation or
chain elongation of polyunsaturated fatty acids. Essential fatty acid deficiency, alco-
holism, cirrhosis, Reye’s syndrome, and chronic malnutrition are accompanied by
abnormal patterns of essential fatty acids in serum phospholipids. Such analyses
point to potential defects in metabolic steps of polyunsaturated fatty acid metabo-
lism, and underline our limited knowledge of the potential for secondary deficits in
essential fatty acid metabolism in many human disorders.

7.4 Relationship to plasma cholesterol

In the past four decades, considerable evidence has emerged to support a correlation
between high dietary intake of saturated fats relative to polyunsaturated fats and the
occurrence of atherosclerosis and coronary disease. There is now reliable evidence
that the risk of coronary disease is proportional to serum cholesterol levels and total

serum cholesterol (particularly in low-density lipoproteins-see Chapters 14, 15 and
16) can be decreased following dietary intake of lipids enriched in polyunsaturated
fatty acyl chains. Factors such as platelet aggregation, blood pressure and vascular
obstruction may be influenced through potent oxygenated derivatives of polyun-
saturated fatty acids (Chapter 10). Although the appropriate balance of polyunsatu-
rated fatty acid intake cannot be defined precisely from available evidence, greater
attention to the overall composition of dietary fats seems to be individually benefi-

8. Future directions

Progress in understanding the desaturation and chain elongation of fatty acyl chains
and the influence of altered fatty acyl composition on lipid-related functions has been
steady and exciting over the last half century. Yet each progressive step reveals a need
to expand knowledge of the mechanisms, regulation and functions of these processes
in higher organisms, and of the extent to which available information applies to dif-
ferent mammalian tissues. Innovative approaches must be added to the cumulative
efforts of many disciplines.
   The function of (n-3) fatty acids remains a major question. A more specific defini-
tion of the essential role of 18:3(n-3) and polyunsaturated fatty acids of the (n-3)
family relative to 18:2(n-6) and its derivatives appears to be evolving. The balance
between (n-3) and (n-6) fatty acids, cells and tissues, in normal and pathological con-
ditions, needs more careful evaluation. It is difficult to generalize dietary require-
ments, considering both species and individual variability and the complex interplay
of a variety of biochemical and physiological functions. Regulation at the cellular
level and associations of desaturation and chain elongation enzymes, possibly as
membrane-bound, multi-enzyme complexes within membrane domains, must be
evaluated through improved isolation, purification, and reconstitution techniques,
coupled with perseverence, ingenuity, and serendipity.
   Studies based on mutant cell lines deficient in one or more components of the desa-
turation-chain elongation sequences, in combination with molecular genetics and
biological techniques and immunochemistry, must be applied to assess regulation at
the level of gene expression. Tissue specificity of expression of distinct desaturase
activities needs more definition. A broader use of normal, neoplastic, and hybrid cell
lines in culture, will facilitate regulation and manipulation of nutritional and environ-
mental factors that impinge on polyunsaturated fatty acid formation and utilization.
Potential alternatives to the classical pathways of (n-3) and (n-6) fatty acid metabo-
lism must be investigated more vigorously. The contribution of retrocoversion of
long chain polyunsaturated fatty acids at the subcellular level, and particularly in
peroxisomes will probably receive greater attention in the years ahead. Highly spe-
cialized biochemical approaches must be extrapolated cautiously to the complexities
of a continuously fluctuating cellular milieu. Despite the wealth of information
gained over the last half century, it is safe to say we have only just begun.

  1. Shinitzky, M. (1984) Physiology of Membrane Fluidity, Vol. 1, CRC Press, Boston, MA.
 2. Stubbs, C.D. and Smith, A.D. (1990) Essential fatty acids in membranes: physical properties and func-
      tion. Biochem. SOC.   Trans. 18,779-781.
 3. Vandenheuvel, F.A. (1965) Study of biological structure at the molecular level with stereomodel pro-
      jections. 11. The structure of myelin in relation to other membrane systems. J. Am. Oil. Chem. SOC.   42,
      48 1492.
 4. Burr, G.O. and Burr, M.M. (1929) On the nature and role of the fatty acids essential in nutriton. J.
      Biol. Chem. 82, 345-367.
 5. Mead, J.F. (1981) The essential fatty acids: past, present and future. Prog. Lipid Res. 20, 1-6.
 6. Bloomfield, D.K. and Bloch, K. (1960) The formation of d9-unsaturated fatty acids. J. Biol. Chem.
 7. Wdkil, S.J. (1961) Mechanism of fatty acid synthesis. J. Lipid Res. 2, 1 4 .
 8. Nugteren, D.H. (1962) Conversion in vitro of linoleic acid into y-linolenic acid by rat liver enzymes.
      Biochim. Biophys. Acta 60,656-657.
 9. Marcel, Y.L., Christiansen, K. and Holman, R.T. (1968) The preferred metabolic pathway from lino-
      leic acid to arachidonic acid in vitro. Biochim. Biophys. Acta 77, 671473.
10. Jeffcoat, R. (1979) The biosynthesis of unsaturated fatty acids and its control in mammalian liver.
      Essays Biochem. 15, 1-36.
1 I . Thompson, G.A., Jr. (1980) The Regulation of Membrane Lipid Metabolism, CRC Press, Boston,
12. Sprecher, H. (1981) Biochemistry of essential fatty acids. Prog. Lipid Res. 20, 13-22.
13. Emken, E.A. (1983) Biochemistry of unsaturated fatty acid isomers. J. Am. Oil Chem. SOC.60, 995-
14. Holloway, P.W. (1983) In: The Enzymes (Boyer, P.D., Ed.), Vol. 14, pp. 63-83, Academic Press, To-
15. Bernert, J.T. and Sprecher, H. (1 975) Studies to determine the role rates of chain elongation and desat-
      uration play in regulating the unsaturated fatty acid composition of rat liver lipids. Biochim. Biophys.
      Acta 398, 354-363.
16. Seubert, W. and Podack, E.R. (1973) Mechanisms and physiological roles of fatty acid chain elonga-
      tion in microsomes and mitochrondria. Mol. Cell. Biochem. 1,2940.
17. Enoch, H.G., Catala, A. and Strittmatter, P. (1976) Mechanism of rat liver microsomal stearoyl-CoA
      desaturase. Studies of the substrate specificity, enzyme-substrate interactions, and the function of lip-
      id. J. Biol. Chem. 251, 5095-5103.
18. Kaestner, K.H., Ntambi, J.M., Kelly, T.J. and Lane, M.D. (1989) Differentiation-induced gene ex-
      pression in 3T3-LI preadipocytes. A second differentially expressed gene encoding stearoyl-CoA de-
      saturase. J. Biol. Chem. 264, 14755-14761.
19. Stukey, J.E., McDonongh, V.M. and Martin, C.E. (1990) The OLE1 gene of Succhuromyces cerevisiue
      encodes the A9 fatty acid desaturase and can be functionally replaced by the rat stearoyl-CoA desatu-
      rase gene. J. Biol. Chem. 265,20144-20149.
20. Brenner, R.R. (1990) Endocrine control of fatty acid desaturation. Biochem. SOC.       Trans. 18, 773-775.
21. De Renobales, M., Cripps, C., Stanley-Samuelson, D.W., Jurenka, R.A. and Blomquist, G.J. (1987)
      Biosynthesis of linoleic acid in insects. Trends Biol. Sci. 12, 364-366.
22. Mead, J.F. (1984) The non-eicosanoid function of the essential fatty acids. J. Lipid Res. 25, 1517-1 521.

23. Fujiwara, Y.,Ishibashi, T. and Imai, Y. (1984) Cytoplasmic location of linoleoyl-CoA desaturase in
    microsomal membranes of rat liver. Arch. Biochem. Biophys. 233,402407.
24. Cook, H.W., Clarke, J.T.R. and Spence, M.W. (1983) Concerted stimulation and inhibition of desatu-
    ration, chain elongation, and esterification of essential fatty acids by cultured neuroblastoma cells. J.
    Biol. Chem. 258, 7586-7591.
25. Sirnopoulous, A.P., Kifer, R.R. and Martin, R.E. (1986) Health Effects of Polyunsaturated Fatty
    Acids in Seafoods, Academic Press, Toronto.
26. Tinco, J. (1982) Dietary requirements and functions of a-linolenic acid in animals. Prog. Lipid Res. 21,
27. Emken, E.A., Adolf, R.O., Rakoff, H., Rohwedder, W.K. and Gulley, R.M. (1990) Metabolism in
    vivo of deuterium-labelled Iinolenic and linoleic acids in humans. Biochem. SOC.   Trans. 18, 766769.
28. Cook, H.W. and Emken, E.A. (1990) Geometric and positional fatty acid isomers interact differently
    with desaturation and elongation of linoleic and linolenic acids in cultured glioma cells. Biochem. Cell
    Biol. 68,653-660.
29. Ernken, E.A. and Dutton, H.J. (1979) Geometrical and positional fatty acid isomers. The American
    Oil Chemists’ Society, Champaign, IL.
30. Holman, R.T. and Johnson, S. (1981) Changes in essential fatty acid profiles of serum phospholipids
    in human disease. Prog. Lipid Res. 20, 67-73.
31. Cook, H.W. and Spence, M.W. (1973) Formation qf monoenoic fatty acids by desaturation in rat
    brain homogenate. Effect of age, fasting and refeeding and comparisons with liver enzyme. J. Biol.
    Chem. 248, 1793-1796.
32. Cook, H.W. (1978) In vitro formation of polyunsaturated fatty acids by desaturation in rat brain:
    some properties of the enzymes in developing brain and comparisons with liver. J. Neurochem. 30,
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemistry oflipids, Lipoproteins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                171

                                                                                      CHAPTER 6

                                         Metabolism of triacylglycerols
                                                                            DAVID N. BRINDLEY

    Department o Biochemistry, Lipid and Lipoprotein Research Group, 328 Heritage Medical
                          Research Centre, University o Alberta, Edmonton, Alta., Canada

I. Introduction
Triacylglycerols play a major role in energy storage in animals, where they are depos-
ited in adipose tissue. When this storage is excessive it is manifested as obesity, and
there is considerable medical interest in trying to understand why some people are so
prone to this condition, whereas others find it equally difficult to gain weight. In
plants, the storage of triacylglycerols is best illustrated by the oil seeds, in which the
triacylglycerols provide energy for growth. These seeds constitute very important
commercial crops. Triacylglycerols are ideally suited to this storage function because
of the highly reduced state of their fatty acids. Triacylglycerols have high energy con-
tents - about 37 kJ/g, compared with 17 kJ/g for protein and 16 kJ/g for carbohy-
drate, including glycogen, which is also used to store energy in mammals. The other
advantange of the triacylglycerols is their insolubility in water, which means that they
do not alter the osmotic pressure of the cell.
   Most of the triacylglycerols in animals are stored in adipose tissue. However, tria-
cylglycerolscan be deposited in liver, heart, and skeletal muscle under several condi-
tions of metabolic stress when the supply of fatty acids from adipose tissue exceeds
the need, or capacity of the cells to oxidize them. The formation of triacylglycerols
removes the potentially toxic effects of excess fatty acids and acyl-CoA esters which
could damage membranes and inhibit enzymes. The process of triacylglycerol synthe-
sis also regenerates CoA. When required, the triacylglycerols are hydrolyzed by in-
tracellular lipases so that the fatty acids can then be oxidized.
   The oxidation of fatty acids in times of fasting, starvation, stress or diabetes pro-
vides a relatively abundant supply of energy and spares the need for glucose oxida-
tion in muscles and the liver. This makes glucose more available for the brain. Fur-
thermore, the liver is able to incorporate some of the acetyl-CoA formed after fatty
acid oxidation into ketones which can provide an alternate form of fuel for the brain
and muscle tissues. The triacylglycerol store has a further function in starvation in

that the glycerol released by adipose tissue during activated lipolysis can be used for
gluconeogenesis by the liver in order to supply the brain with glucose.
   A major function of triacylglycerols is to transport fatty acids. Since these com-
pounds are insoluble in the aqueous environment of the cells and of the blood and
lymph, the triacylglycerol must be stabilized by association with other lipids and pro-
teins. Such aggregates are called lipoproteins. There are two major classes of triacyl-
glycerol-rich lipoproteins, namely, chylomicrons and very low density lipoproteins
(VLDL), as shown in Table I. Chylomicrons carry absorbed dietary fat from the
small intestine to other organs, whereas VLDL mainly carry triacylglycerol from the
liver to other organs.
   Lipoprotein metabolism is of considerable clinical relevance. A high rate of VLDL
secretion from the liver, if coupled with a relatively low rate of removal from the
circulation, leads to hypertriglyceridemia. This condition, in combination with a low
circulating concentration of high-density lipoprotein (HDL), is considered to be a
major risk factor in the development of premature atherosclerosis (see also Chapter
16). VLDL are eventually converted to intermediate density lipoproteins (IDL) and
then to low density lipoproteins (LDL) after the triacylglycerol is removed, as de-
scribed in Chapters 15 and 16. The IDL and LDL particles retain much of the choles-
terol that was originally used to package and stablize the VLDL. Thus, if the number
of VLDL particles that are secreted increases, the flux into LDL is also increased.
Hypercholesterolemia results if this flux is not balanced by an increased rate of re-
moval of LDL and VLDL from the circulation. An increase in the concentration of
LDL in the serum, especially when accompanied by a low concentration of HDL, is
thought to provide one of the best indications of an increased risk for atherosclerosis
(Chapters 15 and 16).

Composition and properties of the triacylglycerol-rich lipoproteins from man

Property                              Chylomicrons                      Very low density

Density (g/ml)                        0.92-0.96                         0.95-1.006
Diameter (nm)                         75-1000                           30-75
Composition (dry weight %)
  Triacylglycerol                    80-95                              5545
  Phospholipids                      3-8                                15-20
  Free cholesterol                   1-3                                10
  Esterified cholesterol             2 4                                5
  Protein                            1-2                                9-10
Major apolipoproteins                A-1, A-11, A-IV                    B l W , C-I, C-I1
                                     B,, C-I, C-11, C-111               C-111, E
Minor apolipoproteins                E                                  A-I, A-11, A-IV

2. Biosynthesis of triacylglycerols

The initial understanding of the pathways of triacylglycerol synthesis began in the
early 1950s with the discovery by Kornberg and Pricer that fatty acids are activated
to acyl-CoA esters before they are esterified to phosphatidate. This discovery was
followed by descriptions from the groups of Kennedy and Shapiro [I] of how phos-
phatidate is converted to various phospholipids and triacylglycerol. In the early
 1960s Hubscher and Clark showed that the small intestine can synthesize triacylgly-
cerols from monoacylglycerols and that this conversion is involved in the transport
of dietary fat across the enterocytes, which are the absorptive cells of the small in-
testine [2]. Later in that decade Hajra and Agranoff provided evidence that dihy-
droxyacetone phosphate can act as an alternative acyl-acceptor to glycerolphosphate
for initiating the synthesis of glycerolipids [I].
   Most of the work in this area of biochemistry is now directed to understanding
how the metabolism and transport of triacylglycerols is controlled. These investiga-
tions have generally not advanced as quickly as those dealing with the metabolism of
water-soluble compounds. Lipids are difficult to use as enzyme substrates, and the
kinetics are very complicated. Most of the enzymes are bound tightly to membranes,
which makes them relatively more difficult to purify and characterize. The rest of this
Chapter will attempt to provide an overview of the routes of triacylglycerol metabo-
lism in mammals and the control of these pathways as understood at present.

2. I . Biosynthesis of phosphatidate

The synthesis of triacylglycerols begins with the activation of fatty acids to acyl-CoA
esters (Fig. l), which involves the input of energy from ATP. During this reaction
ATP is cleaved to AMP and pyrophosphate. Acyl-CoA is then used by the various
acyltransferases in the synthesis of carboxylic ester bonds. Triacylglycerols generally
contain long-chain fatty acids (C,6 or greater). Consequently, the most important
activating enzyme is the long-chain acyl-CoA synthetase (Chapter 3), which is found
mainly in the endoplasmic reticulum and mitochondria of mammalian cells. How-
ever, many cells also contain distinct acyl-CoA synthetases that can activate short-
and medium-chain acids. These acids can also be esterified in some situations. For
example, the milk of many species contains short- and medium-chain acids, and this
may facilitate the digestion of the triacylglycerol by young animals (Section 3.1). Co-
conut fat also has a high proportion of lauric acid (C,J.

Fatty acid   +   ATP   F
                       -           Fatty acyl-AMP + pyrophosphate

Fatty acyl-AMP     +   C o A d F a t t y acyl-CoA   + AMP

Fig. 1. Reaction mechanism of acyl-CoA synthetase.
       Glvcerol                                                 Glucose
       -   'I
            4                   O2           H2°                  1
                                      1                           I
                                                                  c =o 0
            I   Q
       H2C0 -       - 0-                                        H2C0 - P - 0 -
                0-                                                     0-

sn -Glycerol 3-phosphate                           Dihydroxyacetone phosphate

       H2C0 - 8R1                                               H CO - CR1
         I                            5                          21
       HOCH            4                                          c-0

         1 7
       H2C0 - 0 - 0-           ,AD6       \ADPH     t Ht        H2L -
                                                                   - 0-
              0-                                                 0-
lyso-Phosphatidate                                 Acyldihydroxyacetone phosphate

           ' 0 C O A

       70CRl                                       Synthesis of alkyl- and alkenyl-
   H2L0 - f - 0-
Fig. 2. Synthesis of phosphatidate. Enzyme activities are indicated by: (1 ) glycerol-3-phosphate dehydro-
genase EC 1 . I .99.5; (2) glycerol-3-phosphate dehydrogenase (NAD') EC; (3) glycerolphosphate
acyltransferase; (4) dihydroxyacetone phosphate acyltransferase; (5) acyldihydroxyacetone phosphate
reductase; (6) monoacylglycerol phosphate (lysophosphatidate) acyltransferase.

   The main acceptor for acyl-CoA in most tissues is thought to be sn-glycerol-3-
phosphate that is formed by glycolysis. In liver the phosphorylation of glycerol is also
an important route of glycerol phosphate production (Fig. 2). Glycerol-3-phosphate
is first acylated to 1-monoacylglycerol-3-phosphate (lysophosphatidate) by glycerol-
phosphate acyltransferase. A different acyltransferase is responsible for esterifying
the hydroxyl at the position-2 of lysophosphatidate.
   In the liver, the glycerolphosphate acyltransferase is divided almost equally be-
tween the mitochondrial and microsomal fractions. By contrast, in heart, kidney,
adrenal glands, and adipose tissue, the mitochondrial activity is only about 10% of
the total [l]. The microsomal activity is found on the cytosolic side of the rough and
smooth endoplasmic reticulum, and the mitochondrial enzyme is on the inner surface
of the outer mitochondrial membrane. The mitochondrial and microsomal acyltrans-
ferases are different enzymes, as is shown by the relative resistance of the mitochon-


Acyltransferases                Substrates                       Effect of N-ethylmaleimide

Endoplasmic reticulum           Glycerol phosphate or
                                dihydrox yacetone phosphate,
                                saturated and unsaturated
                                acyl-CoA                         Inhibition
Mitochondria1 outer membrane    Glycerol phosphate,
                                saturated acyl-CoA               None
Peroxisomes                     Dihydroxyacetone phosphate
                                saturated acyl-CoA               Slight stimulation

drial acyltransferase to inhibition by heat, proteolytic enzymes, and N-ethylmalei-
mide (Table 11). The mitochondrial enzyme also has a lower K than the microsomal
acyltransferase for acyl-CoA esters and for glycerolphosphate, and the mitochon-
drial enzyme prefers saturated, rather than unsaturated, acyl-CoA esters, for exam-
ple, palmitoyl-CoA. By contrast, the acyltransferase from the endoplasmic reticulum
is able to use a variety of saturated and unsaturated acyl-CoA esters.
   The esterification at the position-2 by monoacylglycerol phosphate (lysophospha-
tidate) acyltransferase is relatively selective for unsaturated fatty acids in both mito-
chondrial and microsomal fractions. However, this specificity is by no means abso-
lute, and the fatty acid composition of the newly synthesized phosphatidate will de-
pend upon the fatty acids available in the cell. The activity of the monoacylglycerol-
phosphate acyltransferase is relatively low in mitochondrial fractions in which lyso-
phosphatidate is often the major product of esterification from glycerolphosphate.
By contrast, phosphatidate is the main product of esterification with microsomal
   The function of the relatively high mitochondrial glycerolphosphate acyltransfer-
ase activity, particularly in liver, is unclear. Phosphatidate is an intermediate in the
synthesis of diphosphatidylglycerol (cardiolipin), which is abundant in the inner mi-
tochondrial membrane (Chapter 7). However, the flux of phosphatidate into diphos-
phatidylglycerol is relatively low when compared with overall glycerolipid synthesis.
The conversion of phosphatidate to triacylglycerol and to phosphatidylcholine is
normally considered not to take place in mitochondria.
   The lower K,,, values of the mitochondrial acyltransferase for glycerolphosphate
and acyl-CoA esters should mean that the mitochondrial system is favored compared
with the enzyme in the endoplasmic reticulum, particularly at low substrate supply.
The physiological function of the mitochondrial acyltransferase is therefore uncer-
tain and remains a major unanswered question in the control of glycerolipid synthe-
sis. Phosphatidate can be hydrolyzed to glycerolphosphate in a substrate cycle (Fig.
3), which might help in regulating the balance between B-oxidation and esterification

      Glycerol phosphote
      lyso -phosphatidote


         1            -      CDP -diocylglycerol +Phosphatidylinositol
         I   II


  @     [::
       II             @ -r   Phosphatidylcholine




 Fig. 3. Metabolism of phosphatidate. Enzyme activities are indicated by: ( I ) phosphatidate phosphohydro-
 lase; (2) diacylglycerol acyltransferase; (3) choline phosphotransferase; (4)ethanolamine phosphotransfer-

 (see Section 4.1). Alternatively, the phosphatidate (or lysophosphatidate) might be
 transferred to the endoplasmic reticulum for the synthesis of triacylglycerol and
    The other route for the biosynthesis of phosphatidate de novo is by the esterifica-
 tion of dihydroxyacetone phosphate (Fig. 2). This reaction can be catalyzed in the
 endoplasmic reticulum by glycerolphosphate acyltransferase, which can use either
 glycerolphosphate or dihydroxyacetone phosphate, which compete for the enzyme
 (Table 11). In addition, there is a peroxisomal dihydroxyacetone phosphate acyl-
 transferase that neither uses glycerolphosphate as a substrate, nor is inhibited by it.
 The peroxisomal enzyme is not inhibited by N-ethylmaleimide (which further indi-
 cates that it is a different enzyme from the microsomal glycerolphosphate acyl-

transferase). The peroxisomal acyltransferase has a higher reaction rate with the
CoA esters of saturated than with unsaturated fatty acids; this specificity therefore
favors the incorporation of saturated fatty acids into the 1-position of acylglycero-
lipids that may be synthesized by this route.
   Acyldihydroxyacetone phosphate can undergo two different biosynthetic modifi-
cations (Fig. 2). First, it can be converted into alkyldihydroxyacetone phosphate,
which can serve as the precursor for the synthesis of the alkyl- and alkenyl-lipids
(Chapter 8). Alternatively, the acyldihydroxyacetone phosphate can be converted to
1-monoacylglycerol-3-phosphate (lysophosphatidate) by a reductase that uses
NADPH. This reductase is found in both peroxisomes and the endoplasmic reticu-
lum, and the same reductase probably acts on alkyldihydroxyacetone phosphate.
   It is generally accepted that acyldihydroxyacetone phosphate is an obligatory in-
termediate in the synthesis of alkyl- and alkenyl-lipids (Chapter 8). However, the
quantitative importance of the esterification of dihydroxyacetone phosphate relative
to glycerolphosphate for the synthesis of glycerolipids is in dispute. On the one hand
it has been argued that the esterification of glycerolphosphate must predominate.
This conclusion takes into account the relatively high concentration of glycerolphos-
phate in the cell and the K,, K,, and V,,, values for glycerolphosphate and dihydrox-
yacetone phosphate for the glycerolphosphate acyltransferase that is located in the
endoplasmic reticulum [3]. These calculations do not consider the acyltransferases
that exist in the mitochondria and peroxisomes. However, it is not clear whether
these organelles can efficiently donate phosphatidate for the synthesis of triacylglyce-
rols and the zwitterionic phospholipids.
   Other arguments are based upon the use of glycerol labeled either with 3H at the
2-position, or with I4C. Conversion of glycerolphosphate to dihydroxyacetone phos-
phate before its incorporation causes the loss of 3H, whereas I4C is retained. The
relative activities of the two pathways can be estimated by comparing the 3H/'4C
ratios in lipids with that in the glycerolphosphate. Using this calculation, 50-75% of
the glycerolipid synthesis by rat liver slices appeared to occur by acylation of dihy-
droxyacetone phosphate [ ] A similar technique showed that about 56% of the phos-
phatidylglycerol and 64% of the phosphatidylcholine were formed from dihydroxy-
acetone phosphate in type I1 cells of lung [5].
   An alternative approach was to incubate cells with a mixture of D-[U-'~C]-        and
~-[3-~H]glucose to measure the labelling of the lipids. This procedure generated
[4-3H]NADPH, which was incorporated into position 2 of glycerolipids by the dihy-
droxyacetone phosphate pathway (Fig. 2). It was then possible to calculate the rela-
tive contribution of the two pathways to the synthesis of glycerolipids on the assump-
tion that the 3H/'4Cratio in alkyl- and alkenyl-lipids results entirely from synthesis by
the dihydroxyacetone phosphate pathway. A correction factor also had to be applied
to compensate for some labeling of glycerolphosphate at the 2-position. From these
results it was estimated that 49-61% of the glycerolipid synthesized by BHK-21 and
BHK-Ts-a/lb-2 cells occurs by acylation of dihydroxyacetone phosphate [6].
   It is also possible to postulate that the dihydroxyacetone phosphate pathway
should be important in glycerolipid synthesis from theoretical considerations of co-
factor requirements. Most anabolic pathways use NADPH as a cofactor in reductive
synthesis. The formation of glycerolipids from glucose via the dihydroxyacetone
phosphate pathway fulfills this requirement (Fig. 2). NADPH production increases
when there is active fatty acid synthesis. This increase could also favor the incorpora-
tion of the fatty acids into triacylglycerols by the dihydroxyacetone phosphate path-
way. By contrast, the conversion of dihydroxyacetone phosphate to glycerolphos-
phate requires NADH, which is normally involved in reductive degradation. In the
liver, glycerolphosphatedehydrogenase, which catalyzes this interconversion has fur-
ther functions in gluconeogenesis from glycerol and in maintaining the redox state.
   The relative importance of the three acyltransferases that are described in Fig. 2 in
initiating glycerolipid synthesis are not well understood; nor do we know whether
their relative contributions change in different physiological conditions. Despite the
possible importance of the dihydroxyacetone phosphate pathway, it is normally as-
sumed that the synthesis of triacylglycerols takes place mainly in the endoplasmic
reticulum by the esterification of glycerolphosphate.

2.2. Conversion o phosphatidate
                 f                to   triacylglycerol

Phosphatidate lies at a branch point in glycerolipid synthesis which provides three
routes for further metabolism (Fig. 3). Hydrolysis of phosphatidate by A-type phos-
pholipases may complete an important substrate cycle back to glycerolphosphate.
This hydrolysis probably prevents the excessive accumulation of phosphatidate in the
membranes. Second, phosphatidate can be converted to CDP-diacylglycerol,which
provides a precursor for the synthesis of acidic phospholipids, phosphatidylinositol,
phosphatidylglycerol,and diphosphatidylglycerol(Chapter 7). Third, phosphatidate
can be hydrolyzed to diacylglycerol, which in turn can serve as a common precursor
in the synthesis of the zwitterionicphospholipids, phosphatidylcholine, and phospha-
tidylethanolamine (Chapter 7) and also for the synthesis of triacylglycerol. We shall
now discuss this latter route of phosphatidate metabolism.
   The conversion of phosphatidate to diacylglycerol is catalyzed by phosphatidate
phosphohydrolase. This enzyme activity has been reported to occur in plasma mem-
branes, lysosomes, mitochondria, the endoplasmic reticulum, and the cytosol [7,8].
However, it is difficult to be certain about the distribution of the phosphohydrolase,
since the assays used in many of these studies measured the release of inorganic phos-
phate or water-soluble phosphates from phosphatidate [8,9]. These phosphates can
also arise by deacylation of the phosphatidate by phospholipase A activities to form
glycerolphosphate, which can be further hydrolyzed by acid or alkaline phospha-
tases. This probably explains why the production of phosphate from phosphatidate
was reported to be high in lysosomal fractions which are known to contain high ac-
tivities of A-type phospholipases and acid phosphatase.

   Subsequent studies failed to detect significant phosphatidate phosphohydrolase
activity in lysosomal fractions [101. In this work phosphatidate phosphohydrolase
activity was measured in rat liver by using phosphatidate labeled with [3H]palmitate
and by following the formation of diacylglycerol [9,10]. The results demonstrated
that at least two different phosphatidate phosphohydrolases exist in the liver. The
first activity was in the cytosol and was dependent upon Mg2+.This activity was in-
hibited by thiol group reagents. The assay therefore relied upon measuring specifi-
cally the phosphohydrolase activity that was inhibited by N-ethylmaleimide [9,10].
The cytosolic phosphatidate phosphohydrolase was able to translocate to the endo-
plasmic reticulum [7,8] and mitochondria [I 11 where phosphatidate can be synthe-
sized. The response of this Mg2+-dependentphosphatidate phosphohydrolase to hor-
mones and metabolites and its subcellular distribution is compatible with a regula-
tory role in glycerolipid synthesis [7,8]. This will be discussed later (Section 4.2).
   The second phosphatidate phosphohydrolase activity in the liver is not inhibited
by N-ethylmaleimide, nor is it stimulated by Mg2+.This activity is located in the
plasma membrane which is compatible with a role in signal transduction following
the agonist-induced stimulation of phospholipase D [101. The action of phospholi-
pase D leads to the conversion of phosphatidylcholine to phosphatidate which can
then be metabolized to diacylglycerol by the phosphohydrolase. The purpose of this
pathway may be to produce a more sustained production of diacylglycerol (an activa-
tor of protein kinase C as discussed in Chapter 7) than can be obtained through the
hydrolysis of phosphatidylinositolbisphosphate after activation of specific phospho-
lipases of the C-type. Furthermore, there is evidence that phosphatidate itself may act
as a second messenger [ 12-1 41. Thus the N-ethylmaleimide-insensitive phosphohy-
drolase may act to regulate the balance between two second messengers in the plasma
membrane [ 101. However, the production of diacylglycerol in the plasma membrane
is unlikely to be directly involved in providing a major part of the diacylglycerol that
is used to synthesize triacylglycerol, phosphatidylcholine and phosphatidylethanola-
mine. It is however, not established whether the pathway of glycerolipid synthesis can
provide some of the phosphatidate and diacylglycerol that is used for cell signalling.
   The final stage of triacylglycerol synthesis is completed by the action of diacylgly-
cerol acyltransferase, which forms the ester bond at position-3 of diacylglycerol (Fig.
3). This enzyme is located on the endoplasmic reticulum and it can be stimulated by
Mg2+.As far as we know, this enzyme competes for a common pool of diacylglycerol
with the choline- and ethanolamine-phosphotransferases       that are responsible for the
synthesis of phosphatidylcholine and phosphatidylethanolemine.

2.3. Conversion of monoacylglycerol to triacylglycerol

The pathways that have been described so far are the possible routes that occur in a
wide variety of cell types for the synthesis de novo of triacylglycerols. However, there
is another route for triacylglycerol synthesis that operates mainly in the enterocytes

of the small intestine. This pathway will be discussed in relation to its role in lipid
absorption and transport (Section 3).
    However, significant activity of this pathway was also detected in neonatal rat liver
[ 151. The route of synthesis involves the esterification of sn-2-monoacylglycero1by an
acyltransferase that is relatively specific for this substrate rather than for rac-1 -mo-
noacylglycerol, or 1- and 2-monoethers of glycerol. This substrate specificity differs
from the monoacylglycerol acyltransferase of the small intestine which can use these
other substrates relatively efficiently. This observation and differences in susceptibil-
ity to heat and to histidine and lysine reagents shows that there are tissue specific
isoenzymes of monoacylglycerol acyltransferase [151.
   The physiological function of the hepatic monoacylglycerol acyltransferase is not
certain, but it appears likely to be related to the high amount of dietary fat consumed
during the suckling period. The source of the monoacylglycerol for the hepatic en-
zyme is not clear. It has been suggested that the substrate comes from hydrolysis of
triacylglycerol within the hepatocytes, and that the acyltransferase may be involved
in recycling monoacylglycerol to triacylglycerol [ 151. A further function for the mo-
noacylglycerol pathway that has been postulated is in the esterification of xenobiotic
organic acids which can be incorporated into triacylglycerol [ 161.
    Monoacylglycerol acyltransferase has also been detected in the livers of chicken
embroys [I71 and in rat adipose tissue [18]. The latter activity was relatively low com-
pared to the rate of glycerol phosphate acyltransferase. However, the monoacylglyce-
rol acyltransferase was measured by using 2-hexadecylglycerol or 2-octadecenylgly-
cerol as model substrates. By analogy with the enzyme in rat liver [15], the activity
from adipose tissue might have been higher had 2-monoacylglycerol been used as a


      HZCOCR3                               H2COH                                  H2COH                  H2COH

      Triacylglycerol                    Diacylglycerol                      2-Monoacylglycerol          Glycerol

B         0                                                                             o
      HZCOCR,                                                                   H2COCRI
      0 1                                                                           1
    R2COCH                                                                      HOCH
        I    0          t                  Phospholipase A 2                       l        o       t
      HzCO - P - OCH2CH2NICH3l3                                                                 -
                                                                                HzCO - P OCH2CH~(CH3I3
             '                                                                         0-

      Phosphatidylchollne                                                       Lysophosphatidylcholme

c     Cholesterol ester               Cho'esterol ester
                                                               Cholesterol t Fatty acid

Fig. 4. Reactions in lipid digestion in the gastrointestinal tract.

3. Digestion, absorption, and transport of lipids

Lipids constitute an important part of the diet, and in the more developed countries
contribute about 4045% of energy intake. This means that many people eat 60-1 30
g of fat each day and that the body has to digest, absorb, resynthesize and transport
this quantity of lipid. Approximately 90-95% of the triacylglycerols from the diet are
absorbed from the intestinal lumen, whereas only about 50% of the dietary cholester-
ol is absorbed. Some of the cholesterol in the intestinal lumen comes from the bile
which also contains bile acids formed from cholesterol in the liver. The incomplete
absorption of cholesterol and bile acids by the intestine thus represents the major
routes of cholesterol excretion.

3.1. Digestion of lipids

Relatively little digestion of triacylglycerols takes place in the stomach, but the action
of the proteolytic enzymes releases lipid from food particles, and a coarse emulsion
is formed by the churning action of the stomach. A gastric lipase, that is distinct from
pancreatic lipase, has been identified in the stomach contents of several mammals,
including man [19-221. This lipase is active at pH 3 4 , and it preferentially releases
short- and medium-chain-length fatty acids from acylglycerols (Fig. 4). These fatty
acids are found particularly at position-3 of the ingested triacylglycerols of the milk
of several species. The partial hydrolysis of triacylglycerols to 1,Zdiacylglycerols
makes the milk fat micelles more susceptible to the subsequent action of pancreatic
lipase in the small intestine. The short and medium chain fatty acids can be absorbed
directly by the gastric muscosa. They are bound to albumin in the blood and carried
to the liver, where they can be readily oxidized to produce energy.
   The lipase that is active in digesting triacylglycerols in the stomach is species specif-
ic. In human beings and baboons the lipase is mainly of gastric origin, but traces of
lingual lipase are secreted from the tongue and travel to the stomach. Mice and rats
have high lingual lipase, but little gastric lipase. By contrast, rabbits and guinea pigs
have no lingual lipase and rely on the gastric lipase [21].
   The remaining lipid enters the proximal part of the small intestine, where the pH
rises to 5.8-6.5. The coarse droplets of oil become coated with bile salts that are
secreted by the liver. These compounds are amphiphilic. They dissolve at the oil-
water interface with their hydrophobic faces pointing into the oil and the hydrophilic
surfaces interacting with the aqueous phase of the luminal contents. The bile salts
help to disperse the oil droplets and donate a negative charge to their surfaces. A
protein called colipase (molecular weight of about 10,000) [20], which is secreted by
the pancreas, is then absorbed onto the surface of the oil droplets (Fig. 5). This pro-
tein acts as an anchor for the attachment of pancreatic lipase at the oil-water inter-
face. The bile salts, colipase, and lipase are thought to interact in a ternary complex
which in the presence of Ca2' is able to hydrolyze the 1- and 3-ester bonds of triacyl-

Complex of  +
            +         OIL DROPLET             Cholesterol esters
lipase, colipase
and bile salts
                                              Vltarnin esters

                                            MICELLAR SOLUTION
                                              Fatty acids
                                              lyso-Phosphol ipids
                                              Bile salts

                        ll\W                ABSORPTIVE PHASE

membrane of
en terocyle
Fig. 5 . Schematic representation of lipid digestion and absorption in the small intestine.

glycerol (Fig. 4A). About 85% of the digestion of dietary triacylglycerol in nonrumi-
nant animals ends with the formation of 2-monoacylglycerol, since pancreatic lipase
is unable to hydrolyze the ester bond at position-2.
   Limited hydrolysis of 2-monoacylglycerols does take place in nonruminant ani-
mals, but it is catalyzed by a nonspecific ester hydrolase that is secreted by the pan-
creas. The same enzyme probably also hydrolyzes cholesterol esters (Fig. 4C), and
the esters of fat-soluble vitamins.
   In ruminants the complete digestion of triacylglycerols is effected by lipases that
are produced by bacteria in the rumen. Bacteria are also responsible for hydrogenat-
ing unsaturated fatty acids that are present in the diets of ruminant animals, which
accounts for the highly saturated nature of their body fats.
   Pancreatic secretions also contains phospholipases of the A,- and A,-types, which
remove fatty acids from the 1- and 2-positions of phospholipids, respectively (Fig.
4B). These phospholipids can be derived either from the diet, or from bile delivered
to the small intestine. The digestion of phospholipids need not be complete: 3 M O %
of the dietary phosphatidylcholine can be absorbed as lysophosphatidylcholine
(Figs. 4B and 5).
   The process of digestion converts lipids that have limited abilities to interact with
water into more polar compounds with amphiphilicproperties. Thus, triacylglycerols
are transformed into 2-monoacylglyerols, cholesterol esters into cholesterol, and

phospholipids into their lyso-derivatives, which are strong detergents. The fatty acids
that are released are ionized at the pH that prevails in the intestinal lumen. The fatty
acids salts are also surface active, that is, they can interact with both lipid and aque-
ous environments.
  As digestion proceeds, the monoacylglycerols,fatty acids, lysophospholipids, chol-
esterol, and bile salts dissociate from the surface of the oil droplets to form a micellar
solution (Fig. 5). The micelles are aggregates of amphiphilic lipids that orient them-
selves with the hydrophobic regions on the inside of the micelles and the polar groups
exposed to the aqueous enviroment (Chapter 1). Further digestion of phospholipids
and cholesterol esters can take place at the microvillus membrane of the enterocytes,
which contain phospholipases and a cholesterol ester hydrolase.

3.2. Absorption of lipidsfrom the small intestine

Most of the lipids are absorbed from the small intestine (in the jejunum) with the
exception of the bile salts, which remain in the lumen of the small intestine to facili-
tate further digestion. The bile salts are finally absorbed in the distal ileum and are
transported back to the liver by the portal blood in a cycle that constitutes the entero-
hepatic circulation.
   Lipid absorption occurs when the micellar solution of lipids comes into contact
with the microvillus membrane of the enterocytes (Fig. 5). The lipids are probably
transported across the membrane by an energy independent process which relies on
the maintenance of an inward diffusion gradient. This gradient can partly be a-
chieved by the attachment of the fatty acids to specific intracellular binding proteins.
However, the ultimate driving force for absorption probably comes from the rapid
re-esterification of the lipids, which is an ATP-dependent process depending upon
activation of fatty acids to acyl-CoA esters (Fig. I).
   In nonruminant animals the major acyl-acceptor is the 2-monoacylglycerol that is
formed in the lumen of the intestine during the partial hydrolysis of triacylglycerol
[2]. Triacylglycerol is resynthesized by the sequential actions of monoacylglycerol

                       Acyl-CoA                          Acyl-CoA

A) 2-MONOACYLGLYCEROL U                 D iAcYtGLYcERoL U        :        K   iAcYtGtYcmt
                      Monoacylglycerol                    Diacylglycerol
                       acyltran sferase                  acyltran sferase


B I 1-MONOACY LGLYCEROPHOSPHORYLCHOL INE                               PHowHAmYtcHotiNE
                                            Lysophosphatidylchol ine

Fig. 6. Major routes for the resynthesis of triacylglycerol and phosphatidylcholine in the enterocytes of the
small intestine.

acyltransferase and diacylglycerol acyltransferase (Fig. 6). This pathway is estimated
to account for 75585% of the total synthesis of triacylglycerol in the enterocytes of
nonruminant animals. The remainder comes from the esterification of glycerolphos-
phate (and perhaps of dihydroxyacetone phosphate), as described earlier (Fig. 2).
Synthesis of triacylglycerol from glycerolphosphate is necessary, since some of the
dietary triacylglycerol is completely hydrolyzed to glycerol. In ruminant animals,
bacterial lipases are responsible for the hydrolysis of 2-monoacylglycerols; therefore
the pathways for the synthesis de novo of the triacylglycerol are required.
   The diacylglycerol that is formed as an intermediate in triacylglycerol synthesis can
also be used to synthesize phosphatidylcholine. However, a prominent route of phos-
phatidylcholine production in enterocytes of nonruminant animals is through the
reesterification of the 1-monoacylglycerophosphorylcholine (lysophosphatidylcho-
line) (Fig. 6B) that is formed by the partial digestion of dietary phosphatidylcholine.
The reactions of glycerolipid synthesis in the enterocytes are very rapid: they take
place within seconds after the precursors are absorbed.
   The transport and metabolism of the absorbed cholesterol is much slower than
that of triacylglycerols. The estimated tY2 for absorbed cholesterol in the enterocyte
is about 12 h. During absorption the cholesterol becomes incorporated into the mem-
branes of the enterocytes and diluted with endogenous cholesterol. A large propor-
tion of the cholesterol that is transported from the enterocyte is esterified, mainly
with oleic acid. This occurs through the action of acyl-CoA:cholesterolacyltransfe-
rase (ACAT). In addition the cholesterol esterase reaction probably also works in
reverse to synthesize some cholesterol ester [2].

3.3. Formation of chylomicrons and VLDL

The transport of lipids from the small intestine requires that they be packaged into a
physically stable form that can exist in an aqueous environment. Thus, the most hy-
drophobic of the lipids (cholesterol esters and triacylglycerols) are coated with a layer
of amphiphilic compounds including phosphatidylcholine, cholesterol, and various
apolipoproteins. Apart from stabilizing the surface of the lipoproteins, the apolipo-
proteins also provide address labels that govern which cells in the body receive and
metabolize these lipoproteins (Chapter 16). The main lipoproteins responsible for
transporting dietary fat from the intestine are the chylomicrons. However, the intes-
tine is also able to secrete VLDL, and is responsible for about 10% of this lipoprotein
that appears in the blood. The chylomicrons are bigger than the VLDL, and they
have a higher content of triacylglycerol (Table I). However, the size and composition
of the lipoproteins can vary according to the composition of the diet, and the rates of
lipid absorption. Thus, with a high intake of dietary fat and at the peak of absorp-
tion, the chylomicrons tend to be larger and contain more triacylglycerol than when
the rate of lipid transport is lower.
   The assembly of chylomicronsand VLDL begins with the resynthesis of triacylgly-

cerols, which takes place in the smooth endoplasmic reticulum in the apical region of
the enterocyte. The triacylglycerols appear in the form of lipid droplets in the cister-
nae of the smooth endoplasmic reticulum within minutes of exposure of the cells to
micellar lipid in the lumen (Fig. 7). The droplets are stabilized by phospholipids and
proteins produced in the rough endoplasmic reticulum. As time progresses, the drop-
lets increase in number as the endoplasmic reticulum extends and pinches off to form
vesicles. It is thought that these lipid-filled vesicles fuse with the Golgi apparatus. The
nascent chylomicrons and VLDL are then carried to the lateral surfaces of the ente-
rocyte by vesicles in the process of exocytosis. Fusion of the Golgi-derived vesicles
and surface membranes takes place, and the chylomicrons and VLDL are secreted
into the intercellular spaces which drain into the lymph vessels (Fig. 7). The lipo-
proteins in these vessels pass via the thoracic duct and enter the circulation at the
level of the jugular vein.
   The hydrolysis of the triacylglycerols of chylomicrons and VLDL involves their
binding through the apo C-I1 on their surfaces to an enzyme called lipoprotein lipase
(Chapter 15), which is found in the capillary beds of various extrahepatic tissues in-
cluding skeletal muscle, cardiac muscle, and adipose tissue. The enzyme is anchored
to polysaccharide chains on the endothelial wall of the capillaries (Fig. 8). Lipopro-

                        Lymph vessel

Fig. 7. Schematic representation of the transport of lipids through the enterocytes and the formation of
chylomicrons. The figure is reproduced from [2] by the permission of Butteworth's Scientific Ltd. SER,
smooth endoplasmic reticulum; RER, rough endoplasmic reticulum. Further details and original referen-
ces can be obtained from [2].


I             Endothelial surface

Fig. 8. Schematic representation of the binding of a chylomicron to lipoprotein lipase. The figure is repro-
duced from [2] by the permission of Butterworth’s Scientific Ltd.

tein lipase is synthesized by the cells that underlie the capillaries. The body is able to
direct chyiomicrons and VLDL to particular organs through its tissue-specific con-
trol of lipoprotein lipase activity. This subject will be considered in more detail later
in this chapter.
   The metabolic fate of released fatty acids depends on the particular cells into which
they are delivered. Fatty acids are readily used for energy production by muscle tis-
sues. If the uptake exceeds the immediate requirement for&oxidation, the fatty acids
are temporarily stored as triacylglycerols within the muscle cells. Storage is, of
course, the major fate of fatty acids that enter adipocytes.

3.4. Partitioning of fatty acids between the portal blood and the lymphatic system

The description so far has dealt with the metabolism and transport of glycerolipids
and cholesterol. Most of the long-chain fatty acids are transported from the intestine
by incorporation into the esterified lipids of the chylomicrons or VLDL. These two
lipoproteins also transport cholesterol, the fat-soluble vitamins, and other hydropho-
bic compounds and drugs. However, fatty acids with 10- to 12-carbon chains are
found both in the triacylglycerols of chylomicrons and in their unesterified forms
bound to albumin in portal blood (Table 111). Shorter-chain-length fatty acids are
transported almost exclusivelyby the latter route to the liver where they are preferen-
tially oxidized.
   There are several reasons for the partitioning of fatty acids between portal blood
and the fluid in the lymphatics on the basis of their chain lengths (Table 111). First,
the short- and medium-chain acids which are often located at the 3-position of tria-
cylglycerol are more readily hydrolyzed. This means that they are unlikely to be re-
tained in the monoacylglycerols that are re-esterified in the enterocytes. The ente-

Partition of fatty acids after absorption between the portal blood and the lymphatic system

Chain length                Portal route (unesterified)               Lymphatics (esterified)

C*-G                        majority                                  little
ClO                         majority                                  significant
Cl,                         significant                               significant
'CI,                        little                                    majority

rocytes also contain ester hydrolases that selectively degrade short- and medium-
chain acylglycerols. Second, the fatty acids in the intestinal lumen partition between
the micellar and aqueous phases. The shorter the chain length, the greater the tenden-
cy to partition into the aqueous phase and to become separated from the bulk of the
lipid. The uptake of short-chain fatty acids into the enterocytes can take place against
a concentration gradient. Finally, the enzymes responsible for re-esterification in the
enterocytes discriminate against the short-and medium-chain-length fatty acids [2].
   Use is made of the body's ability to absorb fatty acids by the portal route in condi-
tions where the digestion, absorption, or transport of triacylglycerols in chylomi-
crons is impaired. The clinical management of these malabsorption syndromes (for
example, pancreatic or biliary insufficiency, damage to the intestinal mucosa, abetali-
poproteinemia) is facilitated by feeding medium-chain triacylglycerols, from which
the fatty acids can be efficiently absorbed by the portal route without the need for
chylomicron formation. This energy-rich diet supplies the liver with a readily oxidiza-
ble substrate.

4. Control of triacylglycerol synthesis

The control of triacyglycerol synthesis varies from tissue to tissue. In the enterocytes
of the small intestine, it is assumed that the load of dietary fatty acids is the major
factor that controls the rate of triacylglycerol synthesis. The availability of 2-mono-
acylglycerol governs the balance between the monoacylglycerol pathway and synthe-
sis de novo. The major function of the enterocyte in this respect is to transport dietary
fats and to synthesize chylomicrons.
   In the liver the situation is more complex. This organ is required to incorporate
fatty acids into triacylglycerols whether these acids are derived primarily from syn-
thesis de novo, from circulating lipoproteins, or from lipolysis in adipose tissue (Fig.
9). In terms of metabolic and hormonal control, these represent quite different condi-
tions. In addition to triacylglycerol synthesis, fatty acids can be used for /?-oxidation
and for the synthesis of phospholipids. These latter lipids are needed for membrane
turnover and for the secretion of bile, VLDL, and HDL.

