Chemotaxis and Haptotaxis of Human Malignant Mesothelioma

Document Sample
Chemotaxis and Haptotaxis of Human Malignant Mesothelioma Powered By Docstoc
					    [CANCER RESEARCH 53.
A-PDF Watermark4376-4.1K2. September 15. 199.1]
                         DEMO: Purchase from to remove the watermark
     Chemotaxis and Haptotaxis of Human Malignant Mesothelioma Cells:
     Effects of Fibronectin, Laminin, Type IV Collagen, and an Autocrine
     Motility Factor-like Substance1
     Julius Klominek,2 Karl-Henrik Robert, and Karl-Costa Sundqvist
     De/hirlment of Lung Medium1 /./. K.¡,Medicine [K-H. R.j. unii Clinical lininiinolog\\ [J. K,, K-(!. S.[. Karolinska Institute til Huililin^e University Hospital,
     S-Ì41 t>Huddinge. Sweden

     ABSTRACT                                                                                  the highly invasive behavior clinically relevant métastasesare not
                                                                                               common in MM. Mesotheliomas are often accompanied by pleural
         A human malignant pleural mesothelioma cell line (STAY) was studied
                                                                                               effusion which may contribute to the local tumor spread. However, in
     with respect to production of the extracellular matrix components lumi
                                                                                               patients without effusion and when the tumor invades adjacent extra-
     nili, type IV collagen, and fibronectin, and interactions with these proteins
     in vitro. We also analyzed S'l'AV cell serum-free conditioned medium with                 thoracal locations, the dissemination probably requires an active lo
     respect to the possible presence of "autocrine motility factor-like" sub                  comotion of mesothelioma cells rather than reimplantation of floating
     stance. Sodium dodecylsulfale-polyacrylamide               gel electrophoresis of bio-    cells.
     synthetically labeled STAY serum-free conditioned medium showed that                         Tumor cell invasion is a complex multistep process that requires
     STAY cells released several proteins into the medium, including compo                     successful infiltration of tumor cells from the primary site through
     nents with molecular weights of 850,000, 540,000 and 440.000. Using                       tissue barriers. Invading MM cells most likely interact both with
     Western blotting we identified these proteins as luminili, type IY collagen,              soluble ECM molecules present in pleural fluid (5, 6) and with in
     and fihronectin, respectively. By immunocytochemistry                  laminin, type IY
                                                                                               soluble forms of such components in pleural and subpleural soft
     collagen, and fibronectin were detected as a matrix surrounding the cells.
     Plastic culture dishes coated with /ug quantities of luminili, type IY colla
     gen, and fibronectin induced attachment and spreading of STAY cells.
                                                                                                  Several substances that stimulate tumor cell locomotion HI vitro
     Laminin, type IY collagen, and fibronectin stimulated directional (chemo-                 have been described. These include factors derived from rcsorbing
     tactic) migration of STAY cells in Boyden chambers Pitted with 8 ¿im bone (7), liver, lung (8). and smooth muscle (9), as well as growth
     filters. The same cells also migrated to insoluble step gradients of filter-              factors such as insulin, insulin-related peptides (10), and bombesin
     bound extracellular matrix components (haptotaxis). When STAY serum-                      (11). Tumor cells can also respond to endogenous substances that in an
     free conditioned medium was separated by using fast protein liquid chro-                  autocrine fashion stimulate locomotion (12-14).
     matography Superóse 6 gel filtration, two motility-inducing protein peaks
                                                                                                  Normal mesothelial cells are known to produce ECM components
     were detected. The first peak contained proteins with molecular weight >
                                                                                               such as collagen type I, III, and IV, as well as elastin, laminin, and
     220,000 that had both chemotactic and haptotactic properties, while the
                                                                                               fibronectin (15). It has been shown that MM cells in vitro produce
     second peak contained material with apparent molecular weights of ap
     proximately 67,000 that had chemotactic and chemokinetic (random mo
                                                                                               collagens type I (16, 17) and IV (17) and the glycosaminoglycan
     tility) but not haptotactic properties. Analysis of the M, 67,000 material                hyaluronic acid ( 18). Pleural effusions of patients with MM have been
     indicated that it was a heat-sensitive and trypsin-digestible protein. The                found to contain high concentrations of soluble fibronectin (5). Fur
     production of both soluble and insoluble extracellular matrix components                  thermore, the amount of fibronectin was higher than in plasma, indi
     by human mesothelioma cells and the motile response to these molecules                    cating local synthesis (5). Another substance often present in pleural
     as well as the production of a M, 67,000 autocrine motility factor-like                   effusions of patients with MM is hyaluronic acid (6, 18-20).
     substance may be important for the highly invasive motile behavior of this                    Fibronectin, laminin, type IV collagen, and thrombospondin have
     tumor.                                                                                    been shown to stimulate locomotion of certain tumors, including
                                                                                               Schwann cell tumor (21), B16 mouse melanoma (22), and A2058
     INTRODUCTION                                                                              human melanoma (23, 24). Hyaluronic acid, which is present in MM
         MM' are rare malignancies arising from mesothelial-lined surfaces.                    both in vitro (18) and in vivo (20), has been proposed to enhance
                                                                                               tumor spread by increasing the fluid content of tissues and physically
     The most common location is the pleural cavity but peritoneum,                            "open up" matrices (25).
     pericardium, and paratesticular soft tissues may also be involved (1.
                                                                                                  Substances that stimulate cell locomotion can be presented to tumor
     2). Mesotheliomas spread rapidly along serosal surfaces to involve
                                                                                               cells either in soluble form or as an insoluble matrix. Chemotaxis,
     pericardium, contralateral hcmithorax, and peritoneal cavity by inva
                                                                                               directional cell movement, describes cell migration toward concen
     sion through the diaphragm. The resulting tumor will often form
                                                                                               tration gradients of solubilized attractants, while haptotaxis describes
     diffuse thickening of involved surfaces (3, 4) rather than solitary
                                                                                               cell movement in response to adhesive gradients on the substratum
     rounded lesions as seen in other neoplasms. MMs often invade
                                                                                               (26, 27). The important distinction between chemotaxis and haptotaxis
     through needle biopsy tracts or incisions in the chest wall (4). Despite
                                                                                               has recently been underlined by evidence that the signals for chemo
                                                                                               taxis and haptotaxis to type IV collagen are mediated by distinct
         Received 3/2/93; accepted 7/13/93.
         The costs of publication of this article were defrayed in part by the payment of page transduction mechanisms (24).
     charges. This article must therefore be hereby marked adtvrtisement in accordance with       The highly invasive behavior of mesothelioma cells taken together
     18 U.S.C. Section 1734 solely to indicate this fact.                                      with local abundance of both soluble and insoluble ECM components
          1This work has been supported by grants from The Swedish Cancer Foundation. The
     Swedish Medical Research Council. The Swedish Society of Medicine, the Swedish Heart      indicate local synthesis of ECM by MM and suggest that those inter
     and Lung Foundation, and the Alex and Eva Wallström Foundation.                          actions between MM and ECM can cause local tumor spread. Several
         2 To whom requests for reprints should be addressed, at Department of Lung Medicine.
     M 53. Huddinge University Hospital. S-14I 86 Huddingc. Sweden.                            investigators have previously described synthesis of ECM by tumor
          'The abbreviations used are: MM. malignant mesothelioma; BSA, bovine serum           cells (28-31). However, most of these studies concern melanomas or
     albumin; DTT. dithiotreitol; ECM, extracellular matrix; FPLC. fast protein liquid chro-   other solitary tumors and to our knowledge there are no in vitro studies
     matography; PBS, phosphate-buffered saline; SFCM. serum-free conditioned medium;
     SS. serum-supplemented medium; SDS-PAGE. sodium dodecyl sulfatc-polyacrylamide            of mechanisms involved in the invasion of MMs. We have previously
     gel electrophoresis; AMF, autocrine motility factor; FCS. fetal calf serum.               described establishment of MM cell lines (18). In the present paper we
                                                         MOTIL.[TY   OÃ--'MALIGNANT   MliSOTIIlil.lOMA   CïLLS