Fatty Ocids from lipoproteins          Fatty acids
(Chylornicron remnants, IDL.LDL        f r o m adipose tissue
and H D L )

                                                   Malonyl-CoA   -,
                                                                  , \

Glycerol phosphate              I
                          Fatty tacyl-CoA

                                                                     ,  I



         I           I               C oA
                                            llNHlBITtONl   *---
                                            / LAcyl-carnitine
                                                     COzand Ketones
Diacylglycerol                  Tri acylglycerol

Fig. 9. The relationship between fatty acid synthesis, oxidation and esterification in the liver.

   The major function of triacylglycerol synthesis in heart and skeletal muscle is to
enable fatty acids to be stored temporarily if their supply is greater than the imme-
diate rate of p-oxidation.
   Long-term storage of triacylglycerols occurs in adipose tissue. In adipocytes the
activities of the enzymes that synthesize triacylglycerols are generally controlled reci-
procally with that of the hormone-sensitive lipase, which is responsible for fatty acid
mobilization [23]. However, because of the high fatty acid availability during lipoly-
sis, some fatty acids can be recycled back to triacylglycerol.

4.1. Control ofphosphutidute synthesis in the liver

One of the obvious sites for the enzymic regulation of fatty acid esterification is at the
level of glycerolphosphate and dihydroxyacetone phosphate acyltransferases (Fig.
2), since at this point fatty acids become committed to glycerolipid synthesis. The
maximum velocity of acyl-CoA synthetase is very much higher than those of the acyl-
transferases, but the flux into acyl-CoA esters is, of course, dependent upon fatty acid
availability. At present there is little evidence for separate pools of acyl-CoA esters
that selectively supply one pathway of fatty acid utilization. Although fatty acids can
be desaturated and elongated, the major competition for glycerolipid synthesis is pro-
vided by ,&oxidation (Chapter 3). Regulation of the relative activities of glycerol-
phosphate acyltransferase and carnitine palmitoyltransferase should therefore con-
trol this branch-point (Fig. 9). The competition for acyl-CoA esters might explain
why there is such an active glycerolphosphate acyltransferase with low K, values for
glycerolphosphate and acyl-CoA esters in liver mitochondria. In addition, there
might be competition between the peroxisomal /I-oxidation system and the specific
dihydroxyacetone phosphate acyltransferase that is present in this organelle. How-

ever, as explained before, the metabolic fate of the majority of the phosphatidate and
lysophosphatidate that can potentially be formed in mitochondria and peroxisomes,
respectively, remains unclear.
    Carnitine palmitoyltransferase is acutely regulated by inhibition with malonyl-
CoA (Chapter 3 and [24]). At present, no such regulator of the acyltransferases has
been found. Consequently, it is tempting to assume that the esterification system ac-
quires those fatty acids not required by carnitine palmitoyltransferase. It might be
wise, however, to reflect that the discovery of the malonyl-CoA inhibition of this
enzyme is relatively recent. Previously, the prevailing view was that fatty acid oxida-
tion responded to fatty acid flux without being regulated itself.
   The mitochondrial glycerolphosphate acyltransferase activity is thought to be in-
creased by insulin. The activity of the mitochondrial enzyme is decreased more in
starvation and in diabetes than is the acyltransferase of the endoplasmic reticulum
[I]. The decrease in the mitochondrial acyltransferase may depend upon the inhibi-
tory effect of glucagon [25]. The acyltransferase from the endoplasmic reticulum
should provide most of the phosphatidate that is used for the synthesis of triacylgly-
cerols. Theoretically, this acyltransferase should operate at high rates of fatty acid
supply to the liver, and presumably the microsomal acyltransferase provides the liver
with its large capacity for triacylglycerol synthesis. This activity remains relatively
constant in different physiological conditions, although small decreases have been
observed during starvation [l]. The physiological significance of such changes is not
clear, since the capacity of the liver to synthesize triacylglycerols is not decreased
during starvation. This fact can be demonstrated by blocking ,&oxidation, which di-
verts an increased flux of fatty acids into triacylglycerol synthesis [24].

4.2. Control ofthe conversion of phosphatidate to triacylglycerol in liver

Phosphatidate phosphohydrolase is in a position to help to regulate the relative rate
of flux of phosphatidate to diacylglycerol (and thus to triacylglcyerol, phosphatidyl-
ethanolamine and phosphtidylcholine) versus the production of CDP-diacylglycerol
and certain acidic phospholipids (Fig. 3). One of the functions of triacylglycerol syn-
thesis in the liver and other tissues is to sequester excess fatty acids and acyl-CoA
esters and thereby to prevent their toxic accumulation which could cause membrane
damage. The liver, in particular, plays an important role in limiting the rise in uneste-
rified fatty acids in the circulation after stimulated lipolysis in adipose tissue. A high
capacity for hepatic triacylglycerol synthesis is therefore important in protecting the
body from fatty acid toxicity. The potential for stimulating triacylglycerol synthesis
in the liver in normal physiological or in pathological conditions is rarely if ever ex-
ceeded. This means that the activities of all of the enzymes of triacylglycerol synthesis
including the activity of phosphatidate phosphohydrolase should also be very high.
However, if a high phosphohydrolase activity were expressed all of the time, the abili-
ty to synthesize acidic phospholipids might be compromised. Thus, tissues have de-

veloped techniques for modulating the amount of phosphatidate phosphohydrolase
that is functionally expressed.
   Phosphatidate phosphohydrolase activity in the liver is increased relative to other
enzymes involved in triacylglycerol synthesis in metabolic stress. These are metabolic
conditions in which the actions of glucagon and glucocorticoids (cortisol or cortico-
sterone) are increased relative to insulin (Table IV). The demonstration of this hor-
monal regulation of phosphatidate phosphohydrolase activity was elucidated using
cultured rat hepatocytes. Dexamethasone (a synthetic glucocorticoid) increased the
phosphohydrolase activity by about 4-fold after 8 h and glucagon alone produced
about a 1.7-fold increase (Table V). There was a synergistic interaction between glu-
cagon and dexamethasone and an increase of 8.3-fold in the phosphohydrolase activ-
ity was obtained. Insulin alone had no significant effect, but it antagonised the ac-
tions of glucagon and dexamethasone [ 10,251. These hormonal effects are compatible

Changes in phosphatidate phosphohydrolase activity in the liver that can be associated with an increased
effect of glucocorticoids and other stress hormones relative to insulin

Treatment or condition                Duration                    Increase in phosphohydrolase

Metabolic Stress
 Starvation                           640h                       1.3-2.3 fold
 Surgical Stress                      6h                         3 fold
 Subtotal hepatectomy                 6h                         5.5 fold
 Diabetes (mildly ketotic)            10 weeks                   I .4 fold
 Diabetes (ketotic)                   48 h                       2.8 fold but reversed by insulin
 Hypoxia                              24 h                       2.5 fold

Toxic conditions produced by:
  Hydrazine                           4-24 h                     2 4 fold
  Morphine                            60-1 80 min                2 fold

Hormone Injections
 Corticotropin                        6h                         3 . 3 4 . 3 fold
 Cortisol                             5 days                     2.4 fold

Genetic obesity (ob/ub) mouse         long term                  2 fold

Ingestion of some nutrients
  Fructose, sorbitol or glycerol      6h                         1.9-2.3 fold
  ethanol                             l h                        6.9 fold

The results are from work with rats with the exception of the ob/ob mouse and the experiments involving
morphine injection which were also performed with mice. Details concerning the experiments and the au-
thorship of the work can be obtained from [8].

Effects of glucagon, dexamethasone and insulin on the activity of phosphatidate phosphohydrolase in
cultured rat hepatocytes

Additions                                                 Relative activity (%)    Significant differences

I      None (Control)                                    100 (10)
I1     Glucagon (10 nM)                                  172 t 32 (10)             1 versus I1
111    Insulin (500 pM)                                  91 4 13 (10)
IV     Glucagon (10 nM) + insulin (500 pM)               106 i 11 (10)             I1 versus IV
V      Dexamethasone (1 00 nM)                           422 t 11 I (9)            I versus V
VI     Dexamethasone (100 nM) + glucagon (10 nM)         830 k 263 (9)             I and I1 versus VI
VII    Dexamethasone (100 nM) + insulin (500 pM)         161 t 29 (6)              V versus VII
VIII   Dexamethasone (100 nM) + glucagon (10 nM)         458 f 139 (7)             VI versus VIII
       + insulin (500 pM)

Rat hepatocytes were incubated for 8 h with the hormones as indicated. The activity of phosphatidate
phosphohydrolase that was measured in the presence of Mg” is shown relative to the incubation in the
absence of hormones. The results are given as means t S.D. for the number of independent experiments
indicated. The P values for the significance of the differences is shown. The information is taken from [25].

with the changes observed in vivo (Table IV). This regulation appears to be brought
about by changes in the synthesis of phosphatidate phosphohydrolase [25]. In addi-
tion, glucagon increases the stability of the phosphohydrolase, whereas insulin has
the opposite effect [8]. These increases in the phosphohydrolase activity in conditions
of metabolic stress provide the liver with extra capacity to esterify the increased con-
centrations of fatty acids that are derived from adipose tissue after stimulation of
lipolysis. Whether or not this increased phosphohydrolase activity is expressed de-
pends upon its acute regulation and the net fatty acid availability.
   Phosphatidate phosphohydrolase exists both in the cytosol and on the endoplas-
mic reticulum [7,8]. It is believed that the cytosolic form of the enzyme is physiologi-
cally inactive until it translocates onto the membranes on which phosphatidate is
synthesized. A signal for this translocation is the increased availability of fatty acids
(Fig. 10, [26]) which leads to an increased production of acyl-CoA esters and phos-
phatidate. Fatty acids, acyl-CoA esters and phosphatidate all partition into mem-
branes and donate a negative charge to the membrane surface [7,8]. This appears to
cause the attachment of the cytosolic phosphohydrolase to the endoplasmic reticu-
lum [7,8], or mitochondria1 fraction [I 11 where it encounters its substrate. Converse-
ly, amphiphilic amines such as the drug chlorpromazine or the natural amine, sphin-
gosine, donate a positive charge to biological membranes. This inhibits phospha-
tidate phosphohydrolase activity by preventing its interaction with the membranes
that contain phosphatidate [8,10]. As a consequence the synthesis of triacylglycerol
and phosphatidylcholine is inhibited by chlorpromazine and phosphatidate accumu-

      t                       _-     ~

                  Fatty acid c o n c e n t r a t i o n

Fig. LO. Activation and translocation of phosphatidate phosphohydrolase by fatty acids in isolated hepato-
cytes. A fatty acid (oleate) increases the total activity of phosphatidate phosphohydrolase and also in-
creases its association with membranes rather than with the cytosol. It is believed that the membrane-
associated enzyme is physiologically active and that the cytosolic enzyme forms a reservoir of potential

lates [27] because the amount of the phosphohydrolase bound to the membranes has
become rate-limiting for diacylglycerol production.
   In addition to causing the translocation of phosphatidate phosphohydrolase to
membranes, very high concentrations of fatty acids (> 1 mM) also increase the total
phosphatidate phosphohydrolase activity within 30-60 min by a mechanism that is
not understood (Fig. 10). This acute activation by fatty acids could provide a further
protection to the liver by providing extra capacity to perform fatty acid esterification.
   The ability of fatty acids to promote the translocation of phosphatidate phospho-
hydrolase from the cytosol to the membrane-associated compartment may be under
hormonal control. CAMP treatment of hepatocytes causes displacement of the en-
zyme from the membranes, but this effect can be overcome by higher concentrations
of fatty acids (Table VI; [ZS]). By contrast, insulin decreases intracellular concentra-
tions of CAMP and therefore should lower the concentration of fatty acids required
to promote translocation [25]. The mechanisms that cause these changes are yet to be
established, but they may involve the reversible phosphorylation-dephosphorylation
of the phosphohydrolase. Translocation of phosphatidate phosphohydrolase is remi-
niscent of the regulation of CTP:phosphatidyIcholine cytidylyltransferase, which
regulates the synthesis of phosphatidylcholine (Chapter 7). Phosphatidate phospho-

Effects of a cyclic AMP analogue and oleate on the location of phosphatidate phosphohydrolase in isolat-
ed hepatocytes

                                                Phosphohydrolase activity (%)

Additions                                       Cytosolic            Membrane-associated

None                                            68                   32
CPT-CAMP(0.5 mM)                                86                   14
Oleate (1 mM)                                   48                   52
CPT-CAMP(0.5 mM) +Oleate (1 mM)                 42                   60

Hepatocytes were incubated for 1 h in the presence or absence of 8-(4-chlorophenylthio) adenosine 3'5'
cyclic monophosphate (CPT-CAMPwhich is a potent analogue of cyclic AMP) and oleate. The informa-
tion is taken from [28].

hydrolase and CTP:phosphocholine cytidylyltransferase can be classified as ambi-
quitous enzymes, which means that they exist in different locations in the cell and can
regulate metabolism by moving from one location to another [29]. The reason for
coordinating the synthesis of phosphatidylcholine and triacylglycerol can readily be
appreciated in terms of the synthesis of VLDL, which contain 20% and 60% (by
weight), respectively, of these lipids (Table I). Both triacylglycerol and phosphatidyl-
choline synthesis are required for VLDL secretion [30].
   Fig. 11 presents a schematic representation of the effects of fatty acids as activators
for the synthesis of triacylglycerols and phosphatidylcholine. An increased supply of
fatty acids and acyl-CoA esters stimulates the synthesis and accumulation of phos-
phatidate in the endoplasmic reticulum. This acts as a feed-forward signal for activat-
ing phosphatidate phosphohydrolase although the fatty acids themselves may act as
translocating agents [8]. The subsequent increase in the production of diacylglycerol
provides substrate for the synthesis of triacylglycerol and phosphatidylcholine (Sec-
tion 4.3). Moreover, the diacylglycerol also appears to be a potent signal for the
translocation and activation of CTP:phosphocholine cytidylyltransferase (Chapter

4.3. Diacylglycerol as a precursor of triacylglycerol, phosphatidylcholine, and phos-

Despite what is indicated about the coordinated control of triacylglycerol and phos-
phatidylcholine synthesis (Fig. 1 l), there is also a competition for diacylglycerol as a
substrate for the synthesis of phospholipids versus triacylglycerols. At low fatty acid
availability and a low rate of formation of diacylglycerol, the major flux from diacyl-
glycerol is directed to the synthesis of phosphatidylcholine and phosphatidylethanol-
amine. This maintains membrane turnover, and bile secretion in the liver.

             Phosphotidote-,                  t


 Triacylglycerol                    Phosphatidylcholine

Fig. 1 I. Proposed scheme for the coordinated regulation of the synthesis of triacylglycerol and phosphati-
dylcholine through the control of the activities of phosphatidate phosphohydrolase and CTP:phos-
phocholine cytidylyltransferase by their association with the endoplasmic reticulum. Both phosphatidate
phosphohydrolase (PAP) and CTP:phosphocholine cytidylyltransferase (CT) are activated by transloca-
tion from the cytosol to the endoplasmic reticulum. This process is promoted by fatty acids and their CoA
esters which appear to act as feed forward activators. However, the effect of these compounds is probably
indirect in cells. The translocation is probably effectd by the increased production of phoshatidate in the
endoplasmic reticulum which causes the translocation and activation of PAP. This in turn increases the
concentration of diacylglycerol in the membranes which causes the translocation and activation of CT. The
figure is reproduced from [8] with permission.

   Diacylglycerol acyltransferase is the first committed step that is specific for the
synthesis of triacylglycerol rather than for other glycerolipids. It might therefore be
expected that this should be a site of regulation. Evidence has been provided that the
availability of diacylglycerol can be rate-limiting in the synthesis of triacylglycerols in
the liver [31]. However, it should also be remembered that the capacity of the liver to
synthesize triacylglycerols is rarely if ever exceeded. The activity of diacylglycerol
acyltransferase may also be decreased by a CAMP-dependent phosphorylation that
would exaggerate this discrimination against triacylglycerol synthesis and favor
phospholipid formation in starvation, diabetes and stress [32]. However, these condi-
tions are also characterized by increased fatty acid availability which would stimulate
the production of acyl-CoA esters and diacylglycerols, the two substrates for the
acyltransferase (Fig. 11). Consequently the inhibition of diacylglycerol acyltransfer-
ase should be overcome at high fatty acid availability. This is important in protecting
the liver against potentially toxic accumulations of fatty acids and acyl-CoA esters.
   Increased fatty acid availability will also stimulate the synthesis of phosphatidyl-
choline and phosphatidylethanolamine. However, this will eventually be limited by
the availability of CDP-choline and CDP-ethanolamine, respectively.
   In order to try to relate these changes in hepatic triacylglycerol synthesis to other
metabolic pathways and to metabolism in other organs, two very different metabolic

situations will be considered. Namely, conditions in which insulin has high and low
activities relative to glucagon, adrenalin, corticotropin, and glucocorticoids.

5. Metabolism o triacylglycerols when the action of insulin is high

A high concentration of insulin is normally seen after the consumption of a carbohy-
drate meal rich in glucose. Such a diet also produces good insulin control, since the
tissues normally respond well to insulin action. If insulin action predominates rela-
tive to that of glucagon and catecholamines in the control of metabolism, then glyco-
lysis, and the synthesis of glycogen [33] and fatty acids in the liver are stimulated
(Chapter 5). At the same time, /I-oxidation and gluconeogenesis are suppressed (Fig.
12). Glucocorticoids, which are also stress hormones, can have a permissive effect on
some of these processes provided that insulin is available. Thus, they augment insulin
action in stimulating the synthesis of glycogen and fatty acids [33].
   The fatty acids that are produced during increased insulin action are readily incor-
porated into triacylglycerols. Glycerolphosphate and dihydroxyacetone phosphate
can be provided by glycolysis, and fatty acids are not diverted into /I-oxidation be-
cause of the inhibition of carnitine palmitoyltranserase by malonyl-CoA (Fig. 9).
Normally, the intracellular concentration of acyl-CoA esters remains relatively low,
but because of the effects of insulin and the relatively low concentration of CAMP,
phosphatidate phosphohydrolase is readily translocated from the cytosol to the
membrane-associated compartment, which facilitates glycerolipid synthesis. Fur-
thermore, if previous glucocorticoid concentrations were relatively high before the

             AD IPOSE

                                                                               SKELETAL MUSCLE

                    Glucose +COZ

Fig. 12. Some effects of insulin on metabolism. The figure shows the metabolic situation after ingesting a
meal rich in starch. The absorption of glucose leads to insulin secretion. The solid lines represent major
routes of metabolism.

meal, the phosphohydrolase activity also increases in the long-term and provides an
increased potential for the synthesis of triacylglycerols [8].
   Triacylglycerols do not accumulate to a large extent in the liver when insulin is
acting effectively, since they are packaged into VLDL and secreted into the blood
(Fig. 12). However, the effect of insulin on the process of VLDL secretion is very
controversial. Some investigators claim that insulin stimulates this process and ex-
perimental support for this comes from studies with perfused liver [33, 341. There is
also an association of high rates of VLDL secretion with hyperinsulinemia [33-351.
However, this latter condition is typical of insulin-insensitivity as seen in type I1
diabetes. In this instance, it is questionable whether insulin is causing the high rates
of triacylglycerol secretion, or whether the lack of effective insulin action is respons-
   Work by several groups with isolated hepatocytes indicates an inhibitory effect of
insulin alone on VLDL secretion [33-351. This can be justified in physiological terms
by considering the condition after eating a mixed meal of carbohydrate and fat. Insu-
lin concentrations will rise, and at the same time chylomicrons will be produced by
the intestine. Chylomicrons are metabolized in the same way as VLDL. It would
therefore be reasonable if insulin inhibited VLDL secretion. Then, as chylomicron
and insulin concentrations subside, VLDL secretion would rise to provide peripheral
tissues, including muscles, with an alternate supply of triacylglycerol [33-351. How-
ever, the effect of insulin in inhibiting the secretion of VLDL is limited by glucocorti-
coids. These hormones, together with insulin, promote fatty acid synthesis [33] and
glucocorticoids stimulate VLDL secretion [34,35]. After a meal there is secretion of
both insulin and glucocorticoids and this hormonal combination is a signal for en-
ergy storage [33,35]. The transport of fatty acids from liver to adipose tissue is an
important component in this process. Glucocorticoids probably play a significant
role in this respect by increasing VLDL secretion after feeding as they do in stress
conditions (see Section 6). The main site of triacylglycerol hydrolysis in the blood-
stream and fatty acid uptake after a meal is in adipose tissue, since lipoprotein lipase
activity in this organ is increased by insulin, and glucocorticoids augment this effect
[36,37]. There are multiple steps at which the regulation of lipoprotein lipase at the
capillary surface can be modulated and this control is still not completely under-
stood. The first stage involves the production of a 5 1 kDa protein which is then con-
verted to a 55.5 kDa high mannose form. This is then converted to a mature 58 kDa
species which is active as a homodimer. This form of the enzyme can either be
degraded within the adipocyte, or secreted and travel to the function site on the capil-
lary endothelium [38].
   In the presence of insulin, glucocorticoids exert an anabolic effect especially in
terms of energy deposition. This may be an important component in the development
of obesity [33]. Glucocorticoids promote adipose tissue differentiation and together
with increased fatty acids they decrease insulin sensitivity in tissues. This is measured
in terms of glucose uptake which may be depressed even in adipose tissue which is

diverting increased quantities of glucose to fat rather than to lactate [33]. Upper
body, or android obesity, is associated with insulin resistance (type I1 diabetes), hy-
perlipidemias and hypertension. Glucocorticoids are thought to be involved in the
development of this condition [33]. In this respect, it is also significant that glucocor-
ticoids suppress the expression of the gene coding for adipsin [39]. This is an adipo-
cyte-specific gene whose expression is normally depressed in obesity. Adipsin is a
member of the serine protease family which may be a systemic regulator of energy
   In lactation some of the VLDL (and chylomicrons) are diverted to the mammary
gland, where prolactin increases lipoprotein lipase activity. The activity of lipopro-
tein lipase in skeletal and cardiac muscle is relatively low compared with that is adi-
pose tissue when insulin is effectively regulating metabolism. This is because muscle
lipase activity is not maintained by insulin, but rather by glucocorticoids, with the
possible involvement of glucagon, catecholamines, and thyroid hormones [36].
   Insulin not only promotes the synthesis of fatty acids and their ultimate transfer as
triacylglycerols from liver to adipose tissue, but it also stimulates glucose entry into
adipose tissue and the activities of the enzymes of triacylglycerol synthesis in this
tissue [23]. This leads to the efficient storage of triacylglycerol. The release of fatty
acids from triacylglycerols in adipose tissue is suppressed because of the inactivation
of hormone-sensitive lipase. This condition is maintained by the predominance of
insulin over the stress hormones and the consequent decrease in CAMP concentra-
tions (Fig. 13). The activation of the hormone-sensitive lipase by a CAMP-dependent
phosphorylation is therefore prevented [40].

Cort icotropin
       t   1    ATP

   diestemse                                                                            glycerol
                  i              Triacyl-                    Diacyl-             Monoacyl-
                                 glycerol   -T               g l y c e r o l T * glycerol ~         *      ~         ‘   y   ~
                                                F a t t y acid         F a t t y acid         F a t t y acid

Fig. 13. The control of the activity of hormone-sensitive lipase in adipose tissue. A plus indicates a stimula-

6. Triacylglycerol metabolism in conditions of metabolic stress

Stress is characterized by the low activity of insulin relative to glucagon, the catechol-
amines, corticotropin, and the glucocorticoids in regulating metabolism. This type of
hormonal balance occurs in starvation, diabetes, trauma, and toxic conditions. To a
lesser extent, nutrients such as fructose, ethanol, and fat present the body with an
energy load without releasing insulin. In fact, fructose and ethanol can provoke the
release of stress hormones, and high-fat diets can cause tissues to become insulin-
insensitive [33].
   The acute-acting stress hormones (glucagon, adrenalin) stimulate adenylcyclase
(adenylate cyclase) in a variety of tissues. In adipose tissue this stimulation leads to
phosphorylation and activation [40] of the hormone-sensitive lipase which initiates
triacylglycerol breakdown to 2-monoacylglycerol. The final stage of degradation is
completed by a monoacylglycerol lipase [41] (Fig. 13). At the same time, the enzymes
of triacylglycerol synthesis are inhibited so as to limit reesterification [23]. However,
the stimulation of lipolysis also increases fatty acid availability and triacylglycerol
synthesis in adipose tissue. This resynthesis is much less than the fatty acid produced
by lipolysis and there is a net release of fatty acid into the blood. It is therefore signif-
icant that the activity of phosphatidate phosphohydrolase on the endoplasmic reti-
culum in adipocytes also rises in response to the increased availability of fatty acids
[23]. This probably helps to prevent an excessive accumulation of fatty acids with the
   During accelerated lipolysis in adipose tissue, fatty acids and glycerol are released
into the circulation. The fatty acids are bound to albumin in the blood and are carried
to other organs. Some of the fatty acids enter muscle tissue to become oxidized and
provide energy (Fig. 14). However, a large proportion of fatty acids is taken up by the
liver. The high concentration of acyl-CoA esters that results, together with the in-
crease in cAMP concentration, decreases the activity of acetyl-CoA carboxylase
(Chapter 4) and the concentration of malonyl-CoA falls (Fig. 9). Consequently, the
inhibition of B-oxidation is relieved, and more of the fatty acids are metabolized to
CO, and ketones. The oxidation process at low fatty acid supply can also be facil-
itated by the effect of CAMP in displacing phosphatidate phosphohydrolase from the
membranes (Table VI), thereby inhibiting triacylglycerol synthesis.
   However, in severe stress the supply of fatty acids and glycerol to the liver nor-
mally increases dramatically, as do the concentrations of acyl-CoA esters and glyce-
rolphosphate in the liver. The high fatty acid availability overcomes the effects of
cAMP in displacing phosphatidate phosphohydrolase (Table VI) and CTP:phos-
phocholine cytidylyltransferase (Chapter 7) from the membranes. The synthesis of
triacylglycerols and phosphatidylcholine is promoted and these can be used for
VLDL secretion. A fatty liver can often develop in stress conditions, when the ca-
pacity to secrete VLDL cannot cope with the increased rate of triacylglycerol synthe-
sis. Fatty liver occurs especially in several toxic conditions where the synthesis of the

                                          W fatty acid
ADIPOSE                                         BRAIN

                                                                             Ketones   +CO2 +Fatty   acid+'

          Fatty acid           Glycerol





Fig. 14. Some effects of high glucocorticoid concentrations on metabolism. The figure shows the results of
metabolic stress such as can occur in starvation, psychological stress, and diabetes. The stress hormones,
glucagon, the catecholamines and the glucocorticoids have a predominant effect over insulin in regulating
metabolism. If the fatty acids that are released from adipose tissue are not used for energy production in
heart and skeletal muscle, they can be used in the liver for ketogenesis or VLDL secretion.

apolipoproteins required in VLDL formation is impaired. Normally, the fatty condi-
tion of the liver is reversed when the fatty acid mobilization from adipose tissue sub-
sides and the triacylglycerols can be secreted, or hydrolyzed so that the fatty acids can
be oxidized by the liver.
   The availability of fatty acids in the liver is a very important factor regulating the
secretion of VLDL. Furthermore, glucocorticoids are potent stimulators of this proc-
ess [35].A stress or diabetic condition is characerized by increases in both fatty acids
and cortisol. Consequently this could account for why VLDL secretion is increased
in ketotic type I diabetes, which is caused by insulin deficiency. Fatty acids and corti-
sol also cause insulin-insensitivity [33] which could also be a contributing factor for
the over-secretion of VLDL that can occur in type I1 diabetes.
   The VLDL that are released during severe stress are preferentially metabolized by
muscle tissue, since its lipoprotein lipase activity is increased by glucocorticoids and
other stress hormones [36]. The relative lack of insulin action during stress ensures
that lipoprotein lipase activity is low in adipose tissue; the fatty acids are therefore
prevented from recycling back into the storage depots (Fig. 14). This relative de-
crease in the ability to degrade VLDL in adipose tissue aggravates the hypertriglyce-
ridemia that can occur in diabetes because of the hypersecretion of VLDL.
   One of the main functions of the liver in stress conditions and diabetes is to dis-
tribute energy to other organs in the form of glucose, ketones, and triacylglycerols
(Table VII). It is significant that the control of phosphatidate phosphohydrolase in

Export of energy from the liver in stress conditions

Metabolite                                        Taken up by

Glucose                                           Brain and erythrocytes
Ketones                                           Skeletal and cardiac muscle and brain
Triacylglycerols (VLDL)                           Skeletal and cardiac muscle

the liver by glucocorticoids, CAMPand insulin resembles that of enzymes involved in
amino acid breakdown and gluconeogenesis, for example, tyrosine aminotransferase,
argininosuccinate synthetase, arginosuccinate lyase, and phosphoenolpyruvate car-
boxykinase [8,25].Lipolysis from adipose tissue supplies the liver with fatty acids and
glycerol for triacylglycerol synthesis, fatty acids for ketogenesis, and glycerol for glu-
coneogenesis (Fig. 14). It is also significant that the activity of the Mg2+-stimulated
phosphatidate phosphohydrolase normally correlates with the accumulation of tria-
cylglycerol in the liver and the circulating concentrations of triacylglycerol [8].
   The hypertriglyceridemia and hyperglycemia that occur in stress and diabetes
therefore seem to have a similar physiological basis. They result from increased he-
patic secretion of triacylglycerol and glucose and the decreased removal of the com-
pounds from the circulation. The latter condition occurs because the uptake systems
for glucose in muscle and adipose tissue, and for triacylglycerols through lipoprotein
lipase in adipose tissue, are insulin-dependent.
   In the case of VLDL metabolism increased fatty acids and glucocorticoids stimu-
late the secretion of VLDL, and lower the responsiveness of the liver to insulin. The
lack of insulin action combined with the effects of glucocorticoids also decrease the
activity of the low density lipoprotein receptor in the liver [42]. This apoB/E receptor
is responsible for removing the major proportion of intermediate density lipoproteins
and low density lipoproteins from the circulation (Chapter 16). The physiological
role of this effect is probably to delay the clearance of intermediate density Iipopro-
teins thus enabling muscle tissues to acquire more fatty acids through the action of
lipoprotein lipase. Muscles would therefore be provided with more energy as part of
a ‘fright and flight’ response. However, in the long-term the action of glucocorticoids
in producing a hypertriglyceridemia and a hypercholesterolemia ought to accelerate
the development of premature atherosclerosis. This probably explains why this dis-
ease is associated with stress, diabetes, obesity and hypertension [33].

7. Future directions

Although progress has been made in understanding the complex control of triacyl-
glycerol metabolism, it is clear that there are many gaps in our knowledge. For in-
                                                                                                      20 1

stance, what are the major functions of the mitochondria1 and peroxisomal systems
that synthesize phosphatidate or lysophosphatidate? What are the relative contribu-
tions of glycerolphosphate and dihydroxyacetone phosphate to glycerolipid synthesis
under different physiological conditions in different organs? Are the initial acyltrans-
ferases of glycerolipid synthesis controlled to a greater extent than is known at pres-
ent? What is the significance of the monoacylglycerol pathway in the liver? How are
the enzymes of triacylglycerol synthesis regulated differentially in different tissues?
What is the exact role of the two phosphatidate phosphohydrolases in liver in regu-
lating glycerolipid synthesis and signal transduction? We also do not fully under-
stand many of the processes that control the distribution of triacylglycerol via VLDL
to different organs. For example, one major outstanding problem is the exact role of
insulin on its own, and in combination with other hormones, on the secretion of
VLDL and on the expression of lipoprotein lipase activity.
   In order to answer some of these questions, several technical problems have to be
overcome. The increased use of tissue- and cell-culture techniques has already meant
that the specific effects of hormones and metabolites on triacylglycerol metabolism
can be investigated. However, it is often difficult to reproduce, or to appreciate the
situation that exists in vivo. Care must be taken in interpreting the results and in
deciding what is, or is not, an artifact. As always, the conclusions have to be compat-
ible with what occurs in vivo.
   One of the biggest problems in lipid metabolism has always been to isolate, charac-
terize, purify, and to raise antibodies against the enzymes which are often tightly
associated with membranes. Until this is done, the mechanisms by which the enzymes
act and are controlled will not be fully understood. It is also difficult to work with
substrates that are insoluble in water. Thus, a strict kinetic analysis of reaction rates
is normally not possible.
   If we can overcome some of these technical problems, the rewards are potentially
high. The regulation of triacylglycerol metabolism is an important part of interme-
diary metabolism. A greater appreciation of this regulation would help to alleviate or
prevent several clinical conditions, including fatty liver, diabetes, obesity, hyperlipi-
demias, and atherosclerosis.

I. Brindley, D.N. and Sturton, R.G. (1982) Phosphatidate metabolism and its relation to triacylglycerol
   biosynthesis. In: Phospholipids: New Comprehensive Biochemistry (Hawthorne, J.N. and Ansell,
   G.B., Eds.) Vol. 4, pp. 179-213, Elsevier Biomedical Press, Amsterdam.
2. Brindley, D.N. (1984) Digestion, absorption and transport of fats; general principles. In: Fats in Ani-
   mal Nutrition (Wiseman, J., Ed.) pp. 85-103, Butterworth Scientific, London.
3. Schlossman, D.M. and Bell, R.M. (1977) Microsomal sn-glycerol3-phosphate and dihydroxyacetone
   phosphate acyltransferase activities of liver and other tissues. Evidence for a single enzyme catalyzing
   both reactions. Arch. Biochem. Biophys. 182,732-742.
 4. Manning, R. and Brindley, D.N. (1972) Tritium isotope effects in the measurement of the glycerol
    phosphate and dihydroxyacetone phosphate pathways of glycerolipid biosynthesis in rat liver.
    Biochem. J. 130, 1003-1012.
 5. Mason, R.J. (1978) Importance of the acyldihydroxyacetone phosphate pathway in the synthesis of
    phosphatidylglycerol and phosphatidylcholine in alveolar type I1 cells. J. Biol. Chem. 253,3367-3370.
 6. Pollock, R.J., Hajra, A.K. and Agranoff, B.W. (1976) Incorporation of D-[~-'H,U-'4C]glucose into
    glycerolipids via acyldihydroxyacetone phosphate in untransformed and viral-transformed BHK-2 1-
    c13 fibroblasts. J. Biol. Chem. 251,5149-5154.
 7. Brindley, D.N. (1984) Intracellular translocation of phosphatidate phosphohydrolase and is possible
    role in control of glycerolipid synthesis. Prog. Lipid Res. 23, 115-133.
 8. Brindley, D.N. (1988) General introduction and phosphatidate phosphohydrolase activity in the liver.
    In: Phosphatidate Phosphohydrolase (Brindley, D.N., Ed.) CRC Series in Enzyme Biology, Vol. 1, pp.
    1-77. CRC Press, Boca Raton, FL.
 9. Martin, A., Gomez-Mufioz, A., Jamal, Z. and Brindley, D.N. (1991) Characterization and assay of
    phosphatidate phosphohydrolase. Methods Enzymol. 197, 553-563.
10. Jamal, Z., Martin, A., Gomez-Muiloz, A. and Brindley, D.N. (1991) Plasma membrane fractions of
    rat liver contain a phosphatidate phosphohydrolase distinct from that in the endoplasmic reticulum
    and cytosol. J. Biol. Chem. 266,2988-2996.
11. Freeman, M. and Mangiapane, E.H. (1989) Translocation to rat liver mitochondria of phosphatidate
    phosphohydrolase. Biochem. J. 263,589-595.
12. Murayama, T. and Ui, M. (1987) Phosphatidic acid may stimulate membrane receptors mediating
    adenylate cyclase inhibition and phospholipid breakdown in 3T3 fibroblasts. J. Biol. Chem. 262,5522-
13. Loffelholz, K. (1989) Receptor regulation of choline phospholipid hydrolysis. A novel source of dia-
    cylglycerol and phosphatidic acid. Biochem. Pharmacol. 38, 1543-1549,
14. Billah, M.M. and Arthes, J.C. (1990) The regulation and cellular functions of phosphatidylcholine
    hydrolysis. Biochem. J. 269,281-291.
15. Coleman, R.A. and Haynes, E.B. (1986) Monoacylglycerol acyltransferase. Evidence that the activi-
    ties from rat intestine and suckling liver are tissue specific isoenzymes. J. Biol. Chem. 261,224-228.
16. Moorhouse, K.G., Logan, C.J., Hutson, D.H. and Dodds, P.F. (1990) The incorporation of 3-phenox-
    ybenzoic acid and other xenobiotic acids into xenobiotic lipids by enzymes of the monoacylglycerol
    pathway in microsomes from adult and neonatal tissues. Biochem. Pharmacol. 39, 1529-1536.
17. Sansbury, K., Millington, D.S. and Coleman, R.A. (1 989) Hepatic monoacylglycerol acyltransferase:
    ontogeny and characterization of an activity associated with chicken embryo. J. Lipid Res. 30,1251-
18. Dodds, P.F., Gurr, M.I. and Brindley, D.N. (1976) The glycerol phosphate, dihydroxyacetone phos-
    phate and monoacylglycerol pathways of glycerolipid synthesis in rat adipose tissue homogenates.
    Biochem. J. 160,693-700.
19. Cohen, M., Morgan. R.G.H. and Hofmann, A.F. (1971) Lipolytic activity of human gastric and duo-
    denal juice against medium and long chain triglycerides. Gastroenterology 60, 1-15.
20. Borgstrom, B. (1974) Fat digestion and absorption In: Intestinal Absorption (Smyth, D.H., Ed.) Vol.
    2, pp. 5 5 5 4 2 0 , Plenum, London.
21. DeNigris, S.J., Hamosh, M., Kasbekar, M., Lee, T.C. and Hamosh, P. (1988) Lingual and gastric
    lipases: species differences in the origin of prepancreatic lipases and in the localization of gastric lipase.
    Biochim. Biophys. Acta 959, 3 8 4 5 .
22. Gargouri, Y., Moreau, H. and Verger, R. (1989) Gastric lipases: biochemical and physiological stud-
    ies. Biochim. Biophys. Acta 1006,255-271.
23. Saggerson, E.D. (1988) Phosphatidate phosphohydrolase activity in adipose tissue. In: Phosphatidate
    Phosphohydrolase (Brindley, D.N., Ed.) CRC Press in Enzyme Biology, Vol. I , pp. 79-129, CRC
    Press, Boca Raton, FL.

24. McGarry, J.D. and Foster, D.W. (1980) Regulation of hepatic fatty acid oxidation and ketone body
    production. Annu. Rev. Biochem. 49,395420.
25. Pittner, R.A., Fears, R. and Brindley, D.N. (1985) Interactions of insulin, glucagon and dexametha-
    sone in controlling the activity of glycerol phosphate acyltransferase and the activity and subcellular
    distribution of phosphatidate phosphohydrolase in cultured rat hepatocytes. Biochem. J. 230, 525-
26. Cascales, C., Mangiapane, E.H. and Brindley, D.N. (1984) Oleic acid promotes the activation and
    translocation of phosphatidate phosphohydrolase from the cytosol to particulate fractions of isolated
    rat hepatocytes. Biochem. J. 220, 825-833.
27. Martin, A., Hopewell, R., Martin-Sanz, P., Morgan, J.E. and Brindley, D.N. (1986) Relationship be-
    tween the displacement of phosphatidate phosphohydrolase from the membrane-associated compart-
    ment by chlorpromazine and the inhibition of the synthesis of triacylglycerol and phosphatidylcholine
    in rat hepatocytes. Biochim. Biophys. Acta 876, 581-591.
28. Butterwith, S.C., Martin, A. and Brindley, D.N. (1984) Can phosphorylation of phosphatidate phos-
    phohydrolase by a cyclic AMP-dependent mechanism regulate its activity and subcellular distribution
    and control hepatic glycerolipid synthesis. Biochem. J. 222,487493.
29. Wilson, J.E. (1980) Brain hexokinase, the prototype ambiquitous enzyme. Curr. Top. Cell. Regul. 16,
30. Yao, Z. and Vance, D.E. (1988) The active synthesis of phosphatidylcholine is required for very low
    density lipoprotein secretion from rat hepatocytes. J. Biol. Chem. 263,2998-3004.
31. Mayorek, N., Grinstein, I. and Bar-Tana, J. (1989) Triacylglycerol synthesis in cultured rat hepa-
    tocytes. The rate-limiting role of diacylglycerol acyltransferase. Eur. J. Biochem. 182, 395400.
32. Haagsman, H.P., De Haas, C.G.M., Geelen, M.J.H. and Van Golde, L.M.G. (1981) Regulation of
    triacylglycerol synthesis in the liver. A decrease in diacylglycerol acyltransferase activity after treat-
    ment of isolated rat hepatocytes with glucagon. Biochim. Biophys. Acta 664,7441.
33. Brindley, D.N. and Rolland, Y. (1989) Possible connections between stress, diabetes, obesity, hyper-
    tension and altered lipoprotein metabolism that may result in atherosclerosis. Clin. Sci. 77,453461.
34. Gibbons, G.F. (1990) Assembly and secretion of hepatic very low density lipoprotein. Biochem. J. 268,
35. Martin-Sanz, P., Vance, J.E. and Brindley, D.N. (1990) Stimulation of apolipoprotein secretion in
    very-low-density and high-density lipoproteins from cultured rat hepatocytes by dexamethasone. Bio-
    chem. J. 271,575-583.
36. Cryer, A. (1981) Tissue lipoprotein lipase activity and its action in lipoprotein metabolism. Jnt. J.
    Biochem. 13,5255541.
37. Speake, B.K., Parkin, S.M. and Robinson, D.S. (1986) Regulation of the synthesis of lipoprotein li-
    pase in adipose tissue by dexamethasone. Biochim. Biophys. Acta 881, 155-157.
38. Ailhaud, G. (1990) Cellular and secreted lipoprotein lipase revisited, Clin. Biochem. 23, 343-347.
39. Spiegelman, B.M., Lowell, B., Napolitano, A,, Dubuc, P., Barton, D., Francke, U., Groves, D.L.,
    Cook, K. and Flier, J.S. (1989) Adrenal glucocorticoids regulate adipsin gene expression in genetically
    obesemice. J. Biol. Chem. 264, 1811-I815.
40. Holm, C., Kirchgessner, T.G., Svenson, K.L.. Fredrikson, G., Nilsson, S., Miller, C.G., Shively, J.E.,
    Heinzmann, C., Sparkes, R.S., Mohandas, T., Lusis, A.J., Belfrage, P. and Schotz, M.C. (1988) Hor-
    mone-sensitive lipase: sequence, expression and chromosomal localization to 19 cent-q 13.3. Science
   '241, 1503-1506.
41. Fredrikson, G.,Tornqvist, H. and Belfrage, P. (1986) Hormone sensitive lipase and monoacylglycerol
    lipase are both required for complete degradation of adipocyte triacylglycerol. Biochim. Biophys. Acta
42. Brindley, D.N. and Salter, A.M. (1991) Hormonal regulatioh of the hepatic low density lipoprotein
    receptor and the catabolism of low density lipoproteins. Prog. Lipid Res. in press.
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemisfry o Lipids, Lipoproteins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                     205

                                                                                           CHAPTER 7

Phospholipid metabolism and cell signalling in
                                                                                   DENNIS E. VANCE

Lipid and Lipoprotein Research Group und Department of Biochemistry, University of Alberta,
                                                                Edmonton, Alta., Canada

I. Introduction
All cells contain phospholipids which make up the essential milieu of cellular mem-
branes and act as a barrier for entry of compounds into cells. Membrane proteins and
enzymes that provide essential functions for the cells such as metabolite transport
and generation of second messengers are contained within the phospholipid bilayer.
As discussed in Chapter 1, phospholipids spontaneously form the bilayer of mem-
branes. Clearly the survival of cells depends on the integrity of the phospholipid
bilayer. The critical function of phospholipids is underscored by the general lack of
inherited disorders in phospholipid metabolism.
   A second function ascribed to phospholipids has only been appreciated in the last
several decades. Phospholipids store second messenger precursors that are released
after a cell has received an appropriate signal. The second messengers generated are
eicosanoids, diacylglycerols and inositol-P,. As an example, inositol-P, is present in
membranes as an inactive component of phosphatidylinositol-4,5-P2.    Within seconds
after a signal is received on the plasma membrane, the inositol-P, is released by a
specific phospholipase C resulting in the mobilization of calcium into the cytosol
from the endoplasmic reticulum. When compared to the formation of inositol-P, via
biosynthesis from inositol, mobilization from a stored precursor offers many advan-
tages to a cell that needs to increase cytosolic calcium rapidly.
   A third and usually overlooked function of phospholipids is to store energy in the
form of fatty acyl components. This function is probably quantitatively important
only under extreme conditions such as starvation.
  The object of this chapter is to provide an overview of phospholipid metabolism at
an advanced level. References to reviews and recent seminal papers will guide the
reader to the current literature.

2. Phosphatidylcholine biosynthesis
2. I . Historical background

Phosphatidylcholine (PtdCho) was first described by Gobley in 1847 as a component
of egg yolk. PtdCho was initially named as ‘lecithin’ after the Greek equivalent for
egg yolk (lekithos).In the 1860s Diakonow and Strecker demonstated that lecithin
contained two fatty acids linked to glycerol and that choline was attached to the third
hydroxyl by a phosphodiester linkage. Chemical synthesis in 1950 by Baer and Kates
confirmed the proposed structure. Elucidation of the chemical structure paved the
way for the biochemistry to begin. The first significant step occurred in 1932 with the
discovery by Charles Best that animals had a dietary requirement for choline. In the
1950s the CDP-choline pathway for PtdCho biosynthesis (Fig. 1) was described by
Eugene Kennedy and coworkers. A key observation was the discovery that CTP,
rather than ATP, was the activating nucleotide for PtdCho biosynthesis [I]. CTP is

Fig. 1. Pathways for the biosynthesis of PtdCho and postulated control mechanisms for regulation of the
CDP-choline pathway. The rate of CDP-choline formation can limit the rate of PtdCho biosynthesis.
CDP-choline is made via the cytidylyltransferase (CT) which is found both in the cytosol as an inactive
enzyme and on the endoplasmic reticulum (ER) where it is active. The movement on and off the ER ap-
pears to be regulated by (1) the concentration of PtdCho, (2) the extent of phosphorylation of CT, (3) the
concentration of diacylglycerol in the membrane, and (4)the concentration of fatty acids in cell cultures.
The rate of PtdCho biosynthesis can also be regulated by the supply of diacylglycerol which is decreased
in hepatocytes treated with CAMP analogues or glucagon. The other abbreviations are: CK, choline ki-
nase; PEMT, PE methyltransferase; AT, lyso-PtdCho acyltransferase; ptase, phosphatase; DAG, diacyl-
glycerol; AdoMet, S-adenosyl-methionine; AdoHcy, S-adenosyl-homocysteine.

required not only for PtdCho biosynthesis but also for the de novo synthesis of all
phospholipids (procaryotic and eucaryotic, excluding phosphatidic acid) at some
step in the biosynthetic pathway.
   An alternative pathway for PtdCho biosynthesis, of quantitative significance only
in liver, is the conversion of phosphatidylethanolamine (PtdEtn) to PtdCho via the
methylation of PtdEtn (Fig. 1). The first indication of this pathway occurred in 1941
when Stetten fed ["jethanolamine to rats and isolated ['5N]choline. Two decades
later Bremer and Greenberg demonstrated the presence of a microsomal enzyme that
converted PtdEtn to PtdCho via transfer of methyl groups from S-adenosylmethio-

2.2. Choline transport and oxidation

There appear to be two distinct transport mechanisms for choline [2]; a high affinity
( K , or K, < 5 pM),sodium-dependent transporter that is coupled to acetylcholine
biosynthesis in neurons and a lower affinity (Kt>30pM), sodium independent trans-
porter found in most animal cells as well as in cholinergic nerve tissues. The high
affinity transporter (Mr=80,000)has been isolated from synaptosomal membranes
and functionally reconstituted into liposomes [3].
   Once choline is inside the cell, its normal fate is phosphorylation by choline kinase.
The major exception is liver and kidney where choline is also oxidized to betaine
(-OOC-CH,-"(CH,),).        Choline dehydrogenase is found on the inner membrane of
rat liver mitochondria and catalyzes the oxidation of the CH20Hgroup of choline to
an aldehyde. Betaine is formed via oxidation of the aldehyde to a carboxyl group by
betaine aldehyde dehydrogenase, reportedly a cytosolic enzyme. Betaine is an impor-
tant donor of methyl groups for methionine biosynthesis and the one carbon pool.