examine one of these cell lines with respect to production of different               uncoated plastic dishes (or coated with 10 ng/ml BSA) and less than 0.1%
ECM components and "motility factors." We further investigate in                      adhered to plastic dishes coated with 10 mg/ml BSA.
                                                                                         Chemotaxis was assayed by using 48-well chemotaxis chamber (Neuro
teractions with ECM components in order to elucidate whether these
                                                                                      Probe, Cabin John. MA) and 8 ¡imNucleopore filters (Pleasanton. CA) coated
could stimulate the locomotion of mesothelioma cells and thus explain
                                                                                      with 100 fig/ml gelatin on both sides. Different concentrations of fibronectin,
their distinctive invasive features.                                                  type IV collagen, and laminin were diluted into serum-free RPMI and added in
                                                                                      triplicate to the lower wells, while upper wells were filled with STAV cells,
                                                                                      2 X 10'Vml in serum-free RPMI containing 1 mg/ml BSA. The chambers were
MATERIALS          AND METHODS                                                        incubated for 15 h in a well humidified incubator at 37°C.Control wells

   Cells. Human malignant mesothelioma cell line STAY PCS (18) was cul                received 1 mg/ml BSA in the lower wells. At the end of the assay, filters were
tured in RPMI 1640 supplemented with 10% heat inactivated fetal calf serum.           removed, fixed in methanol, and stained with Giemsa. The filters were then
This cell line grows as a substrate adherent monolayer and displays a sarco-          placed onto glass slides and the cell pellets corresponding to the upper wells
matuus growth pattern. After reaching conflucncy the cells in some    areas form      were wiped with cotton swabs. Cells that had migrated through pores to the
a multilayer. Normal mesothclial cells were generously provided        hy Dr. P.      lower side of the filter were counted by light microscopy under high power
Heldin (Biomédical Center, Uppsala, Sweden) and cultured as          previously      field (400 X). For each triplicate, the number of cells in three high power fields
described (32).                                                                       was determined and the counts were averaged.
   Preparation of Serum-free Conditioned Media. STAY PCS              cells were          Haptotaxis was assayed in triplicate by using 48-well migration chambers
grown in SS medium until they were nearly confluent. SS medium was re                 and the same cell concentration as in chemotaxis assays. Before the motility
moved and the cell monolayer was extensively washed with Dulbecco's PBS               assays the uncoated 8 /im Nucleopore filters were placed in the chamber, the
in order to remove remaining serum proteins. The monolayer was then incu              dull side of the filter facing the upper wells. The upper wells were filled with
                                                                                      different concentrations of ECM components in serum-free RPMI, control
bated for 48 h in fresh RPMI without serum or any other protein. MeUibolically
labeled SFCM for gel electrophoresis was prepared in the same way, but                wells were filled with 1 mg/ml BSA. The lower wells remained empty. The
methionine-free RPMI was used, and supplemented with 10 /xCi/ml of [15SJ-             chambers were left overnight at 4°C,the filters were then removed and ex
                                                                                      tensively washed on both sides, first in Dulbecco's PBS, then in distilled water,
methionine (50 mCi/ml; Amersham Laboratory, Buckinghamshire, England).
                                                                                      air dried, and finally placed in 48-well chemotaxis chambers with the coated
   Separation of Serum-free Conditioned Medium. One liter of SFCM was
                                                                                      spots of the filler facing exactly toward the lower wells that were filled with
prepared, eentrifuged at 300 X g for 15 min, filtered through a 45 /¿mfilter to
remove cellular debris, concentrated 50-fold by using Amicon membranes YM             RPMI containing 1 mg/ml BSA. This procedure was not difficult as there were
                                                                                      clear pressure made marks on the filters indicating coated spots. We have also
30 (Amicon, Danvcrs, MA) that retain molecules a M, 30,000. dialyzed
                                                                                      tested previously (24) described coating technique by floating 13-mm diameter
against distilled water, and lyophilized. The lyophilized material was sus
                                                                                      filters on different concentrations of ECM components in serum-free RPMI,
pended in serum-free medium and used in chemotaxis and haptotaxis assays,
                                                                                      washed them as described above, and used in haptotaxis assays. We found that
or it was suspended in PBS, pH 7.4. and applied on FPLC Superóse 6 column
                                                                                      our coating technique performed in 48-well chambers did not affect the results
(Pharmacia, Stockholm, Sweden). The column was run ¡it ml/min and 1-ml
                                                                                      (not shown) as compared to coating achieved by floating 13-mm filters on
fractions were collected, dialyzed against serum-free medium, and used in
                                                                                      ECM components. Thus we used our coating method because it was easier,