2.3. Enzymes of the CDP-choline pathway

In most cells and tissues the major fate of choline is phosphorylation by the cytosolic
enzyme choline kinase. The enzyme exists as a dimer of 42 kDa subunits. ATP is the
donor of the phosphate group and CTP, GTP, ITP or UTP are inactive [2]. The pure
enzyme phosphorylates ethanolamine equally well. The expression of choline kinase
has been induced in male roosters by the injection of estrogen and in rat liver by CCl,
and 3-methylcholanthrene [2]. Interestingly, the rat liver-induced forms of choline
kinase are different from the normally expressed enzyme. The function of the induced
forms of choline kinase is unknown.
  The second enzyme in the pathway is CTP:phosphocholine cytidylyltransferase
(CT). Although this enzyme neither catalyzes the committed step in the pathway nor
operates at a branch point, many studies indicate that CT can be the rate-limiting and
regulating enzyme in the biosynthesis of PtdCho [4,5]. After more than a decade of
frustrating attempts in many laboratories to purify CT, only in 1987 were Weinhold

and Feldman finally successful. In contrast, the cloning of the cDNA for CT and its
expression was achieved in 9 months (see ref. 5). The enzyme has a subunit molecular
weight of 41,720 and is recovered from both the cytosol and membrane fraction of
cells. As discussed in Section 3, the membrane-bound form of the enzyme is the active
 form whereas cytosolic CT appears to exist as an inactive reservoir. Cytosolic CT
 exists as a dimer of identical subunits. The arrangement of the subunits in the ER has
not been demonstrated. The central domain of CT from liver has close sequence
homology to the enzyme from yeast. However, there is a unique sequence in the rat
liver enzyme                   that appears to be an amphipathic helix that may be the
membrane-binding domain of CT.
   The third enzyme in the CDP-choline pathway is CDP-choline:1,2-diacylglycerol
choline phosphotransferase. This enzyme is found as an intrinsic protein on the ER
 and the Golgi [6]. Despite intense efforts from many laboratories over the past 35
years, the cholinephosphotransferase has never been purified from any source. How-
ever, the cDNA for the enzyme has been cloned from yeast and the predicted mole-
cular weight of the protein is 46,305 with 7 transmembrane helices. Choline phos-
photransferase acts at a branchpoint in metabolism of diacylglycerol which can also
be converted to PtdEtn, triacylglycerol or phosphatidic acid. Most studies indicate
that there is an excess of choline phosphotransferase in cells; hence the amount of
active enzyme does not appear to limit PtdCho biosynthesis. On the other hand, the
in vivo activity of choline phosphotransferase appears to be regulated by the supply
of the substrates CDP-choline and diacylglycerol. The supply of CDP-choline is
regulated by the activity of CT as discussed in Section 3. The supply of diacylglycerol
in liver seems to be controlled largely by the supply of fatty acids and the activity of
phosphatidic acid phosphohydrolase (Chapter 6). Excess diacylglycerol not utilized
for PtdCho or PtdEtn biosynthesis is stored in the liver as triacylglycerol.

2.4. Phosphatidylethanolamine-N-methyltransferase

The conversion of PtdEtn to PtdCho (Fig. 1) is catalyzed by an intrinsic protein ex-
posed on the cytosolic surface of the ER and Golgi. In liver the three transmethyla-
tion reactions are catalyzed by a single enzyme with a molecular weight of approx-
imately 18 kDa [5,7]. The enzyme is expressed at high levels only in liver and is 1% or
less in nonhepatic cells and tissues [7]. The major function of the methyltransferase
appears to be the de novo formation of choline via the methylation of PtdEtn to
PtdCho and subsequent hydrolysis of the latter to choline. An interesting feature of
the methyltransferase is its high pH optimum (10.2). The significance of this is un-
known, particularly as the pH around the ER is 7.4. As is the case with choline phos-
photransferase, there appears to be an excess of the enzyme present in liver and its
expression seems to be constitutive under a variety of metabolic perturbations [5,7].
The rate of the conversion of PtdEtn to PtdCho appears to be regulated by the supply
of the substrates PtdEtn and S-adenosylmethionine[7].

3. Regulation of phosphatidylcholine biosynthesis

3.1. The rate-limiting reaction

Considerable evidence has demonstrated that the rate of the CT reaction can limit the
rate of PtdCho biosynthesis. The first evidence in favor of this conclusion was the
measurement of pool sizes of the aqueous precursors (in rat liver choline=0.23 mM,
phosphocholine= 1.3 mM, CDP-choline=0.03 mM). These measurements assume
that 1 g wet tissue is 1 ml and there is no compartmentation of the pools. As we will
see in Section 3.4, however, there does appear to be compartmentation of these Ptd-
Cho precursors. Nevertheless the relative amounts of these compounds might be cor-
rect in the biosynthetic compartment. The concentration of phosphocholine is ap-
proximately 40-fold higher than CDP-choline. The accumulation of phosphocholine
is consistent with a bottleneck in the pathway at the reation catalyzed by CT.
   Pulse-chase experiments demonstrate this bottleneck more vividly. After a half-
hour pulse of hepatocytes in culture with [CH3-3H]choline,more than 95% of the
radioactivity in the precursors of PtdCho was in phosphocholine (Fig. 2), the remain-
der in choline and CDP-choline. A subsequent chase with unlabeled choline in the
medium showed that the labeled phosphocholine was quantitatively converted to
PtdCho (Fig. 2). The radioactivity in CDP-choline remained low during the chase

             Chase Time (h)

Fig. 2. Incorporation of [CH,-3H]cholineinto cholinephosphate and PtdCho as a function of time. Hepato-
cytes from rat liver were incubated with labeled choline for 30 min. Subsequently, the cells were washed
thoroughly and incubated (chased) for various times with unlabeled choline. The disappearance of radio-
activity from cholinephosphate (dashed line) and its appearance in PtdCho (solid line) are shown. Adapted
from Fig. 1 of Pelech et al. 1983, J. Biol. Chem. 258,6783, with permission.

period since its concentration is very low and CDP-choline is rapidly converted to
PtdCho. There was minimal radioactivity in choline which suggests that the choline
is immediately phosphorylated after it enters the cell.
   One additional point should be made. If a cell or tissue is in a steady state, pool
sizes and reaction rates are not changing. Thus, although the rate of PtdCho synthe-
sis may be determined by the CT reaction, the rates of the choline kinase and the
choline phosphotransferase reactions will be the same as that catalyzed by CT. If this
were not the case, changes in the pool sizes of precursors would occur. For example,
if the choline kinase reaction were faster than the CT reaction, there would be an
increase in the amount of phosphocholine. Thus, in many instances, CT sets the pace,
but the other reactions proceed at the same rate.

3.2. The translocation hypothesis

CT is recovered from cells and tissues in both the cytosol and microsomal fraction.
However, in the early 1980s evidence from several laboratories strongly suggested a
close correIation between CT activity on the microsomal membranes and the rate of
PtdCho biosynthesis [4,5]. The hypothesis was that the active form of the enzyme was
on cellular membranes and CT in the cytosol acted as a reservoir (Fig. 1). In agree-
ment with this proposal, there is usually no phospholipid in cytosolic fractions and
CT requires phospholipids for activity. A vivid confirmation of the hypothesis was
performed in HeLa cells. Under normal culture conditions, 65% of CT activity is
recovered in the cytosol and 35% in the microsomes. When HeLa cells were incubat-
ed in the presence of 1 mM oleate, PtdCho synthesis was stimulated at least 10-fold.
When these cells were fractionated, almost all of the CT activity was on the microso-
ma1 membranes. Addition of albumin (which avidly binds fatty acids) to the medium
removed the fatty acids concomitantly with release of CT into the cytosol and a re-
turn of PtdCho biosynthesis to normal rates. Thus, cells have a facile and rapid mech-
anism for altering the rate of PtdCho biosynthesis by a reversible translocation of CT
between cytosol and membranes.
   Recently an aggregated form of CT has been discovered in cytosolic preparations
from lung, alveolar type I1 cells and HepG2 cells, but not from rat liver [8] or HeLa
cells. Since phospholipid was associated with this aggregate, the enzyme was active.
The role of this aggregated form of CT in PtdCho biosynthesis is currently under

3.3. Regulatory mechanisms

The major mechanisms that have been demonstrated for regulating CT translocation
and PtdCho biosynthesis are indicated in Fig. 1. Until recently, there was no indica-
tion that the expression of CT was regulated at the level of gene expression. Recent
studies have shown that addition of colony stimulating factor to a macrophage cell
                                                                                     21 1

line resulted in a 4-fold increase in the mRNA for CT within 1 h. The incorporation
of choline into PtdCho was enhanced approximately 6-fold and a 40% increase in CT
activity was demonstrated in homogenates (S. Jackowski and C.O. Rock, unpublish-
ed results). Similarly a 1.5-fold increase in the amount of CT was found 30 h after a
partial hepatectomy of rat liver compared to sham-operated control animals (L.B.M.
Tijburg, 1991). With probes for mRNA of CT as well as antibody to the enzyme
readily available, we can expect that other examples of regulation of CT at the level
of gene expression might be uncovered.

3.3.1. Fatty acids are important regulators of phosphatidylcholine biosynthesis in
        cultured cells
As discussed above for HeLa cells, a large number of examples show a stimulation of
PtdCho biosynthesis when fatty acids are added to cells in culture and this stimula-
tion correlates with translocation of CT to cellular membranes [5,9]. Whether or not
this is a physiologically relevant mechanism is undecided. At the present time there
is no hard evidence that the supply of exogenous fatty acids such as from plasma
directly regulates CT translocation or PtdCho biosynthesis in intact animals. An al-
ternative possibility is that fatty acids might increase the levels of diacylglycerol
which in turn stimulates PtdCho biosynthesis and CT translocation.

3.3.2. Diacylglycerol may also regulate phosphatidylcholine biosynthesis
The concentration of diacylglycerol may alter the rate of PtdCho biosynthesis both
as a substrate and as a modulator of CT binding to membranes. When HeLa cells
were permeabilized with digitonin, the rate of PtdCho biosynthesis was dependent on
the rate of diacylglycerol biosynthesis at saturating levels of CDP-choline [9]. In vitro
an increase in the content of diacylglycerol in membranes has been shown to enhance
the binding of CT. Similarly, treatment of HeLa cells with the tumor promoter tetra-
decanoyl phorbol acetate stimulates CT translocation and PtdCho biosynthesis [9].
The phorbol ester appears to be acting indirectly via protein kinase C since the effect
is abolished in cells down regulated for protein kinase C. Yet CT is not a substrate for
protein kinase C. However, there is a positive correlation between an increased level
of diacylglycerol, CT translocation to membrane and PtdCho biosynthesis in HeLa
cells treated with phorbol esters. Treatment of rat hepatocytes with phospholipase C
stimulates PtdCho biosynthesis and CT translocation and this seems to correlate
with an increased level of diacylglycerol in the cells. Finally, a synthetic diacylglyce-
rol with 2 octanoyl groups can be incorporated into cells in culture. This dioctanoyl-
diacylglycerol stimulated PtdCho biosynthesis with a concomitant translocation of
CT from cytosol to membranes (R.N. Kolesnick, 1990). Taken together, there is con-
siderable evidence to suggest that diacylglycerol is an important regulator of CT
binding to membranes and PtdCho biosynthesis in vitro and in cell cultures.
Evidence that diacylglycerol has an important regulatory function for CT transloca-
tion and PtdCho biosynthesis in animals remains to be provided.

3.3.3. Phosphorylation-dephosphorylation reactions
Do these modulate the rate of PtdCho biosynthesis and CT translocation? Con-
siderable evidence supports this proposal. Treatment of hepatocytes with cAMP
analogues (which activate CAMP-dependent protein kinase) or with okadaic acid
(an inhibitor of protein phosphatase 112A) results in a decrease in the rate of
PtdCho biosynthesis. In some instances these changes have been correlated with a
decreased association of CT with cellular membranes [5,9]. In addition, CT is a sub-
strate for the CAMP-dependent protein kinase in vitro. However, under the best con-
ditions only 0.2 mol of phosphate were incorporated per mol of enzyme. The phos-
phorylation was probably nonspecific since the state of phosphorylation of CT is
unchanged in rat hepatocytes incubated with cAMP analogues or glucagon (H.
Jamil, 1992).
   Since cAMP does not alter the state of phosphorylation of CT in hepatocytes, the
inhibition of PtdCho biosynthesis appears to be indirect. It is well established that in
hepatocytes the rate of fatty acid biosynthesis is inhibited by phosphorylation of ace-
tyl-CoA carboxylase in response to cAMP (Chapter 4 . Consequently, the rate of
diacylglycerol biosynthesis would be decreased which could reduce PtdCho biosyn-
thesis by decreasing the supply of substrate for the choline phosphotransferase reac-
tion, or decrease the binding of CT to membranes, or both. Since the amount of CT
bound to membranes is not significantly altered in CAMP-treated hepatocytes, the
cAMP inhibition of PtdCho biosynthesis appears to be due to a decreased supply of
diacylglycerol for the choline phosphotransferase reaction.
   The CAMP-mediated inhibition of PtdCho biosynthesis might have physiological
relevance since the levels of CAMP in cells are altered under various conditions. Fast-
ing causes an increase in cAMP levels in liver. Fasting of rats for 48 h has been shown
to cause an increase in the binding of CT to membranes [lo]. With the availability of
an immunoprecipitating antibody against CT, it should soon be possible to see if the
phosphorylation of CT is altered in fasted compared to control animals. Alterna-
tively, the increased binding of CT to membranes in fasted animals could be due to
the observed decrease in PtdCho [lo] as discussed in the next section.

3.3.4. Feedback regulation
The level of PtdCho appears to feedback regulate PtdCho biosynthesis in a physio-
logically relevant model system. The regulation of a metabolic pathway by product
inhibition of its biosynthesis is commonly observed. Suprisingly, such regulation had
never been described for PtdCho biosynthesis until recently [5]. In livers or hepato-
cytes derived from choline-deficient rats, the rate of PtdCho biosynthesis was in-
hibited by approximately 70%compared to choline-supplemented rats and there was
a corresponding increased binding of CT to cellular membranes. The CT appeared to
recognize a need for increased PtdCho biosynthesis and was sitting poised on the
membrane prepared for catalysis. However, the attempted increase in the rate of
PtdCho biosynthesis was futile due to the lack of substrate (cholinephosphate). When

hepatocytes under this condition were supplied with choline, there was a nearly
perfect positive correlation between the increase in the level of PtdCho in the cells and
the release of CT into the cytosol. Similar correlations were observed when the level
of PtdCho was increased either by providing methionine for enhanced conversion of
PtdEtn to PtdCho, or by providing lyso-PtdCho which is taken into hepatocytes and
acylated to PtdCho. This was the first study where a mechanism for regulation of
PtdCho biosynthesis in cell cultures could be directly related back to a physiologi-
cally relevant animal model.

3.3.5. CTP
CTP may also regulate the rate of PtdCho biosynthesis. As noted above, the supply
of choline (hence cholinephosphate) can limit PtdCho biosynthesis in choline-defi-
cient animals. There is some evidence that the concentration of CTP, the other sub-
strate for CT, might limit PtdCho biosynthesis under extreme conditions of hypoxia
in the heart (P.C. Choy, 1990). Further studies will be required to see how important
CTP is in regulating PtdCho biosynthesis in other cells and tissues.

3.4. Substrate channeling

An intriguing and unexpected finding is that the cytosolic, water soluble intermedi-
ates in PtdCho biosynthesis appear to be compartmentalyzed in the cell (Fig. 3) [l 11.
This conclusion was reached in experiments performed with permeabilized glioma
cells. Labeled phosphocholine and CDP-choline were taken up by these cells, but not
by intact cells, and were expected to have been incorporated into PtdCho if there
were equilibration of the water soluble metabolites. This did not occur whereas la-
beled choline was incorporated into PtdCho of the permeabilized cells. The current
model to explain these results is shown in Fig. 3. The suggestion is that choline used
for PtdCho biosynthesis must enter the cell via the choline transporter. Phosphocho-
line and CDP-choline that enter the cells as a result of the permeabilization are com-
partmentalized from endogenously made phosphocholine and CDP-choline, and
cannot be used for PtdCho biosynthesis. How this channeling actually occurs is not
presently understood and whether or not other cell types and tissues have similar
compartmentation remains to be demonstrated. The concept of channeling has impli-
cations for mechanisms that regulate PtdCho biosynthesis. For example, does trans-
location of CT occur within this compartment? How do ATP and CTP access this
compartment? The interested student will have to go to the current literature in order
to know the status of these and related questions.
   There is now considerable evidence to suggest a channeling of newly made, rather
than pre-existing, diacylglycerol for PtdCho biosynthesis [6]. Similarly, newly made
PtdEtn seems to be preferentially used for methylation to PtdCho [5]. Moreover, Pt-
dEtn derived from decarboxylation of phosphatidylserine (PtdSer) seems to be pref-
erentially directed toward lipoproteins secreted from hepatocytes whereas PtdEtn de-

Fig. 3. A model of PtdCho metabolism in mammalian cells. The barrier represents a functional compart-
mentation of the CDP-choline pathway of PtdCho biosynthesis in which intermediates are not freely dif-
fusible in the cell. Linkage between choline transport and PtdCho biosynthesis is proposed. The question
as to whether or not choline derived from PtdCho degradation recycles intracellularly or is exported from
the cell followed by re-uptake is posed. From [l 11 with permission. CT, cytidylyltransferase; CPT, choline

rived from CDP-ethanolamine is preferentially not used [5]. How this channeling of
lipids occurs is not understood.

4. Sphingo~yelin

Early work showed that cell free extracts were capable of forming sphingomyelin
from ceramide and CDP-choline. Not suprisingly it has been assumed for several
decades that this was the pathway that occurred in intact cells and tissues. However,
recent studies have clearly shown that the majority of sphingomyelin arises via a
transfer of phosphocholine from PtdCho to ceramide [12]:

      PtdCho+ceramide+ sphingomyelin+diacylglycerol

Evidence for this pathway came from in vitro assays of mouse liver microsomes and
pulse-chase experiments. In the latter studies, incubation of cells with [methyl-
3H]choline showed a tremendous flux of tritium through CDP-choline to PtdCho

with no label transferred into sphingomyelin. After a few hours when PtdCho be-
came labeled, a precursor-product relationship was demonstrated between PtdCho
turnover and sphingomyelin biosynthesis. Equally persuasive was the isolation of a
temperature sensitive CHO (Chinese hamster ovary-derived cell line) mutant defec-
tive in CDP-choline synthesis. The biosynthesis of PtdCho in this mutant was cur-
tailed but sphingomyelin synthesis, using pre-existing PtdCho, was not affected.
   The site in the cell where sphingomyelin is made has generated some controversy.
Several studies implicated the plasma membrane as the location of sphingomyelin
synthase. Interestingly the plasma membrane is relatively enriched in sphingomyelin
compared to other cellular membranes. However, recent studies demonstrate that, at
least in rat liver, sphingomyelin synthase is largely located on the luminal side of cis
and medial Golgi preparations. In contrast, the enzymes of PtdCho and PtdEtn bio-
synthesis associated with Golgi are on the cytosolic aspect. Whether or not an en-
zyme is located on the cytosolic or luminal side of ER or Golgi is usually demonstrat-
ed by its susceptibility to protease digestion. Choline phosphotransferase of Golgi
from liver is inactivated by trypsin digestion whereas sphingomyelin synthase is not.
   Sphingomyelin synthase has not been purified from any source. This is an impor-
tant objective for future research and would set the stage for cloning of the cDNA
and the genomic DNA. The next decade should also bring insight into how the bio-
synthesis of sphingomyelin is regulated. A likely possibility is that sphingomyelin
biosynthesis is regulated by the supply of ceramide, which is in turn controlled by the
activity of 3-ketosphinganine synthase (Chapter 1 l), the first enzyme in the biosyn-
thesis of sphingosine. Thus, the availability of precursors palmitic acid and serine
stimulate the formation of sphingosine and sphingomyelin in isolated hepatocytes
(A.H. Merrill 1990).

5. Phosphatidylserine biosynthesis

5.1. Historical developments and biosynthesis

PtdSer accounts for 5-15% of the phospholipids in eucaryotic cells. The lower con-
centration of PtdSer compared to PtdCho and PtdEtn is probably why it was not
discovered as a separate component of ‘kephalin’ (which was originally identified to
be only PtdEtn in 1930) until 1941 by Folch. The correct structure was proposed by
Folch in 1948 and confirmed by chemical synthesis in 1955 by Baer and Maurukas.
  PtdSer is made in procaryotes via the CDP-diacylglycerol pathway (Chapter 2).
This route does not exist in animals or plants but clearly does in yeast. In animal cells
PtdSer is made via a route that does not directly include CTP-derived precursors
utilized in the biosynthesis of the other phospholipids. Instead PtdSer is made by a
base-exchange reaction first described by Hiibscher in 1959 (reaction 4, Fig. 4) in
which the head group of a pre-existing phospholipid is exchanged for serine. He sub-


                                                                      II       +

                                                            0       0
                                                            I    II
           0                                                I     I
           II                                               0-      0-
  0   CHflCRI
  I1   I                                                         CDPethonolomine
R$O-CH        0                H
       I       I
               1               I +
      CH2-O-P-OCH              C-NH,




                                                                             Re-C-0-C- H


                                                    HO-C-H        0
                                                         I        II
                                                        CH2-0-P-OCH2CH2NH,         +
Fig. 4. Pathways for the biosynthesis of PtdEtn. The numbers indicate the enzymes involved: 1, ethanola-
mine(cho1ine) kinase; 2, CTP:ethanolaminephosphate cytidylyltransferase; 3, CDP-ethanolamine: 1,2
diacylglycerol ethanolaminephosphotransferase; PtdSer synthase; 5 , PtdSer decarboxylase; 6, phos-
pholipase A?; 7, acyl-CoA:lysoPtdEtn acyltransferase.

sequently demonstrated that the enzyme was microsomal, required calcium, was en-
ergy independent and had a K,,, for serine of 0.5 mM. Recent studies by Voelker have

shown the Ca” requirement can be circumvented by ATP. The enzyme occurs on the
cytosolic surface of microsomes. Suzuki and Kanfer purified an enzyme from brain
microsomes that showed both ethanolamine and serine base exchange activity [ 131.
However, base exchange enzymes are considered to be of minor quantitative im-
portance for the biosynthesis of PtdCho or PtdEtn. The base exchange activity is not
due to phospholipase D activity [I 31.

5.2. CHO mutants and regulation

Surprisingly little information is available on the regulation of PtdSer biosynthesis
except for studies with CHO cells. Nishijima, Akamatsu and coworkers have utilized
CHO mutants that were auxotrophic for PtdSer to demonstrate that the cells have
two base exchange enzymes involved in PtdSer biosynthesis [14].

                              serine-exchange enzyme I
                    PtdCho+serine                 PtdSer+choline

                                       A   7

                                   PtdSer decarboxylase
                          PtdSer                  PtdEtn+CO,

                              serine-exchange enzyme I1
                 PtdEtn+serine        7

It appears that PtdSer is initially made via enzyme I on the ER. The PtdSer is trans-
ported to the mitochondria where it is decarboxylated to PtdEtn and PtdSer is then
regenerated possibly on the ER with the release of ethanolamine. The sum of the
reactions is as follows:

    PtdCho+2 serines    -                 PtdSer+choline+ethanolamine+CO,

Why do these cells have such an apparently complex pathway for PtdSer biosynthe-
sis? The answer is not known but the coupled enzymes yield PtdSer at the expense of
PtdCho and generate both choline and ethanolamine which could be recycled into the
biosynthesis of PtdCho (Fig. 1) and PtdEtn (Fig. 4), respectively. As a result, PtdSer
and PtdEtn are both increased in the cell without a decline in the amount of PtdCho
assuming the choline is reutilized for PtdCho biosynthesis.
   Nishijima, Akamatsu and coworkers have shown that addition of exogenous Ptd-
Ser to the medium of CHO cells feedback inhibits the biosynthesis of PtdSer [14].
Screening CHO cells fortuitously resulted in the isolation of a mutant that did not
show this feedback inhibition [14]. In vitro assays of the particulate fractions derived
from the mutant showed the serine exchange enzyme was 5-fold more resistant to
inhibition by PtdSer compared to normal CHO cells. The precise defect in feedback
21 8

regulation of PtdSer biosynthesis in these mutants is not presently understood. These
studies were the first example of feedback inhibition of the biosynthesis of any phos-
pholipid in an animal cell model.

6. Phosphatidylethanolamine biosynthesis

6.1. Historical background

Thudichum published a book on the chemical composition of the brain in 1884 in
which he described ‘kephalin’,a nitrogen- and phosphorus-containing lipid different
from lecithin. In 1913, Renal1 and Bauman independently isolated ethanolamine
from kephalin. In 1930 Rudy and Page isolated what was probably the first pure
preparation of PtdEtn. The structure was confirmed by chemical synthesis in 1952 by
Baer and colleagues.
  The biosynthesis of PtdEtn in eucaryotes can occur via four pathways (Fig. 4 .The
route via CDP-ethanolamine constitutes the de novo synthesis of PtdEtn. The other
pathways arise as a result of the modification of a pre-exisiting phospholipid. The
CDP-ethanolamine pathway was first described by Kennedy and Weiss in 1956,
when they identified two enzymes, CTP:phosphoethanolamine cytidylyltransferase
and CDP-ethanolamine:1,2-diacylglycerol ethanolaminephosphotransferase. The
decarboxylation of PtdSer to yield PtdEtn was shown in 1960to occur in animal cells.
This is the only route for PtdEtn biosynthesis in E. coli (Chapter 2). Previously, it was
thought that serine could be converted to ethanolamine. However, serine is not a
direct precursor of ethanolamine; rather, the ethanolamine moiety appears to be
generated only via PtdSer decarboxylation. The calcium-mediated base exchange en-
zyme and the reacylation enzyme were both discovered in the early 1960s at the same
time as the corresponding enzymes for PtdCho biosynthesis were described.

6.2. Enzymes of the CDP-ethanolamine pathway

As mentioned previously, the phosphorylation of ethanolamine is catalyzed by cho-
line kinase. In the animal tissues examined it appears that ethanolamine kinase and
choline kinase are the same enzyme [2]. The lack of specificity by the kinase suggests
that the enzyme does not have a particular regulatory role in tissues but simply acts
as a trap for phosphate derivatives of choline and ethanolamine that are transported
into the cell. A kinase has been isolated from soya beans that phosphorylates eth-
anolamine but not choline. The soya bean kinase was reported to have a subunit
molecular weight of 18 kDa, but is also found as a dimer.
   CTP:ethanolaminephosphate cytidylyltransferase was partially purified from rat
liver cytosol in 1975 by Sundler. Unlike CT, the ethanolaminephosphate cytidylyl-
transferase is found exclusively in the cytosol, does not associate with cellular mem-

branes and has no lipid requirement for activity. Similar to CT, this enzyme appears
to catalyze the rate-limiting step for PtdEtn biosynthesis under certain conditions, as
discussed in the next section.
   CDP-ethanolamine: 1,2-diacylglycerolethanolaminephosphotransferaseis an inte-
gral membrane protein found on the ER and Golgi and appears to be different from
choline phosphotransferase. Ethanolaminephosphotransferase has been solubil-
ized from rat liver membranes but has resisted further purification. This enzyme
shows a distinct preference for a diacylglycerol species that has 1-palmitoyl-2-doco-
sahexaenoyl (22:6) fatty acids. In hepatocytes in culture, nearly 50% of the PtdEtn
made via the ethanolaminephosphotransferasereaction is this species. The purpose
of this extraordinary selectivity is unknown.
   Studies have also been done on the ethanolaminephosphotransferasein yeast [151.
Hjelmstad and Bell have isolated 9 mutants defective in this enzyme that belong to 5
different complementation groups. The E P T f mutant exhibited a 30- to 90-fold
reduction in ethanolaminephosphotransferase activity whereas the EPT2 mutant ex-
hibited only a 2- to 3-fold reduction in activity. It appears that the EPTZ gene prod-
uct has both ethanolamine-and choline-phosphotransferase activities. Thus, al-
though the enzyme has still not been purified from any source, molecular genetic
studies in yeast has produced novel information.
   PtdSer decarboxylase has been purified from E. coli (Chapter 2). The enzyme has
been localized to the outer aspect of the inner membrane of mitochondria of the liver
but has not been extensively purified. The difficult task of purification of this enzyme
remains an objective for the future.

6.3. Regulation of the CDP-ethanolarninepathway

6.3. I. Regulation at the cytidylyltransferase reaction
There was minimal knowledge about the regulation of PtdEtn biosynthesis until
Akesson and Sundler tackled the problem in the 1970s. From studies on intact ani-
mals, and subsequently with cultured hepatocytes, they provided evidence that the
ethanolaminephosphate cytidylyltransferase was usually rate-limiting for PtdEtn
biosynthesis (Fig. 4. The concentrations of the precursors to PtdEtn in freeze-
clamped liver were 1.09, 3.83 and 0.24 ,umol/liver for ethanolamine, ethanolamine-
phosphate and CDP-ethanolamine, respectively. From the high molar ratio of etha-
nolaminephosphate to CDP-ethanolamine, the cytidylyltransferase reaction appears
to be the bottleneck in this pathway. In agreement with this proposal, incubation of
freshly prepared hepatocytes with various concentrations of ethanolamine resulted in
a corresponding increase in the amount of phosphoethanolamine in the cell with a
negligible change in the CDP-ethanolamine concentration (Fig. 5) [16]. If the eth-
anolaminephosphotransferase were limiting the rate of PtdEtn biosynthesis, an ac-
cumulation of CDP-ethanolamine would have been expected. Thus, at least in rat
liver, the rate of PtdEtn synthesis can be regulated by the ethanolaminephosphate

                           , ,            J
            0.05 0.10    0.15 0.20 0.25
               Ethonolarnine (pM]

Fig. 5. Effect of ethanolamine on the levels of ethanolaminephosphate and CDP-ethanolamine. Hepato-
cytes were incubated with various concentrations of ethanolamine for 30 min. The cells were subsequently
harvested and the amount of ethanolaminephosphate (0) CDP-ethanolamine (0)
                                                         and                         measured per 30 mg
of cellular protein. Adapted from [16] (Fig. 4)with permission.

cytidylyltransferase-catalyzed reaction, which is analogous to the regulation of Ptd-
Cho biosynthesis. Unlike the cytidylyltransferase involved in PtdCho biosynthesis,
there is almost no literature on the mechanisms that control the activity of ethanola-
minephosphate cytidylyltransferase. One possibility is that the cytidylyltransferase is
present in excess and the supply of CTP is controlling the biosynthesis of PtdEtn.
What regulates the ethanolaminephosphate cytidylyltransferase reaction will be an
important area of investigation during the next few years.

6.3.2. Diacylglycerol
The supply of diacylglycerol can limit the rate of PtdEtn biosynthesis. Treatment of
hepatocytes in culture with glucagon causes an increase in the concentration of
CAMP and an inhibition of PtdEtn biosynthesis via the CDP-ethanolamine pathway
[17]. No corresponding effect on the cytidylyltransferase reaction could be demon-
strated either in vitro or with intact cells. In contrast there was a positive correlation
between the level of diacylglycerol in the cells and the rate of PtdEtn biosynthesis. It
thus appears that the supply of diacylglycerol can limit the rate of PtdEtn biosynthe-
sis when the levels of CAMP are elevated (Fig. 6). Recall that diacylglycerol may also
regulate the biosynthesis of PtdCho (Section 3.3.2.). Whether decreased diacylglycer-
01 is due to a decreased supply of fatty acid or reduced activity of phosphatidate
phosphohydrolase remains to be determined.
   From the studies cited in this and the previous section we can conclude that the
rate of PtdEtn biosynthesis can be regulated by either the supply of diacylglycerol or
                                                                                                     22 1



                                    e-      +Fattyc Acid


         P t dS e r

Fig. 6. The CDP-ethanolamine pathway indicating the possible role of the supply of diacylglycerol (DAG)
for regulation of PtdEtn biosynthesis. Incubation of hepatocytes with glucagon decreases the level of dia-
cylglycerol which appears to limit the rate of PtdEtn biosynthesis [17].

6.4. PtdSer decarboxylcrtion and the relative importance of the various pathways for
     PtdEtn biosynthesis

In essence we do not know the amounts of PtdEtn derived from the various path-
ways. Cultured cells are usually maintained in medium without ethanolamine (a
small amount might be derived from the 10% serum in the culture medium). In such
cases the PtdSer decarboxylase is the major source of PtdEtn. However, in liver there
are high concentrations of precursors of PtdEtn made via the CDP-ethanolamine
pathway suggesting that this pathway is important for the biosynthesis of PtdEtn.
Base exchange may also play a role. An experimental approach to understand the
relative contributions of each pathway remains to be devised. Another unknown is
how the PtdSer decarboxylase pathway is regulated.

6.5. N-acyE-phosphatidylethanolamine

N-acyl-PtdEtn (Fig. 7) occurs naturally in many plants, certain microorganisms, the
central nervous system of fish, mammalian epidermis and in erythrocytes. Biosynthe-
sis occurs with both microsomal and mitochondria1 membrane preparations. The
Ca*+-requiring transacylase transfers a fatty acyl moiety primarily from the sn-1 posi-
tion of PtdCho, PtdEtn or cardiolipin to the ethanolamine moiety of PtdEtn (Fig. 7).
As expected from fatty acids arising from the sn-1 position of PtdEtn and PtdCho,

Fig. 7. The biosynthesisand catabolism of N-acyl-PtdEtn. Plipase D, phospholipase D; PtdA, phosphati-
dic acid.

16:0, 18:0, 18:1 and 18:2 (from the sn- 1 position of cardiolipin) were the major fatty
acyl substituents [18]. The N-acyl-ethanolamine moiety can be removed by a phos-
pholipase D which has no activity on PtdCho or PtdEtn.
   N-Acyl-ethanolamine accumulates in infarcted areas of canine hearts [ 181. In such
areas, the mitochondria release abnormal amounts of Ca". N-Acyl-ethanolamine
attenuates this release of Ca2+by an unknown mechanism [18]. One possibility is that
accumulation of these lipids in damaged heart tissue may be a defense mechanism
against cell damage as a result of ischemia.

7. Po lyglycer ophospholipids

7.1. Historical developments and biosynthetic pathways

Diphosphatidylglycerol, commonly known as cardiolipin, was discovered in 1942 in
beef heart by Pangborn [19]. The correct structure (Fig. 8) was proposed in 1956-57
and confirmed by chemical synthesis in 1965-66 by de Haas and van Deenen. Phos-
phatidylglycerol (Fig. 8) was first isolated in 1958 from algae by Benson and Mauro.
The structure was confirmed as sn-1,2-diacylglycerol-3-phospho-sn-l'-glycerol       by
Haverkate and van Deenen in 1964-65. The third lipid in this class, bis(monoacy1-
glycero1)phosphate was recovered from pig lung by Body and Gray in 1967. The two
fatty acyl esters were later assigned to the two primary hydroxyl groups (Fig. 8). The

R-C-0-CH,                  H2COH
   I1       I                I
R-C-0-C-H           0 H-C-OH
            I       II    I

   0                                    0
   I1                                   II
R- C-0-CH2
                                               I        II
          H2C-O-P-   8
                           O-CH              H&-0-C-R


        HO-C-H       0 HO-C-H
                I     II           I
          H&-O-P-O-               CH,

Bls (monoocylg1ycaro)phosphate

Fig. 8. Structures of polyglycerophospholipids.

stereochemistry was shown to differ from the other two lipids since bis(monoacy1-
glycero1)phosphate contains sn-(monoacyl)glycerol-l-phospho-sn-l’-(monoacyl)-glycer-
01 [19], rather than a sn-glycerol-3-phospho linkage. These three lipids are widely
distributed in animals, plants, and microorganisms. In animals diphosphatidylglycer-
01 is found in highest concentration in cardiac muscle (9-15% of phospholipids) [19].
   Phosphatidylglycerol is generally present at a concentration of less than 1% of to-
tal cellular phospholipids, except in lung, where it comprises 2-5% of the phospholip-
id. In pulmonary surfactant and alveolar type I1 cells, phosphatidylglycerolis 7-1 1%
of the total lipid phosphorus. Bis(monoacyLglycero1)phosphateis less than 1% of the
phospholipids in normal animal tissue, except for alveolar (lung) macrophages,
where it is 1 4 18% of the phospholipid. Diphosphatidylglycerol and phosphatidyl-
glycerol are quantitatively more important lipids in plants and bacteria than in ani-
mals; in two blue-green algae, phosphatidylglycerol is the only phospholipid [191.
Similarly, in Acholeplusmu luidlawii, phosphatid ylglycerol is the only phosphoglycer-
ide present.
   The biosynthesis of phosphatidylglycerol was elucidated by Kennedy and co-
workers in 1963 in chicken liver as follows:

      CDP-diacylglycerol+sn-glycerol-3-phosphate + phosphatidylglycerol
      phosphatidylglycerolphosphate + phosphatidylglycerol+Pi

This pathway was also described in E. coli in 1964 by Kanfer and Kennedy.
  Establishment of the pathway for diphosphatidylglycerol biosynthesis required
several years. It is now well accepted that phosphatidic acid is transferred from CDP-
diacylglycerol to phosphatidylglycerol, according to the following reaction:

      phosphatidylglycerol+CDP-diacylglcyerol diphosphatidylglycerol+CMP

The reaction for diphosphatidylglycerol synthesis in E. coli (Chapter 2) differs and
involves the condensation of two molecules of phosphatidylglycerol:

      2 Phosphatidylglycerol3 diphosphatidylglycerol+glycerol

The E. coli pathway for diphosphatidylglycerol synthesis does not seem to occur in
mammalian mitochondria [19].

7.2. Enzymes and subcellular location

Phosphatidylglycerol can be made in mitochondria and microsomes from various
animal cells and, except for lung, appears to be primarily converted into diphosphati-
dylglycerol. Diphosphatidylglycerol is biosynthesized exclusively in the mitochon-
dria and is found only in this organelle. Bis(monoacylg1ycerol)phosphatecomprises
7% of the phospholipid in rat liver lysosomes, where enzymes for the synthesis of this
lipid are found. In macrophages from lung alveoli it appears that phosphatidylglycer-
01 derived from lung surfactant is the precursor of bis(monoacylglycero1)phosphate.
The pathway, currently under investigation, for the conversion of phosphatidylglyc-
erol to bis(monoacylglycero1)phosphate is interesting since it involves a sterochem-
ical change. Both glycerol residues are linked to the phosphate via their sn-1 carbons.

8. Inositol phospholipids

8. I . Historical developments

The first report of an inositol-containing lipid was in 1930 from Mycobacteria [20].
That discovery is humorous because inositol (Ins) lipids are rarely found in bacteria.
Brain is the richest source of these lipids, as first discovered by Folch and Wooley in
1942. In 1949, Folch described phosphatidylinositol phosphate (PtdIns-P), which
was later found to include phosphatidylinositol (PtdIns) and phosphatidylinositol


      I    II

Fig. 9. Structure of PtdIns-Pl.

bisphosphate (PtdIns-Pa). The chemical structures of PtdIns, PtdIns-P, and PtdIns-
P2 (Fig. 9) were determined by Ballou and co-workers between 1959 and 1961.
  The scheme for the biosynthesis of these three inositol phospholipids is as follows:

     phosphatidic acid+CTP F=+ CDP-diacylglycerol+PPi
     CDP-diacylglycerol+inositol    +
     PtdIns+ATP 3 PtdIns-P+ADP
     PtdIns-P+ATP 3 PtdIns-P2+ADP

Agranoff et al. published the first experiments in 1958 on the incorporation of [3H]in-
ositol into PtdIns. The scheme postulated phosphatidic acid, CDP-choline, and
CDP-diacylglycerol as precursors. Subsequently, Paulus and Kennedy showed that
CTP, rather than CDP-choline, was the preferred nucleotide donor.
  PtdIns is present in rat liver at a concentration of 1.7pmol/gliver [21]PtdIns-P and
PtdIns-P2 are present at much lower concentrations (1-3% of PtdIns [21]).

8.2. Biosynthetic enzymes

There are three potential sources for cellular inositol: diet or uptake from the plasma,
de novo biosynthesis and recycling of inositol. Biosynthesis of inositol from glucose
can occur in the brain and testes and other tissues to a lesser extent. The rate-limiting
step appears to be the synthesis of inositol-3-phosphate (Ins-3P) from glucose-6-
phosphate [22]. Ins-3P is converted to inositol by inositol monophosphate phospha-
   Myo-inositol is transported into liver cells by a carrier-mediated process that does
not involve active transport. Inositol is converted to the polyphosphoinositides as
shown in Section 8.1. The enzymes involved are located on the cytosolic surface of
the endoplasmic reticulum and the plasma membrane.
   PtdIns synthase has been purified from rat liver after solubilization from micro-
somes by sodium cholate (K. Egawa 1977). The enzyme requires Mg2+or Mn2+and

phospholipid. There is a report that this enzyme activity is feedback inhibited by the
product, PtdIns (M.C. Gershengorn 1987).
   The phosphorylation of PtdIns at the 4 position appears to be catalyzed by several
different isoenzymes. Two PtdIns 4-kinases have been identified from bovine brain.
One is an integral membrane protein that has been purified to homogeneity [22]. This
isozyme has a low apparent K, for ATP (20-70 pM) and a molecular mass of 55 kDa.
The second kinase has a high K, for ATP (250-750 pM) and an apparent molecular
mass of 200 kDa. Two additional PtdIns 4-kinases have been detected and character-
ized [22]. The role of the different isoenzymes of PtdIns 4-kinase remains to be eluci-
   PtdIns-4P 5-kinase which converts PtdIns-P to PtdIns-P2 can be isolated from
both cytosolic and membrane sources. This kinase appears to be a peripheral mem-
brane enzyme. The relationship between the cytosolic and membrane forms of the
5-kinase has not been determined. Since there is also a PtdIns-4P 3-kinase, it is im-
portant to show which of the hydroxyl groups on the inositol moiety is phosphorylat-
ed [22].

9. Phospholipids as precursors of cellular second messengers

It is now overwhelmingly clear that a major function of phospholipids in cells is to
store the precursors of second messengers that activate specific processes in the cell.
The eicosanoids were the first phospholipid-derived second messengers identified
and will be discussed thoroughly in Chapter 10. This section will deal with the inosi-
to1 phosphates and diacylglycerols.

9.1. Discovery of the phosphatidylinositol cycle

Over 30 years ago, Hokin and Hokin observed what was later called the ‘PtdIns cy-
cle’ [20]. Incubation of pigeon pancreas with the neurotransmitter acetylcholine
caused the release of the digestive enzyme amylase. If 32P were included in the incuba-
tion, there was a rapid labeling of PtdIns and phosphatidic acid. It is now known that
binding of acetylcholine to its cell surface receptor results in hydrolysis of inositol
lipids to diacylglycerol and inositol phosphates. The 32Pincorporation was the result
of resynthesis and rephosphorylation of the lipids. This sequence of reactions is
known as the PtdIns cycle (Fig. 10). Many other agonists (for example, vasopressin,
thrombin) cause similar rapid labeling of PtdIns from 32Pi many different types of
cell. Much work has been expended since 1953 to understand the relationship be-
tween receptor stimulation and inositol lipid metabolism. Despite intensive efforts,
our understanding of these events had not progressed substantially until the early
1980s. It is now recognized that the initial event in inositol lipid metabolism occurs
within 20-30 s of the binding of the agonist to the receptor and involves primarily the




                                                                 -   I-P3

                  CTP                                                 OH
Fig. 10. The PtdIns cycle. I stands for inositol.

catabolism of PtdIns-P2 to diacylglycerol a i d inositol-l,4,5- trisphosphate (IP, in
Fig. 10). For example, 20% of cellular PtdIns-P, is degraded within 30 s of exposure
of hepatocytes to vasopressin or exposure of platelets to thrombin [23]. As shown in
Fig. 10, the diacylglycerol is converted to CDP-diacylglycerol, and the inositol tris-
phosphate is degraded to inositol. The inositol reacts with CDP-diacylglycerol to
yield PtdIns, which can be converted to PtdIns-P,, thus completing the cycle.
   There appear to be two reasons why scientists originally overlooked the involve-
ment of PtdIns-P, and PtdIk-P after receptor activation [23]. First, the reactions
occur extremely rapidly. Second, the polyphosphoinositides are not extracted by the
normal lipid extraction procedures that involve CHCl, and CH,OH. If concentrated
HCl(1 ml) were included with 100 ml of CHC1, and 100 ml of CH,OH, PtdIns-P, and
PtdIns-P were extracted [21].
   The degradation of PtdIns-P, (and stimulation of the PtdIns cycle) is caused by
hormones that increase the concentration of cytosolic Ca2+[21]. The agonist (for ex-
ample, vasopressin) binds to a cell surface receptor which somehow activates a phos-
pholipase C that degrades PtdIns-P, to diacylglycerol and inositol trisphosphate. The
release of inositol trisphosphate and the increase in cytosolic Ca2+ precedes the
activation of phosphorylase, a well-described Ca2+-mediatedeffect of vasopressin.
Thus, inositol trisphosphate is thought to be the intracellular signal which is re-
sponsible for the vasopressin-induced increase in cytosolic Ca2' from intracellular

9.2. Degradation o phosphatidylinositol-P, by phospholipase C

After an agonist binds to a receptor on the cell membrane, a PtdIns-specific phospho-
lipase C appears to be activated via a GTP-binding protein. Evidence for GTP in-
volvement is that a nonhydrolyzable analogue GTPyS (where a sulphur atom

replaces oxygen between the /3 and y phosphate residues) stimulates phospholipase C
in vitro. Secondly, in vitro reconstitution experiments with phospholipase from
turkey erythrocyte cytosol and turkey erythrocyte membranes demonstrated the re-
quirement of GTP for hydrolysis of Ptdlns-P,.
   Four immunologically distinct forms of PtdIns-specific phospholipase C have been
identified. The molecular masses are 56.6 kDa for a,138.2 kDa for / , I148.4 kDa for
y and 85.8 kDa for 6 [24]. All isoenzymes require Ca2+and degrade PtdIns, PtdIns-P
and PtdIns-P,. However, which lipid is degraded depends on the conditions em-
ployed. When the phospholipase was assayed in the presence of a lipid mixture which
resembled the inner leaflet of the plasma membrane, the hydrolysis of PtdIns- P2was
100-fold higher than of PtdIns. This appears to be true for all of the isoen-
zymes. PtdEtn and PtdCho were not substrates. Although the phospholipases C have
been largely purified from the cytosol of tissues homogenates, the substrate is in the
plasma membrane and this is the likely site for hydrolysis.
   The y isozyme of phospholipase C appears to be regulated by direct phosphoryla-
tion [22]. Both platelet-derived growth factor and epidermal growth factor stimulate
PtdIns-P, hydrolysis in certain cells. The receptors for these factors have tyrosine
kinase activity. In cells with mutant receptors, in which the tyrosine kinase activity
was deficient, hydrolysis of PtdIns-P, did not occur. The regulation of the phospho-
lipase by this kinase may apply only to the y isoenzyme which is a substrate for tyro-
sine kinase both in vitro and in vivo. The y phospholipase C is also a substrate for
protein kinase C and CAMP-dependentprotein kinase. An increase in cAMP causes
an increase in phosphorylation of y phospholipase which leads to a decreased PtdIns-
P2hydrolysis induced by an agonist. In contrast, phosphorylation of the phospholip-
ase by the tyrosine kinase stimulated PtdIns-P, hydrolysis. One complication, how-
ever, is that in vitro phosphorylation of the phospholipase by cAMP kinase has no
effect on the phospholipase C activity assayed in vitro.

9.3. Metabolism of the inositol phosphates

Myo-inositol is the major isomer of hexahydroxy cyclohexane found in eucaryotic
cells. For myo-inositol all of the hydroxyl groups are in the equatorial position except
for the substituent on carbon 2 which is axial. It has been estimated that there are 63
possible distinct species of inositol monophosphate [22]. Twenty-five of these have
been reported to occur naturally. The inositol phosphates are found in the cytosol of
tissue extracts. As would be expected, the location of the phosphate residues is critical
to function. Ins(1,4,5)P3 appears to function in cells by binding to a receptor on the
endoplasmic reticulum which mediates the release of Ca2+intothe cytosol. The iso-
mer Ins(1,3,4)P3 binds to the receptor with a 100-fold lower affinity.
   The metabolism of Ins(l,4,5)P3 is shown in Fig. 11 and involves the action of kina-
ses and phosphatases. Two Ins(l,4,5)P3 phosphatases have been identified [22]. Type
I has a molecular weight of 66 kDa and both Ins(1,4,5)P3 and Ins(1,3,4,5)P4 are

    I         I n d l .4)P2
                                       I n s (I, . 4 ) P 3

                              I n s ( 1 , 3 ) P2     Ins ( 3.4IP2


                                Ins-1P                 Ins-3P
                               Inosi to1

Fig. 11. Possible metabolic fates of Ins(1,4,5)P3.

substrates. Type I1 has a molecular weight of 160 kDa and is specific for Ins(1,4,5)P3.
Ins( 1,4,5)P3 3-kinase has been purified from rat brain and its cDNA is cloned. This
kinase appears to be activated by Ca” and a calcium-binding protein (calmodulin).
Eventually much of the Ins( 1,4,5)P3 is converted to inositol for recycling into PtdIns.
Ins-P phosphatases are inhibited by lithium and thereby prevents the cycle with a
concomitant increase in the concentration of inositol phosphates in cells, Interesting-
ly, lithium is used in the treatment of patients with manic depression.
    The story of inositol phosphate metabolism is even more complex since there are
additional inositol polyphosphates. Both Ins( 1,3,4,5,6)P5 and Ins-P6 have been iden-
tified at levels of 5-15 p M in all tissues examined. Millimolar levels of Ins-P5 have
been identified in avian red blood cells. The Ins-P5 binds to hemoglobin and may
regulate oxygen binding.

9.4. Function o inositol phosphates

The Ins( 1,4,5)P3 appears to cause the release of Ca2+from non-mitochondria1 stores,
largely from the endoplasmic reticulum. A receptor for Ins(l,4,5)P3 has been identi-
fied as a 313 kDa protein that has both Ca” channel activity and a ligand binding site
1251. The receptor binds the D-isomer of Ins(l,4,5)P3 with 3000-fold greater affinity
than the L-isomer. The receptor is phosphorylated by CAMP-dependent protein ki-
nase which causes a decrease in both Ins(1,4,5)P3 binding and Ca” release.
Ins(1,3,4,5)P4 does not bind to the receptor hence phosphorylation at the 3 position
of Ins( 1,4,5)P3 inactivates the Ca2’-releasing activity.
   Recent results have demonstrated another pathway in platelets, macrophages and
neutrophils in which PtdIns is converted to PtdIns(3,4)P2 [22]. As this is not a sub-
strate for the phospholipases C discussed previously, it is possible that the inositol

lipid itself may have an unidentified function. Interested students should survey the
current literature to find out recent developments on PtdIns(3,4)P2.

9.5. Diacylglycerol and protein kinase C

The other product of PtdIns-P2 hydrolysis is diacylglycerol which activates protein
kinase C. This kinase, which was discovered in 1977 by Nishizuka and co-workers
[26], phosphorylates a variety of cellular proteins. Protein kinase C appears to be
involved in regulating many cellular processes including secretion, exocytosis, down-
regulation of receptors, gene expression and cell proliferation. As an example, the
epidermal growth factor receptor is a substrate for protein kinase C. When the recep-
tor is phosphorylated, it exhibits decreased binding of epidermal growth factor and
decreased tyrosine phosphorylation activity.
   Seven isoenzymes of protein kinase C have now been identified [26]. All species
have similar structures with a regulatory domain and a kinase domain. Protein ki-
nase C appears to exist in the cytosol in an inactive form and, on the generation of
diacylgycerol in the plasma membrane, is translocated to this membrane and activat-
ed. Protein kinase C also requires Ca2+and PtdSer for activity.
   Protein kinase C can also be activated by a variety of phorbol esters [26]. These
compounds, which have some structural similarity to diacylglycerol, are co-carcino-
gens that enhance the formation of tumors. Phorbol esters are not tumorogenic by
themselves. Like diacylglycerol they appear to act by binding directly to protein ki-
nase C. Phorbol esters are useful for identifying cellular reactions in which protein
kinase C is involved. Prolonged incubation of cells with phorbol esters causes down
regulation of protein kinase C thus decreasing or eliminating the kinase activity from
the cell.
   A surprising discovery by Bell, Merrill and others was that sphingosine and related
compounds are potent inhibitors of protein kinase C [27] (see also Chapter 11).
Sphingosine appears to be a competitive inhibitor with respect to diacylglycerol and
phorbol esters. Treatment of cells with low concentrations of sphingosine (5-25 pM)
interferes with protein kinase C-mediated events such as secretion by platelets, plate-
let aggregation and the neutrophil respiratory burst (release of superoxide and other
oxygen radicals that kill ingested bacteria). Whether or not the sphingosine effects
have physiological relevance remains controversial at the present time. However,
both sphingomyelinase and sphingosine have been identified in the plasma mem-
   A ‘sphingolipid cycle’ has been proposed in which an effector might activate sphin-
golipid hydrolysis to yield ceramide which in turn is degraded to sphingosine. The
possibility also exists that ceramide, which has a structure (Chapter 11) reminiscent
of diacylglycerol, may function as a second messenger. Evidence has been presented
for a possible role for ceramide in differentiation of HL-60 cells (promyelocytic leu-
kemia-derived cell line) to monocytes (Y. Hannun 1991).
                                                                                 23 1

9.6. Phosphatidycholine cycles andformation o diacyglycerol

It is now quite clear that PtdIns is not the only source of diacylglycerol which can
activate protein kinase C; PtdCho, and possibly PtdEtn, can serve this role [28]. As
discussed in Chapter 10 PtdCho is also a major source of arachidonic acid for the
biosynthesis of eicosanoids. Most agonists that cause PtdIns-P, hydrolysis also cause
the catabolism of PtdCho. Since there is an abundance of PtdCho in the plasma mem-
brane of cells (Chapter 1) compared to PtdIns-P,, PtdCho could provide a richer and
more sustained source of diacylglycerol. Compared to PtdIns-P, hydrolysis, the hy-
drolysis of PtdCho has the potential advantage of providing diacylglycerol for me-
diating the activation of protein kinase C without the concomitant mobilization of
Ca2+.There are a few examples where PtdCho is catabolized to diacylglycerol with-
out catabolism of PtdIns-P, such as in T lymphocytes incubated with interleukin 1
[28]. There is evidence for the involvement of GTP binding proteins, protein kinase
C, CaZ+ tyrosine kinases in regulation of agonist-induced PtdCho hydrolysis [28].
   Two potential pathways exist by which diacylglycerol could be generated from
PtdCho: directly via phospholipase C or via the combined action of phospholipase D
and phosphatidate phosphohydrolase (Fig. 12) Recent studies favor the phospholi-
pase D pathway. A phosphatidate phosphohydrolase that differs from the enzyme
involved in glycerolipid biosynthesis (Chapter 6) has now been described (D.N.
Brindley 1991). Phosphatidate itself has been implicated as a possible second messen-
ger in cells. It is expected that the catabolites of PtdCho would be recycled via a
putative PtdCho cycle as shown in Fig. 12.

10. Phospholipid Catabolism and remodeling of the acyl substituents

As indicated in Fig. 12 the catabolism of PtdCho involves more than the formation
of phosphatidic acid or diacylglycerol. These two compounds appear to be minor
degradation products of PtdCho and glycerophosphocholine appears to be the major
product. An enzyme that converts glycerophosphocholine to choline and glycerol-3-
phosphate has been described. Both of these products can be recycled into PtdCho as
shown in Fig. 12. Similarly, glycerophosphoethanolamine appears to be the major
catabolic product of PtdEtn. Our knowledge about the regulation of PtdCho and
PtdEtn catabolism is very limited at this time.
   Another process depicted in Figs. 12 and 13 is the ‘remodeling’ of the fatty acid
composition of PtdCho. After lyso-PtdCho is formed, it can be either degraded or
reacylated to reform PtdCho. If the latter reaction occurs and a new fatty acid is
introduced, the fatty acid composition of PtdCho is said to have been ‘remodeled’.
How the cell decides to degrade or to reacylate lyso-PtdCho is poorly understood. As
indicated in Fig. 13 it is important to note that remodeling occurs on both the sn-1
and sn-2 positions of PtdCho. For example, a major molecular species of PtdCho


               CT      4



Fig. 12. PtdCho cycles. The possible catabolic fates ot'YtdCho (PC) are shown. Diacylglycerols (DAG) can
be generated directly via phospholipase C (PLC) or by the action of phospholipase D (PLD) to yield phos-
phatidic acid (PA), which is cleaved to DAG by PA phosphobydrolase (PAP). Alternatively, the phos-
pholipase A2 (PLA,)-catalyzed hydrolysis of PC provides lyso-PC and arachidonic acid (C20:4), an im-
mediate precursor of eicosanoids as described in Chapter 10. The lyso-PC may be re-esterfied to PC or
catabolized to glycerophosphocholine (GPC), which can be further degraded to glycerol-3-phosphate
(G3P) and choline. G3P can be converted back into DAG via PA synthesis. DAG can react with CDP-
choline to complete the PC cycle. From Pelech, S.L. and Vance, D.E. (1989) Trends Biochem. Sci. 14,

formed from the methylation of PtdEtn is 16:0-22:6-PtdCho [29]. This species has a
half-life of less than 6 h and appears not to be significantly degraded but rather con-
verted to other molecular species, particularly those with 18:O on the sn-1 position
and 20:4, 18:2 or 22:6 on the sn-2 position. Deacylation-reacylation appears to be the
major mechanism for the formation of these species of PtdCho.
   Interestingly, PtdEtn derived from the CDP-ethanolamine pathway does not ap-
pear to be significantly remodeled in rat hepatocytes [29]. Newly made PtdEtn is
preferentially converted to PtdCho via the methylation pathway. At later times (after
2 h) PtdEtn is no longer methylated but is degraded within 10 h. The function of this
relatively rapid catabolism of PtdEtn is unknown. (The catabolism of PtdCho ex-
tends over several days.) Whether or not the same reactions occur in other cell types
and whether or not PtdEtn derived from PtdSer is similarly metabolized, are the sub-
ject of current investigations.

Fig. 13. The fatty acids at both the sn-l and sn-2 positions of PtdCho can be deacylated by phospholipases
and reacylated by acyltransferases. Palmitic acid (16:O) can be removed from the sn-1 position and replaced
with stearic acid (1 8:O). The fatty acid at the sn-2 position is depicted as docosahexaenoic acid (22:6) which
can be replaced with 20:4 or 18:2. If the fatty acid were oleic acid at the sn-2 position, it could also be
deacylated and reacylated. The above sequence of reactions has not been shown to be correct and deacyla-
tionlreacylation could occur at the sn-2 position initially. Plipase, phospholipase; 1-AT, acyl-
CoA:lysoPtdCho 1-acyltransferase; 2-AT, acyl-CoA:lysoPtdCho 2-acyltransferase; Cho, choline.

11. Phospholipid biosynthesis in yeast

The pathways for the biosynthesis of phospholipids in yeast were largely elucidated
by Lester and co-workers in the late 1960s [15]. These pathways are similar to those
found in other eucaryotes except PtdSer is made via a pathway similar to that found
in E. coli where CDP-diacylglycerol reacts with serine to yield PtdSer and CMP. Re-
call from Section 5 that animal cells make PtdSer by a base exchange mechanism.
   Considerable interest in yeast as a model system has developed over the past
decade. Reasons for choosing Succharomyces cerevisiae include a large knowledge

base in classical genetics, recent developments in molecular genetics and the ability to
grow large amounts easily [ 151. Unlike other eucaryotes regulation of phospholipid
biosynthesis at the level of gene expression has been demonstrated in yeast. Thus,
when yeast cells are grown in the presence of choline and inositol, the enzyme ac-
tivities involved in the conversion of phosphatidic acid to PtdCho (via PtdSer, its
decarboxylation and methylation of PtdEtn) are repressed [15]. Why inositol is re-
quired for the repression of the enzymes involved in PtdCho biosynthesis is not clear.
A parallel regulation has not been detected in animal cells. For a more comprehen-
sive discussion on phospholipd biosynthesis in yeast, the reader is referred to the
excellent review by Carman and Henry [IS].

12. Phosphonolipids

12.1. Historical perspective

The natural occurrence of an analogue of phosphoethanolamine (Zaminoethylphos-
phonic acid) that contains a carbon-phosphorus bond was reported in 1959 by Hori-
guchi and Kandatsu [30]. Rouser and colleagues in 1963 reported the first phospho-
nolipids, N-acyl-sphingosyl-1-0-aminoethylphosphonate 14), in a sea anemone.
  The phosphonolipids are present in only trace amounts in mammalian tissues, but

    0  CH2-OCR,
   II  I
                      *         +


    0   CH *-OC,, H ,
    II  I
                 I1    *
        C H2-0-P- C H2-C H2-N H3
  CH, (CH21,, ~-C-CH-CH-CY-O-P-CH,-CH,-NH,
                                             *            +
                    I    I          I
              H     OH NH           0-


Fig. 14. Structures of several phosphonolipids. The asterisk indicates the carbon-phosphorus bonds.

occur extensively in molluscs, coelenterates, and protozoa. Glycerophosphonolipids
(Fig. 14) were first isolated from the protozoan Tetrahymenapyriformis by Liang and
Rosenberg in 1966. Glycerophosphonolipids are present in high concentrations in
Tetrahymena (about 23% of phospholipid) and are the major phospholipid in the
ciliary membrane. The glycerophosphonolipidsare absent from molluscs and coelen-
terates, which instead contain large amounts of N-acyl-sphingosyl-aminoethylphos-
phonate (5-20% of the phospholipids).

12.2. Biosynthesis

The unique feature of the phosphonolipids is the carbon-phosphorus single bond.
Aminoethylphosphonic acid appears to arise from phosphoenolpyruvate according
to the following reaction:

     coo- 0
     I    II           "
                      i'                 II
                                                       L                   qH3
     CHz     0-                            I
                                      -0-P=0                          -0-P=O
phosphoenolpyruvate                        I                               I
                                           0-                              0-

                             2-phosphonoacctaldcnyde         2-arninoethylphosphonic acid

  Once aminoethylphosphonate is formed, it is apparently incorporated into phos-
pholipid-like compounds via the CMP derivative:

      2-arninoet hylphosphonic acid            CMP-arninoethylphosphonic       acid

                      +   diacylglycerol       -  d iacyl gl yceroarn Inoet hyl phosphonate   +   CMP

 CMP-arninoethylphosphonic acid

  These reactions are analogous to those for the biosynthesis of PtdEtn via CDP-
ethanolamine (Fig. 4). It has not been determined whether or not the same enzymes
are involved for the biosynthesis of both lipid classes. Detailed studies on the biosyn-
thesis of N-acyl-sphingosylamino-ethylphosphonate      have not been reported.
  The phosphonolipids are degraded by phospholipase C activities. The subsequent
degradation of the phosphono (carbon-phosphorus) bond presents a unique prob-
lem to the biochemist. Interestingly, an enzyme which cleaves this bond has never
been reported in any of the sources that contain these compounds in significant quan-
tities. However, such an enzyme, phosphonatase, which cleaves the C-P bond of ami-
noethylphosphonate, has been isolated from Bacillus cereus even though bacteria do
not contain phosphonolipids. The aminoethylphosphonate is first transaminated to
form the corresponding aldehyde by a transaminase; then, phosphonatase cleaves the
carbon-phosphorus bond to yield acetaldehyde and inorganic phosphate.

                           pyruvate   alanine
                                                                     -    0
                                                                         HC-CH3   t P,

2-ominoethyl phosphonate                 2-phosphonoacetaldehyde

   The specific functions of the phosphonolipids as distinct from other phospholipids
have not been clearly defined. Studies on the subcellular distribution of the glycero-
phosphonolipids in Tetrahymena showed an enrichment of these lipids in the cilia
and pellicles of these organisms relative to internal membranes such as mitochondria
and microsomes. This may be a protective adaptation due to the relative resistance of
the phosphonolipids to chemical and enzymatic degradation. Tetrahymena do not
have the protective coating that fungal and many bacterial cells have.
   Growth of Tetrahymena at low temperatures causes an enrichment of polyunsatu-
rated fatty acids (particularly 18:2) in the phosphonolipids. However, the phosphono
bond shows no significant difference from phosphate bonds in the physical behavior
of these membrane lipids as a function of temperature. Hence, the temperature re-
sponse is similar to that of phospholipids seen in bacteria and eucaryotes which do
not contain phosphonolipids. As a result, no selective advantage is apparent for the
phosphonolipids in temperature acclimatization of membranes.
   The selection of mutants of Tetrahymena without phosphonolipids would demon-
strate that these lipids are not essential for life of this organism. Such mutants would
allow comparative studies with wild-type strains which might provide definitive in-
formation on the specialized function, if any, of the phosphonolipids.

13. Glycosyl phosphatidylinositols for attachment of cell surface
So far we have seen that phospholipids have a variety of roles in cellular functions
many of which were not appreciated even a decade ago. An even more recent discov-
ery is that a variety of cell surface proteins are linked covalently to a glycosyl PtdIns

(GPI) in the plasma membrane [31,32]. These GPI anchor a wide variety of proteins
including hydrolytic enzymes such as lipoprotein lipase (Chapter 16), cell surface an-
tigens, protozoan antigens (best described is the variant surface glycoprotein of Try-
panosoma brucei) and proteins involved in celkell interactions [31,32]. The advan-
tage to the cell of linking such a wide variety of proteins to the cell via GPI is not
clear. However, proteins linked in this manner can be released from the cell surface
after treatment with a PtdIns-specific phospholipase C or D. Thus, one potential
function of the GPI link would be to allow the selective release of proteins from the
cell surface by phospholipases. The physiological advantage of selective release of
GPI-linked proteins is largely a matter of speculation.
   The core structure of GPI consists of PtdIns linked to a tetrasaccharide which has
a phosphoethanolamine attached to a mannose residue (Fig. 15). The ethanolamine
moiety is linked to the protein via the CI carboxyl group of the C-terminal amino acid
residue. This core structure is found in a wide variety of cells from I: brucei to brain
and erythrocytes. There are wide variations in structures of GPI from different sour-
ces that include the addition of other saccharides, fatty acids or another phosphoeth-
anolamine residue to the core structure [31,321.
   It is thought that the proteins may become linked to GPI in the endoplasmic reticu-
lum soon after protein synthesis is completed. The signal for GPI attachment to the
protein has not been completely elucidated but probably involves hydrophobic and
hydrophilic regions in the C-terminal region [32].
   A cell-free system for the biosynthesis of the core GPI structure has been developed
with extracts from I: brucei [32]. The process begins with the transfer of N-acetylglu-
cosamine (GlcNAc) to PtdIns. After deacetylation, the GlcN-PtdIns is modified by
sequential addition of 3 mannose residues from GDP-mannose. A dolichol interme-
diate linked to mannose may be involved in the transfer [32]. In the final step ethano-
laminephosphate is added, the donor of which does not appear to be CDP-ethanola-
mine, but possibly PtdEtn. The GPI of I: brucei is unusual in that the fatty acids
substituents of PtdIns are myristate (14:O). The addition of myristate appears to in-
volve myristoyi-CoA and occurs by remodeling reactions after the initial core GPI is



Fig. 15. Structure of the core glycosyl PtdIns (GPI). The protein is linked to ethanolamine (Etn) via an u
carboxyl group at the C terminus. The other abbreviations are: P, phosphate; Man, mannose; GLcN,
glucosamine; Ins, inositol; DAG, diacylglycerol.

14. Future directions
Since the first edition of this book was published in 1985 there has been astonishing
progress in the area of phospholipid metabolism. Many of the enzymes involved in
phospholipid biosynthesis have now been purified and some of the cDNAs have been
cloned. We can anticipate further progress along these lines in the future. Control
mechanisms involved in regulation of phospholipid biosynthesis are now better un-
derstood through classical approaches and use of genetic mutants. The concepts that
are emerging now, such as feedback regulation of phospholipid biosynthesis and
feedforward regulation by diacylglycerol, should be more firmly established in the
years to come. In the next decade the role of phosphorylatioddephosphorylation
reactions for regulation of phospholipid biosynthesis should be better understood.
The role of channeling of biosynthetic precursors and the apparent compartmentali-
zation of phospholipid metabolism will be vigorously investigated and new insights
can be anticipated. We can also anticipate that much progress will be made in our
understanding of the control of phospholipid catabolism and the enzymes involved.
   Phospholipids as a source of cellular second messengers has literally exploded
upon the biomedical research community. Research throughout the world in this
area probably costs in the vicinity of $100 million annually. We can expect dramatic
strides in our understanding of the enzymes involved, and the discovery of additional
pathways and processes whereby cells regulate the generation of the second messen-
   It seems likely that the 1990s will be the most important decade for understanding
the function and regulation of phospholipid metabolism since this area of research

1. Kennedy, E.P. (1989) Discovery of the pathways for the biosynthesis of phosphatidylcholine. In: Phos-
   phatidylcholine Metabolism (Vance, D.E., Ed.) pp. 1-9, CRC Press, Boca Raton, FL.
2. Ishidate, K. (1989) Choline transport and choline kinase. In: Phosphatidylcholine Metabolism (Vance,
   D.E., Ed.) pp, 9-32, CRC Press, Boca Raton, FL.
3. Knipper, M., Boekhoff, I. and Breer, H. (1989) Isolation and reconstitution of the high-affinity choline
   carrier. FEBS Lett. 245, 235-237.
4. Vance, D.E. (1989) CTP:Cholinephosphate cytidylyltransferase. In: Phosphatidylcholine Metabolism
   ('Vance, D.E., Ed.) pp. 3345, CRC Press, Boca Raton, FL.
5. Vance, D.E. (1990) Phosphatidylcholine metabolism: masochistic enzymology, metabolic regulation,
   and lipoprotein assembly. Biochem. Cell Biol. 68, 1151-1 165.
6. Cornell, R. (1 989) Cholinephosphotransferase. In: Phosphatidylcholine Metabolism (Vance, D.E.,
   Ed.) pp. 47-64, CRC Press, Boca Raton, FL.
7. Ridgway, N.D. (1989) Phosphatidylethanolaine N-methyltransferase. In: Phosphatidylcholine Me-
   tabolism (Vance, D.E., Ed.) pp. 103-120, CRC Press, Boca Raton, FL.
8. Feldman, D.A., Rounsifer, M.E., Charles, L. and Weinhold, P.A. (1990) CTP:phosphocholine cytidy-

     lyltranserase in rat lung: relationship between cytosolic and membrane forms. Biochim. Biophys. Acta
  9. Vance, D.E. (1989) Regulatory and functional aspects of phosphatidylcholine metabolism. In: Phos-
     phatidylcholine Metabolism. (Vance, D.E., Ed.) pp. 225-239, CRC Press, Boca Raton, FL.
10. Tijburg, L.B.M., Geelen, M.J.H. and Van Golde, L.M.G. (1989) Regulation of the biosynthesis of
     triacylglycerol, phosphatidykholine and phosphatidylethanolamine in the liver. Biochim. Biophys.
     Acta 1004, 1-19.
I 1. George, T.P., Morash, S.C., Cook, H.W., Byers, D.M., Palmer, F.B.St.C. and Spence, M.W. (1989)
     Biochim. Biophys. Acta 1004,283-291.
12. Spence, M.W. (1989) Sphingomyelin biosynthesis and catabolism. In: Phosphatidylcholine Metabo-
     lism (Vance, D.E., Ed.) pp. 185-203, CRC Press, Boca Raton, FL.
13. Kanfer, J.N. (1989) Phospholipase D and the base exchange enzyme. In: Phosphatidylcholine Metabo-
     lism (Vance, D.E., Ed.) pp. 65-86, CRC Press, Boca Raton, FL.
14. Hasegawa, K., Kuge, O., Nishijima, M. and Akamatsu, Y (1989) Isolation and characterization of a
     Chinese hamster ovary cell mutant with altered regulation of phosphatidylserine biosynthesis. J. Biol.
     Chem. 264, 19887-19892.
15. Carman, G.M. and Henry, S.A. (1989) Phospholipid biosynthesis in yeast. Annu. Rev. Biochem. 58,
16. Sundler, R. and Akesson, B. (1975) Regulation of phospholipid biosynthesis in isolated rat hepato-
     cytes. Effects of different substrates. J. Biol. Chem. 250, 3359-3367.
17. Tijburg, L.B.M., Houweling, M., Geelen, M.J.H. and van Golde, L.M.G. (1989) Inhibition of phos-
     phatidylethanolamine synthesis by glucagon in isolated rat hepatocytes. Biochem. J. 257,645-650.
18. Schmid, H.H.O., Schmid, P.C. and Natarajan, V. (1990) N-acylated glycerophospholipids and their
     derivatives. Prog. Lipid Res. 29, 1-43,
19. Hostetler, K.Y. (1982) Polyglycerophospholipids:phosphatidylglycerol, diphosphatidylglycerol, and
     bis (monoacylglycero) phosphate. In: Phospholipids (Hawthorne, J.N. and Ansell, G.B., Ed.) pp. 215-
     26 I, Elsevier, Amsterdam.
20. Hawthorne, J.N. (1982) Inositol phospholipids. In: Phospholipids (Hawthorne, J.N. and Ansell, G.B.,
     Ed.) pp. 263-278, Elsevier, Amsterdam.
21. Creba, J.A., Downes, C.P., Hawkins, P.T., Brewster, G., Michell, R.H. and Kirk, C.J. (1983) Rapid
     breakdown of phosphatidylinositol-4-phosphateand phosphatidylinositol4,5-bisphosphate rat he- in
     patocytes stimulated by vasopressin and other calcium-mobilizing hormones. Biochem. J. 212, 733-
22. Downes, C.P. and MacPhee, C.H. (1990) Myo-inositol metabolites as cellular signals. Eur. J. Biochem.
     193, 1-18.
23. Fisher, S.K., Van Rooijen, L.A.A. and Agranoff, B.W. (1984) Renewed interest in the polyphospho-
     inositides. Trends Biochem. Sci. 9,53-56.
24. Rhee, S.G.,Suh,P.-G., Ryu, S.-H. and Lee, S.Y. (1989) Studies of inositol phospholipid-specific phos-
     pholipase C. Science 244, 546-550.
25. Ferris, C.D., Huganir, R.L., Supattapone, S. and Snyder, S.H. (1989) Purified inositol 1,4,5-tris-
     phosphate receptor mediates calcium flux in reconstituted lipid vesicles. Nature, 342, 87-89.
26. Nishizuka, Y.(1988) The molecular heterogeneity of protein kinase C and its implications for cellular
     regulation. Nature 334, 661-665.
27. Merrill, A.H. and Stevens, V.L. (1989) Modulation of protein kinase C and diverse cell functions by
     sphingosine-a pharmacologically interesting compound linking sphingolipids and signal transduc-
     tion. Biochim. Biophys. Acta 1010, 131-139.
28. Exton, J.H. (1990) Signalling through phosphatidylcholine breakdown. J. Biol. Chem. 265, 1 4 .
29. Samborski, R.W., Ridgway, N.D. and Vance, D.E. (1990) Evidence that only newly made phosphati-
     dylethanolamine is methylated to phosphatidylcholine and that phosphatidylethanolamine is not sig-
     nificantly deacylated-reacylated in rat hepatocytes. J. Biol. Chem. 265, 18322-18329.

30. Moschidis, M.C. (1985) Phosphonolipids, Prog. Lipid Res. 23,223-246.
31. Low, M.G. (1989) The glycosyl-phosphatidylinositolanchor of membrane proteins. Biochim. Bio-
    phys. Acta 988,427454.
32. Doering, T.L., Masterson, W.J., Hart, G.W. and Englund, P.T. (1990) Biosynthesis of glycosyl phos-
    phatidylinositol membrane anchors. J. Biol. Chem. 265,621414.
D.E. Vance and J. Vance (Eds.) Biochemistry of Lipids, Lipoproleins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                  24 1

                                                                                        CHAPTER 8

Metabolism, regulation, and function of ether-
linked glycerolipids and their bioactive species
                                                                                    FRED SNYDER

 Medical Sciences Division, Oak Ridge Associated Universities, Oak Ridge, TN 37831, U.S.A .

1. Introduction

Recent advances in the field of ether-linked g1ycerolipids can largely be attributed to
the renewed interest in this novel subclass of lipids that developed following the 1979
discovery of the chemical structure of platelet activating factor (PAF), the most po-
tent biologically active cellular mediator known. Several complete books [l-51 and a
variety of recent reviews [6-101 on ether lipids/PAF, including two up-to-date mini-
reviews on PAF [ 1 1,121 are recommended reading material for those who would like
a more detailed account of this subject area. These articles are also sources of a com-
prehensive listing of the published papers used to prepare this chapter.

2. Nomenclature

Ether linkages associated with glycerolipids occur in nature primarily as 0-alkyl and
0-alk- 1-enyl moieties; however, there have been some reports of glycerolipids con-
taining thio-ethers and dialkyl moieties as minor components of certain mammalian
cells. Lipids with 0-alkyl groups are often referred to as ‘glycerol ethers’ and those
with 0-alk-1-enyl groups as ‘plasmalogens’ or ‘vinyl ethers.’ The double bond in the
alk-1-enyl linkage of native plasmalogens has a cis configuration. In most cells, ether-
linked aliphatic moieties have a limited range in carbon chain length and number of
double bonds; the major groupings normally encountered are 16:0, 18:0, and 18:l
carbon chains at the sn-1 position. In view of the cumbersome length of the chemical
terms for the U-alkyl and 0-alk- 1-enyl complex glycerolipids (e.g., 1-0-alk-1-enyl-2-
acyl-sn-glycero-3-phosphoethanolamine), terms plasmanyl and plasmenyl are re-
commended by IUB-IUPAC for the l-alkyl-2-acyl-sn-glycero-3-phospho- l-alk-  and
l-enyl-2-acyl-sn-glycero-3-phospho-    radicals, respectively. These names are the
equivalents of ‘phosphatidyl’, which can be only used to designate the diacylglyce-

        H2COCH2CH2R                                H2COCH2CH2R

            l o
            I    II      +
                                                     l o
                                                     I    It     t
I           0-                       II
      1.alkyl-Z-acyl-sn-glycero-3-        1-alkyl-Z-aceh/l-8_glycero-3-
             phosphocholine                         phosphocholine

         H,COCH=CHR                                H,COCH=CHR

            l o
            I    00
                                                     l o
                                                     I    II
         H2COPOCHJ2&NH2                            H2COPOCHF:YNH2
III          0-                      m                    0-
1-alk-l'-enyl-Z-acyl-sn-glycero- 1-alk-l'-enyl-Z-acetyl-~-glycero-
    3-phosphoethanolamine                    3-phosphoethanolamine


                                               ?   ,f""

                                             c%cOr HFOH

Fig. 1 . Chemical structures of important types of ether-linked lipids found in mammals. Common names
for these compounds are I, plasmanylcholine;11, PAF; 111, ethanolamine plasmalogen; IV, ethanolamine
analog of PAF; V, glyceryl ether diester; VI,acetylated ether lipid analog of diglyceride.

rophospho- radical in glycerophospholipidssuch as phosphatidylcholine and phos-
phatidylethanolamine. Thus, plasmanic acid and plasmenic acid represent the alkyl
and alk- 1-enyl analogues of phosphatidic acid, respectively. Similarly, plasmanyl-
choline designates l-alkyl-2-acyl-sn-glycero-3-phosphocholine plasmenyletha-
nolamine denotes 1-alk-1-enyl-2-acyl-sn-glycero-3-phosphoethanolamine, etha-  (i.e.,
nolamine plasmalogen). In mammals the alkyl and alk-1-enyl chains are located at
the sn-1 position of glycerolipids (L-stereoisomer configuration based on Baer-
Fischer nomenclature). There have been several reports of phospholipids in bovine
heart and spermatozoa that have 0-alkyl groups at both the sn-1 and sn-2 positions.
Halophilic bacteria contain very large amounts of dialkylglycerolipids but these have
the opposite stereochemical configuration (sn-2,3 or D-series) of other native ether
lipids. The chemical structures of several most widely encountered types of ether
lipids in the neutral lipid and phospholipid fractions of mammalian cells are il-
lustrated in Fig. l.
   An interesting class of synthetic phospholipids that have structures similar to PAF
has received considerable attention in recent years because of their potent and selec-
tive cytotoxic properties towards certain cancer cells. Fig. 2 illustrates the structure



            (PAF antitumor analog)

Fig. 2. The methoxy analog of PAF, an unnatural phospholipid with potent and highly selective antitumor

of the methoxy analog of PAF which has been investigated in a variety of cell culture
systems, intact animals, and in clinical trials. Many other analogs, including the sn-2
acetamide analogue of PAF and thio-ethers of similar chemical structures, have also
been studied as potential therapeutic drugs in the treatment of cancer. The novel
feature of these ether-linked lipids is that they are plasma membrane targeted instead
of locating in the nucleus as other conventional chemotherapeutic agents. Their abil-
ity to inhibit protein kinase C and the transport of nutrient molecules across cellular
membranes is well established. However, their biochemical mechanism of action is
still a mystery.

3. Historical events
The presence of alkyl ether lipids in liver oils of various saltwater fish was originally
described in 1920 by the Japanese scientists Tsujimoto and Toyama. The common
names ‘chimyl’ [16:0 alkyl], ‘batyl’ [18:0 alkyl], and ‘selachyl’ [18:1 alk-9-enyl] alco-
hols for the alkylglycerols are based on the fish species from which they were origi-
nally isolated. Complete proof of the chemical nature of the alkyl linkage at the sn-1
position in these glycerolipids did not become apparent until 13 years later as report-
ed by Davies, Heilbron, and Jones from England.
   In 1924, plasmalogens were detected, quite by accident, in Germany by Feulgen
and Voit. After first preserving a variety of fresh tissue slices, including aortas and
kidneys of rats, horse muscle, and protozoa in a HgCl, solution, the specimens were
erroneously treated the following day with a fuchsin-sulfurous acid solution without
the normal fixation and related histological processing with organic solvents. Under
these conditions the cytoplasm of cells, but not the nuclei, was stained a red-violet
color. This phenomenon led Feulgen to conclude that an aldehyde was present in the
cell plasma, and he called this substance ‘plasmal.’ If the histological preparations
were treated with a lipid-extracting solvent before exposure to the dye, no colored
stain appeared in the cytoplasm. The unknown precursor of the cytosolic aldehyde
that reacted with the dye was called plasmalogen, a name that continues to be re-

tained as the generic term for alk- l-enyl-containing glycerolipid classes. Despite the
efforts of many different researchers to identify the chemical structure of the plasma-
logens discovered by Feulgen and Voit, it was not until the late 1950s that the precise
structural features of the alk- 1-enyl linkage in ethanolamine plasmalogens was prov-
en primarily through the combined efforts of Rapport and Marinetti in the United
States, Gray in England, Klenk and Debuch in Germany, and their various co-
   During the late 1960s and early 1970s the development of new analytical and
chromatographic methods for the characterization of both the alkyl-and alk-l-
enyl-glycerols made it much easier for scientists to investigate the ether-linked lipids.
The technological advances made during this period provided new approaches and
more sensitive and specific methods for studies of ether lipids that have led to our
current knowledge about their widespread occurrence, metabolism, and possible
   Between 1969 and 1972, the first cell-free enzyme systems that synthesized the alkyl
and alk-1-enyl linkages were discovered along with other enzymatic steps associated
with these pathways. As with the initial discovery of ether-linked lipids, elucidation
of the biosynthesis of the alkyl ethers by groups directed by Hajra, Friedberg, and
Snyder (U.S.A.) preceded that of the alk-l-enyl ethers. In fact, the close metabolic
interrelationship between the alkyl and alk-l-enyl types of lipids did not become ap-
parent until 1972, when it was documented by Wykle, Blank, and Snyder (U.S.A.)
and Paltauf and Holasek (Austria) with cell-free systems that the alkyl moiety of
phospholipids was the direct precursor of the alk-l-enyl grouping. Perhaps the most
exciting development in the ether-lipid field has been the more recent discovery of
certain acetylated forms of alkylglycerolipids known as PAF (Fig. 1) that possess
potent biological activities.
   As with other progress in the ether lipid field, information about the possible func-
tions of plasmalogens continues to lag far behind that of the alkyl lipids. In fact, no
specific cellular role has yet been identified for plasmalogens. However, the relatively
high proportion of plasmalogens in the ethanolamine phospholipid fraction of ner-
vous tissue and certain other cells suggests that these unique lipids must have an
essential function in at least some biological systems. Their high content of arachi-
donic acid indicates the ethanolamine plasmalogens could be an important source of
eicosanoid mediators. Also, it has been proposed that plasmalogens can serve as
scavengers for oxygen-generated free radicals. The possible cellular importance of a
deficiency in ethanolamine plasmalogens has been emphasized most recently in stud-
ies of a genetic disorder in infants known as the Zellweger syndrome. This disease is
a lethal, recessive autosomal inborn error of metabolism where cells lack peroxi-
somes. Although the disease symptoms might not be directly linked to the lack of
plasmalogens, the enzyme activities responsible for the synthesis of the alkyldihy-
droxyacetone-P are generally located in the peroxisomes of many cells and therefore,
the lack of peroxisomes in Zellweger patients would explain the plasmalogen deficit.

Unfortunately, the role of plasmalogens and related ether lipids during early em-
bryonic and postnatal development is not yet understood.

4. Analytical approaches

Adsorption chromatography, either thin-layer or column procedures, can resolve the
neutral lipid fraction into the various types of acyl- and ether-linked lipid classes of
analogues; the order of migration or elution of a particular series is alk-l-
enybalkybacyl. In contrast, phospholipid subclasses of ether and ester lipids are not
easily separated by adsorption chromatography, although it is sometimes possible to
isolate certain individual molecular species by reverse phase high-performance liquid
chromatography. However, generally it is necessary to separate and quantitate the
diacyl, alkylacyl, and alk- 1-enylacyl fractions of a particular phospholipid class after
the phosphobase moiety is hydrolyzed by phospholipase C and various derivatives of
the diacyl-, alkylacyl-, or alk- 1-enylacyl-glycerols are prepared for chromatographic
analysis. These diradylglycerol derivatives can then be easily resolved into their sub-
classes and specific molecular species of each subclass by normal adsorption and re-
verse phase chromatography, respectively. Benzoates or other chomophoric deriva-
tives (nitrobenzoates, dinitrobenzoates, and dinitrophenylurethanes) offer an advan-
tage over acetates for analysis because they can be quantitated on the basis of their
UV-absorbing properties.
   Some of the most useful chemical reactions for the identification and analysis of
ether-linked lipids are summarized in Table I. Ether linkages in glycerolipids are un-
affected by chemical reduction with Vitride or LiAIH,, and therefore, these reducing
agents are excellent for removing esterified groupings (e.g., acyl or phosphobase)
from lipids without loss of the ether-linked alphatic moiety attached to the glycerol.
The alkylglycerols and alk- 1-enylglycerols produced by chemical reduction are im-
portant derivatives in the characterization of the ether-linked aliphatic moieties and
are a starting point for subsequent preparation of various substituted derivatives for
chromatographic and mass spectral analysis. Either phospholipase A, or mild alka-
line hydrolysis is usually used to remove the sn-2 acyl moiety of ether-linked phos-
pholipids. Monomethylamine is a useful reagent for hydrolyzing acyl moieties at the
sn-2 position of ether-linked phospholipids, since this reagent is much easier to
remove from the reaction mixture than other reagents; the products are the same as
those obtained with other types of mild alkaline hydrolysis (Table I), except that
N-methyl fatty acid amides are formed instead of fatty acids.
   Complete hydrolysis of all esterified fatty acids in ether-linked neutral lipids (e.g.,
alkyldiacylglycerols, Fig. 1) can be achieved by either mild alkaline or monomethyla-
mine hydrolysis. If selective hydrolysis of only the sn-3 acyl groups is desired, either
pancreatic or Rhizopus lipases can be employed. Pancreatic or Rhizopus lipases (Ta-
ble I) are also useful enzymes for removing diacyl phospholipids as contaminants of
The analysis of ether-linked commonly found i mammalian cells: chemical reactions and lipid products

                                                   PRIMARY LIPID PRODUCTS FORMED BY CHEMICAL OR ENZYMATIC REACTIONS
        Starting              NaAlH2(OCYCH20CH,)2]             Mild alkaline                                                                          Pancreatic o r
          lipid                             or                  hydrolysis                  PhoSphOl ipase A2              Phospholipose C           Rhizopus I I pase
        sample                  L i A I H 4 reduction

       H2C phosphobase
                                      I + RCH OH
                                                               HO H
                                                                       + RCOOH
                                                                HzC phosphobase
                                                                                               I + RCOOH
                                                                                             H2C phosphobase

                                                                                                                                                         no reaction

 1-olkyl-2-acyl-sn-glycero-    1- a1kyl-sn - gIycerol     -alkyl-2-lyso-sn-glyerol-3     alkyl-2-l yso-sn-glyceml-3.    -alkyl-2-acyl-sn-glycerol
     3-phosphobase                + f a t t y alcohol     ihasphobase + f a t t y acid   lhosphobase + t a t t y acid

       H2COCH=CHR                   H$XCH=CHR                   H2COCH=CHR                    HgOCH=CHR                         Y C O C H = CH R

     RCOCH                          HOCH
                                        I    + RCHflH          HO€H
                                                                   I   + RCOOH               HOCH
                                                                                                 I   + RCOOH                  RCOCH                       no reaction
       H2C phosphobose              HgOH
                                        I                         I
                                                                H2C-phosphobase                K
                                                                                              H phosphobase

 -alk-1-enyl-Z-ocyf-sn-glyaro-alk-l-enyl-sn-glyceml                                 ulk-lenyC2-lyso-sn-glyero .afk-l-enyl-2-acyl-sn-glycerol
     3-phosphobase                + fatty alcohol         I-phosphobase + fatty ocid     -phosphobase + fatty acid


                                             + RCH2OH

                                                                   I   + RCOOH                 no reaction                   no reaction
       HSOCR                         H2COH
 ' olkyC2.3-diocyl-sn-glycem 1-alkyl-sn-glycerol            1-alkyl-sn-gl ycerol                                                                    -alkyl-2-acyl-sn-glycerol
                                   + fatty alcohol               + f a t t y acid                                                                         + f a t t y acid

ether-linked lipid classes since the sn-2-lyso phospholipids produced possess com-
pletely different chromatographic properties from the alkylacyl-and alk- 1-enylacyl
types. The lipases from both sources exhibit positional specificity for sn-1 and sn-3
acyl groups in glycerolipids.
   Ether-linked phospholipids also serve as substrates in phospholipase C- or D-cata-
lyzed reactions. The glycerolipid products formed by phospholipase C treatment of
ether phospholipids can then be derivatized for subsequent analysis. Benzoate and
acetate derivatives are especially important in the analysis of alkylglycerols, alk-1-
enylglycerols, alkylacylglycerols, and alk-1-enylacylglycerols by high-performance
liquid and gas-liquid chromatography. Acetolysis (with acetic acid and acetic anhy-
dride) of any ether lipid replaces all ester groupings (acyl and phosphate) with ace-
tate. Isopropylidenes or trimethylsilyl ethers of alkylglycerolsare also valuable deriv-
atives for gas-liquid chromatographic analysis, especially when it is combined with
mass spectrometry. Alk- 1-enylglycerols can be analyzed in the same manner after
they have been converted to alkylglycerols by catalytic hydrogenation.
   Acid hydrolysis of the alk- 1-enyl linkage produces the corresponding fatty alde-
hyde. The free aldehydes can be measured directly by gas-liquid chromatography if
they are analyzed immediately after being generated. If methanol is present during
acid hydrolysis of plasmalogens, a more stable derivative of the aldehyde, the dime-
thylacetal, is produced.

5. Physical properties

Replacement of ester linkages in glycerolipids with ether bonds mainly affect hydro-
phobic-hydrophilic interactions. Nevertheless, the closer linear packing arrangement
attainable with ether-linked moieties also are capable of influencing the polar head
group regions of phospholipids. The novel placement of the d l double bond in plas-
malogens can also exert effects on stereochemical relationships and therefore, the
presence of an ether-linked phospholipid can modify the configuration of membranes
in a variety of ways.
   In model membranes, ether-linked lipids have been shown to decrease ion permea-
bility, surface potential, and lower the phase temperature of membrane bilayers when
compared to their diacyl counterparts. However, one should be cautious in making
too many generalizations in this area, since it is known that the di-0-alkyl analog of
phosphatidylcholine has a higher phase transition temperature than the correspond-
ing alkylacyl analog (i.e., plasmanylcholine).
   In studies of membranes from Closrridium buryricum, the cells appear to be able to
regulate the stability of the bilayer arrangement by altering the ratio of ether vs acyl
type of ethanolamine phospholipids in response to changes in the lipid unsaturation
of the membranes. The experiments with bacteria indicate the substitution of plasme-
nylethanolamine for phosphatidylethanolamine in biomembranes would have only

small effects on lipid melting transitions, whereas the tendency to form non-lamellar
lipid structures would be significantly increased.

6. Occurrence in nature

Ether-linked lipids occur throughout the animal kingdom and are even found as mi-
nor components in higher plants. Some mammalian tissues, as well as tissues from
avian, marine, molluscan, protozoan, bacterial, and other species can contain signifi-
cant proportions of ether-linked lipids. The highest levels of ether lipids in mammals
occur in nervous tissue, heart muscle, testes, kidney, preputial glands, tumor cells,
erythrocytes, bone marrow, spleen, skeletal tissue, neutrophils, macrophages, plate-
lets, and lipoproteins. The large quantities of ethanolamine plasmalogens associated
with various lipoproteins from rat serum and human plasma (36% and 50% respec-
tively, of the total ethanolamine phosphatides) is of particular interest since the liver
contains relatively low amounts of ether lipids. Although the dietary consumption of
ether lipids by humans has largely been ignored by nutritionists, it is clear that certain
meats and seafoods can contain relatively high amounts of these lipids.
   Some membranes, such as the myelin sheath, are highly enriched in ethanolamine
plasmalogens (e.g., up to 8070 of the lipid phosphorus). In general, the alkyl group-
ings are primarily associated with the choline glycerophospholipids, and the alk-l-
enyl moieties are almost exclusively found in the ethanolamine-containing phospho-
lipids, except in heart tissue of certain species, where choline plasmalogens can be
   Analogs of triglycerides have also been described. 1-Alkyl-2,3-diacyl-sn-glycerols
(Fig. 1) are characteristically elevated in tumor lipids and 1-alk-1-enyl-2,3-diacyl-sn-
glycerols (neutral plasmalogens) have also been detected in tumors and adipose tissue
of mammals and in fish liver oil.
   1-Alkyl-2-acyl-sn-glycero-3-phosphocholine 1) is a significant component of
platelets, neutrophils, and macrophages and is a precursor of PAF (I-alkyl-Zacetyl-
sn-glycero-3-phosphocholine;     Fig. 1). Thus, this lipid class appears to be a constit-
uent of all cells known to produce PAF by the remodeling pathway ( e g , eosinophils,
basophils, monocytes, and endothelial, mast, and HL-60 cells). PAF is also found in
saliva, urine, and amniotic fluid, which indicates other cell types might be the source
of PAF in these fluids.
   Dialkylglycerophosphocholines have been reported as minor constituents of bo-
vine heart and spermatozoa. Moreover, heart tissue is unique with respect to its plas-
malogen content, since in some animal species, this is the only mammalian tissue
known to contain significant amounts of choline plasmalogens instead of the usually
encountered ethanolamine plasmalogens.
   As mentioned in the section on Nomenclature, halophilic bacteria contain an unu-
sual dialkyl type of glycerolipid (a diphytanyl ether analogue of phosphatidylglyce-

rophosphate) that has an opposite D-type stereochemical configuration from all oth-
er known ether-linked lipids, i.e., the ether linkages are located at the sn-2 and sn-3
positions. The biosynthetic pathway for the formation of the ether bond in halophiles
is still unknown.
   Moreover, despite the significant amounts of ethanolamine plasmalogens in anae-
robic bacteria such as Clostridium butyricum, no information is yet available about
how the alk-l-enyl ether bond is biosynthesized in this organism. This is probably
one of the few examples where it is has not been possible to use bacteria rich in a
specific lipid as a model for investigating the enzymes responsible for its production.

7. Biologically active ether lipids

In 1979, a potent biologically active phospholipid was discovered that could aggre-
gate platelets at a concentration of lo-'' M and induce an antihypertensive response
when as little as 60 ng were administered intravenously to hypertensive rats. The
chemical formula of the semisynthetic lipid tested in these experiments was l-alkyl-2-
acetyl-sn-glycero-3-phosphocholine    (Fig. 1, structure 11) or now commonly referred
to as PAF. The chemical structure of PAF has been very well characterized in rabbit
basophils and it is known to be produced by a large variety of cells. Studies of enzyme
activities involved in PAF metabolism have demonstrated that many blood cells, var-
ious healthy tissues, and tumor cells can synthesize and degrade PAF. The ubiquitous
distribution of these enzymes and the diverse biological properties of PAF indicate
that this bioactive phospholipid in mammals serves as a multifunctional cellular me-
diator. PAF exerts many different types of biological responses and has been impli-
cated as a contributing factor in the pathogenesis of such diverse disease processes as
asthma, hypertension, allergies, inflammation, and anaphylaxis, to name only a few.
Specific PAF receptors have been documented on the surface of a number of cells and
based on studies with PAF receptor antagonists it is thought most of the biological
responses are receptor-dependent (see Functional Roles, Section 1 1). However, there
is also some reason to believe that PAF might behave as an intracellular mediator.
   Other acetylated glycerolipids with biological activities that have also been de-
scribed include 1-alkyl-2-acetyl-sn-glycerols and the plasmalogen analogues of both

A c y l - COP.                        Fatty                                Fatty
                                   a Ide h y d e                       alcohol

Fig. 3. Biosynthesis of long-chain fatty alcohols by acyl-CoA reductase.
choline- and ethanolamine-containing PAF. All have considerably lower biological
activities than PAF itself but both the alkylacetylglycerols and choline plasmalogen
analog of PAF can mimic the actions of PAF. In fact, the blockage of platelet
aggregation by the choline plasmalogen analogue of PAF by PAF receptor antag-
onists indicates it can occupy the same receptor as PAF, The ethanolamine plasma-
logen analog of PAF has been reported to be able to act synergisticallywith PAF, but
little is known about its significance.