ehemotaxis and haptotaxis assays (see below).
   Antibodies. Monoclonal anti-fibronectin antibody IST-1 was obtained                faster, and required smaller amounts of ECM.
                                                                                          FPLC fractions were either placed in the lower compartment of the Boyden
from Sera Laboratory (Crowley Down, England), working dilution 1:100. A
monoclonal anti-laminin antibody was purchased from Boehringer Mannheim               chamber or were coated on Nucleopore filters as described above. Protein
(Mannheim. Germany), working dilution 1:20; anti-laminin polyelonal anti              content was determined as described (35). In all adhesion and motility assays
                                                                                      the subconfluent cells were detached by brief incubation in 0.25% trypsin/0.
body was purchased from Collaborative Research (Bedford. MA), working
dilution 1:300(1. A monoclonal anti-type IV collagen antibody was purchased           02% EDTA and then allowed to regenerate for at least l h in SS medium at
                                                                                      room temperature. In experiments involving synthetic peptides or pertussis
from Dakopatts a/s (Glostrup. Denmark), working dilution 1:30; anti-type IV
                                                                                      toxin the cells were preincubated for I h with indicated concentrations of
collagen polyelonal antibody was from Collaborative Research, working dilu
                                                                                      GRGDS, GRGES. or pertussis toxin (List Biological Laboratory, Campbell,
tion 1:500. A monoclonal anti-CD-3 antibody was from Becton Dickinson
                                                                                      CA) in scrum-free RPMI. The motility assays were then performed in contin
Immunocytochemistry Systems (Mountain View, CA). working dilution 1:100.
                                                                                      ued presence of the peptides or pertussis toxin. The peptides were made hy
   Gel Electrophoresis and Western Blotting. Metabolieally labeled SFCM
                                                                                      using synthesizer (Model %00: Bioresearch. San Rafael, CA) and generously
(50 ml), was prepared as described above. SFCM was then separated on
                                                                                      provided by Dr. E. C. Kohn, (Bethesda, MD).
3.5-8% gradient SDS-PAGE gels under reducing (+50 mM DTT) or nonre-
ducing (-DTT) conditions. The gels were prepared hy using the buffer system
of Laemmli (33). In order to characterize different bands, the gels were elec-        RESULTS
troblotted to nitrocellulose as previously described (34) and stained with the           Column Chromatography of Motility-stimulating           Factors in
use of antibodies to ECM components. The reactions were detected by using             STAV SFCM. Unseparated mesothelioma cell SFCM was both che-
the biotin-avidin-peroxidase   system (Vectastain PK 4002 ABC Kit, Vector
                                                                                      motactic and haptotactic for STAV cells in a dose-dependent fashion
Laboratories, Burlmgame. CA). Enzyme label was demonstrated             by using
3-amino-y-ethyl-earbazole (Sigma, St. Louis, MO).                                     (Fig. 1). The strongest motility response was found when SFCM was
    Immunocytochemistry.       Mesothelioma cells were cultured on glass slides       concentrated 100 times, but even unconccntratcd SFCM augmented
inserted into the 6-well culture plates until confluence was reached. The cells       the motility of STAV cells. In order to investigate the possibility that
growing on glass slides were fixed in ice cold methanol and stained by using          the motility-inducing activity in SFCM was a scrum contaminant we
anti-fibronectin. anti-laminin. and anti-type IV collagen monoclonal antibod          used different dilutions of SS medium (20% FCS-0.01% PCS) as
ies. Monoclonal anti-CD-3 antibody Leu-4 was used as a negative control. The          attractants in motility assays. None of these concentrations of SS
reactions were detected by using the biotin-avidin-peroxidase      system (see        medium induced motility of STAV cells which excludes that serum
above).                                                                               factors were responsible for the motile response of STAV cells (not
    Cell Adhesion and Motility Assays. Adhesion assay was carried out in              shown). In order to characterize the factors responsible for cell trans-
triplicate in Petri dishes coaled overnight at 4°Cwith 10 fig/ml of fibronectin
                                                                                      location we separated mesothelioma SFCM proteins by using molecu
(Boehringer Mannheim), laminin (Sigma), or type IV collagen (Sigma). Con
trol dishes were coated with 10 mg/ml BSA (negative control) or 10 /xg/ml
                                                                                      lar sieve Chromatography on a FPLC Superóse 6 column. All fractions
                                                                                      were assayed for motility-inducing activity in chemo- and haptotactic
BSA (positive control). One million STAV cells in RPMI containing 1 mg/ml
BSA were added to each dish and incubated. After 40 min the nonadherent cells         assays. Two peaks that induced significant motile responses were
were removed and counted by using trypan blue exclusion, and the percentage           found (Fig. 2). The first peak corresponded to proteins with molecular
of adherent cells was calculated. In all experiments 100% cells adhered to            weights >200,000. These fractions caused both chemotactic (a) and
                                                                 MOTILITY OF MALIGNANT         MESOTHELIOMA          CELLS