8. Biosynthesis

8.1. Ether lipid precursors

8.1.1. Acyl-CoA reductase
Fatty alcohol precursors in ether lipid biosynthesis are derived from acyl-CoAs in a
reaction sequence catalyzed by a membrane-associated acyl-CoA reductase (Fig. 3).
A cytosolic form of the reductase from bovine heart has also been described. Other
soluble aldehyde reductases are known to exist, but their lack of substrate specificity
suggests they do not play a significant physiological role in the formation of fatty
   The acyl-CoA reductases associated with membrane systems use only acyl-CoA
substrates, and in mammalian cells, they exhibit a specific requirement for NADPH.
Although only traces of fatty aldehydes can normally be detected in these reactions,
the use of trapping agents such as semicarbazide has documented that aldehydes are
indeed formed as intermediates. Acyl-CoA reductase prefers saturated substrates
over acyl-CoAs that are unsaturated; in fact, the enzyme in brain microsomes is not
able to convert polyunsaturated moieties to fatty alcohols. Acyl-CoA reductase has
been investigated in various tumors, mouse preputial glands, brains and hearts of
rats, Euglena gracilis, and certain bacteria. Some evidence indicates that, at least in
brain, acyl-CoA reductase might be localized in microperoxisomes instead of the mi-
crosomal fraction. Topographical studies of both microsomal vesicles and microper-
oxisomal particles have revealed the active center of acyl-CoA reductase is located at
the cytosolic surface of these membranes.
   Formation of fatty alcohols has also been observed in rabbit harderian glands and
E. gracilis, where neither acyl-CoA nor the free acid is a substrate for the reductase.
In these systems, the NADPH-dependent reductase appears to be closely coupled
with fatty acid synthase. It has been suggested that the fatty acid bound to acyl car-
rier protein, rather than acyl-CoA, is the substrate for this type of acyl reductase.

8.1.2. Dihydroxyacetone-P acyltransferase
Presumably, dihydroxyacetone-Pacyltransferase (see Chapter 6) is present in all cells
that synthesize alkylglycerolipids, since the acylation of dihydroxyacetone-P is an
                                                                                                   25 1

obligatory step in the biosynthesis of the ether bond in glycerolipids. On the other
hand, the quantitative importance of the dihydroxyacetone-P versus sn-glycerol-3-P
routes in the biosynthesis of glycerolipid esters is not yet firmly established in most
  Current evidence suggests that the dihydroxyacetone-P acyltransferase and related
enzymes of the ether lipid pathway are localized in microperoxisomes. Therefore, the
original discovery and subsequent finding of these enzymes in microsomes and/or
mitochondria could be explained by the fact that microperoxisomes sediment with

           A1 k y l                                         Alkyl
           0      +    NADPH   + Ht              ~

AcYl   [   Alkyl



           Alkyl                                            Alkyl

               + CDP-Ethanolamine
                      (or CDP-Choline)

           OH                                               PEthonolamine
                                                             (P- Choline)

ACYI   {   Alkyl

                 t    O2 t Cyt. b5
                                      -  X
                                                            A l k - 1 -enyl

                                                            PEt hanolamine

Fig. 4. Biosynthesis of membrane phospholipids from alkyldihydroxyacetone-P, the first detectable in-
termediate formed in the biosynthetic pathway for ether-linked glycerolipids. Enzymes responsible for
catalyzing the reactions shown in this figure are: I - alkyldihydroxyacetone-Psynthase; 11, NADPH: alkyl-
dihydroxyacetone-P oxidoreductase; 111, acyl-CoA: 1 -alkyl-2-lyso-sn-glycero-3-P    acyltransferase; IV, 1-
alkyl-2-acyl-sn-glycero-3-P                                                   CDP
                            phosphohydrolase;V, 1-alkyl-2-acyl-sn-glycerol: ethanolamine (or CDP
choline) ethanolamine (choline) phosphotransferase; and VI, 1-alkyl-2-acyl-sn-glycero-3-phosphoethanol-
amine A'-desaturase.

microsomes, and large peroxisomes sediment with mitochondria under the usual
preparation conditions of these subcellular fractions. Nevertheless, Ehrlich ascites
cells which are essentially devoid of peroxisomes are a rich source of ether-linked
lipids and the enzyme alkyldihydroxyacetone-P synthase (Fig. 4, reaction I) that syn-
thesizes the alkyl ether bond. Thus, it is clear that much more work is required in this
area before the quantitative significance and role of the different organelle systems in
the biosynthesis of ether-linked glycerolipids among different cell types is fully under-
   Investigations of the topographical location of dihydroxyacetone-P acyltransfer-
ase in membrane preparations from rabbit harderian glands and rat brains indicate
that unlike most other enzymes in glycerolipid metabolism, dihydroxyacetone-P
acyltransferase appears to be located on the internal side of microsomal vesicles, but
their topographical distribution in membranes of other cells has not been investigat-
ed. Moreover, two forms of dihydroxyacetone-P acyltransferases apparently exist in
mammals, a peroxisomal and a microsomal type. The peroxisomal activity is specific
for dihydroxyacetone-P and does not utilize sn-glycerol-3-P as a substrate, whereas
the microsomal dihydroxyacetone-P acyltransferase from fat cells, liver, and several
other rat tissues can acylate either dihydroxyacetone-P or sn-glycerol-3-P. Further-
more, the properties of the enzyme in peroxisomes differ from the microsomal form
in that the latter is inhibited by N-ethylmaleimide. Substrate competition experiments
also support these differences. Unfortunately the reason for these differences is not

8.2. Ether lipids in membranes

8.2.1. Biosynthesis o the ether bond
The initial formation of the alkyl ether bond in glycerolipids is catalyzed by alkyldi-
hydroxyacetone-P synthase. This reaction (Fig. 4, reaction I) is unique among bio-
chemical reactions, since it is the only one known where a fatty alcohol is substituted
for an acyl moiety. Alkyldihydroxyacetone-P synthase has been primarily investigat-
ed in microsomal preparations; however, as with the dihydroxyacetone-P acyltrans-
ferase, recent evidence indicates that the synthase activity is also located in the per-
oxisomes, at least in some cells.
   Alkyldihydroxyacetone-P synthase has been solubilized from Ehrlich ascites cell
microsomes and subsequently purified approximately 1000-fold. Kinetic experiments
with the partially purified enzyme have demonstrated that the reaction mechanism of
alkyldihydroxyacetone-P synthase is of a ping-pong type (Fig. 5), with an activated
enzyme-dihydroxyacetone-P intermediate playing a central role. The existence of this
intermediate explains the reversibility of the reaction, since the enzyme-dihydroxy-
acetone-P complex can react with either fatty alcohols (forward reaction) or fatty
acids (back reaction). This mechanism is also consistent with the overall characteris-
tics of alkyldihydroxyacetone-P synthase listed below.

                           RC02H                                              HkCOR'

                                                                                IC=O       +   E   +   X

   I                                                                            I

                                     1    "E-DHAP"   i                       Al kyl-DHAP

Fig. 5. The molecular reaction catalyzed by alkyldihydroxyacetone-P (alkyl-DHAP) synthase in the forma-
tion of the alkyl ether bond is thought to proceed via a ping-pong mechanism. The abbreviation DHAP
designates dihydroxyacetone-P. Upon binding of acyl-DHAP to alkyl-DHAP synthase (E), the pro-R hy-
drogen at C-I is exchanged by an enolization of the ketone, followed by release of the acyl moiety to form
an activated E-DHAP complex. C-1 is thought to carry a positive charge that may be stabilized by an
essential sulfhydryl group of the enzyme; the incoming alkoxide ion reacts at C-1 to form alkyl-DHAP. X
is a nucleophilic cofactor, possibly an amino acid functional group at the active site that covalently binds
the DHAP portion of the substrate.

   The following features are characteristic of the reaction that produces alkyldihy-
droxyacetone-P, the first detectable intermediate in the formation of complex ether-
linked glycerolipids:
  (1) The pro-R hydrogen at C-1 of the dihydroxyacetone-P moiety of acyldihy-
      droxyacetone-P exchanges with water.
  (2) The configuration of the C-1 carbon of the dihydroxyacetone-P moiety of acyl-
      dihydroxyacetone-P is preserved.
  (3) The acyl group of acyldihydroxyacetone-Pis cleaved before addition of the fatty
  (4) Either fatty acids or fatty alcohols can bind to the activated enzyme-dihy-
      droxyacetone-P intermediate to produce acyldihydroxyacetone-P or alkyldihy-
      droxyacetone-P, respectively.
  (5) A Schiff s base intermediate is not formed.
  (6) The oxygen in the ether linkage of alkyldihydroxyacetone-P is donated by the
      fatty alcohol.
  (7) Both oxygens in the acyl linkage of acyldihydroxyacetone-P are found in the
      fatty acid product.
  (8) Acyldihydroxyacetone-P acylhydrolase activity is not associated with purified
      alkyldihydroxyacetone-P synthase.
  (9) Alkyldihydroxyacetone-P synthase activity is sensitive to modifiers of sulfhy-
      dry1 and amino functional groups.
(10) A nucleophilic cofactor (possibly an amino acid functional group at the active
      site) covalently binds the dihydroxyacetone-P portion of the substrate acyldi-
      h ydrox y acetone-P.
(1 1) A ketone function is an essential feature of the substrate, acyldihydroxyacetone-

   Alkyldihydroxyacetone-P synthase exhibits a very broad specificity for fatty al-
cohols of different alkyl chain lengths. On the other hand, the specificity of this en-
zyme for acyldihydroxyacetone-P with different acyl chains is less well understood,
primarily because of the unavailability of acyldihydroxyacetone-P possessing acyl
chains with different chains lengths and degrees of unsaturation.
   Topographical studies of alkyldihydroxyacetone-P synthase in membranes from
rabbit harderian glands revealed that this enzyme activity is located on the lumenal
side of microsomal vesicles. However, since this gland functions primarily as a se-
cretory organ, it would not be appropriate to formulate a general conclusion about
the membrane sidedness of alkyldihydroxyacetone-P synthase in either microsomes
or peroxisomes from other cells.

8.2.2. Biosynthesis o the alkyl analogue o phosphatidic acid
                     f                     f
Once alkyldihydroxyacetone-Pis synthesized, it can then be readily converted to the
alkyl analogue of phosphatidic acid (Fig. 4,reactions I1 and 111) in a two-step reac-
tion sequence involving NADPH:alkyldihydroxyacetone-P oxidoreductase and acyl-
CoA:1-alkyl-2-lyso-sn-glycerol-3-P acyltransferase. The NADPH-dependent oxido-
reductase is capable of reducing both the alkyl and acyl analogues of dihydroxyace-
tone-P. Dietary ether lipids can also enter this pathway, since alkylglycerols resulting
from the catabolism of ether-linked lipids during absorption are known to be phos-
phorylated by an ATP:alkylglycerol phosphotransferase to form 1 -alkyl-2-lyso-sn-
glycerol-3-P (Fig. 6).

8.2.3. Biosynthesis o complex neutral glycerolipids and phospholipids with 0-alkyl
The alkyl analogue of phosphatidic acid plays a central role in the formation of lipids
that contain ether bonds. Reaction steps beginning with l-alkyl-2-acyl-sn-glycerol-3-
P in the routes leading to the more complex ether lipids (Fig. 4,reactions IV and V)
are thought to be catalyzed by the same enzymes involved in the pathways estab-
lished by Kennedy and co-workers in the late 1950s for the diacylglycerolipids
(Chapter 7).

  H,COR                        H,COR
  HOCH i A T P      -             I
                               HOCH      +ADP

Alkylglycerol                Alkylglycero-P

Fig. 6.How dietary products of ether lipids can enter the intermediary metabolismof ether-linked glycero-
lipids. The reaction shown is catalyzed by ATP:alkylglycerol phosphotransferase.

   1-Alkyl-2-acyl-sn-glycerols,derived from the alkyl analogue of phosphatidic acid
by the action of a phosphohydrolase (Fig. 4, reaction IV), can serve as substrates for
cholinephosphotransferase, ethanolaminephosphotransferase, or acyl-CoA acyl-
transferase in the formation of the alkyl analogues of phosphatidylcholine, phospha-
tidylethanolamine, or triacylglycerols, respectively. Thus, the alkylacylglycerols par-
ticipate at a very crucial branch point in the ether lipid pathway in a manner identical
to their acyl counterparts, the diacylglycerols.
   The two major classes of ether-linked phospholipids formed via the choline- and
ethanolamine-phosphotransferases serve as precursors for two other important
classes of ether lipids. One of the products, l-alkyl-2-acyl-sn-glycero-3-phosphocho-
line, is the membrane source of the ether lipid precursor of the potent biologically
active phospholipid, PAF. The other phospholipid product, 1-alkyl-2-acyl-sn-gly-
cero-3- phosphoethanolamine (an alkyl analogue of phosphatidylethanolamine),
serves as the direct precursor of ethanolamine plasmalogens in an unusual desatura-
tion reaction where this intact alkylacyl phospholipid is the substrate for a A'-alkyl
desaturase (Fig. 4, reaction VI), a microsomal mixed-function oxidase that is de-
scribed in the next section.

8.2.4. Biosynthesis o plasmalogens
The desaturase system responsible for the biosynthesis of ethanolamine plasmalo-
gens from alkyl lipids (Fig. 4, reaction VI) was characterized in the early seventies by
Snyder, Wykle and Blank (U.S.A.) and Paltauf and Holasek (Austria). The reverse
of this reaction (that is, conversion of an alk-1-enyl moiety to an alkyl grouping) has
never been demonstrated. The A'-alkyl desaturase, which produces the alk- 1-enyl
grouping, is a unique enzyme, since it can specifically and stereospecifically abstract
hydrogen atoms from C-1 and C-2 of the 0-alkyl chain of an intact phospholipid
molecule, l-alkyl-2-acyl-sn-glycero-3-phosphoethanolamine. the intact etha-
nolamine phospholipid is known to serve as a substrate for the A'-alkyl desaturase.
  A'-Alkyl desaturase, like the acyl-CoA desaturases (see Chapter 5), exhibits the
typical requirements of a microsomal mixed-function oxidase: molecular oxygen, a
reduced pyridine nucleotide, cytochrome b, reductase, and a terminal desaturase
protein that is sensitive to cyanide. The precise reaction mechanism responsible for
the biosynthesis of the ethanolamine plasmalogens is unknown. However it is clear
from an investigation with a tritiated fatty alcohol, that only the IS and 2s (erythro)
labeled hydrogens are lost during the formation of the cis alk-1-enyl moiety of etha-
nolamine plasmalogens. The A'-alkyl desaturase does not utilize 1-alkyl-2-acyl-sn-
glycero-3- phosphocholine as a substrate. In fact, the biosynthesis of significant
quantities of choline plasmalogens that occurs in some heart tissue remains an en-
igma, although recent evidence strongly implicates they are derived from the ethano-
lamine plasmalogens. It has been speculated that the ethanolamine plasmalogens
might serve as precursors of choline plasmalogens by reactions involving a CDP-
choline reverse substitution of P-choline for P-ethanolamine, base exchange,

methylation of ethanolamine plasmalogens, or a coupled reaction sequence catalyzed
by phospholipase C and CDP-choline cholinephosphotransferase.

8.3. Bioactive ether lipids

PAF is known to be synthesized by two separate sequences of reactions, the remodel-
ing and de novo pathways (Fig. 7). A preformed subclass of membrane phospholip-
ids, alkylacylglycerophosphocholines,is the stored precursor in the remodeling se-
quence, whereas a branchpoint intermediate in the general ether lipid pathway, 1-
alkyl-2-lyso-sn-glycero-3-P, is the direct precursor of PAF formed de novo (Figs. 4
and 7).
   The remodeling pathway consists of a closely integrated two step reaction cata-
lyzed by a transacylase/phospholipase A, activity to produce lyso-PAF from the al-
kylacylglycerophosphocholine pool (Fig. 8) and an acetyltransferase activity that
utilizes acetyl-CoA in the acetylation of the lyso-PAF to form PAF. Recent evidence
has indicated the transacylase/phospholipase A, is highly specific for alkylacylgly-
cerophosphocholines possessing an arachidonoyl moiety at the sn-2 position. PAF
synthesis by the remodeling route is stimulated by various inflammatory agents and
bioactive arachidonic acid metabolites can also be generated. The fact that both eico-
sanoid and PAF mediators can be formed via the remodeling pathway stresses the

    De Novo PAF       De Novo Synthesis of    Remodeling PAF
     Synthesls       Membrane PAF Precursor     Synthesis
         OR                                              OR
HOIp                                                bCho
         I       1
         OR                                              P
                                               Ho    1   PCho

4    0       H                                  Lyso-PAF

         1a                                              !m
         OR                                              OR
         PCho                                 ’‘1        PCho
     PAF                                            PAF
Fig. 7. Biosynthesis of PAF by the remodeling and de novo pathways. The center set of reactions illustrates
the sequence of reactions that form the membrane precursor (alkylacylglycerophosphocholines) PAF.  of
Enzymes responsible for catalyzing the various reactions are: I, acetyl-CoA: 1 -alkyl-2-lyso-sn-glycero-3-P
acetyltransferase; 11, 1-alkyl-2-acetyl-sn-glycero-Pphosphohydrolase; 111, CDP-choline:l-alkyl-2-acetyl-
sn-glycerol cholinephosphotransferase (DTT-insensitive); IV, acyl-CoA: 1-alkyl-2-lyso-sn-glycero-3-P
acyltransferase; V, I-alkyl-2-acyl-sn-glycero-3-P    phosphohydrolase; V1, CDP-cholincl -alkyl-2-acyl-sn-
glycerol cholinephosphotransferase (DTT-sensitive); VII, 1-alkyl-2-acyl-sn-glycero-3-phosphocholine
phospholipase A,/transacylase; VIII, acetyl-CoA: 1-alkyl-2-lyso-sn-glycero-3-phosphocholine   (lyso-PAF)
acetyltransferase; acy, acyl; ace, acetate; cho, choline; DTT, dithiothreitol.



 Fig. 8. The transacylation reaction between alkylacylglycerophosphocholines and ethanolamine lyso-
 plasmalogens or other ethanolamine- and choline-containing lyso-glycerophosphatides can generate lyso-
 PAF and PAF as shown in this reaction scheme. The reaction sequence is catalyzed by a tightly associated
 transacylase/phospholipaseA, activity (reaction I) and a phospholipase A, (reaction 11) associated with
 membranes. The lyso-PAF formed as an intermediate is acetylated by an acetyl-CoA dependent acetyl-
 transferase (reaction 111). It is possible that the enzyme activities responsible for catalyzing steps I and I1
 could reside in a single protein or as a multienzyme complex. Similar types of transacylation reactions
 could also explain the production of acyl and alk-I-enyl analogs of PAF from their corresponding diacyl
 or alk-I-enylacyl precursor phospholipids.

 complexity in interpreting biological responses following stimulation of this route. It
 is thought the remodeling pathway of PAF synthesis is the primary contributor to
 hypersensitivity reactions and for this reason this route has been implicated in most
 pathological responses involving PAF.
    In contrast, the de novo pathway of PAF biosynthesis (unaffected by inflamma-
 tory stimuli) is thought to be the primary source of endogenous physiological levels
 of PAF (Fig. 7 ) .The sequence of reactions in the de novo route include (1) acetylation
 of 1-alkyl-2-lyso-sn-glycero-3-P an acetyl-CoA-dependent acetyltransferase, (2)
 dephosphorylation of 1-alkyl-2-acetyl-sn-glycero-3-P, (3) the transfer of phos-
 phocholine from CDP-choline to 1-alkyl-2-acetyl-sn-glycerol by a cholinephospho-
 transferase [dithiothreitol (DTT)-insensitive] to form PAF. The DTT-insensitivity of
 the cholinephosphotransferase involved in PAF biosynthesis is in contrast to the in-
 hibitory action of DTT on the cholinephosphotransferase that forms phosphatidyl-
 choline and plasmanylcholine from diacylglycerols and alkylacylglycerols, respec-
 tively. In addition, the cholinephosphotransferases that synthesize PAF and phos-
 phatidylcholine exhibit different pH optima and respond differently to detergents,
 ethanol, temperature, and substrates. Thus, it would appear that the DTT-insensitive
 cholinephosphotransferase represents an activity that is distinctly separate from the
 one responsible for the production of phosphatidyl-and plasmanyl-choline.Proof of
 whether they are actually different enzymes must await their purification from the
 membrane environment where they are found.

   The enzymes in both the de novo and remodeling pathways exhibit high substrate
specificity; however, the acyl analog appears to be able to replace the 0-alkyl substit-
uent of the substrates utilized by the acetyltransferase, phosphohydrolase, and cho-
linephosphotransferase in de novo synthesis and by the acetyltransferase in the re-
modeling synthesis of PAF. In contrast, the two different acetyltransferases (Fig. 7,
enzymes I and VIII) possess completely different properties and substrate specifici-
ties and the phosphohydrolase (enzyme 11) and cholinephosphotransferase in the de
novo route (enzyme 111) appear to be distinctly different activities from those in-
volved in the de novo synthesis of alkylacylglycerophosphocholines     (enzymes V and
VI, respectively), the membrane precursor of PAF in the remodeling pathway (Fig.
   Although the quantitative importance or the physiological roles of the two path-
ways for PAF biosynthesis are unknown at the present time, it is clear that their
significance is probably dependent upon the specific cell types and whether physio-
logical or pharmacological stimuli are involved. Physiological factors such as fatty
acids and neurotransmitters can stimulate the de novo synthesis of PAF, but this
pathway does not generate any bioactive eicosanoid metabolites. The regulation and
rate-limiting steps for both pathways of PAF biosynthesis are discussed in Section 10
on Regulatory Controls.

9. Catabolism

9.1. Ether lipid precursors

9.1.1. Long-chainfatty alcohols
Fatty alcohols are oxidized to fatty acids via an NAD+:fatty alcohol oxidoreductase,
a microsomal enzyme found in most mammalian cells. The high activity of this en-
zyme probably accounts for the extremely low levels of fatty alcohols generally found
in most tissues or blood. The detection of fatty aldehyde intermediates (by trapping
them as semicarbazidederivatives) during oxidation of the alcohol suggests that the
fatty alcohol oxidoreductase catalyzes a two-step reaction. The difficulty in purifying
the membrane-associated enzymes of fatty alcohol metabolism has prevented much
progress in this area, except for the demonstration of the reaction sequence and the
characterization of some of the properties.

9.1.2. Dihydroxyacetone-P and acyldihydroxyacetone-P
Dihydroxyacetone-Pcan be channeled from the ether lipid pathway through the for-
mation of sn-glycerol-3-P via NADH:glycerol-3-P dehydrogenase. An alternate by-
pass of alkyldihydroxyacetone-Pformation occurs if acyldihydroxyacetone-P is re-
duced by the NADPH-dependent oxidoreductase, since the product, I-acyl-2-lyso-
sn-glycerol-3-P, is then directed into diacylglycerolipids. The metabolic removal and/
                                   p j-

 HzCOCH,CH,R                            H,COCCH,R

HOCHI       + Pte.H, __F
                      0,                                               RCH,CHO   + glycerol + Ple-H,+   H,O
 H,COI I                                H,COH

  Alkylglycerol                          Hemiacetal                 Fatty Aldehyde

Fig. 9. Cleavage of the 0-alkyl linkage in glycerolipids is catalyzed by a tetrahydropteridine(Pte*H,)-
dependentalkyl monooxygenase,a microsomal enzyme found primarily in liver and intestinal tissues. Only
glycerolipids containing at least one free hydroxyl group in the glycerol moiety appear to serve as sub-
strates; a hemiacetal has been proposed as an intermediatein this reaction but it has never been isolated.

or formation of fatty alcohols, dihydroxyacetone-P,or acyldihydroxyacetone-Pfrom
the ether lipid precursor pool represent impxtant control points in the ether lipid
pathway. Regulatory aspects of these reactions are discussed later.

9.2. Ether lipids in membranes

9.2. I . AIkyI cleavage enzyme
Oxidative cleavage of the 0-alkyl linkage in glycerolipids is catalyzed by a microso-
ma1 tetrahydropteridine (Pte.H,)-dependent alkyl monooxygenase (Fig. 9). Alkyl
cleavage enzyme activities are highest in liver and intestinal tissue, whereas most
other cells/tissues possess very low activities. Tumors and other tissues that contain
significant quantities of alkyl lipids generally have very low alkyl cleavage enzyme
activities, an observation that has suggested the cellular amounts of ether lipids are
inversely related to the levels of alkyl cleavage enzyme activities.
   Fatty aldehydes produced in the cleavage reaction can be either oxidized to the
corresponding acid or reduced to the alcohol by the appropriate enzymes and the
Pte.H, is regenerated from the Pte.H, by an NADPH-linked pteridine reductase, a
cytosolic enzyme. Oxidative attack on the ether-linked grouping in lipids is similar to
that known for the hydroxylation of phenylalanine.
   Structural requirements of glycerolipid substrates utilized by the alkyl cleavage
enzyme are (a) an 0-alkyl moiety at the sn-1 position, (b) a free hydroxyl group at the
sn-2 position, and (c) a free hydroxyl or phosphobase group at the sn-3 position. If
the hydroxyl group at the sn-2 position is replaced by a ketone or acyl grouping, or
when a free phosphate is at the sn-3 position, the 0-alkyl moiety at the sn-1 position
is not cleaved by the Pte.H,-dependent monooxygenase. Thus, 1-alkyl-2-lysophos-
pholipids (e.g., lyso-PAF) are substrates for the cleavage enzyme, but they are at-
tacked at much slower rates than the alkylglycerols. The ether linkage in alkylglycols
is cleaved in the same manner as the alkylglycerok., but more slowly.

9.2.2. Plasmalogenuses
Microsomal enzymatic activities have been described that hydrolyze the alk-1-enyl
grouping of plasmalogens or lyso-plasmalogens; the products are a fatty aldehyde
and either I-lyso-2-acyl-sn-glycero-3-phosphoethanolamine choline) or sn-
glycero-3-phosphoethanolamine(or choline), depending on the chemical structure of
the substrate utilized by the plasmalogenase (Fig. 10). Plasmalogenase activities have
been found in microsomal preparations of liver and brain membranes from rats,
cattle, and dogs. The possibility that a plasmalogenase works in concert with lyso-
phospholipase A, to release the rich source of arachidonic acid generally associated
with plasmalogens is an intriguing thought, since the regulation of the release of ara-
chidonic acid from such phospholipids could be controlled by the substrate specifici-
ties of two separate enzymes. Since plasmalogenases have not yet been purified, their
role in lipid metabolism still remains obscure.

9.2.3. Phospholipases and lipases
In general, the ester groupings associated with either the alkyl or alk-1-enyl glycero-
lipids are hydrolyzed by lipolytic enzymes with the same degree of specificity as their
acyl counterparts. However, the presence of an ether linkage at the sn-1 position of
the glycerol moiety generally impairs the overall reaction rate to the extent that cer-
tain lipases have been successfully used to remove diacyl contaminants in the purifi-
cation of some ether-linked phospholipids. Table I1 lists various lipases that have
been investigated with ether lipids as substrates and identifies the products of such
reactions. The only lipase (other than those that cleave the ether linkages) known to
exhibit an absolute specificity for ether-linked lipids is lysophospholipase D, an en-
zyme discovered by Robert Wykle (U.S.A.).The uniqueness of lysophospholipase D
is that it exclusively recognizes only 1-alkyl-2-lyso-sn-glycero-3-phosphobases 1-or
alk- l-enyl-2-lyso-sn-glycero-3-phosphobases substrates (Fig. 11, reaction 11).
   The lypolytic enzymes of ether lipid metabolism have received far less attention
that those associated with the biosynthetic pathways. Certainly, more knowledge
about this area of research is required before regulation of the metabolic steps that
degrade the ether-linked lipids can be adequately understood.


                            -             YCOH

                                                                    + RCHO
          dH                                  dH
1-Alk-I ‘-enyl-Z-acyl-sn-glycero-   Lyso-phosphatidylethanolamine     Fatty
   3-phosphoethanolamine                                            Aldehyde

Fig. 10. Reaction illustrating the cleavage of the 0-alk-I-enyl linkage in an ethanolamine plasmalogen by
a plasmalogenase.
Ether lipids as substrates for various lipases

Enzyme                                  Products formed from

                                        1-Alkyl-(or 1-alk-l-enyl)-2,3    1-Alkyl-(or 1-alk- 1-enyl)-       I -Alkyl-(or 1-alk-1-enyl)-
                                        diacyl-sn-glycerols              -2-acyl-sn-glycero-3-             -2-lyco-sn-glycero-3-
                                                                         -phosphobase                      -phosphobase

Pancreatic lipase or                    1-Alkyl-(or 1-alk-l-enyl)-2,3-   No reaction                       No reaction
  Rhizopus lipase*                        sn-glycerols + RCOOH
Phospholipase A,                        No reaction                      1-Alkyl-(or I-alk-1-eny1)-        No reaction
                                                                           -phosphobase + RCOOH
Phospholipase C                         No reaction                      I-Alkyl-(or 1-alk-I-eny1)-        I-Alkyl-(or 1-alk-1-enyl)-
                                                                           -2-acyl-sn-glycerol+              -sn-glycerol + phosphobase
Phospholipase D                         No reaction                      1 -Alkyl-(or 1-alk-I-eny1)-       1-Alkyl-(or I-alk-l -enyl)-
                                                                           -2-acyl-sn-glycero-3-P + base     -sn-glycero-3-P + base

Lysophospholipase D                     No reaction                      No reaction                       I-Alkyl-(or I-alk-1-eny1)-
                                                                                                             -sn-glycero-3-P + base
Pte.H4-dependent                        No reaction                      No reaction                       sn-Glycero-3-phosphobase
  monooxygenase                                                                                              + RCHO
(specific for alkyl group
Plasmalogenase                          2,3-Diacyl-sn-glycerols + RCHO   1-Lyso-2-acyl-sn-glycero-3-       sn-Glycero-3-phosphobase
(specific for alk-I-enyl group)                                            -phosphobase + RCHO               + RCHO

*Rhizopus arrhizus or delemar

                                                        4               H , r          I‘,   ~     H,iBFI
CH,COCH                          HOCH                          b        HOCH                       HOCH
      I0          +
                                    I?            +
          6-                           6-
          PAF                       Lyso-PAF                          Alkylglycero-P             Alkylglycerol

Fig. 11. Inactivation of PAF by hydrolysis of the acetate moiety is catalyzed by an acetylhydrolase (I). The
product, 1-alkyl-2-lyso-sn-glycero-3-phosphocholine    (lyso-PAF) can be further degraded by a lysophos-
pholipase D (11) and a phosphohydrolase (111).

9.3. Bioactive ether lipids

Inactivation of PAF is achieved via l-alkyl-2-acetyl-sn-gIycero-3-phosphocholine
acetylhydrolase, the enzyme that catalyzes the hydrolysis of the acetate moiety
to produce the inactive 2-lyso form of PAF (Fig. 1 1, reaction I). This type of hydro-
lytic activity is catalyzed by an acetylhydrolase that occurs in both intracellular
(cytosolic fraction) and extracellular forms (serum). The serum enzyme has been
purified to homogeneity; it differs from the intracellular form in that it is resistant
to proteases and has a higher molecular weight. Since all other characteristics of the
two enzyme activities appear to be identical, it is likely that the intracellular form
is processed in the cell for export (for example, by glycosylation or protein
modification) to the blood compartments. The serum acetylhydrolase is associated
with LDL and HDL, but only the portion residing with the LDL is catalytically ac-
   Acetylhydrolase’s (serum or intracellular forms) properties clearly indicate that it
differs from the usual type of activity described for phospholipase A, in tissues, al-
though phospholipase A2 can also hydrolyze the acetate moiety of PAF. Other phos-
pholipids (such as acylacetyl types) with an acetate at the sn-2 position also are sub-
strates for the acetylhydrolase, but not those phospholipids with long-chain acyl
moieties at the sn-2 position. Acetylhydrolase activities are widely distributed in a
variety of cells, tissues, and blood throughout the animal kingdom.
   The product of the acetylhydrolase reaction, lyso-PAF, can be either reacylated to
the membrane precursor (alkylarachidonoylglycerophosphocholine)of PAF, de-
graded by the Pte.H, alkyl monooxygenase, or utilized by a lyso-phospholipase D
(Fig. 1 1, reaction 11) to form an alkyllysoglycero-P. The phosphorylated intermediate
can be further degraded to alkylglycerols by a phosphohydrolase (Fig. 11, reaction
111). The lyso-PAF, alkyllysoglycero-P, or alkylglycerols could be further degraded
by the Pte.H, alkyl cleavage enzyme and thus, be completely eliminated from the
ether lipid pool.

10. Regulatory controls

Regulatory mechanisms that control the metabolism of ether-linked lipids are still
poorly understood. In fact, most progress in this area has concerned PAF metabo-
lism, primarily because of the high degree of interest in this potent mediator. How-
ever, a variety of factors are known to influence the overall rates of ether lipid me-
tabolism, but such studies have mainly been of the descriptive type and almost none
have dealt with molecular mechanisms at the enzyme level of regulation. Regulatory
controls that must be considered in the metabolism of ether-linked lipids are those
that influence (a) the enzymes responsible for catalyzing the formation and catab-
olism of the ether lipid precursors (fatty alcohols and dihydroxyacetone-P), (b) the
alkyldihydroxyacetone-P synthase responsible for the synthesis of alkyldihydroxya-
cetone-P, (c) branch point enzymes and (d) those enzymes responsible for the degra-
dation of the ether lipids.
   Glycolysis plays an important role in controlling the levels of ether lipids at the
precursor level since the high glycolytic rate of tumors that generates significant
quantities of dihydroxyacetone-P would appear to explain the relatively high levels of
ether lipids in tumor cells. Such a correlation has been observed in a series of trans-
plantable hepatomas that possess high rates of glycolysis, low glycerol-P dehydro-
genase activities, and high levels of ether-linked lipids. Accompanying the higher
levels of dihydroxyacetone-P in tumor cells are also elevated amounts of long chain
fatty alcohols. Thus, the glycolytic state of a cell would seem to be important in pro-
viding the precursors required for the formation of the ether bond by alkyldihy-
droxyacetone-P synthase.
   The coordinated biosynthetic and catabolic enzyme activities that catalyze specific
reaction steps in the metabolic pathways for ether-linked lipids are obviously under
very complex control. Although the rate-limiting steps are poorly understood, two
important intermediary branch points in the biosynthesis of ether-linked lipids in-
volve 1-alkyl-2-lyso-sn-glycero-3-P. 1-alkyl-2-lyso-sn-glycero-3-P be ultim-
                                      The                                 can
ately converted to either PAF or to 1-alkyl-2-acyl-sn-glycerols. The latter is also
at a branch point since the alkylacylglycerolscan be the direct precursors of any one
of the following products: plasmanylcholines, plasmanylethanolamines or alkyldia-
cylglycerols. Factors or conditions that influence either branch point would have
profound effects on the proportion of the different types of ether-linked lipids
formed. Since the same enzyme appears to be able to utilize both diacyl- and alky-
lacyl-glycerols, it is apparent that the availability of diradylglycerols would be an
important factor in controlling the membrane lipid composition with respect to
diacyl and alkylacyl species.
   Studies of the regulation of PAF metabolism are still in the early stages of develop-
ment. Rate-limiting steps in the de novo pathway of PAF biosynthesis are the acetyl-
CoA: 1-alkyl-2-lyso-sn-glycero-3-P  acetyltransferase and the cytidylyltransferasethat
forms CDP-choline for the cholinephosphotransferase catalyzed step (Fig. 7; see
Chapter 7). Any factor that stimulates these rate-limiting reactions (e.g., activation of
cytidylyltransferase by fatty acids) will greatly enhance the de novo synthesis of PAF.
    In the remodeling pathway, it is clear arachidonic acid can influence the formation
of PAF at the substrate level since cells depleted of alkylarachidonoylglycerophos-
phocholines loose their ability to synthesize PAF. Therefore, the transacylase/phos-
pholipase A, step (Fig. 8) can be rate-limiting. On the other hand, the acetyl-
CoA:lyso-PAF acetyltransferase (Fig. 7, reaction VIII) can also be rate-limiting since
it is under the tight control of a phosphorylating/phosphohydrolasesystem; only the
phosphorylated form of the acetyltransferase in this route is active. Unfortunately,
little is known at the present time about the cellular factors responsible for regulating
the activation and inactivation of this acetyltransferase. Moreover, detailed charac-
terization of the protein kinase responsible is lacking, but both a cyclic AMP-depend-
ent kinase and a calcium calmodulin-dependent kinase have been implicated in differ-
ent cell types. Some results have also indicated the phospholipase A, (Fig. 7, reaction
VII) in the remodeling route requires activation by protein kinase C, but the exact
mechanism for the reactions catalyzed by the transacylase/phospholipase A, (Fig. 7)
is not yet well understood.
    Calcium ions are required for the biosynthesis of PAF in the remodeling route,
whereas Ca2+inhibits the activities of all three enzymes in the de novo route. There
also is evidence to indicate Ca2+can inhibit the deacetylation of PAF and the transa-
cylation of lyso-PAF but in intact cells these effects appear to be through some indi-
rect mechanism rather than at the enzyme level per se.
    Acetylhydrolase (Fig. 11, reaction I) and other catabolic enzymes in PAF metabo-
lism also have an important regulatory role in controlling PAF levels since it is
known that the activity of acetylhydrolase can drastically change during various dis-
eases, pregnancy, and macrophage differentiation. Furthermore, endogenous inhibi-
tors of PAF responses require more extensive investigation to determine whether
they affect specific enzymes involved in PAF metabolism.

11. Functional roles

11.1. Membrane components

Cellular functions of ether-linked glycerolipids are still poorly understood, but their
ability to serve as both membrane components and as cellular mediators is now well
established. Both the alkyl and alk- 1-enyl phospholipids that contain long-chain acyl
groups at the sn-2 position appear to be essential structural components of many
membrane systems. Some species of the ether lipids associated with membranes act
as storage reservoirs for polyunsaturated fatty acids. The apparent protective nature
of ether-linked groups is undoubtedly due to their ability to slow the rate of hydroly-
sis of acyl moieties at the sn-2 position by phospholipase A*. The preferential seques-

tering of polyunsaturated fatty acids in phospholipids has been observed in essential
fatty acid deficiency. Under this condition the ether lipids of testes from deficient rats
retained arachidonic acid, whereas the diacyl phospholipids were rapidly depleted of
their arachidonoyl moieties.

11.2. Cell mediators (activities and mechanisms of action)

The multifaceted responses generated by PAF in vivo and in target cells has empha-
sized the important role of such lipids to act as diverse regulators of metabolic and
cellular processes. Specific PAF receptors on the plasma membrane have been docu-
mented for a number of cells (e.g., platelets, neutrophils, smooth muscle cells, mac-
rophages, brain and liver membranes, differentiated HL-60 cells) and a stereospecific
requirement of the receptors for native PAF (1-alkyl-2-acetyl-sn-glycero-3-phospho-
choline) has been established. An array of highly specific PAF receptor antagonists
(natural and synthetic) have been developed by a variety of pharmaceutical firms.
Recently the PAF receptor from guinea pig king and human neutrophils was cloned
and analyzed by a Japanese group headed by Shimizu and Seyama. The primary
structure of the PAF receptor is associated with the general family of G protein-
coupled receptor agonists. Interestingly, the cytoplasmic tail of the PAF receptor
consists of four serine and five threonine residues that could serve as phosphate ac-
   In neutrophils, PAF causes aggregation and degranulation responses, induces
chemotaxis and chemokinesis, increases cell adherence, enhances respiratory bursts
and superoxide production, and stimulates the production of arachidonic acid me-
tabolites formed via the lipoxygenase pathway. PAF also causes aggregation and
degranulation of platelets and stimulates the uptake of calcium, protein phosphoryla-
tion, the phosphatidylinositol cycle, and arachidonic acid metabolism (potentiates
the formation of hydroxy fatty acids and thromboxanes). In other systems, PAF in-
creases vascular permeability, induces the constriction of the ileum and lung strips,
stimulates hepatic glycogenolysis, causes bronchoconstriction, increases pulmonary
resistance, causes pulmonary hypertension and edema, decreases dynamic lung com-
pliance, produces neutropenia and thrombocytopenia, causes intestinal necrosis, and
produces systemic hypotension. Many of the biological activities described for PAF
are associated with hypersensitivity reactions, such as allergies and inflammation.
The myriad events that can occur when PAF reaches sufficiently high levels in blood
can end in death by anaphylactic shock.
   The mechanism explaining these diverse actions of PAF is thought to involve a
cellular signal via a C protein since PAF is known to stimulate GTPase activity.
However, since neither cholera nor pertussis toxins alter PAF responses, it appears
that a G protein other than Ni and N, is involved. The activation of phosphatidylino-
sitol-specific phospholipase C by the binding of PAF to its receptor elevates the intra-
cellular levels of Ca2+and diacylgycerols, events that activate protein kinase C. The

latter catalyzes the phosphorylation of specific proteins and it is clearly documented
that 20 and 40 kDa proteins are phosphorylated in PAF-treated rabbit platelets. 0th-
er factors also need to be considered in any proposed biochemical mechanism for
PAF actions. For example zinc ions inhibit the aggregation of platelets by PAF but
not by thrombin. Results have also been presented to indicate a chymotryptic type of
serine protease and vicinal sulfhydryl groups participate in the activation of platelets
by PAF. Although many important components in the mechanism have been de-
scribed, considerably more studies must be forthcoming to establish firmly how PAF
participates in cellular signalling processes.

12. Future directions

Future studies of ether lipids are needed to address the many unanswered questions
about their regulatory controls and functions. Significant progress has already been
made in our understanding of the mechanism of formation of the alkyl ether bond
and of the biological activities, metabolism, and some regulatory aspects of the alkyl-
acetylated phospholipids, but little is known about the mechanism of action of these
unique bioactive phospholipids. Studies on the mode of action of a group of unnatu-
ral ether-linked phospholipids that are analogues of PAF (e.g., I-alkyl-2-methoxy-
sn-glycero-3-phosphocholine)      with highly selective antineoplastic properties could
provide exciting dividends for those who pursue this challenging area of research.
   Despite the large quantities of ethanolamine plasmalogens found in nervous tissue
and other cells, we still have no idea how they function. Prospective investigators
concerned with the biological role of plasmalogens will need to consider the molecu-
lar mechanism of the desaturation step responsible for their biosynthesis and the way
in which plasmalogen levels are regulated. The biosynthesis of choline plasmalogens
is not yet fully understood, although they appear to originate from the ethanolamine
plasmalogens. Another segment of research on ether lipids that has virtually been
untouched is their dietary significance.
   In order to understand more fully the regulation of the synthesis and degradation
of the ether-linked lipids, it will be necessary to purify key enzymes of the biosynthet-
ic and catabolic pathways and to use molecular biology approaches so that their reac-
tion mechanisms can be delineated. Antibodies against these enzymes could be
prepared for use as probes to evaluate specific regulatory steps in cellular systems.
Information is also lacking about the physical properties of specific molecular species
of ether lipids and their functional role as membrane constituents.
   Knowledge about the cellular functions of ether lipids will continue to expand as
scientists from different disciplines enter this field. The impact of such interdiscipli-
nary efforts has been well demonstrated in the PAF field after its discovery in 1979.
The established importance of the biologically active ether-linked phospholipids as
potent and diverse cellular mediators with both physiological and pharmacological

effects should be looked on as only the first step in developing our understanding of
the cellular function of the various types of ether-linked lipids.

This work was supported by the Office of Energy Research, U.S. Department of
Energy (Contract No. DE-AC05-760R00033), the American Cancer Society (Grant
BC-70V), the National Heart, Lung and Blood Institute (Grant 27109-10, and the
National Cancer Institute (CA-4164245).

 1. Snyder, F. (Ed.) (1972) Ether Lipids: Chemistry and Biology, pp. 1433, Academic Press, New York,
 2. Mangold, H.K. and Paltauf, F. (Eds.) (1981) Ether Lipids: Biochemical and Biomedical Aspects, pp.
    1439, Academic Press, New York, NY.
 3. Snyder, F. (Ed.) (1987) Platelet-Activating Factor and Related Lipid Mediators, pp. 1471, Plenum
    Press, New York, NY.
 4. Winslow, C.M., Lee, M.L. (Eds.) (1987) New Horizons in Platelet Activating Factor Research, pp.
    1-373, Wiley, New York, NY.
 5. Barnes, P.J., Page, C.P. and Henson, P.M. (Eds.) (1989) Platelet activating factor and human disease.
    Frontiers in Pharmacology and Therapeutics, pp. 1-334, Blackwell Scientific Publications, Oxford.
 6. Lee, T.-C. and Snyder, F. (1985) Function, metabolism, and regulation of platelet activating factor
    and related ether lipids. In: Phospholipids and Cellular Regulation (Kuo, J.F. Ed.) pp. 1-39, CRC
    Press, Boca Raton, FL.
 7. Snyder, F., Lee, T.-C. and Wykle, R.L. (1985) Ether-linked glycerolipids and their bioactive species;
    enzymes and metabolic regulation. In: The Enzymes of Biological Membranes (Martonosi, A.N., Ed.)
    pp. 1-58, Plenum Press, New York, NY.
 8. Hanahan, D.J. (1986) Platelet activating factor. A biologically active phosphoglyceride. Annu. Rev.
    Biochem. 55,483-509.
 9. Braquet, P., Touqui, L., Shen, T.Y.and Vargaftig, B.B. (1987) Perspectives in platelet activating factor
    research. Pharmacol. Rev. 39, 97-145.
10. Snyder, F. (1989) Biochemistry of platelet-activating factor; a unique class of biologically active phos-
    pholipids. Proc. SOC. Exp. Biol. Med. 190, 125-135.
1I. Snyder, F. (1990) PAF and related acetylated lipids as potent biologically active cellular mediators.
    Am. J. Physiol. 259, C697-C708.
12. Prescott, S.M., Zimmerman, G.A. and Mclntyre, T.M. (1990) Platelet-activating factor. J. Biol.
    Chem. 265, 17381-17384.
13. Goldfine, H. and Langworthy, T.A. (1988) A growing interest in bacterial ether lipids. Trends Biol.
    Sci. 13,217-221.
14. Honda, Z., Nakamura, M., Miki, I., Minarni, M., Watanabe, T., Seyama, Y., Okado, H., Toh, H., Ito,
    K., Miyarnoto, T. and Shimizu, T. (1991) Cloning by functional expression of platelet-activating factor
    receptor from guinea pig lung. Nature 349, 342-346.
This Page Intentionally Left Blank
D.E. Vance and 3. Vance (Eds.) Biochemistry o Lipids, Lipoproreins and Membranes
0 1991 Elsevier Science Publishers B.V. All rights reserved.                                   269

                                                                                         CHAPTER 9

                                                                                   MOSELEY WAITE

       Department of Biochemistry, Bowman Gray School of Medicine, Wake Forest University,
                                                         Winston-Salem, NC 27103, U.S.A .

I . Overview

I. I. Definition of phospholipases

The phospholipases are a group of enzymes widely distributed throughout nature
whose generic name indicates their common property of hydrolyzing phospholipids.
Since the substrates hydrolyzed normally exist in an aggregated state, these enzymes
also have the common property of acting at a water-lipid interface. In this regard,
they are distinct from the general class of esterases that exhibit normal saturation
kinetics, as shown in Fig. 1. Phospholipases in general have a weak capacity to hy-
drolyze soluble monomeric phospholipids below their critical micellar concentration
(cmc) and become fully active only when aggregated structures of lipids are formed
above their cmc. The activity of phospholipases on aggregated phospholipids can be
more than a 1 000-fold higher than that observed with soluble substrates. In this chap-
ter, the term ‘aggregated phospholipid’ refers to phospholipid above its cmc without
reference to a specifically defined form such as micelle, bilayer, or hexagonal array
(see Chapter 1).
   Beyond these general similarities in properties of phospholipases there is a great
diversity in their characteristics. These differences are found in their site of attack on
the phospholipid molecule, distribution, function, mode of action, and hydrophobic-
ity. The classification of the phospholipases based on their site of attack is given in
Fig. 2.
   The phospholipases A are acyl hydrolases classified according to their hydrolysis
of the 1-acyl ester (phospholipase A,) or the 2-acyl ester (phospholipase AJ. Some
phospholipases hydrolyze both acyl groups and are termed the phospholipases B. In
addition, lysophospholipases remove one or the other acyl groups from monoacyl-
(1yso)phospholipids. Phospholipases B also have high lysophospholipase activity, as
might be expected. Cleavage of the glycerophosphate bond is catalyzed by phospho-
lipase C, while the removal of the base group is catalyzed by phospholipase D. The

Fig. 1 . Dependence of phospholipase and nonspecific esterase activity on substrateconcentration. Esterase
exhibits Michaelis-Menten kinetics on soluble substrates, whereas phospholipase becomes fully active
above the crnc of the substrate.

phospholipases C and D are phosphodiesterases. Under the appropriate conditions
some phospholipases use an organic hydroxyl rather than H,O in substrate cleavage.
This type of reaction results in transesterification rather than hydrolysis.
   As a broad generalization, two types of functions can be ascribed to phospholipa-
ses. First, many phospholipases are digestive enzymes found in high concentration in
venoms, bacterial secretions, and digestive fluids of higher organisms. These enzymes
are soluble in an aqueous environment even though they avidly adsorb to lipid inter-
faces. The second group of phospholipases has regulatory function whose activity
leads to the production of cellular mediators such as diacylglycerols, inositol tris-
phosphate, platelet activating factor, and the eicosanoids. This role is quite broad

                 C       D
Fig. 2. Site of hydrolysis by phospholipases.
                                                                                     27 1

and still poorly defined. Many of the regulatory phospholipases are membrane asso-
ciated and this membrane association may reflect one of the mechanisms by which
phospholipases are regulated. Perhaps one of the clearest examples of a regulatory
function is that found in the arachidonate cascade in which a phospholipase(s) con-
trols the formation of the bioactive eicosanoids at the level of deacylation from the
phospholipid (see Chapter 10 for a complete description of the eicosanoids). Like-
wise, the phospholipase C that degrades phosphatidylinositols in the phosphatidyl-
inositol cycle is thought to be regulatory in Ca2’ metabolism and in activation of
protein kinase C (Chapter 7).