                        80-                                                                        The substance used to coat the filters was important for the detec
                                                                                               tion of chemotactic activity. Thus, the first peak of motility-inducing
                                                                                               activity could be detected on both gelatin- and type IV collagen-
                        60-                                                                    coated filters. However, in order to identify the second peak of mo
                                                                                               tility-inducing activity it was necessary to use filters coated on both
                                                                                               sides with 40 /ng/ml of type IV collagen. Furthermore, if gelatin-
                        40-                                                                    coated filters were used the motile responses recorded were weak or
                                                                                               absent (not shown in Fig. 2), indicating that the presence of type IV
                                                                                               collagen on the filters was necessary for motility of MM cells to the
                                                                                               M, 67,(HH)material. In control experiments we found that coating of
                                                                                               filters on both sides with type IV collagen did not induce motility of
                        10-                                                                    mesothelioma cells as compared to haptotactic migration to type IV
                                                                                               collagen (see below).
                         0                                                                         The motility-inducing activity in M, 67,000 fractions was inacti
                              I            10            100            1000
                                                                                               vated by heat and protease. We further observed that if STAV cells
                                     Concentration of SFCM (x)                                 were pretreated with ng concentrations of pertussis toxin the motile
    Fig. I. Migration of STAV cells to their own serum-free conditioned medium. Condi          response to fractions containing high molecular weight material (first
tioned media were prepared as described in "Materials and Methods." Lyophilized SFCM
was dissolved in fresh serum-free medium (2(K) x concentrated SFC'M) and then diluted.         peak) remained virtually unaffected (not shown), whereas the motile
For determination of haptotaxis (D), different dilutions of SFC'M were used to coal the        response to fractions within the second peak was completely abol
filters,    while for determination of chemotaxis ( 0 ). different dilutions of SFCM were      ished. This observation indicates that the motility-inducing signals
placed      in the lower compartment of Boyden chambers. Migration to control proteins
(BSA,       1-1000 ngjml, or different concentrations of RPMI + PCS) was 0-1 cells/high        delivered by molecules in the first and the second peak are mediated
power      field (HPF).                                                                        by different transduction mechanisms. The total inhibition of chemo
                                                                                               taxis by pertussis toxin to fractions of molecular weight «=67,000
                                                                                               indicates that G-proteins are involved in signal transduction.
                       M r            200 116 90 67 43      13                                     Detection of ECM Components in STAV SFCM by SDS-PAGE.
                                                                                               In order to characterize the high molecular weight substances respon
                                                                                               sible for the chemotactic and haptotactic migration of mesothelioma
                                                                                               cells we separated STAV [35S]methionine-labeled SFCM on 3.5-8%
                                                                                               SDS-PAGE gels under reducing and nonreducing conditions (Fig. 4).

                                                                                                                             SF          10 ug        20 ng         40 ug

                                                                                                                        v     5

    Fig. 2. Determination of motility-inducing activity in STAV serum-free conditioned
medium separated by using FPLC Superóse t>column chromatography. The dark areas
denote fractions containing motility-inducing activity. Numbers within the graph, number                           so
of chemotactically and haptotactically migrating cells. For determination of chemotaxis                                                    ^6027Vv\702624V\7^
(a), the fractions were pooled and placed in the lower compartment of the Boyden                              8
chamber. In determinations of haptotaxis (b). STAV cells migrated to pooled fractions
coated onto the distal surface of filters. Top abscissa, molecular weight in thousands.

haptotactic (b) responses in a concentration-dependent                      fashion (not
shown). The second peak had an apparent molecular weight of 67,000.
To assess whether the motility induced by the M, 67,000 peak was                                   Fig. 3. Checkerboard analysis of migration induced by pooled FPLC Superóse 6
directional or random we performed a checkerboard analysis. In this                            fractions with M, 67,(KH).Varying concentrations of pooled FPLC fractions were added to
assay cells are exposed to different gradient conditions obtained by                           the upper chamber with the cells or to the lower chambers, as indicated. The M, 67,tKM(
                                                                                               protein acts in a dose-dependent fashion and is both chemotactic (values below the
varying the amounts of attractant in the upper and lower compart                               diagonal) and chemokinetic (values on the diagonal). SF, serum-free medium.
ments of the Boyden chamber. Pooled fractions within this peak
induced both chemotactic (to a gradient) and chemokinetic (in the                                                                     o
                                                                                                                  Table I Properties• f pooled M, 67,000 FPLC fractions
absence of the gradient) motility (Fig. 3). The M, 67,000 peak was
                                                                                                          Treatment                                                 No. of migrated cells
further characterized as shown in Table 1.
                                                                                                                                                                          57 ±7°
   We also examined if mesothelioma-derived material corresponding                                 None (control)
                                                                                                   1WC 5 min*                                                                5±2
to M, 2(X),()00 and M, 67,000 motility peaks affected the migration of                             Trypsin, KM)u,g/ml'                                                       5±2
normal mesothelial cells. In this experiment normal mesothelial cells                              PMSF, 5 mm                                                               55 ±6
                                                                                                   Pertussis toxin, (1.5 u.g/ml''                                            6±2
were placed in the upper compartment of Boyden chambers and
                                                                                                  " Mean ±SD.
exposed to pooled FPLC fractions containing Mr 200,000 and Mr                                     h Pooled fractions were boiled for 5 min.
67,000 material, respectively. Normal mesothelial cells showed a                                  ' Pooled fractions were incubated with trypsin for l h at 37°Cand then 5 ÕTIM
weak motile response to the high molecular weight ECM-containing                               phenylmethylsulfonyl fluoride (PMSF) was added to inactivate the enzyme. PMSF alone
                                                                                               was used as a control.
peak, whereas no migration was observed to M, 67,000 material (data                               '' STAV cells were preincubated with pertussis toxin for I h and placed in the upper
not shown).                                                                                    compartment of a Boyden chamber; pooled FPLC fractions were used as chemoattraclant.
                                                                  MOTILITY      OF MALIGNANT      MESOTHELIOMA       CELLS