1.2. Assay ofphosphoripases

The subject of phospholipase assays has been concisely reviewed by Van den Bosch
and Aarsman [I] and by Waite [2]. For thorough coverage of this aspect of phospho-
lipases, these works are recommended. There are many approaches to the assay of
phospholipases. Factors such as sensitivity, site of attack, and equipment available
will dictate to an extent the assay of choice.
   The simplest procedure is to measure the proton released during hydrolysis with a
titration apparatus; in this way, a rapid continuous recording of activity can be made.
The usual substrate used in this assay is egg yolk lipid emulsion prepared in the ab-
sence of a buffer. This undoubtedly is the most economical approach. However, this
technique lacks the sensitivity and specificity of other types of assays. The lower limit
of activity is about 0.1 pmol of substrate hydrolyzed per minute. Therefore, this assay
has been limited primarily to the better-characterized digestive phospholipases that
have high turnover numbers, such as snake venom and pancreatic phospholipases.
   A second procedure that is equally rapid yet has greater sensitivity and specificity
employs phospholipid analogues that have fluorescent moieties or moieties that can
be reacted with chromophores such as Ellmann’s reagent. The assays are usually per-
formed on monomeric substrates or clear emulsions. An example of the former sub-
strate is 1,2-bis[4-(1-pyreno)butynoyl]-sn-glycero-3-phosphorylcholine;     hydrolysis of
this compound yields the soluble monomeric product 4-( 1-pyreno)butyrate, which
emits energy at 382 nm when excited at 332 nm, while the micellar substrate emits
energy at 480 nm [Hendrickson & Rauk, 198 11.

Structure of 4-( 1 -pyreno)butyrate.

  Examples of compounds whose products are reactive with Ellmann’s reagent are
phosphatidylcholineswith thioacylesters that upon hydrolysis release a reactive sulf-
hydryl group [ 11.

        NO,    NO,
-0ocQ          @coo-


Structure of 5,5’-dithiobis(2-nitrobenzoate)(Ellmann’s reagent).

   These assays can be at least 100 times more sensitive than titration procedures and
can be quite useful in the assay of intracellular phospholipases that have low turnover
numbers. The lack of availability of these substrates, plus complications in the assay
when aggregated phospholipids are used, can limit their usefulness in some cases.
   A third approach is the use of phospholipid molecules with isotopes incorporated
preferentially into specific positions in the molecule [2]. The various products of hy-
drolysis are separated from the substrate by extraction and chromatographic proce-
dures. These substrates can be prepared by either chemical or enzymatic methods. By
the appropriate choice of labeling, the specificity of the enzymes can readily be estab-
lished, and as little as a few picomoles of substrate produced can be detected. The use
of isotopes has been helpful in the measurement of phospholipase activity in the
membranes of whole cells or isolated subcellular fractions. In this case, the label is
incorporated into membranous phospholipid in vivo followed by stimulation of the
cell or organelle by an agent that activates the phospholipase. This approach has the
advantages of specificity plus sensitivity. While radioactive labeling of substrates is
utilized widely in the study of phospholipases with low turnover numbers, the meth-
od is limited by the laborious nature of the assay and the expense of the isotopic
   A fourth and very elegant method of phospholipase assay employs a monomolecu-
lar film of phospholipid [3]. With this technique the interfacial properties of the lipid
substrate are carefully controlled and sensitive zero-order kinetics are obtained. The
major drawback of this technique is limitation of equipment available.

1.3. Interaction o phospholipases with interfaces

The distinction of phospholipases from the general esterases implies that special con-
sideration must be given to the interfacial interaction of the enzymes with aggregated
lipids. The increase in activity found in the hydrolytic rate when phospholipids are
present above their cmc (Fig. 1) indicates that all phospholipases have a certain hy-
drophobic character regardless of their origin or type. Therefore, two aspects of en-
zyme activity with phospholipid aggregates must be considered: (1) the hydrophobic

interaction and (2) the formation of the catalytic Michaelis complex. Four factors
can be considered responsible for the enhanced hydrolysis at interfaces:
(1) Increased effective substrate concentration.
(2) Orientation of the phospholipid molecule at the interface.
(3) Enhanced diffusion of the products from the enzyme.
(4)Conformational change of the enzyme upon binding to the interface.
   The first factor can be understood most readily by the example of dihexanoyl phos-
phatidylcholine as substrate, which has a cmc of about 10 mM. Above this interfacial
concentration, micelles form that have an effective concentration in the surface that
is several molar, or three orders of magnitude higher than the free monomers in solu-
tion. Since the enzyme binds to the interface, the high local concentration of substrate
saturates the enzyme. Once the enzyme is bound to the micelles (or in the case of
long-chain phospholipids, bilayers, or hexagonal arrays), the increased concentra-
tion of substrate available markedly increases activity, assuming that the enzyme
does not dissociate from the aggregate after the hydrolysis of a phospholipid mole-
cule. The validity of this assumption was thoroughly examined by Jain and Berg [4]
who developed a model in which the phospholipase ‘scoots’ on the surface of a single
lipid vesicle to hydrolyze the total surface before ‘hopping’to a new vesicle.
   The second factor, orientation of the phospholipid at the interface, was established
using ‘H-NMR measurements that demonstrated that the orientation of the acyl es-
ters is restricted in a bilayer system. Presumably, this restriction of the phospholipid
facilitates interaction of the substrate molecule with the active site in the enzyme. Fig. 3
[5] shows that the 1-acyl ester is more deeply buried in the hydrophobic region than
the 2-acyl ester. It is not yet clear just why such an orientation of the acyl esters favors
enzymatic hydrolysis or why this would be the case for all types of acyl hydrolases.
Likewise, polar head group interaction influences the activity of the phosphodiester-
ase-type phospholipases C and D. However, little information on the relationship of

                0- CH,-       CH,-   N(CHJ,
      0 0- P = O
    Water  0                             Water
Interface H * C

                     -   0
                             - - 7.. Interfaoe

                c =o


Fig. 3. Orientation of ester bonds of phosphatidylcholineacyl chains. (Adapted from Dennis [5]).

these phospholipases to the orientation of the moieties under attack is available. An
elegant study by Thuren et al. (1984) did, however, show that the positioning of the
acyl esters of phospholipid regulated the relative activities of phospholipases A, and
   The third factor postulated to favor activity, enhanced diffusion of the products
from the enzyme, is dependent on both the nature of the substrate and the site of
attack. When long-chain phospholipids, as found in membranes, are substrates for
hydrolysis, the products have a low cmc and remain associated with the hydrophobic
region of the enzyme if only an aqueous environment were present. This is true for all
types of phospholipases, although in the case of phosphodiesterases one of the prod-
ucts is water soluble and less hydrophobic binding occurs between products and en-
zyme than occurs with the acyl hydrolases. The presence of the substrate aggregate
favors the solubilization of the nonpolar product in a hydrophobic environment,
thereby favoring diffusion of product from the enzyme. If, however, short-chain
phospholipids were substrates, the products would be relatively water soluble, so the
lipid diffusion factor would be expected to be less predominant.
   The fourth point, conformational changes in phospholipases, has been postulated
to account in part for the activity on aggregated lipid [6]. The structural and spectral
properties of some digestive acyl hydrolases have been found to change when Ca2+or
substrate binds. Also, kinetic studies with the pancreatic enzyme suggest that an acti-
vation process, probably linked to a conformational change, is necessary for maxi-
mal activity. This process could be related to a dimerization of the enzyme that has
been shown for the activation of a snake venom phospholipase [5].
   The nature of the aggregated lipid markedly influences the activity of phospholipa-
ses on the basis of the following parameters:
(1) Charge at the lipid-water interface.
(2) Packing of the molecules within the aggregate.
(3) Polymorphism of aggregate.
(4) Fluidity of phospholipid.
   The ionic charge of the lipid has long been known to influence the activity of phos-
pholipases. The interfacial charge is a reflection of the surface pH, which can be dif-
ferent from the bulk pH and, therefore, can influence the pH optimum of the phos-
pholipase. There are various techniques to measure surface charge, for example, mi-
croelectrophoresis [7]. Ionic amphipaths as well as the ionic content of the aqueous
environment influence this surface charge. In addition to the charge imparted by the
polar head group of the individual phospholipid molecules, the chemical structure of
the polar head group is also important in the formation of the Michaelis complex, as
shown by the ability of phospholipases to distinguish amongst phospholipids in ag-
gregates of the same charge.
   The effect of molecular packing of lipid molecules within the aggregate on phos-
pholipase activity has long been recognized as a factor which regulates activity. While
this effect is most pronounced for the acyl hydrolases, it also occurs with the phos-

phodiesterases. The influence of packing is most precisely determined using mono-
molecular films of phospholipid with defined surface pressures [8]. In this system, the
phospholipids are present as a monomolecular film at the air-water interface. The
pressure of the film and, therefore, the area occupied by the individual phospholipid
molecules can be regulated by a mechanical bar placed at the end of the film [6].    In
an interesting study, the ability of various phospholipases to penetrate films of vari-
ous pressures was used to estimate the surface pressure and concentration of phos-
pholipids in erythrocytes [9].The monolayer system, therefore, appears to be a good
model for the study of natural bilayers.
   The polymorphic states that have been considered primarily include micelles, vari-
ous bilayer liposome structures, and hexagonal arrays. The first two of these are of
considerable physiological interest and more recently evidence has emerged that local
regions of hexagonal arrays or cubic phases may enhance action by phospholipases.
The micellar mixture of phospholipids and bile salts found in the intestine is a prime
example of the degradation of micellar phospholipid. Likewise, the attack of mem-
branous bilayer systems by both digestive and regulatory phospholipases occurs
widely. The attack of hexagonal arrays could possibly have some interesting func-
tions including an involvement in membrane fusion. For example, a phospholipase
that preferentially attacks a transitory hexagonal array at a point of fusion could be
responsible for removal of the fusative lipid and the reestablishment of a bilayer
   As is true for many membrane-associated enzymes, the state of fluidity also regu-
lates phospholipase activity. It is well established that many phospholipases are most
active at the phase transition of a pure phospholipid or in mixed phospholipid sys-
tems that exhibit phase transition. It appears that phospholipases recognize and pen-
etrate fissures between gel and liquid crystalline phases that favor activity. Interest-
ingly, phospholipases in some model systems appear to degrade phospholipid in the
gel phase more rapidly than in a liquid-crystal structure [5,6]. example, the phos-
pholipase A2 from Nuju nuju nuju venom hydrolyzed bilayers of disaturated phos-
phatidylcholines more rapidly below, than above, the phase transition [6].
   Kinetic analysis of phospholipases using aggregated phospholipid is very complex
and must be approached using models other than those used with water-soluble sub-
strates. The commonly used Michaelis-Menten type kinetics, although often report-
ed, have little value except for comparative purposes. As noted previously, varying
substrate concentration above the cmc determines only the number of lipid aggre-
gates present, not the concentration of phospholipid within the aggregate. The model
of Jain and Berg [ ]who used pure lipid vesicles showed that the initial rate of hy-
drolysis became independent of excess vesicles when a ratio of vesicles to enzyme
molecules exceeded 3 4 . In this system, all enzyme molecules became bound to the
interface and the enzyme was limited by the number of substrate molecules in the
vesicle. This is evidence for the tight binding of phospholipase to the lipid interface
with slow release of enzyme once the substrate in the vesicle becomes exhausted.
   Two basic approaches have evolved to provide a large available excess of substrate
to the enzyme. One, the monolayer system, in which the concentration of phospholip-
id is regulated by the pressure applied, permits true zero-order kinetics to be studied.
Unlike the assays that use vesicles of substrate, only a few percentages of the enzyme
molecules become bound to the interface [3]. The second, a mixed-micelle system in
which the phospholipids are diluted within the micelle by the addition of increasing
amounts of detergent [5], also allows for a vast excess of substrate since the substrate
and detergent rapidly exchange between micelles. In the mixed micelle system, the
concentration of the phospholipid in an individual micelle can be regulated by the
ratio of phospholipid to detergent. By varying the ratio of lipid substrate to detergent
within a fixed and saturating number of micelles, it is possible to establish the maxi-
mal velocity of hydrolysis and the affinity of the enzyme for the substrate phospho-
lipid molecule. This affinity constant is most closely akin to the Michaelis-Menten
affinity constant. On the other hand, the absolute phospholipid concentration can be
varied with a constant phospholipid to detergent ratio. The use of a constant ratio of
phospholipid to detergent that varies only the number of micellar substrate particles
yields an affinity constant for bulk lipid [lo].

2. The phospholipases
Many phospholipases have been at least partially purified and characterized. Since it
is impossible to cover all enzymes, only a few examples will be given. These represent
a wide range of functions and organisms.

2. . Phosphoiipase A ,

The phospholipases A, comprise a large group of 1-acyl hydrolases some of which
can also degrade neutral lipids (lipases) or remove the acyl group at position 2 in
addition to that at position 1 (phospholipase B). Thus far, too little information is
available on the protein structure of these enzymes to establish genetic relationships
except for those that also hydrolyze neutral lipids. The function of most phospholipa-
ses A, remains obscure although some functions for those with lipase activity can
now be assigned. In addition, phospholipases A, may play a role in remodeling the
acyl group at the sn-1 position, especially in the endoplasmic reticulum of liver where
phospholipase A, predominates.
   Two phospholipases A, have been purified from Escherichia coii based on their
sensitivity to treatment with detergents; one is resistant to detergents while the other
loses activity in the presence of detergents [2,5]. The former is localized in the outer
membrane, whereas the latter is found on the cytoplasmic membrane and in soluble
fractions. Both have been purified and extensively studied. The outer membrane en-
zyme has a broad substrate specificity, hydrolyzing all phospholipids and many neu-

tral glycerides. On the other hand, the detergent-sensitive enzyme preferentially de-
grades phosphatidylglycerol. The detergent-sensitive enzyme also acts as a transacy-
lase. Transacylation reactions presumably occur when an acyl-enzyme intermediate
is formed in a two-step reaction. In the second step, the acyl group is transferred
nonspecifically to the hydroxyl acceptor of a soluble alcohol or of a lipid such as a
monoacyl lipid. The Escherichia coli transacylase-acting phospholipase forms methyl
esters of the fatty acid in the presence of methanol. Escherichia coli mutants deficient
in either one or both phospholipases have been developed. These mutants have
normal growth characteristics and phospholipid turnover that leaves open the ques-
tion of the function of these enzymes.
   Rat liver lysosomes contain a soluble phospholipase A, that is a glycoprotein and
has multiple forms, perhaps due to variation in the carbohydrate content [2,1 I]. It
has optimal activity at pH 4.0 and does not require Ca2+for activity. The apparent
substrate specificity is highly dependent on the physical structure of the substrate. As
shown in Fig. 4 [ 121, phosphatidylethanolamine is preferentially degraded when no
Triton X-100 (or WR1339) is present. Triton stimulates the activity toward all phos-
pholipids tested except phosphatidylethanolamine. In that case, activity is inhibited,
which suggests that in the absence of Triton the hexagonal arrays formed by phos-
phatidylethanolamine are more readily attacked than the bilayer liposomes of the
other phospholipids. When Triton is added and mixed micelles are formed, the en-
zyme is optimally active on phosphatidylglycerol. Studies with charged amphipaths
and divalent cations demonstrated that this phospholipase A, is optimally active on
substrate with a slight negative surface charge. Although the function of this enzyme
is not established, it probably degrades materials that have been phagocytized, in-
cluding cellular membrane phospholipids.


     I]                                   /pG

     0                 2            4            6
             Molar Ratio of Triton to Phospholipid

Fig. 4. Effect of Triton on lysosomal phospholipase substrate specificity. Triton stimulates the hydrolysis
of phosphatidylcholine and phosphatidylglycerol liposomes but inhibits hydrolysis of phosphatidyletha-
nolamine hexagonal arrays. PE, phosphatidylethanolamine;PC. phosphatidylcholine; PG, phosphatidyl-
glycerol. (Adapted from Robinson and Waite [12]).

   There are two lipases that primarily degrade triacylglycerols and phospholipids in
lipoproteins. Their function in lipoprotein metabolism is described in Chapters 15
and 16 and therefore will be mentioned only briefly here. The extrahepatic lipopro-
tein lipase is optimally active on triacylglycerol when activated by apoprotein CII,
although monoacylglycerols, diacylglycerols, and phospholipids are also degraded.
This enzyme has optimal activity at slightly alkaline pH values (8.0-9.0) and is stimu-
lated by Ca2+.However, in the absence of apoprotein CII,this enzyme maintains full
activity on monoacylglycerol and phospholipids but loses almost all activity on tri-
acylglycerol. It would appear, therefore, that the function of the activator is to pro-
mote interaction of the enzyme with the highly hydrophobic triacylglycerol. Physio-
logically, this enzyme is responsible for the degradation of the neutral lipids in tri-
acylglycerol-rich chylomicrons and very low density lipoproteins, although it could
also be involved in the degradation of phospholipid in these and other lipoproteins.
   The hepatic lipase, like lipoprotein lipase, has broad substrate specificity. It does
not appear to be as dependent upon apoproteins as the lipoprotein lipase and works
well on triacylglycerol emulsions and on pure phospholipid and monoacylglycerol.
Optimal activity of the hepatic lipase is found at pH 8.0-9.0 and functions without
Ca2+.Under the appropriate conditions, the preferred reaction is a transacylation in
which two molecules of monoacylglycerol are converted to a molecule each of diacyl-
glycerol and free glycerol. This strongly suggests that an acyl-enzyme intermediate is
   As a phospholipase, hepatic lipase has a marked preference for phosphatidyletha-
nolamine and phosphatidic acid. Phosphatidylcholine, the most abundant phospho-
lipid on lipoproteins, is poorly degraded. One possible function of the enzyme is to
degrade part of the phospholipid in high density lipoprotein which may provide a key
shuttle of cholesterol (ester) to the liver (Fig. 5 ) [ 131. In this postulate, the phospholip-
id in high density lipoprotein that is cholesterol-ester rich is degraded by the hepatic
lipase, which reduces the amount of surface lipid. No particular mechanism for cho-
lesterol (ester) uptake has been established although physical interaction such as
transient fusion could occur, as shown in Fig. 5. Pittman et al. (1987) have shown that
cholesterol ester of high density lipoprotein is taken up by liver by a non-endocytotic
mechanism. Other studies showed that apolipoprotein E activates the hydrolysis of
phospholipid 3-fold and, therefore, those lipoproteins rich in apoprotein E should be
good substrates for hepatic lipase [14]. Interestingly, hepatic lipase does not act on
pure triacylglycerol monolayers but can be activated as a lipase by phospholipid.
This suggests that hepatic lipase may have two lipid binding sites, one that is regula-
tory and one that is catalytic.

2.2. Phospholipase B and lysophospholipases

The distinction between phospholipase B and lysophospholipases is not clear [ 151,
since the phospholipase B from Penicillium notatum that has been purified and thor-

                      Cell Interior


Fig. 5. Postulated role of hepatic lipase in cholesterol ester uptake by hepatocytes. Hydrolysis of high
density lipoprotein, (HDL,) phospholipids leads to deposition of cholesterol ester in the hepatocyte. The
resulting particle, HDL,, is smaller in size than its precursor as the result of the process. A and C, apopro-
teins; PL, phospholipid; Ch(E), cholesterol (ester); FA, free fatty acid. Grey rectangles, hepatic lipase.
(Adapted from Van 't Hooft et al. [13]).

oughly characterized has high lysophospholipase activity [161. Such activity would be
expected, since phospholipase B deacylates at both the sn-1 and sn-2 positions, with
a lysophospholipid as an intermediate that is subsequently deacylated. Lysophos-
pholipases, on the other hand, need not attack diacylphospholipids, although most
have some activity on diacylphospholipids.These enzymes are widely distributed and
found in microorganisms, bee venoms, and mammalian tissues. The phospholipase B
purified from Penicillium notatum and the lysophospholipases from liver and heart
are described here.
   One of the crucial factors in establishing the dual activity of an enzyme is the puri-
fication of the enzyme to homogeneity [16]. This is particularly true of phospholipase
B, since it is known that the combination of phospholipase A, and phospholipase A,
will catalyze the complete deacylation of diacylphosphoIipid. Phospholipase B is op-
timally active on a wide range of substrates at pH 4.0 and does not have a metal ion
requirement. The proposed sequence of events for phospholipase B activity is as fol-

                                phospholipase A,
    Diacylphospholipid          Pfatty                           acid+ 1-acyl-lysophospholipid
1-Acyl-lysophospholipid                                *     glycerophosphoryl base+fatty acid

  The sum of the two activities of the enzyme, phospholipase A2and lysophospholi-

pase, gives the total activity of phospholipase B. Detailed studies with substrate ana-
logues have allowed a proposed but yet unproven mechanism for this enzyme:
(1) The enzyme has two binding sites: site I, which binds diacylphospholipids, and
     site 11, which binds monoacyl(1yso)phospholipids.
(2) The acyl group at position 2 is transferred covalently to a moiety X in the active
     site of the enzyme when diacylphospholipid is bound to site I.
(3) The lysophospholipid thus formed is transferred to site 11.
(4) The acyl group bound to moiety X is transferred to H,O to liberate the fatty
(5) The acyl group at position 1 is transferred to moiety X and subsequently trans-
     ferred to H,0 to effect the second deacylation.
   In the absence of detergents, phospholipase B is roughly 100 times more active on
lysophospholipid than on diacyl lipid. However, when Triton X-100 is present, the
hydrolysis of diacyl lipid is increased, while that of the lysolipid is inhibited, resulting
in roughly equal activity on the two substrates. Under optimal conditions, lysophos-
pholipase activity (no detergent) is 16-fold higher than phospholipase activity (plus
detergent). The inhibition of lysophospholipasesby detergents is a general effect and
has been used to block this activity when measurement of phospholipase A activity
is sought. This inhibition by detergents can be misleading, however, since detergents
can stimulate certain phospholipases.
   Lysophospholipases, specific for the l-acyl group, have been purified from both
heart and liver and may be the same or similar enzymes [ 15,171. Both tissues have two
forms, a large size (63 kDa) that also catalyzes transacylations, and a small form (22
kDa) that only catalyzes hydrolysis. The larger enzyme from heart transacylated
monoacylphospholipids near their cmc which suggests transacylation between two
monomers of monoacylphospholipids. This is unlike some lysophospholipases that
transacylate only when micelles of substrate are present. Comparison of these differ-
ent activities of lysophospholipasespresent the possibility that three types exist: those
that are hydrolytic only, those that transacylate substrates in micelles, and those that
transacylate with monomers of substrate. If this were the case, distinct physiologic
functions for these enzymes would be expected. Transacylation reactions are now
known to be important physiologic events in many cells and provide a means for the
redistribution of acyl groups between phospholipids without a deacylation-reacyla-
tion cycle. This will be discussed further in the latter part of this chapter.

2.3. Phospholipase A ,

Phospholipases A, have been thoroughly studied and are well understood at the
structural level. While phospholipases A, are ubiquitous in nature, until recently
most emphasis was placed on the venom and pancreatic enzymes, since these are
abundant, easily purified, and stable to a variety of manipulations. Over the past few
years, a number of cellular phospholipases A, from mammals have been purified and
                                                                                                 28 1

related to their venom and pancreatic counterparts. The phospholipases A, were the
first of the phospholipases to be recognized. Over a century ago, Bokay (1 877-1 878)
recognized that phosphatidylcholine was degraded by some component in pancreatic
fluid. This component is now known to be the pancreatic phospholipase A,. At the
turn of the century, it was recognized that cobra venom had hemolytic activity direct-
ed toward the membranes of erythrocytes [18]. A decade later, the lytic compound
produced by the venom phospholipase was identified and termed lysocithin (later,
lysolecithin). These studies spurred further investigation of this intriguing class of

Fig. 6. Amino acid sequence of bovine pro-phospholipase A,. Proteolytic cleavage removes the heptapep-
tide exposing the N-terminal alanine (1). (From Verheij et al. [6]).

enzymes and their mechanism of attack on water-insoluble substrates. Sufficient
quantities of natural and mutant enzymes have been obtained for X-ray crystallo-
graphic analysis. Also, the peptide sequences of over 50 of this class of phospholipa-
ses are known, that demonstrates their structural, functional, and evolutionary close-
   The pancreatic phospholipases are synthesized as zymogens that are activated by
the cleavage of a heptapeptide by trypsin. Both the zymogen and the processed en-
zyme are active on monomers of phospholipid. Fig. 6 shows the amino acid sequence
of the bovine pancreatic pro-phospholipase A,. The seven disulfide bonds provide
the stability observed. Cleavage of the Arg-Ala bond at the position indicated expo-
ses the necessary a-helical hydrophobic site and allows binding of the enzyme with
lipid aggregates or monolayers. The processed enzyme has a molecular weight of
about 14 kDa, typical of all phospholipases A2 of this type.
   The degree of homology of the structures of 40 venom and pancreatic phospholi-
pases has been used for a classification system of the phospholipases [19]. It was
proposed that the pancreatic phospholipases are closely related to the venom en-
zymes of the Elapids (Nu-a and Bungarus amongst others), while the phospholipases
of the Viperids (Crotulus) appear to be unrelated. The only other venom phos-
pholipase A, that has been sufficiently characterized, the bee venom enzyme, has five
disulfide bonds, as compared with seven for the other phospholipases A, of this type.
Therefore, the nature of the disulfide bridges is used to define the groups of phos-
pholipases. Both Group I and Group I1 contain seven disulfides that are conserved.
The Group I enzymes have a disulfide between Cys" and Cys77      whereas Group I1 has
the seventh disulfide between CysSo a Cys at the C-terminus of a six amino exten-
sion to the enzyme. The presence or absence of the Cys" has been useful in char-
acterizing cellular phospholipases that are purified in small quantities that limit the

Fig. 7. Model of pancreatic phospholipase A, binding to and hydrolysis of a monolayer of phospholipid.
E*, enzyme (activated); S , substrate; P, product. E is in the aqueous subphase and E* is absorbed to the
monomolecular film. k, denotes dissociation of the enzyme from the interface. k , denotes binding of the
substrate molecule to the enzyme active site while k-, denotes substratedissociation.k,,, is the catalytic rate
constant for hydrolysis. (From Verger et al. [ZO]).

possibility of complete amino acid sequence analysis. Davidson and Dennis [19] give
a thorough discussion of the possible evolutionary reasons for the variations
amongst the venom phospholipases. Apparently, minor modifications in sequence
have pronounced effects on their function.
   The mechanism of phospholipase A, action has been studied extensively. Verger et
al. [20] postulated a model for the binding and activation of the pancreatic enzyme to
monolayers of phosphatidylcholine, as shown in Fig. 7. The initial binding (k,) of the
enzyme (E) in the subphase is a slow phase that precedes the formation of the active
enzyme (E') absorbed to the monolayer of phospholipid. The EDforms the catalyti-
cally active complex (E'S) with the interfacial substrate (S). In this monolayer system
employing short-chain phosphatidylcholine, the short-chain, water-soluble fatty acid
(P) diffuses into the subphase. Products that are more hydrophobic would diffuse
away from the enzyme but remain in the liposome or aggregate. Interestingly, the
venom phospholipases bind more rapidly to the monolayer than the pancreatic phos-
pholipases, as shown by their relatively low value for k,.
   The hydrolysis of monomolecular films by phospholipases increases with increas-
ing pressure to a critical point; beyond this point the enzyme is no longer capable of
penetrating the film, and activity ceases (Fig. 8). The increase in activity is expected,
since the surface concentration of substrate increases. On the other hand, the amount

          3.4                                              20

          3.3                                              15   F


          3.2                                              10   4

          0.1                                              5

                9               10                14
                S u r f a c e p r e s s u r e (dyneskrn)

Fig. 8. Dependence upon monolayer surface pressure of enzyme binding, rate of hydrolysis, and lag in
activity. (Adapted from Verger and Pattus [8]).

of enzyme bound decreases with increasing pressure, since less space is available for
its penetration.
   There is clear evidence that the catalytic site is distinct from the lipid binding site
in both the pancreatic and venom phospholipases, as has been demonstrated by
chemical modification of the processed phospholipases and nuclear magnetic relaxa-
tion studies. Compounds such as bromphenacyl bromide that react with His48in the
active site do not prevent binding of the substrate to the processed enzyme. Likewise,
the presence of distinct catalytic and binding sites would account for both the zymo-
gen and the processed enzyme acting on monomeric substrates, while only the proc-
essed enzyme binds and acts at interfaces. The conversion of zymogen to processed
enzyme is thought to result in conformational changes that not only expose more
hydrophobic residues for binding to lipid but also align the active site with the inter-
   A different model is proposed by Dennis [5] based on the finding that the Naja nuja
naja enzyme requires an activator molecule that contains phosphorylcholine. For ex-
ample, sphingomyelin (which is not a substrate for the phospholipase A2) can stim-
ulate the hydrolysis of phosphatidylethanolamine, ordinarily not a good substrate.
Fig. 9 illustrates the proposed mechanism of activation by mixed micelles. Two
monomers of enzyme are shown, one that binds the activator molecule and one that
binds the substrate. Interaction of the two monomers activates the catalytic mono-
mer which promotes hydrolysis. It is possible that the dimeric form, known to exist
for the Crotalus enzymes, could interact with the activator and substrate sequentially
with the same effect. Recent analysis of Naju venom phospholipase A2 crystals indi-

       +@                                                                     &    +


                                                              Substrate      0    Inactive Enzyme
                                                     - -r
                                                      - . g   Products

Fig. 9. Proposed two-site dimer model for Nuju nuju nuju venom phospholipase. The activator monomer
(square) or the inactive catalytic monomer (circle) can bind its phospholipid with subsequent binding of
and interaction with the other monomer activating the catalytic monomer (octagon). As presented, both
the activator and catalytic proteins have bound Ca”. (From Dennis [5]).

cates that only one of the two enzyme molecules of the dimer binds substrate. How-
ever, centrifugation of the enzyme with substrate under conditions where hydrolysis
is measured showed that the dimer of Crotalus phospholipase A, is the active form.
It is possible, therefore, that even though dimers form, only one enzyme molecule is
   The venom and pancreatic phospholipases have an absolute requirement for Ca2',
which is bound adjacent to His48,as shown by Ca2' blockage of the binding of brom-
phenacyl bromide to this residue. The binding of Ca2+lowers the pK of the essential
His4*from 7 to 5.7.
   A proton relay system that employs a molecule of water as the nucleophile attack-
ing the ester bond has been proposed as the mechanism of catalysis. A serine is not
involved, as is sometimes the case with esterases. The Aspg9-His4*     pair removes a
proton from bound water, producing the nucleophilic hydroxyl group (Fig. 10) [6].
Ca2' interacts with both the phosphate and the carbonyl groups of the ester under-
going hydrolysis as well as the carboxyl of Asp49and binds the free fatty acid formed
until the fatty acid diffuses from the active site. Studies with H,"O showed that the
enzyme acts through an 0-acyl cleavage mechanism. The availability of pancreatic
phospholipase A, mutated at position 69 (Tyr + Phe or Lys) indicates that this resi-
due determines the stereospecificity of the enzyme through its interaction with the



       4            I

R,-C-O-CH~(               'cH,-o-~-o-x

              r f
       L                       OH    Rp

                          = '7\2= _ -@-
                          2                         -root

   0            H\.   C
                      '   (O               0-
   II                                           I
R2-C    -0-   CH2         \CH2      - 0-    P-0-X

Fig. 10. Proposed proton-relay mechanism of hydrolysis by venom and pancreatic phospholipases. (From
Verheij et al. [6]).

Fig. 11. Schematic representation of the proposed relative positions of the Ca2' ion, residue 69 of wild-type
and mutant phospholipase At, and the phosphate moiety of lecithin. (A) Interaction of lecithin (left) with
wild-type phospholipase A2. (B) Interaction of lecithin with Phe69mutant phospholipase A,. The charges
of the Ca2+ion and of the phosphate are not indicated for reasons of clarity. X stands for choline. (From
Kuipers et al. [21]).

phosphate moiety of the substrate. Substrate analogs used with the mutant phospho-
lipases A2 reinforce the essentiality of Ca2+in catalysis, either by stabilization of the
enzyme substrate complex or by polarization of the carbonyl group as depicted in
Fig. 11, or both [21].
   Cellular phospholipases A, from several mammalian sources have now been puri-
fied to homogeneity and their amino acid sequence established sufficiently to classify
them in either Group I or Group 11. The very low copy number of these enzymes in
mammalian tissues presented a formidable task; in some cases the enzyme had to be
purified over 1,000,000-fold [22]. These enzymes, like the pancreatic and venom en-
zymes, have seven disulfide bonds that provide considerable stability, including re-
sistance to acid. Consequently, many investigators have treated cell extracts with acid
(pH 1) as an initial step in purification [23]. This procedure undoubtedly selects for
this class of phospholipases A2. Enzymes purified thus far fall into either Group I or
Group I1 (Table I). While most fit into the non-pancreatic Group 11, examples of
Group I exist. It is of interest to note that a single tissue, spleen, contains both
groups, although the cellular origin may differ. It is not yet established if Group I and
Group I1 phospholipases play a role in the signal-couplingprocesses that lead to lipid
mediator formation.
   The structure of the amino terminus of these phospholipases A, is important in
determining the ability of the enzyme to interact with a 'helper' protein in digestion
of bacteria [24]. Those enzymes with a large cluster of basic residues that align on one
side of the N-terminal helical region can interact with the bacterial permeability in-
creasing protein from leukocytes and degrade bacterial lipids, an interaction that fa-
cilitates bacterial killing. It appears therefore that the variable regions of phospholi-
pases A2 will account for the different physiologic functions of these enzymes.

The NH2-terminal amino acid sequence of selected Group I and Group I1 phospholipases A2 of mamma-
lian origin”


                                      1            10             20

Group I
  Rat spleen (soluble)               AVWQFRNMIKCTIPGSDPLREYNNYGC
  Human lung                         AVWQFRKMIKCVIPGSDPFLEYNNYGC

Group II
    OA synovial fluid                NLVNFHRMIKLTTGKEAALSYGFYGX
    RA synovial fluid                XLVNFHRMIKLTTGXEAALSYGFYGX
    Platelet                         NLVNFHRMIKLTTGKEAAL
    Placenta                         NLVNFHRMIK LTTG
    Spleen                           NLVNFHRMIKLTTGKEAALSYGFYGC

  Other Mammals
    Rat platelet                      SLLEFGQMILFKTGKRADVSYGFYGC
    Rat ascitic fluid                 XLLEFGQMILFKTGKRADVSYGFYGC
    Rat spleen                        XLLEFGQMILFKTGKRADVSYGFYGC
    Rat liver                        DLLEFGQMIL FKTGKRADVSYGFY
    Rabbit platelet                  HLLDFRKMIRYTTGKEATTSYGAYGC
    Rabbit ascitic fluid             HLLDFRKMIRYTTGKEATTSYGAYGC
    Rabbit leukocyte                 ALLDFRKMIRYTTGKEATXSYGAYG
    Pig ileum                        DLLNFRKMIKLKTGKAPVPMYAFYGC

“Adapted from Kuipers et al. [21].
OA, osteoarthritis; RA, rheumatoid arthritis; X, a nonidentified residue.

   Not all cellular phospholipases are in Group I or Group 11. Phospholipase A, that
is 100 kDa has been purified from a cultured macrophage cell line. While little is
known about the structure or function of this type of phospholipase A,, its preference
for substrates with arachidonic acid and its response to very low Ca2+concentrations
      M) may indicate a role in signal transduction [25].

2.4. Phospholipase C

Phospholipases C have been associated with bacteria since the classic demonstration
by Macfarlane and Knight (1941) that the clostridial a-toxin was phospholipase C.
The most common source of the enzyme is from culture filtrates of Clostridiumper-
fringens. This enzyme exhibits microheterogeneity based on electrofocusing, and
each form has a constant ratio of a-toxin and phospholipase C activity. Although the

enzyme has a broad specificity, phosphatidylcholine is the preferred substrate. In
studies to define the effect of substrate charge on enzyme activity, Bangham and
Dawson (1958) found that a slight positive charge favored activity. This finding
would explain, at least in part, why neutral phosphatidylcholine is a better substrate
than the acidic phospholipids.
   The most extensively studied phospholipases C are those from Bacillus cereus.
Three distinct enzymes have now been identified and purified from the culture media
of this organism: one specific for phosphatidylinositol, one having broad specificity
similar to the enzyme from Clostridium perfringens, and a sphingomyelinase. The
enzyme with broad specificity, which is active on phosphatidylcholine, has 2 mol of
Zn2+tightly bound to histidine of the enzyme. Removal of the Zn2' causes the revers-
ible loss of activity. If Zn2' is replaced by Co2+, specificity of the enzyme changes
somewhat; sphingomyelin, not normally degraded, becomes a substrate. This enzyme
is extremely resistant to degradation even though it does not have disulfide groups.
Zn2+apparently maintains the structure of the enzyme rather than being involved in
catalysis directly. The phospholipase C with broad specificity does not appear to be
related closely to the phosphatidylinositol-specific phospholipase C from the same
organism. The latter enzyme lacks Zn2' and its molecular weight is 29 kDa (versus 23
kDa for the former). Although these two enzymes are devoid of toxic activity, the
phosphatidylinositol-specific enzyme does cause the release of alkaline phosphatase
from cellular membranes, resulting in phosphatasemia, the release of alkaline phos-
phatase into circulation.
   Phospholipase C is also found in a wide range of mammalian tissues. One of the
earliest reports of a mammalian phospholipase C came from Sloane-Stanley (1953),
who demonstrated the release of inositol from phosphatidylinositol, catalyzed by a
brain preparation. Phospholipases C have now been purified from the cytosolic frac-
tion of muscle, brain, platelets, and ram seminal vesicles. Most of the phospholipases
C that have been purified and cloned at this time are those that act on phosphatidyl-
inositol or its phosphorylated derivatives (see Chapter 7). Mammalian phospholipa-
ses involved in signal transduction are also known that remove the phosphobase
group from phosphatidylcholine. A phospholipase C in the soluble fraction from ly-
sosomes is unusual in that a wide spectrum of phospholipids is attacked. The enzyme
purified from muscle and from platelets hydrolyzes mono-, di-, and triphosphoinosi-
tides, although the diphosphoinositide appears to be the preferred substrate. Most of
the phospholipases C, except the lysosomal one, appear to require Ca2+,but the pH
optimum ranges from 4.5 in the lysosomes to the neutral range for the cytosolic en-
   The properties of phospholipases C purified from diverse sources have been com-
pared and a more systematic nomenclature developed by Rhee et al. [26]. Thus far
five basic types are listed, each with distinct immunologic properties (Table 11) and
show little homology in their predicted amino acid sequences. Also, their catalytic
properties differ with respect to both the Ca2+requirement and substrate specificity.

Isozymes of phospholipase C purified from different tissues

                               Molecular mass                    Nomenclature

Source                         SDS-PAGE         cDNA             Original        Proposed
                               (kDa)            Wa)              report

Rat liver                       68                               I               U

Sheep seminal                   65                               I               U
  vesicular gland             -85                                I1              6 or E

Bovine brain                   150              138.2"           I               8-1
                               140                                               8-2
                               100                                               8-3
                               145               148.4"                          Y
                                85                85.8"                          6

Guinea pig uterus               62               56Ab            I1              U
Bovine brain                   154              138.6            PLC-154         8-1
                                88                                               6

Rat brain                       85                               I1              6
                                85                               111             E

Human platelets                 61                               mPLC-I1

Rat liver                       87

From Rhee et al. [26].
"Calculated molecular mass based on amino acid sequence deduced from rat brain cDNA.
bCalculated molecular mass based on amino acid sequence deduced from rat basophilic leukemic cell
SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; PLC, phospholipase C.

Three of the four most carefully characterized enzymes have two conserved regions
of sequence. However, since the fourth, the a-form, did not have these conserved
regions, either these regions are not involved in catalysis or the mechanism of action
of a is different from the other three. Phospholipase C y has sequence regions that are
homologous to the regulatory domains of tyrosine kinases. The possibility that this
and perhaps other phospholipases C may be regulated by phosphorylation and by
G-proteins seems likely [27]. One of the major tools employed in these studies is the
sensitivity of phosphatidylinositol degradation to pertussis toxin that interacts with
G-proteins. These regulatory actions are both positive and negative depending on the
system studied.
   The major pathway of sphingomyelin degradation involves a special phospholi-
pase C, sphingomydinase. Although some phospholipases C that act on phosphati-
dylcholine also work on sphingomyelin, a number of distinct sphingomyelinasesexist
[2]. Recent interest has focused on sphingomyelinase in the plasma membrane that is
active at neutral pH values that, when coupled to ceramidase action, yields sphingo-
sine, a negative regulator of protein kinase C [28]. The sphingomyelinase of the plas-
ma membrane is distinct from that in the lysosomes since the lysosomal enzyme func-
tions at acid pH values.

2.5. Phospholipase D

Classically, plants have been the major source of the phospholipases D. A number of
phospholipases D exist in bacteria also and some of these have been purified and
cloned. Further, the presence of phospholipase D in mammalian tissues is now well
documented. A rather specialized Mg?+-requiringphospholipase D has been de-
scribed in brain and other tissue. The substrate for the brain enzyme is lysophos-
phatidylcholine or lysophosphatidylethanolamine, with an ether rather than acyl
group at position 1 of the glycerol. This specificity suggests that this phospholipase
D is involved in the catabolism of ether lipids (Wykle et al., 1980) and may be one
mechanism for the catabolism of platelet activating factor.
  The plant phospholipases D, first identified in carrots by Hanahan and Chaikoff
(1947), are found in a wide variety of plant tissue, with Savoy cabbage, Brussels
sprouts, and peanut seeds being the most commonly used sources. The enzyme as
usually isolated has a molecular weight of about 1 15 kDa, which might represent an
aggregate, since under the appropriate conditions a minimum molecular size of 20
kDa is found. For reasons that are not yet clear, rather high concentrations (20-100
mM) of Ca” are required for full activity. Phospholipases D readily hydrolyze the
base group of phospholipids with the sn-1 stereoconfiguration unlike other phos-
pholipases that have little, if any, activity on substrates with sn-1 configuration. The
phospholipases D also have a broad substrate specificity.
   It is known that the enzyme acts by a phosphatidate exchange with a covalent
phosphatidyl-enzyme intermediate. For this reason, the enzyme can catalyze a base-
exchange reaction in which alcohols can substitute for water as the phosphatidate
acceptor. Although ‘base exchange’ is somewhat misleading, and the term ‘trans-
phosphatidylation’ is more accurate, this activity has been used in the laboratory for
the preparation of a variety of phospholipids. For example,

                               - phosphatidylglycerol+choline
    Phosphatidylcholine+glycerol +

  The products of the exchange reaction have the same acyl group composition as
the starting substrate which is useful when a series of phospholipids that differ in the
polar head group only are needed. Alcohols are better than water as phosphatidate
acceptors, since about 1% of an alcohol (for example, glycerol) in water yields an
                                                                                                  29 1

equal mixture of phosphatidic acid and exchanged phospholipid. The presumed
phosphatidate intermediate is a thioester, since activity is inhibited by thiol reagents.
When lysophospholipids are substrates, the enzyme also forms 1-acyl-sn-2,3-phos-
phoglycerol using the hydroxyl at position 2 as its own phosphatidate acceptor.

H,C   -OCR

Structure of 1 -acyl-sn-2,3-phosphoglycerol.

  The function of phospholipase D in plants is not well understood although it may
be involved in cell turnover and energy utilization during different cycles in plant life.
Bacterial phospholipases D in some cases are toxins and can lead to severe cellular
damage either alone or in combination with other proteins secreted from bacteria.
These bacterial enzymes may also serve to provide nutrients for the cell, as do the
bacterial phospholipases C [2]. On the other hand, mammalian phospholipases D
appear to be involved in signal transduction since the action of phospholipase D is
only observed during cell stimulation. The action of phospholipase D is often cou-
pled to a phosphatidate phosphatase that produces diacylglycerol, an activator of
protein kinase C. Experimentally, it is difficult in intact cells to differentiate between
phospholipase D plus phosphatase and phospholipase C (Fig. 12). Billah and cowork-
ers [29] developed an approach to differentiate between these various pathways by
taking advantage of the transphosphatidylation activity of the phospholipase D. In

       H2 C 4 - C H 2 -R   ( 3H)          CHOLINE                     PLD + PA ( 3% 3H)
    OII     I                      +PLD - * PA
R2 4 - 0 4 2 - H
            l        o                    CHOLINE
                                                      C b CH2 OH
                                                                \ PLD + PA-OCH,Cb          ( 32P/ 3H)
        Hp C-0- 32:~-CH2 CH2 N’(CH3)


                                              ( H) DG + choline phosphate ( 32P)
Fig. 12. This scheme depicts how [3H]alkyl-[’2P]phosphatidylcholinecan be used to differentiate between
phospholipase C and phospholipase D action in HL-60 granulocytes. DG, diacylglycerol; PA, phosphatid-
ic acid; PLC, phospholipase C; PLD, phospholipase D. (From Billah et al. [29]).

the presence of ethanol, phosphatidylethanol is formed that is not significantly fur-
ther degraded. Therefore, by the measurement of diacylglycerol and phosphatidyl-
ethanol when cells are incubated in a small percentage of ethanol and a stimulant, the
relative activities of phospholipases C and D are established.
   Phospholipases C and D that act on phosphatidylinositol anchored proteins on the
cell surface are known to be involved in the remodeling of the cell surface as exempli-
fied by Trypanosoma brucei and in metabolism of membrane receptors such as the
insulin receptor [30]. These proteins are anchored to the membrane via phosphatidyl-
inositol that is coupled to the protein linked by a glycan and ethanolamine. While
some phospholipases C that act on phosphatidylinositol, such as that from Staphylo-
coccus aureus, can release these cell surface proteins, distinct phospholipases C that
only function on phosphatidylinositol-glycan-proteins    have been purified (Fox et al.,

2.6. Phospholipases in signal transduction

Throughout this book and especially in the following chapter, the function of lipids
as cellular mediators is described. The action of lipid mediators is exquisitely sensi-
tive and under the rigid control of a variety of anabolic and catabolic enzymes. How-
ever, when this regulation is not balanced, a number of pathologic states such as
inflammation, asthma, and hypertension may result. As seen in Fig. 13, phospholi-
pases play an essential role. Often, the phospholipases initiate a cascade of metabolic
events that lead to the formation of mediator. This figure, while complex, divides
these stimulus response events into two categories: those catalyzed by the phospho-
diesterases at the bottom, and those catalyzed by the acyl hydrolase at the top. Those
at the bottom give rise to events involved in Ca2’ mobilization and protein kinase C
action while those at the top are involved in the ‘arachidonate cascade’ that yields the
eicosanoids and that produce platelet activating factor (Chapter 8). While this is
oversimplified and considerable cross talk between the top and bottom halves of this
scheme exists, this does serve to emphasize the essential role of phospholipases in
cellular regulation.

3. Future directions

Since the first edition of ‘Biochemistry of Lipids and Membranes’ remarkable prog-
ress has occurred in our knowledge of phospholipases. This includes a better under-
standing of the molecular architecture and mechanism of phospholipases as well as
their regulation and function. The exciting examples of our knowledge of the phos-
pholipases A, demonstrate this. The comparison of natural modifications of the en-
zyme between species and the use of protein engineering has set the stage for the
elucidation of both structurexatalysis relationships and protein-lipid interactions.

                      FA          TX      >4     AA                   Lipoxin

  RECEPTORH       GF~*~'*  -   Plf p2 )          PC                   Alkyl-PC


                                          @ PL + TG synthesis 4   '

Fig. 13. The central role for phospholipases in signal transduction. The underlined terms are enzymes;
effectors or responses are in stippled boxes. PG, prostaglandins; TX, thromboxins; CO, cyclooxygenase;
LO, lipoxygenase; HETE, hydroxyeicosatetraenoic acid; LKT, leukotrienes; PAF, platelet activating fac-
tor; PC, phosphatidylcholine; PI, phosphatidylinositol; PA, phosphatidic acid; AA, arachidonic acid; CP,
choline phosphate; DG, diacylglycerol; PL, phospholipase; TG, triacylglycerol; PKC, protein kinase C;
IP,, inositol Tris-phosphate; FA, fatty acid, L.., lyso...

We will see this work become the prototype for understanding how enzymes function
at lipid-water interfaces.
   This, however, is only one class of phospholipase, even within the family of phos-
pholipases A,. A major question that remains in the study of phospholipases A, is
which enzyme(s) is(are) involved in signal transduction and what are the molecular
mechanisms of regulation. An intriguing question exists as to the mechanism of acyl
rearrangement amongst lipids: how much acyl modification is the result of deacyla-
tion-reacylation and how much results from transacylation? Do phospholipases cat-
alyze transacylations in situ? The combined efforts of enzymologists, cell biologists,
biophysicists, and molecular biologists will be required to resolve these problems.
   The other phospholipases are less well understood and, in many cases, no pure
enzyme is available for study. For example, the new interest in phospholipases D
requires that we know something of their structure. Does this enzyme play a role in
signal transduction alone through the formation of phosphatidic acid or must it be
coupled with the phosphatidic acid phosphatase to exert its effect?
   These questions are aimed primarily at the role of mammalian phospholipases yet
these are ubiquitous enzymes. Very little is known of phospholipases in the plant

kingdom. Pragmatically, these enzymes may be involved in cell regulatory events as
found with mammalian enzymes. If so, can plant development be regulated by engi-
neering of phospholipases? The broad approaches of comparing purified phospho-
lipases from a range of sources has yielded valuable information on their molecular
mechanisms and evolutionary relationships. This broad approach will give us new
insights into some of the less well understood systems.


 1. Van den Bosch, H. and Aarsman, A.J. (1979) A review on methods of phospholipase A determination.
    Agents Actions 9,382-389.
 2. Waite, M. (1987) The Phospholipases. In: Handbook of Lipid Research (Hanahan, D.J., Ed.), Vol. 5,
    332 pages, Plenum, New York, NY.
 3. Verger, R. (1980) Enzyme kinetics of lipolysis. Methods Enzymol. 64B, 340-392.
 4. Jain, M.K. and Berg, O.G. (1989) The kinetics of interfacial catalysis by phospholipase A, and regula-
    tion of interfacial activation: hopping versus scooting. Biochim. Biophys. Acta 1002, 127-1 56.
 5. Dennis, E.A. (1983) Phospholipases. In: The Enzymes (Boyer, P., Ed.), Vol. XVI, pp. 307-353, Aca-
    demic Press, New York, NY.
 6. Verheij, H.M., Slotboom, A.J. and DeHaas, G.H. (1981) Structure and function of phospholipase AI.
    Rev. Physiol. Biochem. Pharmacol. 91,91-203.
 7. Dawson, R.M.C. (1966) The metabolism of animal phospholipids and their turnover in cell mem-
    branes. In: Essays in Biochemistry (Campbell, P.N. and Greville, G.D., Eds.), Vol. 2, pp. 69-1 15,
    Academic Press, New York, NY.
 8. Verger, R. and Pattus, F. (1982) Lipid-protein interactions in monolayers. Chem. Phys. Lipids 30,
 9. Demel, R.A., Geurts Van Kessel, W.S.M., Zwall, R.F.A., Roelofsen, B. and Van Deenen, L.L.M.
    (1981) Relation between various phospholipase actions on human red cell membranes and the interfa-
    cial phospholipid pressure in monolayers. Biochim. Biophys. Acta 406,97-107.
10. Hendrickson, H.S. and Dennis, E.A. (1984) Kinetic analysis of the dual phospholipid model for phos-
    pholipase A, action. J. Biol. Chem. 259, 57345739.
11. Van den Bosch, H. (1980) Intracellular phospholipases A. Biochim. Biophys. Acta 604, 191-246.
12. Robinson, M. and Waite, M. (1983) Physical-chemical requirements for the catalysis of substrates by
    lysosomal phospholipase A,. J. Biol. Chem. 258, 14371-14378.
13. Van 't Hooft, F.M., Van Gent, T. and Van Tol, A. (1981) Turnover and uptake by organs of radioac-
    tive serum high-density lipoprotein cholesteryl esters and phospholipids in the rat in vivo. Biochem. J.
14. Thuren, T., Wilcox, R.W., Sisson, P. and Waite, M. (1991) Hepatic lipase hydrolysis of lipid monolay-
    ers: regulation by apolipoproteins. J. Biol. Chem., 266,48534861.
15. Waite, M. (1991) Phospholipases, enzymes that share a substrate class. In: Advances in Experimental
    Medicine and Biology (Mukherjee, A.B., Ed.), Plenum Press, New York, NY, Vol. 279, pp. 1-22.
16. Kawasaki, N. and Saito, K. (1973) Purification and some properties of lysophospholipase from Peni-
    cillium notatum. Biochim. Biophys. Acta 296,426-430.
17. Van den Bosch, H. (1982) Phospholipases. In: Phospholipids (Hawthorne, J.N. and Ansell, G.B.,
    Eds.), Vol. 4, pp. 313-357, Elsevier, Amsterdam.
18. Kyes, P. (1902) Ueber die Wirkungsweise des Cobragiftes, 1,II. Berlin Klin. Wochenschr. 39,889-890,

19. Davidson, F.F. and Dennis, E.A. (1990) Evolutionary relationships and implications for the regula-
    tion of phospholipase A, from snake venom to human secreted forms. J. Mol. Evol. 31,228-238.
20. Verger, R., Mieras, M.C.E. and DeHaas, G.H. (1973) Action of phospholipase A at interfaces. J. Biol.
    Chem. 248,40234034.
21. Kuipers, O.P., Dekker, N., Verheij, H.M. and DeHaas, G.H. (1990) Activities of native and Tyr69
    mutant phospholipases A, on phospholipid analogues. A reevaluation of the minimal substrate re-
    quirements. Biochemistry 29, 60946102.
22. Kramer, R.M., Johansen, B., Hession, C. and Pepinsky, R.B. (1990) Structure and properties of a
    secretable phospholipase A, from human platelets. In: Phospholipase A,: Role and Function in In-
    flammation (Wong, P.Y.-K. and Dennis, E.A., Eds.), Vol. 275, pp. 35-53, Plenum, New York, NY.
23. Elsbach, P., Weiss, J., Franson, R.C., Beckerdite-Quagliata, S., Schneider, A. and Harris, L. (1979)
    Separation and purification of a potent bactericidaVpermeabilityincreasing protein and a closely asso-
    ciated phospholipase A, from rabbit polymorphonuclear leukocytes. J. Biol. Chem. 254, 11000-1 1009.
24. Weiss, J. and Wright, G. (1990) Mobilization and function of extracellular phospholipase A, in inflam-
    mation. In: Phospholipase A,: Role and Function in Inflammation (Wong, P.Y.-K. and Dennis, E.A.,
    Eds.), Vol. 275, pp. 103-1 13, Plenum Press, New York, NY.
25. Clark, J.D., Milona, N. and Knopf, J.L. (1990) Purification of a 110-kilodalton cytosolic phospholi-
    pase A, from the human monocytic cell line U937. Proc. Natl. Acad. Sci. U S A . 87,7708-7712.
26. Rhee, S.G., Suh, P.-G., Ryu, S.-H. and Lee, S.Y. (1989) Studies of inositol phospholipid-specific phos-
    pholipase C. Science 244, 546550.
27. Harden, T.K. (1 990) G-protein-dependent regulation of phospholipase C by cell surface receptors.
    Am. Rev. Respir. Dis. 141, S119-Sl22.
28. Merrill, A.H. (1989) Lipid modulators ofcell function. Nutr. Rev. 47, 161-169.
29. Pai, J.-K., Siege], M.I., Egan, R.W. and Billah, M. (1988) Phospholipase D catalyzes phospholipid
    metabolism in chemotactic peptide-stimulated HL-60 granulocytes. J. Biol. Chem. 263, 12472-12477.
30. Low, M.G. and Saltiel, A.R. (1988) Structural and functional roles of glycosylphosphatidylinositol in
    membranes. Science 239,268-275.
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemistry o Lipids, Lipoproteins and Membranes
8 1991 Elsevier Science Publishers B.V. All rights reserved.                                                                297

                                                                                                             CHAPTER 10

                      The eicosanoids: cyclooxygenase,
               lipoxygenase, and epoxygenase pathways
                                    WILLIAM L. SMITH”, PIERRE BORGEATband
                                                    FRANK A. FITZPATRICK“

   aDepartment o Biochemistry, Michigan State University, East Lansing, MI, 48824, U.S.A.,
      bLe Centre Hospitalier de I’Universite Laval, Laurier, Que., Canada and Department of
    Pharmacology, University o Colorado Health Sciences Center, Denver, CO, 80262, U.S.A ,

1. Introduction

The term ‘eicosanoids’ is used to denote a group of oxygenated, twenty carbon fatty
acids (Fig. 1; [ 1,2]).The major precursor of these compounds is arachidonic acid (all
cis 5,8,11,lCeicosatetraenoic acid), and the pathways leading to the eicosanoids are
known collectively as the ‘arachidonate cascade.’ There are three major pathways
within the cascade, including the cyclooxygenase, lipoxygenase, and epoxygenase
pathways. In each case, these pathways are named after the enzyme that catalyzes the
first committed step. The prostanoids, which include the prostaglandins and throm-
boxanes, are formed via the cyclooxygenase pathway. The first part of our discussion
will focus on the prostanoids. Later in this chapter, we will describe the lipoxygenase
and epoxygenase pathways.

                                                       5-,12-,or 15-
                                                                IHpETEs)                                           1

                                                                              +NASE               5 . 6 - , 8.9- ,11,12-, or 14,15-

                                  HYDROXY EICOSATETRAENOIC                       DIHYDROXY        EPOXYEICOSATRIENOIC ACIDS
                                                                                                     (EpETrEs lor EETs))
                                     ACIDS (HETEs)                             EICOSATETRAENOIC
                                                                               ACID (DIHETEsI
   (                                )
        PGD2 ,PGE2,PGF2.PGi2 ,TxAZ

                                                            LEUKOTRIENE A 4
                                      LEUKOTRIENES                           LEUKOTRIENE Eq
                                    (LTC4, LTD4, LTE4 1

Fig. 1 . Pathways leading to the formation of eicosanoids from arachidonicacid. From [2], with permission.

                \                                                  CELL SURFACE

                     \ PHOSPHOLIPASE   PHOSPHOLIPID

                                       ARACHIDONIC ACID
             CYCCOOXYGENASE                                            SYNTHASE

             PEROXI DASE




                                                                       bH   TxAp
     PGEp      6H

                       Ho -o HI3
                               PGF2,      bH

?ig. 2. Structures and biosynthetic relationships among prostanoids.

2. Prostanoids

2.1. Structures and nomenclature

The structures and biosynthetic relationships of the most important prostanoids are
shown in Fig. 2 [1,2]. PG is the abbreviation for prostaglandin, and Tx (or TX) is the
abbreviation for thromboxane. Naturally occurring prostaglandins contain a cyclo-
pentane ring, a trans double bond between C-13 and (2-14, and an hydroxyl group at
C-15. The letters following the abbreviation PG indicate the nature and location of

the oxygen-containing substituents present in the cyclopentane ring. Letters are also
used to label thromboxane derivatives (e.g. TxA and TxB). The numerical subscripts
indicate the number of carbon-carbon double bonds present in the side chains
emanating from the cyclopentane ring (e.g. PGE, vs. PGE,). In general, those pros-
tanoids with the ‘2’ subscript are derived from arachidonate; the ‘1’ series prosta-
noids are formed from 8,11,14-eicosatetraenoate, the ‘3’ series compounds are
derived from 5,8,11,14,17-eicosapentaenoate.   Greek subscripts are used to denote
the orientation of ring hydroxyl groups (e.g. PGF,).

2.2. Prostanoid chemistry

All the prostaglandins are soluble in lipid solvents below pH 3.0, and are typically
extracted from acidified aqueous solutions with ether, chlorofodmethanol, or ethyl
acetate. PGE, PGF, and PGD derivatives are relatively stable in aqueous solution at
pH range 4-9, but above pH 10, both PGE and PGD are subject to dehydration.
PGI,, which is also known as prostacyclin, contains an enol-ether sensitive to acid-
catalyzed hydrolysis; PGI, is unstable below pH 8.0. The stable hydrolysis product of
PGI, is 6-keto-PGFI,. PGI, formation is usually monitored by measuring 6-keto-
PGF,, formation. TxA,, which contains an oxane-oxetane grouping in place of the
cyclopentane ring, is hydrolyzed rapidly (t4=30 s at 37°C in neutral aqueous solu-
tion) to TxB,.

2.3. Prostanoid biosynthesis

Unlike many hormones, eicosanoids are not stored by cells, but rather are syn-
thesized and released rapidly (within 5-60 s) in response to extracellular stimuli. The
pathway for stimulus-induced prostanoid formation as it might occur in a model cell
is illustrated in Fig. 2 [1,2]. Prostanoid formation occurs in three stages: (a) release of
arachidonate from membrane phospholipids; (b) conversion of arachidonate to the
prostaglandin endoperoxide PGH,; and (c) cell-specificconversion of PGH, to one of
the major prostanoids.

2.4. Arachidonate release

Prostaglandin synthesis is initiated by the interaction of a stimulus with the cell
surface. Depending on the cell type, the stimulus can take the form of a hormone such
as bradykinin, angiotensin 11, or antidiuretic hormone, and/or a protease such as
thrombin. Presumably, these agents operate through specific cell surface receptors or
target proteins. Interaction of a stimulus with the cell results in the activation of one
or more cellular phospholipases. The steps which intervene between the stimulus and
phospholipase activation have not been resolved. However, in all cases, phos-
pholipase activation either requires or is accompanied by an increase in intracellular

[Ca,']. Importantly, release of arachidonate is selective. Other fatty acids normally
found in phospholipids are not mobilized [3,4].
   That cell which has been studied most extensively with respect to arachidonate
release is the human platelet. When platelets are stimulated with thrombin, about
15% of the total esterified arachidonate pool of 100 nmol/109 cells is mobilized.
Released arachidonate comes from two glycerophospholipid pools, phosphatidylino-
sitol (PtdIns) and phosphatidylcholine (PtdCho) [4].
   In the case of platelet PtdIns, release of arachidonate involves the sequential ac-
tions of a PtdIns-specific phospholipase C (PLC), a diglyceride lipase, and a mono-
glyceride lipase. Because PtdIns is highly enriched at the sn-2 position with arachi-
donate, this PLC pathway provides for a relatively selective release of arachidonate.
About half of the arachidonate released by platelets arises from hydrolysis of PtdIns
   Thrombin treatment of platelets also causes arachidonate to be released from
PtdCho via a phospholipase A, (PLA,). Both soluble and membrane-bound PLA,
activities have been described. A cDNA for a membrane-bound platelet PLA, has
been cloned [5].The molecular weight is about 14 kDa, and the deduced amino acid
sequence resembles that of secreted venom PLA,s. The role of this membrane-bound
PLA, in thrombin-induced arachidonate release is not known. A soluble platelet
PLA, with a molecular weight of about 60 kDa becomes membrane-associated when
the intracellular [Ca2+]  increases to about 250 nM [6]. A [Ca"] of this magnitude is
achieved when platelets are stimulated with agonists such as thrombin. The soluble
platelet PLA, is somewhat selective for plasmenylphosphatidylcholine (Chapter 8).
Although the question has not been investigated with the platelet enzyme, a soluble
PLA, isolated from the RAW 264.7 macrophage cell line causes the selective release
of arachidonate from the sn-2 position of phospholipids.
   It is clear that in platelets arachidonate can be mobilized from both PtdIns and
PtdCho and that there are phospholipase systems capable of releasing arachidonate
from each of these lipids. Nonetheless, it is still not known which phospholipid and
which phospholipase system is actually involved in the release of that arachidonate
which is converted to TxA,. Clouding the picture is another eicosanoid, a lipoxy-
genase product called 12-HpETE (see below), which is formed in significant
quantities following stimulation of platelets with thrombin. It is likely that one pool
of released arachidonate is used primarily for TxA, synthesis via the cyclooxygenase
pathway while another pool is used for 12-HpETE formation via the 12-lipoxygenase
pathway, but it is unknown which arachidonate pool gives rise to which eicosanoid
product [I].

2.5. Prostaglandin endoperoxide formation

Once arachidonate is released, it can be acted upon by PGH synthase [1,7,8]. PGH
synthase exists in an already active form in the endoplasmic reticulum. This enzyme
                                                                                              30 1

 exhibits two different catalytic activities (Fig. 2): (a) a cyclooxygenase (bis-oxygen-
 ase) which catalyzes the formation of PGGz from arachidonate and (b) a peroxidase
 (or hydroperoxidase) which facilitates the reduction of PGG, to PGH2.As discussed
 below, these activities occur at distinct, but interactive, sites within the protein. PGH
 synthase has several unusual and intriguing properties. First, a hydroperoxide is re-
 quired to activate the enzyme; second, the enzyme undergoes a ‘suicide’ inactivation;
 and third, the enzyme is the target of nonsteroidal anti-inflammatory drugs. We will
 discuss each of these properties in the context of PGH, synthesis.
    The initial step in the action of the cyclooxygenase is the stereospecific removal of
 the 13-pro-S hydrogen from arachidonate. As depicted in Fig. 3 [1,7], the enzyme is
 thought to orient an arachidonate molecule by inducing a kink in the carbon chain at
 C-10. Abstraction of the 13-pro-S hydrogen and subsequent isomerization leads to a
 carbon-centered radical at C-11 and attack of molecular oxygen at C-11 from the
 solvent side. The resulting 1 1-hydroperoxyl radical adds to the double bond at C-9
 leading to intramolecular rearrangement and formation of another carbon-centered
 radical. This radical reacts with another molecule of oxygen at C-15 with the forma-
 tion of PGG,. Newly formed PGG2 can undergo a two electron reduction to PGHz
 catalyzed by the peroxidase activity of PGH synthase.
    In order for the cyclooxygenase to function, a hydroperoxide is required. As dis-
 cussed further below, the hydroperoxide oxidizes a heme prosthetic group at the per-
 oxidase active site of PGH synthase. This, in turn, leads to the oxidation of a protein

10        H%.. H

 k                                    0

          loo“                  -H’                                     +02


     0.                                   0      -       -      T   W

     ‘J      - jCOOH
           jCOOH                          ‘O*”

                   /                                 /
                   COOH                              COOH

 Fig. 3. Model for the mechanism of the cyclooxygenase reactioc. From [I], with permission.

tyrosine residue producing a tyrosine radical which is apparently involved in the
abstraction of the hydrogen from the 13-pro-S position of arachidonate.
   The cyclooxygenase is inactivated during catalysis as the result of a nonproductive
breakdown of an active enzyme intermediate. On average, one cyclooxygenase active
site is lost per 1400 catalytic turnovers. The chemical change in the protein that ac-
companies this suicide inactivation is unknown. However, the suicide process is a
crude regulatory mechanism which places an upper limit on cellular prostaglandin
biosynthetic activity.

2.6. Physico-chemical properties of PGH synthase

PGH synthase is a membrane-associatedprotein with a subunit molecular weight of
72 kDa. The detergent-solubilized enzyme exists as a homodimer. The sequences of
cDNA clones for PGH synthases from sheep, mouse, and human sources indicate
that the protein initially contains a signal peptide of 24-26 amino acids which in all
species is cleaved to yield a mature protein of 576 amino acids [7]. The presence of the
signal peptide suggests that the enzyme traverses the endoplasmic reticulum during
its synthesis, but it is unclear how the mature PGH synthase is associated with the
membrane since it contains no obvious transmembrane sequences. The native en-
zyme is also glycosylated at asparagine residues 68, 114, and 410 [7]. PGH synthase
is a hemoprotein containing one protoporphyrin IX per subunit. There is immu-
nochemical evidence that both the cyclooxygenase and peroxidase active sites of
PGH synthase are on the cytoplasmic surface of the endoplasmic reticulum.

2.7. PGH synthase and nonsteroidal anti-inflammatory drugs

Prostaglandin synthesis can be inhibited by both nonsteroidal anti-inflammatory
drugs and anti-inflammatory steroids. The best known of the nonsteroidal anti-in-
flammatory drugs is aspirin. Aspirin competes with arachidonate for binding to the
cyclooxygenase active site, but the binding of arachidonate is about 10,000 times
more efficient than that of aspirin. However, once bound, aspirin can acetylate a
specific serine residue of PGH synthase: Ser530   (Fig. 4). Acetylation of SerS3'causes
irreversible cyclooxygenase inactivation. It was initially thought that the hydroxyl
group of Sers3* required for catalysis, However, studies using site-directed muta-
genesis have shown that replacement of SerS3'with an alanine residue yields an active
enzyme, whereas replacement with an asparagine, which is about the same size as an
acetylated serine, yields an inactive enzyme. Apparently, acetylation of SeP' by
aspirin results in steric hindrance at this position; the bulky acetyl group protrudes
into the cyclooxygenase active site and prevents arachidonate from binding [1,7].
   Acetylation of PGH synthase by aspirin has important pharmacological conse-
quences. Besides the analgesic, anti-pyretic, and anti-inflammatory actions of
aspirin, low dose aspirin treatment (one 'baby' aspirin daily or one regular aspirin

+fl-!CH3       aspirin          -                       + F salicylic acid


           530SER-OH                                 530SER-O-CCH3

                  COOH                                       COOH

            PGH synthose                              PGH synihase
             (active)                                  ( inacii v e )

Fig. 4. Acetylation of Ser’”’ of PGH synthase by aspirin.

every three days) is a useful anti-platelet cardiovascular therapy [9]. This low-dosage
regimen leads to selective inhibition of platelet thromboxane formation (and platelet
aggregation) without appreciably affecting the synthesis of other prostanoids in other
cells. Circulating anucleate blood platelets, unlike most other cells, are unable to syn-
thesize new protein. Exposure of the PGH synthase of platelets to aspirin causes irre-
versible inactivation of the platelet enzyme. Of course, PGH synthase inactivation
also occurs in other cell types, but cell types other than platelets can resynthesize
PGH synthase protein relatively quickly. For new PGH synthase activity to appear
in platelets, new platelets must be formed. Since the replacement time for platelets is
5-10 days, it takes a considerable amount of time for the circulating platelet pool to
regain its original complement of active PGH synthase.
   There are many nonsteroidal anti-inflammatory drugs. In fact, this is by far the
largest niche in the pharmaceutical market, currently accounting for 3-5 billion dol-
lars in annual sales in the United States. Most other nonsteroidal anti-inflammatory
drugs also act by inhibiting the cyclooxygenase activity of PGH synthase. However,
unlike aspirin, most of these drugs cause reversible enzyme inhibition by competing
with arachidonate for binding. A well-known example of a reversible nonsteroidal
anti-inflammatory drug is ibuprofen (Advil).

2.8. Anti-inflammatory steroids

For many years, it was thought that anti-inflammatory steroids attenuated pros-
tanoid synthesis by inhibiting stimulus-induced arachidonate release. More recent

evidence suggests that anti-inflammatory steroids may function principally by inhib-
iting transcription of the PGH synthase gene. Presumably, these steroids interact
with a receptor which binds to a negative regulatory element in the promoter region
of the PGH synthase gene [8].

2.9. PGH synthase active site

Depicted in Fig. 5 is a model of the PGH synthase active site [ 1,7]. A heme group is
shown interacting on one side (the distal position) with an alkyl peroxide and on the
other side (the axial position) with a histidine (His309).A hydroperoxide binds to
the heme oxidizing it to an 0x0-ferry1 oxidized intermediate. This intermediate under-
goes a rearrangement in which an electron from Tyr3” is transferred to the oxo-
ferry1 group [1,7,10]. The resulting Tyr3*’ radical then abstracts the 13-pro-S
hydrogen from arachidonate bound at the cyclooxygenase active site, thereby ini-
tiating the cyclooxygenase reaction. Sers3’ is shown neighboring the cyclooxygenase
active site.

2.10. Regulation of PGH synthase gene expression

The regulation of PGH synthase gene expression is currently receiving considerable
experimental attention. There are at least two general levels of regulation [1,8]. The
first is developmental, and an example is the induction of PGH synthase in ovarian
follicles by luteinizing hormone (LH). Ovaries contain little or no detectable PGH
synthase until exposed to circulating LH; the levels of PGH synthase mRNA and
protein then increase dramatically. The molecular basis for these changes is un-

Fig. 5 . Model of the active site of PGH synthase.

known. A second type of regulation can be termed ‘tuning’. In contrast to the case of
ovarian follicles, human vascular endothelial cells and mouse 3T3 fibroblasts
normally express PGH synthase at constitutively high levels. However, when growth
factors such as platelet-derived growth factor (PDGF) or interleukins such as IL-1
are added to cultures of these cells, there is a further significant increase in enzyme
activity and gene transcription [8]. There is evidence that activation of protein kinase
C is a prerequisite for increases in PGH synthase gene expression in fibroblasts.
Responses to PDGF and IL-1 are relatively rapid. Increases in PGH synthase activ-
ity and mRNA levels occur within 2-3 h. In addition, PGH synthase is ‘super-
induced’ by cycloheximide [8]. This latter property is common to a class of proteins
encoded by ‘immediate early’ genes. The best-known immediate early genes are c-fos
and c-myc. Like the expression of c-fos and c-myc, PGH synthase expression may be
an important early event in cell growth.
   Recently, the intron/exon structures of both the mouse and human PGH synthase
genes have been described. The genes are quite similar. There are 11 exons spanning
about 25 kb of genomic DNA [8].

2.11. PGH, metabolism

Although all the major prostanoids are depicted in Fig. 2 as being formed by a single
cell, prostanoid synthesis appears to be cell specific [I]. For example, platelets form
mainly TxA,, endothelial cells form PGI, as their major prostanoid, and PGE, is the
major prostanoid produced by renal collecting tubule cells. The formation of PGE,,
PGD,, PGF,,, PGI,, and TxA, from PGH, is catalyzed, respectively, by PGE
synthase, PGD synthase, PGF, synthase, PGI synthase, and TxA synthase. Synthesis
of PGF,, involves a net two electron reduction of PGH,, and a PGF, synthase uti-
lizing NADPH can catalyze this reaction. All other prostanoids are formed via iso-
merization reactions involving no net change in the oxidation state of PGH,.
   PGI synthase and TxA synthase are hemoproteins with molecular weights of 50-
55 kDa. Ullrich has presented evidence suggesting that these proteins are cytochrome
P450s. Both enzymes, like PGH synthase, undergo suicide inactivation during cat-
alysis. TxA synthase is found in abundance in platelets and lung. PGI synthase is
localized to endothelial cells and vascular and nonvascular smooth muscle.
   PGE synthase activities are present in several different tissues, but there are dif-
ferences among these proteins from different tissues. PGE synthases are unique in
that each requires reduced glutathione (GSH) as a cofactor. GSH appears to facili-
tate cleavage of the endoperoxide group and formation of the 9-keto group [l].
   PGF, synthase activity has been partially purified from lung. Structurally, the en-
zyme is a member of the aldose reductase family of proteins.
   GSH-dependent and -independent PGD synthases have been isolated. The GSH-
dependent forms also exhibit GSH-S-transferase activity. A GSH-independent form
of PGD synthase has been purified from brain.

2.12. Catabolism: prostanoids as local hormones

Once a prostanoid is formed, it exits the cell, probably via carrier-mediated
transport. Prostanoids are local hormones that act very near their sites of synthesis
[1,2]. Unlike typical circulating hormones which are released from one major tissue
site, prostanoids are synthesized and released by virtually all organs. In addition, all
prostanoids are inactivated rapidly in the circulation. The initial step of inactivation,
depicted in Fig. 6 for PGE,, is oxidation to a relatively inactive 15-keto compound in
a reaction catalyzed by a family of 15-hydroxyprostaglandin dehydrogenases.
Further catabolism involves reduction of the double bond between C-13and (2-14
(Fig. 6), w-oxidation, and /?-oxidation.
   Prostanoids exhibit both autocrine and paracrine activities. That is, their target
sites include both the cell in which they are formed and neighboring, different cell
types. Those examples which have been studied in the most detail are the platelet-
vessel wall interaction involving PGI, and TxA, [9] and the renal collecting tubule-
thick limb interaction involving PGE, synthesized by the collecting tubule [2]. We
will present this latter example as a paradigm for the cellular and molecular actions
of prostanoids. In all cases, prostanoids can be viewed as operating through cell sur-
face receptors which, when occupied, interact with and stimulate guanine nucleotide-
binding regulatory (G) proteins (Fig. 7; [2]). As discussed below, this general model
also applies to lipoxygenase products.
   In the renal collecting tubule, antidiuretic hormone or arginine vasopression
(AVP) acts through a V, receptor coupled to G, and adenylate cyclase (A.C.) to
stimulate cAMP formation (Fig. 7). Increases in cAMP lead to reabsorption of water
from the renal tubule. PGE, is synthesized by collecting tubules in response to ADH
acting through a V, receptor. At low concentrations, PGE, binds and activates an

  &          bH       Po DEHVDROGENASE

           - O
H 0'

             A -REDUCTASE
           .         COOH
                               13.1 4-DIHY DRO-
H 0'                           1CKETO-PGEz


Fig. 6. Catabolism of PGE,.


                                         INTERSTITIUM                                    TUBULAR

Fig, 7. Mechanism of action of PGEz in the renal collecting tubule and thick ascending limb. ADH, anti-
diuretic hormone; A.C., adenylate cyclase. From [2], with permission.

inhibitory receptor (R,) which, acting through a pertussis toxin sensitive G protein,
Gi, attenuates the response of the collecting tubule to ADH. Thus, PGE, is acting as
an autocoid and functioning as a biochemical governor to regulate the ADH
   For water to be reabsorbed from the collecting tubule, there must be an ap-
propriate osmotic gradient. This gradient occurs because the thick ascending limb of
Henle’s loop pumps NaCl (but not water) from the lumen of the thick limb into the
surrounding interstitium. ADH potentiates NaCl reabsorption by the thick limb,
thereby helping to build the gradient for water reabsorption by the collecting tubule.
The ability of ADH to stimulate NaCl reabsorption in the thick limb involves ADH
acting through a Vz receptor to stimulate CAMPformation. And again, PGE, acting
through a receptor, Ri, present on thick limb cells, can attenuate this response. This
is an example of a paracrine activity of PGE,.
   In addition to the G,-linked PGE receptor (Ri), there is also a G,-linked PGE re-
ceptor (R,) coupled to adenylate cyclases of both thick limb and collecting tubule
cells. This latter receptor has a relatively high KDvalue, so it is only occupied at high
concentrations of PGE,. The stimulatory receptor is apparently involved in feedback
inhibition of tubular PGE, formation when PGE, is overproduced.
   There is also another PGE receptor (not depicted in Fig. 7) in the collecting tubule.
Occupancy of this receptor by PGE, causes the mobilization of intracellular Ca2’.
Increases in [Ca”] also attenuate the hydroosmotic effect of ADH.

  Thus, there are three distinct receptors for PGE. These receptors are functionally
and mechanistically homologous to /?-, and a,-adrenergic receptors which operate
through different G proteins to stimulate adenylate cyclase, inhibit adenylate cyclase,
and stimulate Ca” mobilization, respectively. It is quite likely that there is a sub-
family of receptors for each prostanoid [2]. A TxA/PGH receptor from platelets has
been purified [l 11.

3. Hydroxy- and hydroperoxy-eicosaenoic acids and leukotrienes
3.1. Introduction and overview

Plant enzymes called lipoxygenases which catalyze the introduction of oxygen into
polyunsaturated fatty acids were discovered almost four decades ago. It is now
known that mammalian lipoxygenase activities exist which will catalyze the insertion
of oxygen at positions 5, 12, and 15 of various eicosaenoic acids. The immediate
products of the action of lipoxygenases on 20-carbon polyunsaturated fatty acids are
hydroperoxy fatty acids. In the case of arachidonic acid, the products are hydro-
peroxy-eicosatetraenoic acids abbreviated HpETEs.
   HpETEs can subsequently undergo one of several different enzymic transforma-
tions. Fig. 8 shows the reactions observed with the 5-lipoxygenase product 5-HpETE.
One reaction is a two electron reduction of the hydroperoxy group to an alcohol
yielding the corresponding hydroxy-eicosatetraenoic acids (HETEs). A second type
of reaction is lipoxygenation at another position in the aliphatic chain yielding (after

/&COOH                                        A     C    O       O   H

      5s- H ETE                                           ‘0 H
                                                  5S,1 5s-dIHETE

                      LEUKOTRIENE A 4

Fig. 8. Major biosynthetic transformations of 5-hydroperoxy-6.8,I 1,14-eicosatetraenoic acid (5-HpETE).

reduction of the two hydroperoxy groups) dihydroxy-eicosatetraenoic acids
(DiHETEs); double dioxygenation reactions can also lead to the formation of tri-
hydroxy-eicosatetraenoic acids, named lipoxins (not shown in Fig. 8) [12]. Finally, a
third type of transformation of HpETEs is a dehydration to produce an epoxy
(oxido) fatty acid (e.g., leukotriene A,; Fig. 8.
   During the last decade, the 5-lipoxygenase has attracted a great deal of interest
because several products of this pathway, named ‘leukotrienes’ (abbreviated LTs),
have been found to have potent biological activities. The term ‘leukotriene’ denotes
the cells (leukocytes) originally recognized to form these products and a structural
characteristic of these compounds, the conjugated triene unit. To date, 5-lipoxy-
genase activity has been documented in neutrophils, eosinophils, basophils, mono-
cytes, macrophages and mast cells from several species, including man.
   As noted earlier, the 12-lipoxygenase producing 12-HpETE is an important en-
zyme in human platelets; it is also present in phagocytes such as mouse peritoneal
macrophages and in porcine neutrophils, but not in human phagocytes. The 15-lip-
oxygenase is a very active pathway in human eosinophils and reticulocytes; it is also
an important pathway of arachidonic acid metabolism in human lungs and porcine
neutrophils. The biological significance of the 12- and 15-lipoxygenasesremains to be
defined; in this regard, recent studies have established that products derived from
both pathways, in particular 12-HETE, 15-HETE, hepoxilins, and lipoxins show
biological activities in various systems [12,13].
   Given the biological importance of the 5-lipoxygenase products, our discussion
will focus on the metabolism of eicosaenoic acids (mainly arachidonic acid) by the
neutrophil 5-lipoxygenase. Among the various cells capable of producing leuko-
trienes, the human blood neutrophil is the cell type that has been most extensively

3.2. Mechanism of leukotriene biosynthesis in human neutrophils

The arachidonate 5-lipoxygenase catalyzes the specific dioxygenation of arachidonic
acid at position C-5 and initiates the synthesis of leukotrienes (Fig. 9) [14]. For ease
of nomenclature in discussing lipoxygenase pathways, we have used the E-Z system
to denote cis (Z) and trans (E) carbon-carbon double bonds. Arachidonic acid is first
transformed into (5S)-5-hydroperoxy-(E,Z,Z,Z)-6,8,11,14-eicosatetraenoic (5-    acid
HpETE). This is a simple dioxygenase reaction; there is no net oxidation-reduction
of either the fatty acid or oxygen. A cis-trans conjugated diene is formed in the reac-
tion. Neutrophils rapidly metabolize 5-HpETE into (SS)-5-hydroxy-
(E,Z,Z,Z)6,8,11,1Ceicosatetraenoic (5-HETE) (Fig. 8) and (5S,6S)-5(6)-oxido-
(E,E,Z,Z)7,9,11,1Ceicosatetraenoic (leukotriene A, or LTA,) (Fig. 9). The en-
zyme responsible for the reduction of 5-HpETE to 5-HETE has not been character-
ized. In naming leukotrienes such as LTA,, the letter A indicates the nature and posi-
tion of the oxygen-containing substituent, and the numerical subscript indicates the


 '     "H_
 ARACHIDONIC ACID                                   5s-HpETE

       5-LIPOXYGENASE                       H20 4 1 - L I P O X Y G E N A S E

      5s-HpETE                                 LEUKOTRIENE A 4

Fig. 9. Mechanisms of the reactions catalyzed by the 5-lipoxygenase.

number of double bonds in the molecule. This nomenclature is analogous to that
used for prostanoids.
   The formation of LTA, from 5-HpETE is also catalyzed by the 5-lipoxygenase and
involves the stereospecific removal of the 10(R) hydrogen atom of 5-HpETE and loss
of water (Fig. 9). Newly formed LTA, is converted rapidly by both enzymic and
non-enzymic reactions into various more polar compounds. As expected for an al-
lylic epoxide, LTA, is highly unstable and undergoes facile nucleophilic substitution
(Fig. 10); in aqueous buffer at pH 7.4 and 25"C, the time for 50% decomposition is

                             LEUKOTRIENE A 4

                             2'/NON   ENZY M I C   v 2'

             5.1 2-DiHETEs                           5,6-DIHETES

Fig. 10. Formation of a carboniurn-ion intermediate resulting from opening the 5.6-epoxide ring of LTA,
and formation of products of non-enzymic hydrolysis.
                                                                             31 1

less than 10 s, and LTA, instantaneously hydrolyzes at acidic pH; LTA, is stabilized
under alkaline conditions and upon binding to serum albumin.
   The (5S,                                                            acid
             12R)-5,12-dihydroxy-(E,E,E,Z)-6,8,10,14-eicosatetraenoic (6-trans-
LTB,) and (5S,    12S)-5,12-dihydroxy-(E,E,E,Z)-6,8,10,14-eicosatetraenoic   acid (6-
trans-12-epi-LTB4) are products of the non-enzymic hydrolysis of LTA,, which
occurs spontaneously in aqueous medium (Fig. 10). Nonenzymic hydrolysis of LTA,
also results in the formation of two isomeric 5,6-dihydroxy-7,9,11,14-eicosate-
traenoic acids (5,6-DiHETEs). The mechanism of formation of these dihydroxy acids
is similar to that of the 5,12-dihydroxy acids with reaction at position C-6 of the
carbonium ion intermediate. The structural identification of these four isomeric di-
hydroxy acids in initial studies on the leukotriene pathway was a key step in the dis-
covery of LTA,.
   The DiHETEs derived from non-enzymic hydrolysis of LTA, are almost devoid of
biological activity; however, LTA, also undergoes enzymic conversions leading to
various compounds with potent biological properties (Fig. 11); (5S, 12R)-5,12-dihy-
droxy-(Z,E,E,Z)-6,8,10,14-eicosatetraenoic (LTB,) is the product of the enzymic
hydrolysis of LTA, catalyzed by a specific enzyme, LTA, hydrolase. For example,
LTB, is formed by human neutrophils incubated with the CaZ+    ionophore A23187 or
with specific neutrophil stimuli such as the formylated chemotactic peptides. The
direct addition of synthetic LTA, to a suspension of human leukocytes also results in
LTB, synthesis. A second pathway for LTA, metabolism (not present in neutrophils)
involves another specific enzyme, the LTA,-GSH (GSH) transferase or LTC,
synthase; (5S,6R)-5-hydroxy-6-S-glutathionyl-(E,E,Z,Z)-7,9,11,14-eicosatetraenoic
acid (LTC,) is the product of the conjugation of GSH and LTA, (Fig. 11).
   LTC, is formed in human polymorphonuclear leukocytes (PMNLs), mainly in eo-
sinophils (see below) stimulated with the calcium ionophore A23187 (or other agents)
or incubated directly with LTA,. Obviously, the formation of LTB, or LTC, in cell

                           LTC4 SYNTHASE        OH
                                                L vCOOH

Fig. 1 I . Enzymic transformations of LTA, to LTB4 and LTC,.

types containing the 5-lipoxygenase will depend on the presence of LTA, hydrolase
or LTC, synthase. Human neutrophils and alveolar macrophages, which contain
LTA, hydrolase, produce mainly LTB,, while eosinophils, which contain LTC,
synthase, produce only LTC,. Other cell types such as blood monocytes, peritoneal
macrophages and mast cells produce both LTB, and LTC,. Furthermore, several cell
types that do not contain the 5-lipoxygenase, but contain either LTA, hydrolase
(such as erythrocytes and lymphocytes) or LTC, synthase (such as platelets and en-
dothelial cells) contribute to the generation of leukotrienes through transcellular me-
tabolism of LTA,.

3.3. The enzymes of the 5-lipoxygenasepathway

Arachidonic acid is not the exclusive substrate of the 5-lipoxygenase; the enzyme will
also transform other fatty acids having methylene interrupted cis double bonds at
C-5 and C-8. Accordingly, 5,8,11,14,17-eicosapentaenoic         acid (EPA), 5,8,1 l-eicosa-
trienoic acid, and even hydroxy fatty acids such as 12-HETE and 15-HETE are effi-
ciently converted by the 5-lipoxygenase into HETEs or DiHETEs. The arachidonate
5-lipoxygenase has now been isolated from human neutrophils and other cells of
myeloid lineage [15]. The human neutrophil enzyme is a cytosolic protein with an
apparent molecular weight of about 80 kDa. The purified enzyme requires Ca2+,
ATP, a fatty acid hydroperoxide, PtdCho, and the presence of several uncharac-
terized cytosolic and membrane-bound factors for full activity. The 5-lipoxygenase
and the LTA, synthase activities reside in a single protein as indicated by the copu-
rification of the two activities, parallel suicide inactivation, the same stability, and the
same requirements for activation.
   The cDNA coding for the human 5-lipoxygenase has been cloned from human
placental and human leukemia (HL)-60 cell cDNA libraries. The DNA sequence en-
codes a protein of 673 amino acids (78 kDa). No consensus membrane-spanning
domains, ATP-binding domains, or Ca2'-binding domains are present. The cDNA
for 5-lipoxygenase has been expressed in mammalian osteosarcoma cells, a baculo-
virushnsect cell system, and E. coli. The recombinant enzyme is identical to the
human neutrophil enzyme with regard to antigenicity, molecular weight, and require-
ments for activity. The structure of the 5-lipoxygenase gene has also been reported;
the gene spans >82 kb and consists of 14 exons; it does not have a TATA box in the
5'-untranslated region. Very little is known about the regulation of 5-lipoxygenase
gene expression. The 5-lipoxygenase mRNA is detectable in HL-60 cells only after
dimethylsulfoxide-induced differentiation, an observation consistent with some form
of developmental regulation.
   The sequences of several plant and mammalian lipoxygenases have been deduced
from their cDNAs. The 5-lipoxygenase shows extensive homologies to other lipoxy-
genases. It is almost identical to the rat neutrophil54ipoxygenase (93% identity) and
has 40% homology with the soybean and pea seed lipoxygenases. The human reticu-

locyte 15-lipoxygenase and the human platelet 12-lipoxygenase display, respectively,
39% and 40% sequence identity (61% and 62% similarity) with the human neutrophil
5-lipoxygenase [ 161.
   The LTA, hydrolase that catalyzes the stereospecific hydrolysis of LTA, to LTB,
is a cytosolic enzyme with a molecular weight of about 68 kDa [ 151; it is a zinc metal-
loenzyme and is clearly distinct from the epoxide hydrolases described previously in
liver. The LTA, hydrolase shows very high substrate specificity. Like the 5-lipoxy-
genase, it is an auto-inactivating enzyme; inactivation occurs following covalent
binding of the substrate at the active site, as demonstrated with [,H]LTA,. LTA,,
formed from EPA, is also a substrate for the LTA, hydrolase, which catalyzes its
conversion to LTB,. LTA,, which can be formed from 5,8,11-eicosatrienoic acid (the
Mead acid of essential fatty acid deficiency; see Chapter 5) is a very poor substrate
for the enzyme but a potent inhibitor, which becomes covalently bound to the active
site. The LTA, hydrolase appears to exist in different isoforms and has a wide dis-
tribution; it has been detected in human neutrophils, macrophages, lymphocytes,
erythrocytes, lung, liver, and plasma, and in various tissues of guinea pigs.
   Lung, spleen and placental cDNA libraries have been used to select cDNA clones
coding for the LTA, hydrolase; the deduced primary structures have predicted a
protein of 610 amino acids with no apparent homology to other known proteins.
When expressed in E. coli, LTA, hydrolase has properties comparable to the isolated
human neutrophil LTA, hydrolase. The ubiquitous distribution of LTA, hydrolase
contrasts with that of 5-lipoxygenase, indicating that the genes controlling the bio-
synthesis of the two enzymes required for the production of LTB, from arachidonate
are not coordinately regulated.
   The synthesis of LTC, involves a GSH-S-transferase named LTC, synthase. The
enzyme has been isolated from RBL-1 cells and guinea pig lungs [ 171. Unlike classical
GSH transferases which a r e soluble enzymes, LTC, synthase is a microsomal
protein. Furthermore, it has a different inhibition profile than the liver GSH trans-
ferases, and exhibits a high degree of specificity for LTA, and its methylester. Thus,
LTC, synthase is considered to be a unique enzyme. LTC, synthase activity is present
in human eosinophils, monocytes, peritoneal macrophages, basophils, and mast
cells; these cell types also contain 5-lipoxygenase and, therefore, possess the enzy-
matic activities necessary for the production of LTC4 from arachidonic acid.

3.4. Regulation o leukotriene synthesis

As with cyclooxygenase products, the synthesis of 5-HETE and leukotrienes depends
on substrate availability. However, in many cell types, such as in human blood neu-
trophils, eosinophils, and monocytes, substrate availability is not the sole require-
ment for synthesis of 5-lipoxygenase products, as these cells will transform exog-
enous arachidonic acid only to a small extent; however, incubation of these cells in
the presence of the CaZ+ ionophore A23 187 leads to synthesis of substantial amounts

of 5-HETE and leukotrienes from endogenous (or exogenous) arachidonic acid.
These observations are of primary importance since they indicate that A231 87 not
only causes the release of arachidonic acid, but also activates the 5-lipoxygenase,
which is a Ca2’-dependent enzyme. Thus, efficient leukotriene synthesis requires both
the availability of arachidonate and activation of 5-lipoxygenase. This is a major
difference between the 5-lipoxygenase and other dioxygenases and implies that there
are additional regulatory mechanisms for the synthesis of leukotrienes. It also
suggests that the 5-lipoxygenase exists in at least two different states having different
levels of activity.
   In this regard, recent studies have indicated that the stimulation of neutrophils by
A23187 induces a Ca2’-dependent translocation of the 5-lipoxygenase from the cy-
tosol to membrane structures, an event which appears to be directly linked to the
activation of the enzyme. Indeed, an indole derivative, MK-886, a novel and highly
specific and potent inhibitor of leukotriene synthesis, which does not inhibit the ac-
tivity of the purified 5-lipoxygenase, blocks the translocation and the activation of
the enzyme in intact neutrophils treated with A23187 [18]. Studies on the mechanism
of action of MK-886 have led to the discovery of a neutrophil membrane protein that
binds MK-886. The protein, named ‘five lipoxygenase-activating protein’ or FLAP,
has a transmembrane domain and is believed to act as a docking protein for the 5-
lipoxygenase on the plasma membrane [19]. Although it is likely that this transloca-
tion process is associated with the activation of the 5-lipoxygenase, it is not clear how
translocation is coupled to activation of 54ipoxygenase.
   Over the past decade, the stimulation of cellular leukotriene synthesis by a variety
of soluble agonists or phagocytic particles has been extensively investigated; it is now
generally recognized that agents that activate phagocytes, basophils, and mast cells
(through interactions with surface receptors), such as the chemotactic peptide M e t -
Leu-Phe, the complement fragment C5a, platelet-activating factor (PAF), opsonized
zymosan, IgE (monomeric or aggregated), or immune complexes, stimulate the ac-
tivity of the 5-lipoxygenase and induce the synthesis of leukotrienes from endogenous
arachidonate. Although the sequence of events leading to stimulation of leukotriene
synthesis is still far from being elucidated, it is distinctly possible that alterations of
Ca2+metabolism induced by cell activation (Ca” influx or release from intracellular
pools) are involved in the activation of the 5-lipoxygenase product synthesis by nat-
ural stimuli. Other factors that modulate the synthesis of leukotrienes include inter-
actions with activated platelets or other cell types which release arachidonic acid or
LTA,, and diets where arachidonic acid is partially replaced by EPA (marine diet or
fish oil supplement) or other polyunsaturated fatty acids in cellular lipids.
   The ‘priming’ effects of cytokines such as the granulocyte/macrophage-colony
stimulating factor (GM-CSF) on neutrophil functions constitute novel mechanisms
of regulation of leukotriene synthesis. Recombinant GM-CSF, at subnanomolar
concentrations, strikingly enhances the synthesis of leukotrienes in eosinophils and
neutrophils stimulated with A23187, the chemotactic peptides, or PAF. Indeed, the

exposure of the neutrophils to GM-CSF confers enhanced activity for the generation
of leukotrienes through increased 5-lipoxygenase capacity and arachidonate availa-
bility. This suggests a crucial role for this cytokine as a modulator of inflammatory

3.5. The metabolism of lipoxygenase products

The first metabolic process reported for lipoxygenase products was the reacylation of
monohydroxy acids, mainly 5-HETE, into cellular lipids in neutrophils. The biologi-
cal significance of this process, which has been documented for 5-,12-, and 15-HETE,
is presently unknown [ 131. LTB, does not undergo reacylation in neutrophils.
   LTB, is rapidly metabolized in human neutrophils in vitro; the major degradation
pathway is through w-oxidation. LTB, is converted to 20-hydroxy-and 20-carboxy-
LTB, in two distinct steps. The enzyme involved is a membrane-associated NADPH-
dependent cytochrome P-450. Recently, 20-oxo-LTB4 has been demonstrated to be
an intermediate in the formation of 20-carboxy-LTB4 suggesting that NAD-depend-
ent alcohol and aldehyde dehydrogenases are involved in the transformation of 20-
hydroxy-LTB, to 20-carboxy-LTB4 [20]. Among human blood cells, only neutrophils
catalyze w-hydroxylation of LTB,. The human neutrophil LTB, w-hydroxylase
shows some specificity for LTB,; various HETEs and stereoisomers of LTB, are hy-
droxylated at slower rates.
   A saturable and temperature-sensitive export mechanism for LTB, exists in human
neutrophils; this process may have an important role in the optimal expression of the
biological activity of LTB, because 20-carboxy-LTB4, the end-product of w-oxida-
tion of LTB, in neutrophils, is biologically inactive.
   Several other transformations of LTB, occur in neutrophils from other species and
in other cell types, including the formation of 18-hydroxy-and 19-hydroxy-LTB4 in
rat neutrophils and reduction of the double bond between C-10 and C-1 1 and oxida-
tion of the hydroxyl group to a keto group at C-12 (followed by stereospecific reduc-
tion and inversion of configuration) in the porcine neutrophils [21]. The reduction of
the double bond at C-10 of LTB, has also been observed in human glomerular me-
sangial cells, human monocytes, and human lungs; the resulting 10,ll -dihydro-LTB,
is a biologically inactive metabolite of LTB,. In rat kidney homogenates, LTB, is
inactivated by isomerization of the double bond at C-6. Rat hepatocyte microsomes
produce 19-hydroxy-, 20-hydroxy-, and 20-carboxy-LTB4, while incubation of LTB,
with cultured rat hepatocytes yields 18-carboxy-19,20-dinor-LTB4 well as te-
tranor and hexanor metabolites), in addition to 20-hydroxy-and 20-carboxy-LTB4,
indicating the involvement of both w- and /?-oxidation pathways (Chapter 3). It is
noteworthy that /?-oxidation of leukotrienes occurs from the w-terminus following
w-oxidation, in contrast to /%oxidation of prostanoids which takes place from the
carboxyl end.
   Very few studies have addressed the question of the metabolism of LTB, in vivo.