                                                                                                  which was used as a control. Other bands detected on top of the gel
      850                                                                                         had molecular weights in the range 500,()0()-2()(),(K)0.
                                                                                                    SDS-PAGE-separated STAY [35S]methionine SFCM was electrob-
                                                                                                  lotted onto nitrocellulose and reacted with antibodies to laminin,
                                                                                                  fibronectin, and type IV collagen. This identification, using antibodies
      400     >•                                                                                against each ECM component, demonstrated that the mesothelioma
                                                                                                  cells secreted laminin, fibronectin, and type IV collagen into the
                                                                                                  culture medium. Under nonreducing conditions, anti-laminin poly
       200                                                                                        clonal antibody recognized a M, 850,000 band. Two different type IV
                                                                                                  collagen antibodies recognized a M, 540,000 band, while fibronectin
                                                                                                  antibody recognized a M, 440,000 band. In all cases the antibody
                                                                                                  reactive bands corresponded to fluorogram bands, indicating that they
                                                                                                  represented endogenous ECM components. Under reducing condi
       116                                                                                        tions laminin was detected as a M, 200.000 band corresponding to the
                     1                                                      8                     ß chain of the molecule. The Mr 400,000 a chain of laminin stained
    Fig. 4. Separation of proteins released by STAY cells using 3.5-8% SDS-PAGE gels.             weakly. Fibronectin was detected as a M, 220,000 band. The mono
Fluorogram of STAY [15S]methionine scrum-free conditioned medium. Lane 1 (-DTT)
                                                                                                  clonal anti-type IV collagen antibody did not detect reduced forms of
and Lane 6 ( + 50 HÃŒM   DTT). Immunoblol stained with anti-CD-3 monoclonal antibody
(Lane 2). Immunoblot stained with anti-laminin polyclonal antibody. Lane 3 (-DTT) and             this molecule (not shown), while the polyclonal anti-type IV collagen
Lane 7 ( + DTT). Immunoblot stained with anti-fibronectin monoclonal antibody. Lane 4             antibody recognized Mr 180,000 and Mr 230,000 bands. A third anti
(-DTT) and Lane 8 ( + DTT). Immunoblot stained with anli-lypo IV collagen polyclonal
antibody. Lane 5 (-DTT) and Lane 9 ( + DTT). Ordinate, molecular weight in thousands.             type IV collagen polyclonal antibody (generously provided by Dr. D.
                                                                                                  Shuppan, Berlin, Germany) also detected M, 180,000 and M, 230,000
                                                                                                  bands (not shown).
Several high molecular weight bands were detected by using fluorog-                                  Immunocytochemistry.      STAV cells cultured on glass slides were
raphy, showing that these components were indeed produced by the                                  reacted with antibodies against fibronectin, type IV collagen, and
mesothelioma cells. The component with highest molecular weight                                   laminin, (Fig. 5). The strongest immunoreactivity was observed with
(>850,000) comigrated with commercially available laminin (Sigma)                                 fibronectin and laminin antibodies. Fibronectin was detected as an

                                                                                  •¿-*-.                                            V       ;"7H              S


                                                            ""*      -- "

             Fig. 5. Immunoslaining   nf STAV cells with anli-fibronectin       (a), anti-type IV collagen (h), anli-laminin (c) and anti-CD-3 monoclonal antibodies (d) (X243).

                                                                MOTILITY    OF MALIGNANT       MESOTHEL1OMA     CELLS

extracellular matrix particularly in areas where cells formed a multi                          25 /u,g,respectively. Substratum-bound laminin was weakly haptotac
layer but was also seen around individual cells. In areas where cells                          tic for mesothelioma cells with peak activity at 12.5 jag. The biphasic
grew sparsely spread on the substratum there was relatively weak                               motile response of STAV cells to ECM components parallels findings
fibronectin immunoreactivity. Laminin immunoreactivity was present                             in B16 and A2058 malignant melanoma cells (22, 24).
as a matrix in areas of multilayer formation and as a diffuse staining                            Coating of the lower surface of the filter with BSA 1-1000 ju.gor
throughout the culture. Type IV collagen immunoreactivity was also                             serum-free RPMI did not induce translocation of mesothelioma cells.
most pronounced in areas were cells formed multilayers and was                                 In several triplicate experiments merely 0-1 cell migrated to any BSA
distinctly located as a matrix on and between individual cells. As with                        concentration or to serum-free RPMI coating of the lower filter sur
the fibronectin matrix, the intensity of the type IV collagen immuno                           face. Coating of filters with ECM components 1-200 fig on both sides
reactivity varied in different areas of the culture, indicating that local                     induced migration of only 0-5 cells to the lower filter surface (not
environmental factors such as cell density and multilayer formation                            shown).
probably play a role for production and assembly of these molecules.                              In order to study if the tripeptide arginine-glycine-aspartic   acid
   Adhesion and Motility of STAY Cells to ECM Components.                                      (ROD) (37), known to constitute a cell-binding site in three of the
STAV cells exhibited adhesion and spreading on ECM components.                                 tested ECM components, was necessary for motile responses of me
The highest number of attached cells was observed when Petri dishes                            sothelioma cells in the Boyden chamber experiments described above,
were coated with fibronectin and type IV collagen with 90 and 85%                              we pretreated the cells with different concentrations of GRGDS or the
attached cells, while only 30% cells adhered to laminin. Virtually all                         control peptide GROES (Fig. 7). Chemotactic (25 ^g/ml) and hapto
attached cells showed pronounced cytoplasmic spreading.                                        tactic migration to fibronectin (100 ;u.g) was inhibited by GRGDS in
   The motility of STAV cells was tested in multiwell chemotaxis                               concentrations 500-31 ju.g/ml, whereas different concentrations of
chambers. Relatively long incubation times were required since no or                           GRGES had no effect. These observations indicate that the ROD
very few cells translocated during initial experiments when the incu                           sequence is an important binding site for mesothelioma cells migrat
bation times were 4—6 The optimal time necessary for STAV cells                              ing to soluble or insoluble fibronectin. No significant inhibition of
to respond to attractants was 15 h. It has previously been reported that                       laminin or type IV collagen-stimulated motility was observed when
tumor cells may need different incubation times in order to translocate                        mesothelioma cells were incubated with the pentapeptides (not
through filter pores (36). STAV cells migrated in a concentration-                             shown), indicating that the ROD sequence present in these molecules
dependent fashion toward soluble fibronectin, laminin, and type IV                             does not comprise a major binding site necessary for migration.
collagen (chemotaxis) (Fig. 6A). The most potent chemoattractant for
mesothelioma cells was fibronectin, which at a concentration of 1.5
/xg/ml caused cell migration as compared with BSA controls. Peak
stimulation was observed at 25 jug/ml whereas higher fibronectin                                  This study shows that the human malignant pleural mesothelioma
concentrations resulted in gradually weaker motile responses. Soluble                          cell line STAV PCS produces fibronectin, laminin, and type IV col
type IV collagen also stimulated mesothelioma cell migration with a                            lagen as detected by biosynthetic incorporation of [1;iS]methionine,
peak motile response at 100 jag/ml. A similar concentration-dependent                          Western blotting, and immunocytochemistry. The results further show
stimulation of STAV cell motility was observed with laminin, al                                that these ECM components in an autocrine fashion can induce ad
though laminin induced a weaker chemotactic response than fibro                                hesion, spreading, and chemotactic and haptotactic responses in the
nectin and type IV collagen.                                                                   producer cells.
   STAV cells also migrated toward step-density gradients of ECM                                  In addition, we have detected an activity with apparent M, 67,000
components bound to the lower surface of filters (Fig. 6B). Fibronec                           which is both chemotactic and chemokinetic and thus similar to the
tin and type IV collagen were the most potent haptotactic agents with                          autocrine motility factors previously described (14, 38), but different
maximal stimulation of motility at a coating concentration of 100 and                          from ECM components which are chemotactic and haptotactic.


                                  80-                                                                80-




                           -.     30-
                           g      20-


                                                     10              100             1000                               10                             1000

                                                       ECM cone ug/ml                                                       Coating cone ng
    Fig. 6. Chemotactic (A) and haptotactic (B) migration of STAVcells in response to ECM components. In chemotactic experiments increasing concentrations of fibronectin (A),
collagen type IV ( 0 ), and laminin (O) were diluted into serum-tree RPMI and added to the lower wells. In the determination of haptotaxis the Nucleopore filters were precoated with
a step gradient of increasing concentrations of fibronectin (A), collagen type IV ( 0 ). and laminin (O). Migration to any concentration of control protein BSA was 0-1 cell/high power
field (HPF).
                                                                  MOTILITY    OF MALIGNANT          MESOTHELIOMA    CELLS


                            80-                                                                          80-

                            70-                                                                          70-

                            60-                                                                     x    60-

                            50-                                                                     ^    50 -

                            40-                                                                     I    40-

                            30-                                                                     ¡ 30-
                            20-                                                                     § 20-

                             10-                                                                         10-

                              0                                                          1                0
                                   10                       100                        1000                    10                       100                        1000

                                                  Peptideconc (Hg/ml)                                                         Peptide cone (u,g/ml)
    Fig. 7. Effects of peplidcs GRGDS (U) and GROES ( 0 ) on chcmolaclic and haptotactic migration of STAV cells to fihronectin. In these experiments STAV cells were preincubated
for l h with indicated concentrations of peptidcs and allowed to migrate to cither soluble fibronectin (25 fig/ml) (A ) or to fibroncctin (10(1 fig/ml) that was coated on the distal surface
of the filters (ß).The peptides were allowed to remain with the cells during the assay. HPF. high power field.