In rats, it appears that both renal elimination and hepatobiliary excretion occur [22];
there is extensive catabolism of infused LTB, via both w- and b-oxidation.
   Two different pathways have been described for the catabolism of the peptidoleu-
kotrienes in polymorphonuclear leukocytes (PMNLs). Inactivation occurs in stimu-
lated neutrophils and eosinophils through a pathway which involves myeloper-
oxidase, H202, and halide ions. Products of this metabolic pathway have been
identified as the sulfoxide stereoisomers of LTC,, LTD, and LTE,, and the 6-truns-
and 12-epi-6-trans-LTB4,formed by elimination of the peptide side chain at C-6. A
specific export mechanism for LTC, has been described for human eosinophils. A
second catabolic pathway for peptidoleukotrienes involves the proteolytic cleavage
of the peptide moiety by y-glutamyl transpeptidase (y-GTP) and dipeptidases
present in PMNLs, leading to LTD, and LTE, (Fig. 12) [17]. In fact, the proteolytic
cleavage of LTC, has been observed in many cell types and tissues and in plasma; it
should be noted that LTE,, the end-product of this pathway, although less potent
than LTC, and D,, shows significant biological activity.
   In rat liver microsomes, hepatocytes, and perfused liver, LTC, undergoes
proteolytic cleavage to LTD, and LTE,, which is further metabolized to N-acetyl-
LTE, and 20-hydroxy-and 2O-carboxy-N-acetyl-LTE4through w-oxidation. In vivo
experiments in rats have indicated the importance of both hepatobiliary excretion

             L vCOOH


             L vCOOH


             L /COOH



Fig. 12. Conversion of LTC, to other peptidoleukotrienes. y-GTP is y-glutamyl transpeptidase.

and renal elimination, and the occurrence of /I-oxidation (from the w-terminus) with
formation of 16-carboxy-17,18,l9,20-tetranor-14,15-dihydro-N-acetyl-LTE4 In    [23].
man, intravenous tritiated LTC, is eliminated in urine, mainly as LTE,, but also as
N-acetyl-LTE,, 20-hydroxy-LTE4, 20-carboxy-LTE4, and tritiated water (indicating
/%oxidation), as well as in feces [24]. Urinary LTE, is considered to be a useful indi-
cator of whole body production of peptidoleukotrienes in man, and assay methods
have been developed for measurement of LTE, in urine [25].

3.6. Biological activities of leukotrienes

‘Slow reacting substance of anaphylaxis’ (SRS-A) was the name coined by Brokle-
hurst in 1953 to describe an active principal released by lung tissue in response to
immunological challenge. The main biological effect of SRS-A was to cause contrac-
tion of respiratory tract smooth muscle. Following studies performed in the mid
1970s by Borgeat and Samuelsson (see [26]), it was recognized (in the early 1980s)
that SRS-A activity prepared from lung tissue consists of a mixture of LTC,, LTD,,
and LTE,. It is now clear from pharmacological studies with purified and synthetic
leukotrienes that LTC,, LTD,, and LTE, are extremely potent constrictors of respi-
ratory tract smooth muscle in man and other species; in addition, these agents are
vasoconstrictors, and they increase vascular permeability. As suggested more than 25
years ago in the early studies on SRS-A, there is evidence that leukotrienes mediate
the bronchoconstriction observed with immediate hypersensitivity reactions, includ-
ing those associated with asthma [17].
   Although the impetus for work on the leukotrienes was the challenge of determin-
ing the chemical nature of SRS-A, it is clear that another leukotriene with no SRS-A
activity, namely LTB,, has important roles in inflammatory responses [27]. Indeed,
LTB, is highly chemotactic towards neutrophils and other phagocytes in vitro and in
vivo; LTB, induces the adherence of neutrophils to the vascular endothelium and
stimulates migration of neutrophils into extravascular tissues, two fundamental
events of the inflammatory process.
   A clear picture of the biological activity profiles of leukotrienes is now emerging.
Synthetic leukotrienes, which have been available for the past 10 years, have made it
possible to define the effects of leukotrienes in animal and human models; further-
more, the recent development, both of specific and sensitive assay methods for leu-
kotrienes and of potent antagonists of the action and synthesis of leukotrienes,
should facilitate the definition of the physiological and pathophysiological roles of
these compounds [ 17,271.
   A considerable amount of experimental attention is currently directed at under-
standing the mechanisms of action of leukotrienes at the molecular level. Specific
high affinity receptors for LTB,, C,, D,, and E, have been characterized in membrane
preparations from various leukocytes, lung, spleen, and other tissues [17]. Guanine
nucleotide binding (G) proteins regulate receptor affinity for LTB, and LTD4 in lung

membranes. The neutrophil is the cell type where the mechanisms of action of leu-
kotrienes have been studied in greatest detail. Signal transduction in LTB,-stimu-
lated neutrophils resembles, but is not identical to, that observed for other soluble
agonists of the neutrophil. Activation of neutrophils by LTB, involves a G-protein
sensitive to pertussis toxin, activation of a PtdIns-specific phospholipase C, elevation
of intracellular [Ca”], and reorganization of the cytoskeleton. Both Ca2+influx and
mobilization of Ca2’ from internal pools are involved in the LTB,-elicited increase in
[Ca”], an event which is believed to play a central role in the initiation of neutrophil
functions, and in particular the locomotor response [28]. In short, leukotrienes, like
prostanoids, are thought to function via a group of receptors coupled to G proteins.

4. Epoxygenase products
4.1. Introduction

The term ‘epoxygenase pathway’ refers to transformations of arachidonic acid in-
itiated via cytochrome P-450 mixed-function oxidases [29-3 11. This terminology des-
ignates chemical features of certain products, and it distinguishes the enzymology
involved in their formation from that of the other pathways of eicosanoid bio-
synthesis. Cofactors required by cytochrome P-450 monooxygenases include the fla-
voprotein cytochrome P-450 reductase, NADPH/NADP+or NADH/NAD+, and
molecular oxygen. Isoenzymes of cytochrome P-450 monooxygenases can display
substrate specificity, and some data indicates that there is an hepatic isoenzyme with
a preference for oxidation of arachidonic acid [30]. However, it is likely that multiple
isoenzymes can participate in the synthesis of epoxygenase products.

4.2. Structures, nomenckuture, and biosynthesis

There are two main types of ‘epoxygenase’ product (Fig. 13): (i) cis-epoxy-eicosa-
trienoic acids (EpEtrEs or EETs*) and (ii) hydroxyeicosatetraenoic acids (HETEs).
Members of each group are biologically active (Fig. 13; Table I). There is little regio-
specificity associated with ‘epoxygenase’ derived HETEs. Different isomers may
originate, nonspecifically, from a single enzyme operating at different rates, or they
may originate from discrete enzymes typified by those present in hepatic, renal,
ocular, and leukocyte microsomes. The stereoselectivity of HETE formation differs
between monooxygenase and lipoxygenase enzymes. HETEs from both pathways
have similar cis-trans olefin configurations; however, the lipoxygenase products are

*The older trivial nomenclature for epoxy-eicosatrienoic acids denotes these compounds as EETs. The
newer nomenclature (Smith et al., 1991) denotes these compounds as EpETrEs (i.e.,Ep=epoxy, E=eicosa,

Fig. 13. Structures of products of epoxygenase pathways of the arachidonate cascade

enantiospecific, with an (S) hydroxyl configuration typically at the 5-, 12-, or 15-
carbons. The ‘epoxygenase’ activity from hepatic microsomes produces six regio-
somers with little enantioselectivity except for 12(R)-5,8,14-cis-lO-truns  eicosate-
traenoic acid. Psoriatic lesions and corneal microsomes [32] produce the biologically
active enantiomer, 12(R)-HETE. The activity of the 12(S) enantiomer is uncertain. In
contrast to hydroxylation, ‘epoxygenase’ catalyzed epoxidation of arachidonic acid
is enantioselective [33]. Purified rodent hepatic microsomal monooxygenase yields
80% 14(R),1S(S)-cis-EpETrE, 97% s(R),s(S)-cis-EpETrE, 97% 1l(S), 12(R)- cis-
EpETrE. 5,6-cis-EpETrE is distribute.d as an equimolar mixture (60140) of RS and
SR enantiomers.
   Hepatic, renal, ocular, and pituitary systems all transform arachidonic acid into
biologically active ‘epoxygenase’ products, but biosynthetic traits vary among the
systems. Renal cortical microsomes from pentobarbital-treated rabbits produce
either #-oxidation products or epoxides and cis-truns dienols. The rate of arachido-
nate conversion by renal cortical microsomes (0.15 nmol converted/min/mg) is low,
relative to that observed for liver microsomes (5-6 nmol converted/min/mg). Such

Biological actions of epoxygenase metabolites

Compound                       Action

5,6-EpETrE                     Somatosatin release in vitro from hypothalamic median eminence

5.6-EpETrE                     Insulin release in vitro from pancreatic islets

14,lS-EpETrE                   Glucagon release in vitro from pancreatic islets

1I . 12-diHETrE                Inhibition of Na'lK' ATPase in vitro

5,6-EpETrE                     Vasodilation of intestinal arteriolar blood flow in vivo
1 I , 12-EpETrE

5,6-EpETrE                     Inhibit vasopressin stimulated water flow in toad bladder in vitro
and diHETrEs

5,6-EpETrE                     Increase jsCa2+loss From canine aortic smooth muscle microsomes in vitro
1 1.12-EpETrE

5,6-EpETrE                                   efflux from anterior pituitary; increased cytosolic free CaZ'
                               Increase 45Ca2+

14,lS-cis-EpETrE               Inhibit cyclooxygenase in vitro

14,154s-EpETrE and             Inhibit platelet aggregation in vitro
stereoisomers; 14,15-cis-
episulfide ET

14,lS-cis-EpETrE               Promotes tumor cell adhesion to endothelium

14.15-cis-EpETrE               Activation of Na'/H+ exchange, mitogenesis

14,lS-cis-EpETrE               Inhibition of renin release

12(R)-HETE                     Inhibits Na'/K' ATPase (corneal epithelium) in vitro

12(R)-14,15-dihydro-HETE Vasolidator, angiogenesis

differences may be organ-, inducer-, and species-dependent; or the proportions of
peroxidative- versus monooxygenase-dependent catalysis may differ.
                                                                                      32 1

   Renal monooxygenase has been characterized at the cellular level [34]. Activity
resides within the TALH cells (thick ascending limb of the loop of Henle). TALH
cells, lacking lipoxygenase activity and possessing low cyclooxygenase activity,
oxidize 4.4k2.1 pg arachidonic acidlmg of protein130 min. Vasopressin, at phys-
iological concentrations, stimulates TALH cell arachidonate metabolism, supporting
a role for these eicosanoids in renal function.

4.3. Occurrence of epoxyeicosatrienoic acids

EpETrEs occur as constituents of liver, kidney, lung, and in urine [35]. Urine and
kidney contain, principally, 8,9- and 14,15-cis-EpETrEs. Lung lavage fluid from
patients with adult respiratory distress syndrome contains 9,lO-epoxy-12-octade-
cenoic acid. Human platelets may contain 14,154s-EpETrE esterified within phos-
pholipids. Data also suggest that EpETrEs, predominately 14,l SEpETrE, is
released by endothelial cells and that low density lipoprotein enhances their genera-
tion [36].Several biological actions of EpETrEs are consistent with their localization
or sites of formation.
   The occurrence of EpETrEs as endogenous cellular constituents is unusual among
the eicosanoids. Cells do not typically reincorporate or accumulate prostaglandins,
thromboxanes, or leukotrienes within membranes or storage vesicles. Cell associa-
tion of the eicosanoids usually involves receptor-mediated processes or nonspecific,
reversible interactions. Occurrence of EpETrEs as intact, stored species opens the
possibility that they have additional non-autacoid roles.

4.4. Metabolism of epoxygenase metabolites o arachidonic acid

Rapid, complex, and comprehensive metabolism of EpETrEs in vitro or in vivo has
complicated efforts to correlate their biosynthesis with putative physiological roles.
Both enzymatic and non-enzymatic hydration converts EpETrEs into their corre-
sponding vicinal diols. Hydrolysis rates for 8,9-, 11,12-, and 14,154s-EpETrE by
purified mouse liver cytosolic epoxide hydrolase exceed the rates with microsomal
enzyme. Thus, isolation of intact cis-epoxides is difficult unless one minimizes en-
zymatic hydration with inhibitors and avoids conditions, such as low pH, which facil-
itate non-enzymatic hydration. Indirectly, quantitation of corresponding vicinal di-
hydroxy metabolites can be a useful index of epoxide formation. The EpETrEs can
also form conjugates with reduced GSH. It is uncertain whether conjugation
produces a metabolite with its own biological properties, analogous to sulfidopeptide
leukotrienes, or whether this process deactivates EpETrEs.
   5,6-cis-EpETrE, which is a poor substrate for cytosolic and microsomal epoxide
hydrolases, has a unique metabolic trait. It retains its 8,ll ,14-cis-olefin substituents;
therefore, PGH synthase (Fig. 2) can convert it to 5,6-epoxy PGG, and 5,6-epoxy-
PGH, . These endoperoxides can subsequently be transformed into corresponding

5,6-epoxy prostaglandins of the E, F, and I series. Other EpETrEs have olefin con-
figurations which are inappropriate for transformation by PGH synthase. However,
by analogy, they are often appropriate substrates for lipoxygenases. Other biochem-
ical fates such as B- or w-oxidation are likely for both EpETrEs and HETEs.

4.5. Biological actions of ‘epoxygenase’ derived EpETrEs and HETrEs

EpETrEs and HETEs derived from the ‘epoxygenase’ pathway have prominent
biological actions (Table I). These include stimulation of peptide hormone release
from endocrine cells; inhibition of renal Na+/K+-ATPase; mobilization of micro-
soma1 Ca2+from aortic smooth muscle and anterior pituitary cells; inhibition of cy-
clooxygenase activity; inhibition of arachidonic acid-induced platelet aggregation;
vasodilation of intestinal microcirculation; inhibition of vasopressin-stimulated
water transport; angiogenesis and vasodilation of arteries; stimulation of endothelial
tumor cell adhesiveness; stimulation of mitogenesis in glomerular mesangial cells;
and inhibition of renin release from renal cortical slices. These activities may be phar-
macological, physiological, or both.
   Mechanisms of action have been defined for few of the biological activities of
EpETrEs or HETEs. Inhibition of Naf/K+ ATPase accounts for the renal actions of
EpETrEs. Inhibition of PGH synthase may account partially, but not completely, for
some actions of 14,15-cis-EpETrE. Activation by PGH synthase may account for
some of the actions of 5,6-EpETrE such as arterial vasodilation. Effects on cellular
or organelle Ca2+homeostasis may be an important feature in EpETrE action. With
the exception of the 8,9-isomer, all other EpETrEs increase the release and inhibit the
incorporation of Ca” from canine aortic smooth muscle microsomes. 5,6-cis-
EpETrE, in anterior pituitary cells, is similar. Effects on Ca” mobilization are an
attractive mechanism for many actions of EpETrEs. Currently, there is little evidence
that EpETrEs work through G protein-linked receptors. Thus, EpETrEs may not
operate through signal transduction mechanisms common to products of the cyclo-
oxygenase and lipoxygenase pathways of arachidonate metabolism.

5. Future directions
5.1. Cyclooxygenase metabolites

Major breakthroughs are likely to occur during the next five years in our understand-
ing of which lipids and which phospholipases are involved in arachidonate mobiliza-
tion. In addition, we should have a clear view of what factors regulate the expression
of the genes coding for all the enzymes of prostanoid biosynthesis. Recent progress
in receptor purification and cloning also suggests that within five years, there will be
detailed information on how many prostanoid receptors there are, their specificities,
and their cellular locations.

5.2. Lipoxygenase metabolites

In the coming years, basic research in the area of the lipoxygenases will address se-
veral important questions on the regulation of the pathways. Among other specific
topics, the role of cytokines and of cellular interactions in the modulation of the syn-
thesis and action of the biologically active products of the lipoxygenases and the
regulation of the genes for the enzymes of the lipoxygenase pathways, should gen-
erate a great deal of interest.
   The development of antagonists and synthesis inhibitors of leukotrienes will re-
main a very active area; indeed the availability of potent and specific compounds
represents the most direct and reliable approach to define the role of leukotrienes in
health and disease and their relative importance among the various mediators of in-
flammation. There is little doubt that in the next few years, there will be a dramatic
increase in clinical studies on leukotriene antagonists and synthesis inhibitors; hope-
fully these studies will lead to the discovery of new classes of anti-inflammatory com-

5.3. Epoxygenase metabolites

Two themes will be important in the future. First, the mechanisms of action of epoxy-
genase metabolites will be clarified, and it is likely that these will not involve receptor-
mediated processes. Second, improved analytical methods, typified by tandem mass
spectrometry, will facilitate direct detection and quantitation of phospholipids con-
taining epoxyeicosatrieneoic acids. Synthesis of novel metabolites and analogs and
assay standards will be vital for continued progress.

 I . Smith, W.L., Marnett, L.J. and DeWitt, D.L. (1991) Prostaglandin and thromboxane biosynthesis.
     Pharm. Ther., 49, 153-179.
 2. Smith, W.L. (1989) The eicosanoids and their biochemical mechanisms of action. Biochem. J. 259,
     31 5-324.
 3. Dennis, E.A. (1987) Regulation of eicosanoid production: role of phospholipases and inhibitors. Bio-
     technology 5, 12941300.
 4. Neufeld, E.J. and Majerus, P.W. (1983) Arachidonate release and phosphatidic acid turnover in stirnu-
     lated human platelets. J. Biol. Chem. 258, 2461-2467.
 5. Kramer, R.M., Hession, C., Johansen, B., Hayes, G., McGray, P., Chow, E.P., Tizard, R. and
     Pepinsky, R.B. (1989) Structure and properties of a human non-pancreatic phospholipase A,. J. Biol.
     Chem. 264,5768-5775.
 6. Loeb, L.A. and Gross, R.W. (1986) Identification and purification of sheep platelet phospholipase A,
     isoforms. J. Biol. Chern. 261, 10467-10470.
 7. Smith, W.L. and Marnett, L.J. (1991) Prostaglandin endoperoxide synthase: structure and catalysis.
     Biochim. Biophys. Acta, 1083, 1-17.

 8. DeWitt, D.L. (1991) Prostaglandin endoperoxide synthase: regulation of enzyme expression. Biochim.
    Biophys. Acta, 1083, 121-134.
 9. Vane, J.R., Flower, R.J. and Botting, R.M. (1990) History of aspirin and its mechanism of action.
    Stroke 21 (Suppl. IV), IV12-IV23.
10. Dietz, R., Nastainczyk, W. and Ruf, H.H. (1988) Higher oxidation states of prostaglandin H synthase:
    rapid electronic spectroscopy detected two spectral intermediates during the prroxidase reaction with
    prostaglandin G,. Eur. J. Biochem. 171, 321-328.
11. Ushikubi, F., Nakajima, M., Hirata, M., Okuma, M., Fujiwara, M. and Narumiya, S. (1989) Puri-
    fication of the thromboxane AJprostaglandin H, receptor from human blood platelets. J. Biol. Chem.
    264, 16496-16501.
12. Rokach, J. and Fitzsimmons, B. (1988) The lipoxins. Int. J. Biochem. 20, 753-758.
13. Spector, A.A., Gordon, J.A. and Moore, S.A. (1 988) Hydroxyeicosatetraenoic acids (HETEs). Prog.
    Lipid Res. 27, 271-323.
14. Borgeat, P. (1989) Biochemistry of the lipoxygenase pathways in neutrophils. Can. J. Physiol. Phar-
    macol. 67,936-942.
15. Samuelsson, B. and Funk, C.D. (1989) Enzymes involved in the biosynthesis of leukotriene-B4. J. Biol.
    Chem. 264, 19469-19472.
16. Funk, C.D., Furci, L. and FitzGerald, G.A. (1990) Molecular cloning, primary structure and expres-
    sion of the human platelet erythroleukemia cell 12-lipoxygenase. Proc. Natl. Acad. Sci. U.S.A. 87,
17. Lewis, R.A., Austen, K F. and Soberman, R.J. (1990) Leukotrienes and other products of the Slip-
    oxygenase pathway-biochemistry and relation to pathobiology in human diseases. N. Engl. J. Med.
18. Rouzer, C.A., Ford-Hutchinson, A.W., Morton, H.E. and Gillard, J.W. (1990) MK886, a potent and
    specific leukotriene biosynthesis inhibitor blocks and reverses the membrane association of 5-lipoxy-
    genase in ionophore-challenged leukocytes. J. Biol. Chem. 265, 1436-1442.
19. Miller, D.K., Gillard, J.W., Vickers, P.J., Sadowski, S., Leveille, C., Mancini, J.A., Charleson, P.,
    Dixon, R.A.F., Ford-Hutchinson, A.W., Fortin, R., Gauthier, J.Y., Rodkey, J., Rosen, R., Rouzer,
    C., Sigal, IS., Strader, C.D. and Evans, J.F. (1990) Identification and isolation of a membrane protein
    necessary for leukotriene production. Nature 343, 278-28 1.
20. Sumimoto, H. and Minakami, S. (1 990) Oxidation of 20-hydroxyleukotriene-B4 to 20-carboxyleu-
    kotriene-B4 by human neutrophil microsomes-role of aldehyde dehydrogenase and leukotriene-B4
    omega-hydroxylase (cytochrome-P-450Ltb-omega) in leukotriene-B4 omega-oxidation. J. Biol. Chem.
21. Wainwright, S., Falck, J.R., Yadagiri, P. and Powell, W.S. (1990) Stereochemistry of leukotriene-B4
    metabolites formed by the reductase pathway in porcine polymorphonuclear leukocytes-inversion of
    stereochemistry of the 12-hydroxyl group. Biochemistry 29, 1180-1 185.
22. Hagmann, W. and Korte, M. (1990) Hepatic uptake and metabolic disposition of leukotriene-B4 in
    rats. Biochem. J. 267, 467470.
23. Perrin, P., Zirrolli, J., Stene, D.O., Lellouche, J.P., Beaucourt, J.P. and Murphy, R.C. (1989) In vivo
    formation of B-oxidized metabolites of leukotriene E4 in the rat. Prostaglandins 37, 5340.
24. Maltby, N.H., Taylor, G.W., Ritter, J.M., Moore, K., Fuller, R.W. and Dollery, C.T. (1990) Leu-
    kotriene-C4 elimination and metabolism in man. J. Aller. Clin. Immunol. 85, 3-9.
25. Westcott, J.Y., Johnston, K., Batt, R.A., Wenzel, S.E., and Voelkel, N.F. (1990) Measurement of
    peptidoleukotrienes in biological fluids. J. App. Physiol. 68,2640-2648.
26. Borgeat, P. and Sirois, P. (1981) Leukotrienes a major step in the understanding of immediate hyper-
    sensitivity reactions. J. Med. Chem. 24, 121-126.
27. Ford-Hutchinson, A.W. (1990) LeukotrieneB4 in inflammation. Crit. Rev. Immunol. 10, 1-12.
28. Naccache, P.H., Sha’afi, R.I. and Borgeat, P. (1989) Mobilization, metabolism and biological effects

      of eicosanoids in polymorphonuclear leukocytes. In: Neutrophil Physiology (Hallett, M.B., Ed.), pp.
       113-139, CRC Press, Boca Raton, FL.
29.   Capdevila, J., Marnett, L., Chacos, N., Prough, R. and Estabrook, R. (1982) Cytochrome P-450 de-
      pendent oxygenation of arachidonic acid to hydroxyeicosatetraenoic acids. Proc. Natl. Acad. Sci.
      U.S.A. 79, 761-770.
30.   Laniado-Schwartzman, M., Davis, K., McGiff, J., Levere, R. and Abraham, N. (1988) Purification
      and characterization of cytochrome P-450dependent arachidonic acid epoxygenase from human liver.
      J. Biol. Chem 263, 25362542.
31.   Oliw, E., Guenguerich, P. and Oates, J. (1982) Oxygenation of arachidonic acid by hepatic monooxy-
      genase: isolation and metabolism of four epoxide intermediates. J. Biol. Chem. 251, 3771-3787.
32.   Murphy, R.C., Falck, J.R., Lumin, S., Yadagiri, P., Zirrolli, J., Balazy, M., Masferrer, J., Abraham,
      N. and Schwartzman, M. (1988) I2(R)-Hydroxyeicosatrienoic acid: a vasodilator cytochrome P-450
      dependent arachidonate metabolite from the bovine cornea epithelium. J. Biol. Chem. 263, 17197-
33.   Capdevila, J., Karara, A,, Waxman, D., Martin, M., Falck, J.R. and Guengerich, P. (1990) Cy-
      tochrome P-450 enzyme-specific control of the regio- and enantiofacial selectivity of the microsomal
      arachidonic acid epoxygenase. J. Biol. Chem. 265, 10865-10871.
34.   Ferreri, N., Schwartzman, M. Abraham, N., Chander, P., and McGiff, J. (1984) Arachidonic acid
      metabolism in a cell suspension isolated from rabbit renal outer medulla. J. Pharmacol. Exp. Ther.
35.   Catella, F., Lawson, J., Fitzgerald, D. and FitzGerald, G. (1990) Endogenous biosynthesis of ara-
      chidonic acid epoxides in humans: increased formation in pregnancy-induced hypertension. Proc.
      Natl. Acad. Sci. U.S.A. 87, 5893-5897.
36.   Pritchard, K.. Wong, P.-K. and Stemerman, M. (1990) Atherogenic concentrations of low-density
      lipoprotein enhance endothelial cell generation of epoxyeicosatrienoic acids products. Am. J. Pathol-
      ogy 136, 1383-1391.
This Page Intentionally Left Blank
D.E. Vance and J. Vance (Eds.) Biochemistry o Lipids, Lipoproteins and Membranes
8 1991 Elsevier Science Publishers B.V. All rights reserved.                                        327

                                                                                         CHAPTER 11

                                                                         CHARLES C. SWEELEY

       Department of Biochemistry, Michigan State University, East Lansing, M I 48824, U.S.A .

I. Introduction
Sphingolipids are nearly ubiquitous constituents of membranes in animals, plants,
and some lower forms of life. They were first described in a remarkable treatise on the
chemical constitution of the brain by Johann L.W. Thudichum, a physician scientist
in London, who published his findings more than 100 years ago. Among the novel
compounds discovered by Thudichum were three related lipids, which he called
sphingomyelin, cerebroside, and cerebrosulfatide (Fig. 1). Some of the substituents of
these lipids, recovered after hydrolysis in aqueous barium hydroxide or sulfuric
acid, were themselves novel, such as two unusual long-chain fatty acids which Thu-
dichum named lignoceric acid and cerebronic acid, and a long-chain aliphatic amine
which he found in all of these lipids and called sphingosine (Fig. 2). Substantial
problems involved in the isolation of brain sphingolipids, and the relative neglect of
this class of lipids by most biochemists in the early twentieth century, prolonged the
time required to establish the complete structures of Thudichum’s originally de-

       +         0-      OH
    )                             CH(CH,),,CH,
                 0     NHCO(CH,),CH,

                                        (cH,),        ,CH,
Fig. 1. Structures of sphingomyelin (top), cerebroside (middle), and cerebrosulfatide (bottom) from human


CH (cH,),,       ~HCOOH


Fig. 2. Structures of lignoceric acid (top), cerebronic acid (middle), and sphingosine (bottom). See text for
abbreviations of structures.

scribed sphingolipids. The impetus to study the chemistry of sphingolipids and, ul-
timately, the mechanisms for their biosynthesis and metabolism was the discovery of
several rare human diseases which could be attributed to the occurrence of abnormal
levels of various sphingolipids such as sphingomyelin (Niemann-Pick disease), cere-
broside (Gaucher’s disease and Krabbe’s globoid cell leukodystrophy), and an acidic
glycosphingolipid called ganglioside, massive amounts of which were found in the
brains of babies with Tay-Sachs disease. Important structural studies of these lipids
by E. Klenk, H.E. Carter, L. Svennerholm, T. Yamakawa, R. Kuhn, H. Wiegandt,
H. Egge, R. Ledeen, R.K. and others can be found in the original literature and
in several reviews [1,2].
   It is now known that there are many different kinds of sphingolipids. More than
300 structures have been reported to occur in nature, and the list is still growing [3].
Structurally, they consist of a hydrophobic portion, called ceramide, that contains a
mixture of fatty acids that are amide-linked to sphingosine (Fig. 2) or other related
long-chain aliphatic amines (sphingoid bases), and a hydrophilic portion (polar head
group). In cells sphingolipids are found predominantly in membranes, where the ce-
ramide (Cer) moiety is imbedded along with the hydrophobic diacylglyceryl chains of
phospholipids and other membrane lipids. Sphingomyelin occurs in both intracellu-
lar and plasma membranes and can be shown to be symmetrically distributed in the
bilayer of the plasma membrane. In contrast, carbohydrate-containing sphingolipids
(glycosphingolipids) usually are found predominantly in the plasma membrane,
where they are asymmetrically disposed so that the polar head groups are extended
into the extracellular environment.
   The glycosphingolipids are the most diverse type in terms of the structures of their
carbohydrate chains. Because of this diversity, and the large number of three-dimen-
sional forms they can assume, the glycosphingolipids are suspected to have funda-
mental roles in membrane phenomena such as cell-cell adhesion, cellkell recogni-
tion, modulation of receptor activity and other kinds of transmembrane signalling,
antigenic specificity, and regulation of growth. The appearance of different glyco-
sphingolipids on the surface of transformed cells has been interpreted in terms of
perturbed cell surface functions of such cells. Changes in composition that ac-

company development and differentiation of organisms and organ systems suggest
that the glycosphingolipids may be involved here as well.
   It has become necessary to develop a system of nomenclature for the sphingolipids
so that individual species can be referred to in a logical manner. Ceramides are de-
fined in terms of the constituent fatty acids and sphingoid bases. A simple abbrevia-
tion can be used to denote the fatty acid chain length, degree of unsaturation, and
presence or absence of a hydroxyl group at C-2. For example, octadecanoic acid
(stearic acid), 2-hydroxytetracosanoic acid, and nervonic acid (9-tetracosenoic acid)
are abbreviated 18:0, h24:O and 24: 1, respectively. The sphingoid bases are abbre-
viated in a similar manner, citing (in order of appearance in the abbreviation)
number of hydroxyl groups (d and t for di- and tri-hydroxy, respectively), chain
length and number of double bonds. For example, Thudichum’s sphingosine is D-
erythro-2-amino-rrans-4-octadecene-   1,3-diol and is abbreviated d l 8: 1. Shorter and
longer sphingoid bases have been discovered in insects (d14:1) and brain (d20:l),
respectively, and plants are especially enriched in 4-hydroxysphinganine (originally
called phytosphingosine), which is abbreviatcd t18: 1.
   The glycosphingolipids are classified into several broad types on the basis of car-
bohydrate composition. Neutral glycosphingolipids contain neutral sugars such as
glucose (Glc), galactose (Gal), N-acetylglucosamine (GlcNAc), N-acetylgalactosa-
mine (GalNAc), and fucose (Fuc). Acidic glycosphingolipids contain charged func-
tional groups such as phosphate (phosphoglycosphingolipids) or sulfate (sulfatogly-
cosphingolipids) as well as charged sugar residues such as glucuronic acid (GlcA) in
some plant glycosphingolipids, or sialic acid in all gangliosides. Further classification
can be made on the basis of shared partial oligosaccharide sequences, sometimes re-
ferred to as ‘root structures,’ as summarized in Table I. Examples of the use of this
nomenclature are shown below. Note that the Roman numeral and Arabic super-

Nomenclature for classification of glycosphingolipids

Root name           abbreviation                Partial structure’

                                                        IV      111    I1      I

ganglio             Gg                                    Ga4?l-3GalNAc~1-4Ga~l-4Glc-Cer
lacto               Lc                                    Gal/3l-3GlcNAc~1-3Ga~l-4Glc-Cer
neolacto            n Lc                                  Gal/?l-4GlcNAc~1-3Ga~1-4Glc-Cer
globo               Gb                                    GalNAc~l-3Galal-4Ga~l-4Glc-Cer
isoglobo            iGb                                   GalNA~l-3Galal-3Ga~l-4Glc-Cer
mollu               Mu                                  GalNAc~l-tManal-3Mar$?I-4Glc-Cer
arthro              At                                             -
                                                  GlcNAySl-4GlcNAc~I 3 M a d I -4Glc-Cer

‘Roman numerals define sugar positions in the root structure.

script refer to the sugar in the root structure that is substituted (counting from the
ceramide toward the non-reducing end) and the position of that substitution, respec-

Forssmon hapten (I93-u-N-ocetylgalactosom~nyl-globotetraosylceram~de~

                       Gal,'31-3GolNAc/?l-4Ga1~1-4Glc/?l    -l'Cer

sM,gang1 ioside         ~
                  ( 1 1 -u-N-ocetylneuramlnosyl-   gong1 i o t e t raosylceramide.
                  TI3 -u-NeuSAc - G g 4 C e r )

   The systematic names and abbreviations used in this chapter were proposed in
1977 by the IUPAC-IUB Commission on Biochemical Nomenclature [4]. Brain gan-
gliosides are also denoted by a system of nomenclature proposed by Svennerholm
(i.e., GM,above). These abbreviations were assigned according to the number of sia-
lic acid residues and the relative position of the ganglioside on a thin-layer chroma-
tography plate. The sugar residues in most glycosphingolipids are assumed to have
six-membered pyranose (p) ring structures although this has been confirmed in only
a few cases, using high resolution proton nuclear magnetic resonance spectroscopy
(NMR). In this chapter, ring size will not be denoted except in an unusual case in
which one of the sugars has the furanose (f) ring structure.

2. Chemistry and distribution

2.1. Sphingoid bases

4-Sphingenine (sphingosine), sphinganine (dihydrosphingosine), and 4-hydroxy-
sphinganine (phytosphingosine) are the most abundant sphingoid bases in sphingo-
lipids of the animal and plant kingdoms (Fig. 3). It is evident, however, that the
sphingoid bases are a very heterogenous component of sphingolipids, with diverse
differences in the structure of the long alkyl chain. Their chain length may vary from
14 to 22 carbon atoms, with branching (methyl groups) at the 0-1 (iso), 01-2 (anteiso),
or internal carbon atoms, and double bonds at remote sites in the chain as well as at
C4. Utilizing gas chromatography and mass spectrometry, more than 60 different
sphingoid bases have been described (K. Karlsson, 1970). Sphingomyelin from
guinea pig Harderian gland has an especially complex mixture of sphingoid bases
(Fig. 4) ranging from d16:O and d16:l to d20:l with 44% containing branched methyl
groups at C8, C9, C10 or C16 [5]. Human brain and epidermis (A. Suzuki, 1989)
contain sphingolipids enriched in eicosasphingenine (d20: 1) whereas most other tis-
sues contain sphingolipids that have mainly C18 bases.
   Proton NMR spectra of sphingolipids reveal the trans double bond of sphingenine
                                                                                                    33 1

     CH20H                    CHzOH
      I                       I
    HCNH2                   HCNHZ
      I                       I

sphingonine              sph i ngeni ne            -
                                                 4 hydroxysphi ngani ne

    d18:O                   dl8:T                             t 18: 0

Fig. 3. Commonly occurring sphingoid bases. Chain-length isomers, branched-chain bases, and multiply
unsaturated bases are also known to occur.

by signals at 5.3-5.4 ppm and 5.L5.8 ppm for the methyne protons (ca. 15 Hz cou-
pling constant). A partial one-dimensional proton NMR spectrum [6]of peracetylat-
ed sphingenine, shown in Fig. 5, gives the assignment of signals for the NH proton
and those at C1 (L'" and Lib), C2 (L2), C3 (L') and the methyne protons (L4and L5).
Signals in the 4.0 to 5.3 ppm region are not easily observed in the NMR spectra of
complex glycosphingolipids, and chain length and degree of unsaturation are diffi-
cult to determine.


                                                              18 1
                                            17 1
                                          17 1
                                           +           18 1
                                                                        19 1

                                          17 0          +
                                                       18 0                    19 0
                              16 1
                                     160         170

0                        10                                    20                     30
                                    Time ( m i d

Fig. 4. Gas chromatogram of the trimethylsilyl derivatives of N-acetyl sphingoid bases from sphingomyelin
of guinea pig Harderian glands. Column: Chemically bonded OV-1 fused silica capillary column (25
mx0.25 min);column temperature, 250°C isothermal; nitrogen f o rate, 1 mllmin; split ratio, 30:l. Num-
bers refer to chain length and number of double bonds. Shaded peaks indicate branched, long-chain bases.
Reproduced with the permission of the authors [5].

                                                         I       LID



Fig. 5. Partial one-dimensional proton NMR spectrum of 1,3-di-O-acetyl-2-N-acetylsphing-4-enine.
symbols are described in the text. Reproduced with the permission of the authors [6].

2.2. Ceramides

Long-chain fatty acids are covalently linked to the sphingoid bases to form N-acyl
derivatives (ceramides). Considering the possibility of random combinations of sev-
eral fatty acids with several sphingoid bases, large numbers of different ceramides
may exist in an otherwise homogenous species of sphingolipid. In the sphingomyelin
fraction of guinea pig Harderian gland, for example, eleven nonhydroxy fatty acids
(16:O to 24:l) and nine 2-hydroxy fatty acids (h16:O to h23:O) were found [5]. If
randomly linked to 22 different sphingoid bases, there could be as many as 440 dif-
ferent ceramide species in this sphingomyelin sample!
   Ceramide composition can be determined by gas chromatographic analysis of the
methylated fatty acids and N-acetylated sphingoid bases after hydrolysis or metha-
nolysis of the sphingolipid. Molecular species analysis requires direct analysis of the
sphingolipid (or ceramide derived therefrom) by mass spectrometry. The molecular
weights of d l 8: 1/16:0 and t 18:0/24:0 ceramides are observed as molecular ions at m/z
537 and 667, respectively, using an appropriate means of ionization such as fast atom
bombardment (FAB) with mass spectrometry (MS). Molecular ions alone cannot
differentiate between compounds that are isobaric (same molecular weight), nor can
substituents such as hydroxyl groups and double bonds be localized in the sphingoid
base or fatty acyl moieties. For such purposes, MS/MS (daughter ion analysis of

selected parent ions) can be used to advantage. Analysis of a derivatized ceramide
from a soft coral by MS/MS (Fig. 6 ) is an especially impressive example of the analyt-
ical power of this approach [7]. The ceramide itself had a molecular weight (by FAB-
MS) of 535, suggesting that it consisted of palmitic acid (1 6:O) and a d l 8:2 sphingoid
base (or d18: 1 with a 16:I fatty acid). The protonated molecular ion of the derivatized
ceramide, where deuterium (D) and OH have been added to the double bonds,
(M+H)+at d z 562 was fragmented in a collision cell by helium gas, and the daughter
ions were recorded. As shown in the structures (Fig. 6, inset), an isolated double bond
between C8 and C9 was deduced by the progression of fragment ions at 14 mass unit
intervals from m/z 490 to lower masses, with tell-tale 30 mass unit interruptions (434
and 404; 419 and 389) at the sites of CHOH and CHD groups in the two isomeric
products of derivatization of the double bond. The sphingoid base was thus shown to
be 4,8-sphingadienine (dl 8:2) and the ceramide was therefore d l 8:2/16:0.

2.3. Glycosphingolipids

It would be difficult for the reader to comprehend the incredible diversity of struc-
tures within this class of sphingolipids from the limited discussion that follows.
Those who require a more complete understanding of the detailed chemistry of these
lipids should consult reviews by the author [l], Stults et al. [3], Wiegandt [8], Yu and
Saito [9], and Kanfer and Hakomori [ 2 ] .

2.3.1. Neutral glycosphingolipids
The most commonly occurring neutral glycosphingolipids containing one or two sug-

        +X   4 0
                    316   419   440                                           404 448

E   .
0   .              31 6

Fig. 6. Partial daughter ion mass spectrum of parent m/z 562 [M+H]' of boron trideuteride-derivatized
Sinumeramide, obtained by fast atom bombardment ionization (positive mode) and tandem mass spec-
trometry. Fragment ions in the region from m/z 389 to 490 allow identification of the position of the remote
double bond in the underivatized ceramide. Reproduced with the permission of the authors [7].

ar units are glucosylceramide (Glcpl-l’Cer), galactosylceramide (Gap1-l’Cer), and
lactosylceramide (Gap1-4Glcp1-1’Cer or LacCer) in higher organisms. The glyco-
sidic linkage to ceramide is of the /?configuration in these lipids. Less widely dis-
tributed are the gala type, such as galabiosylceramide (Galal-4Ga~l-l‘Cer),     which
is found primarily in kidney and pancreas, and fucosylceramide (Fuca 1-l’Cer),
which was isolated from human colon carcinoma. Simple neutral glycosphingolipids
of nonvertebrates are more diverse in nature; mannosylceramide (Ma@l- 1’Cer) was
first isolated from a freshwater bivalve, Hyriopsis schlegelii, and a dimannosylcera-
mide (Manp1-2ManPl-l’Cer) was found in the hepatopancreas of the same organ-
ism. A mannosyl-glucosylceramide, Ma@1-4Glcpl-l’Cer, occurs in plants (wheat
flour) and bivalves.
   More complex neutral glycosphingolipids are generally derived from either Lac-
Cer or Ma@1-4Glcj?I-l’Cer, although there are a few exceptions. Numerous com-
pounds containing the oligosaccharide sequence, -Galj?l-4GlcNAcp1-3-,are found
in erythrocytes, leukemia cells, spleen, granulocytes, liver and colon carcinoma cells,
meconium, and elsewhere. This sequence may be at the non-reducing end of the car-
bohydrate chain, as in paragloboside (Gal/?l-4GlcNAc~l-3Ga~l-4Glc~l         -l’Cer), or
internal as in a neolacto type murine leukemia cell glycosphingolipid (Gala1-3Galpl -
4GlcNAc,8l-3[GalNAc~1-4]Ga~1-4G1~~1-1’Cer). may be repeated, as
                                                    The sequence
in a complex glycosphingolipid of rabbit erythrocytes, Gala1-3Gal~1-4GlcNAc/31-
3Gapl- 4GlcNAcpl- 3[Galal- 3 G a p l - 4GlcNAq91- 6lGav1- 4GlcNAcp1- 3Gaw1-
4Glcpl- l’Cer, and is common in fucose-containing neutral glycosphingolipids such
as a Lewis’ blood group-specific antigen from human adenocarcinoma, Galpl-
4[Fucal-3]GlcNAc~i-3Gal/3l-h;4Gl~l-1‘Cer, extremely large blood group
                                                 and an
A-active glycosphingolipid from human erythrocyte membranes, shown below. L-
Fucose (Fuc) is 6-deoxy-~-galactose.
Ga I NAcul-3Galpl-   4GI cNAcp1- 3Ga Ipl- 4GlcNAc,?l/
              2                    2
              I                    I
         Fucul                Fucul

2.3.2. Acidic glycosphingolipids Gangliosides. Gangliosides are essential membrane-bound glycosphin-
golipids of all vertebrates, occurring in highest concentrations in the central nervous
system. The carbohydrate chains of gangliosides are diverse in structure, but have in
common one or more units of an acidic sugar called sialic acid (Fig. 7). Sialic acid,
which is biosynthetically derived from N-acetylmannosamine and pyruvic acid, is a
C9 sugar containing a pyranose ring formed by a hemiketal linkage of the carbonyl
unit at C2 with an OH group at C6. In glycosphingolipids (and glycoproteins) sialic


          y 2            HO\$

CH,CONHFH                   H’HO   H

       HOF H         CHaCONH
        HFOH                             H

Fig. 7. Fischer projection (left) and Haworth (right) structures of N-acetylneuraminic acid.

acid residues have a glycosidic linkages to other sugars. Sialic acids may have N-
acetyl or N-glycolyl groups at C5, and are distinguished by the names N-acetylneura-
minic acid (NeuSAc) and N-glycolylneuraminic acid (NeuSGc). There may be addi-
tional acetyl substituents on sialic acid at C7, C8 or C9 [lo].
   The simplest ganglioside contains one sialic acid and one hexose molecule, which
may be galactose (brain) or glucose, as found in an abundant glycosphingolipid in
eggs from sea urchin, Anthocidaris crassipina:

     NeuSGca2-6Glc/31-l’Cer               human brain
     NeuSAca2-3Galpl-l’Cer                sea urchin

   The complete chemical structure of the sea urchin ganglioside was deduced by a
combination of one-dimensional and two-dimensional proton NMR, FAB mass
spectrometry, neuraminidase digestion, gas chromatographic analysis of the constit-
uents of the ceramide fraction, and gas chromatographic analysis of the partially
methylated glucitol acetate obtained by permethylation, hydrolysis, borohydride re-
duction, and acetylation. Full discussion of the methods of structural analysis is
beyond the scope of this textbook. The reader is referred to the original paper [I 13 on
the structure proof of this simple ganglioside and to its bibliography for details.
   Most gangliosides are derived from lactosylceramide. The addition of one sialic
acid gives GM3                                                      which
                 ganglioside, NeuSAca2-3Ga~1-4Glc~l-l’Cer, is a common
component of the ganglioside fraction from many biological sources. Gangliosides
frequently contain a string of two or three sialic acid residues, attached to each other
in a2-8 glycosidic linkages, examples of which are GD3 GT3:and

     Neu5Aca2-8Neu5Aca2-3Ga1/31-4Glcj3                                                         GD3
     NeuSAca2-8NeuSAca2-8Neu5Aca2-3Gal~   1-4GlcD1-l’Cer                                       GT3

  Alternatively, sialic acid residues may occur singly, substituted at different posi-

tions on the oligosaccharide, such as in a tetrasialosyl neolucto type ganglioside from
human placenta (S. Hakomori, 1989).
                                                        Neu5Acu2- 3Galpl- 4GlcNAcpr

                             Neu5ArmZ- 3GalPl-4GlcNAcpi                                   ‘3/6Gal~1-4GicNAcl)l-3Gol~l-4Gic/)l-l’Cer
Neu5Aca2 - 3GaIp1-4GIcNAcpl                                        ‘3/6GaIpl-4GlcNAcl)l
Neu5Aca2 -3Galp1-4GIcNAcpl

   Until recently, gangliosides could be classified into a few related series, such as the
gunglio and neolucto types, with the sialic acids usually being substituted onto one or
both root galactose residues of the gunglio type, as for example in the disialosyl gan-
gliotetraosylceramide from human brain (GD,J. However, the discovery of ‘unusual’
arrangements, such as that in a frog brain ganglioside containing a sialic acid residue
linked to N-acetylgalactosamine, was an important advance by Ohashi and others,
signalling the possibility of other kinds of naturally occurring structures. Such gan-
gliosides are being found with increasing frequency using immunostaining of thin-
layer chromatography plates with selective antibodies and highly sensitive methods
of structural analysis. Three structures of this type are now known to be constituents
of adult and fetal brain (G,,, and G,,,), frog brain (G,,,), dogfish brain (GT,), rat
ascites hepatoma (G,,, and G,,,), and subpopulations of murine spleen lymphocytes
(GILllh ‘2nd G-rla).

                 Gal/31-3GalNAc/?1-4Gal/?1-4Glc~1-1                            ‘Cer
                         I                              I
 NeuSAcu2        -3                 Neu5Acu2-3                                                GD1,

                Gal/31--3GalNAc/?1-4Gal~1-4Glc/31 - l ’ C e r
 N e u 5 A c u 2 -3                                                                           GM,,

                 Gal/?l-3GalNAc/31-4Ga1/?1-4GIcpl-1                           ‘Cer            GDl cl

                Neu5Aca2 - 6
               GalPl - 3 G a l N A c / ? l - 4 G a l / ? l - 4 G l c P 1 - l ’ C e r          GTX
                     I                              I
N e u 5 A c n 2 -3               Neu5Aca2- 3 Phosphorus-containing glycosphingolipids. Several kinds of acidic glyco-
sphingolipid containing phosphorus have been found in lower organisms. The gly-
cosphingolipids of the blowfly, Calliphora vicina, are all of the urthro type; two sub-

stances a