   STAV cells secrete a large number of biosynthetically labeled pro                             previously described for Schwanoma and human melanoma cells (21,
teins into the medium. Three of the proteins were characterized as                               22). Furthermore, when STAV SFCM was separated on a FPLC Su
disulfide-linked ECM molecules located in the high molecular weight                              peróse 6 column, high molecular weight fractions containing ECM
region on SDS-PAGE gels. The largest molecule (A/r 850,000 before                                components were found to stimulate both chemotaxis and haptotaxis,
and M, 200,000 and M, =400,000 after reduction) was identified as                                indicating that these molecules have the ability to stimulate mesothe-
laminin. Laminin is known to have a M, of 850,000 and after reduc                                lioma cell motility in an autocrine fashion. Control experiments with
tion of S—S bonds to form a Mr =400,000 a chain and two ß       chains                        BSA or serum-free medium alone unequivocally showed that ECM
                                                                                                 components are "true" chemotactic and haptotactic agents.
with M, 230,000 and 240,000, respectively, which often form a single
broad band (39). The a chain is usually difficult to detect on Western                               Human mesothelioma cells in vivo are often imbedded in a diffuse
blots and often yields fainter staining than the ß  chains or may even                          collagenous matrix (1,2). The tumor and the surrounding matrix form
be absent (39). The next high molecular weight component secreted                                flat sheets that spread along serosal surfaces. Thus, malignant meso
by STAV cells was identified as type IV collagen. The monoclonal                                 thelioma matrix has a different distribution than desmoplasia, an ex
anti-type IV collagen antibody used did not detect reduced forms of                              cessive matrix formation that encapsulates certain solid tumors (42).
this molecule, probably due to the fact that it is prepared for identi                           The possible origin of increased matrix formation that surrounds some
fication of the intact molecule in tissue sections rather than in Western                        tumors is a subject of controversy. It has been suggested that the
blotting. After reduction the polyclonal type IV collagen antibody                               invasive breast carcinoma cells secrete the extracellular matrix (43),
recognized M, 230,000 and MT 180,000 bands. The M, 180,000 com                                   but it has also been argued that source of desmoplasia were host cells
ponent corresponds to the previously described reduced form of type                              (44). Due to a different distribution of the matrix surrounding some
IV collagen (28). The M, 230,000 component is probably insuffi                                   tumors such as breast carcinomas, which are often encapsulated, and
ciently reduced collagen or a contaminant. The third high molecular                              that seen in mesotheliomas, we consider it likely that mesothelioma
weight band detected by fluorography had a molecular weight of                                   cells may use self-secreted matrix as a substrate for invasion.
440,000 and after reduction of disulfide bonds a molecular weight of                                 Pleural fluid of patients with mesothelioma contains high concen
220,000. These bands were detected on Western blots with anti-                                   trations of fibronectin. Our data supports this evidence (5) that fibro
fibronectin monoclonal antibody and are in good agreement with                                   nectin is locally secreted and indicate that the source is the malignant
previously described staining patterns of this molecule (40).                                    mesothelioma tumor itself.
   STAV cells also produced an insoluble matrix of fibronectin, lami                                 An additional observation in the present study is the partial char
nin, and type IV collagen associated with the cell surface, as detected                          acterization of M, 67,000 FPLC Superóse 6 fractions that induced both
by immunocytochemistry. This matrix of fibronectin, type IV colla                                chemotactic and chemokinetic responses in mesothelioma cells. This
gen, and laminin immunoreactivity was associated with areas of high                              activity might be a serum contaminant that remained on the cells and
cell density and multilayer formation. This observation is consistent                            in the culture dish despite extensive washing with PBS. However,
with previous reports of presence of fibronectin preferentially at the                           different dilutions of serum containing medium did not stimulate the
surface of confluent cells as compared to sparse cultures (reviewed by                           motility of STAV cells. Therefore we conclude that the motility factor
Hynes, Ref. 41). In addition to the matrix formation in some areas                               in M, 67,000 fractions must be produced by STAV cells. An important
laminin showed a homogeneous distribution throughout the culture.                                difference which distinguishes the M, 67,000 factor from the high
From these data we conclude that STAV cells produce both soluble                                 molecular weight peak of locomotor activity is the absence of hapto
and insoluble forms of fibronectin, laminin, and type IV collagen.                               tactic activity, and the presence of a chemokinetic activity. The mo-
   The data presented in this paper demonstrate a direct role of ECM                             tility-inducing activity of pooled M, 67,000 fractions was heat labile
components in mediating adhesion, spreading, and chemotactic and                                 and degradable by trypsin. When STAV cells were pretreated with
haptotactic responses of STAV cells in concentrations similar to those                           nanogram concentrations of pertussis toxin the motility toward M,
                                                                   MOTILITY    OF MALIGNANT         MESOTHELIOMA      CELLS

67,000 fractions was lost, indicating that G proteins are involved in                               17. Kataoka, H., Wickström, B., Klominek. J.. Gay, R. E., Gay. S., and Hjerpe, A.
signal transduction while motility toward high molecular weight ECM                                      Immunocytochemical demonstration of collagen types I and IV in cells isolated from
                                                                                                         malignant mesothelioma and in lung cancer cell lines. Lung Cancer, 6: 16-27, 1990.
components became largely unaffected. From this observation we                                      18. Klominek. J.. Robert. K-H., Hjerpe, A.. Wickström. B., and Gahrton. G. Serum-
conclude that the M, 67,000 fractions contain a protein with properties                                  dependent growth patterns of two, newly established human mesothelioma cell lines.
                                                                                                         Cancer Res., 49: 6118-6122. 1989.
similar to a previously described autocrine motility factor (14, 38, 45).
                                                                                                    19. Mayer, K., and Chaffee. E. Hyaluronic acid in the pleural (luid associated with a
This autocrine motility factor is defined as a substance that in an                                      malignant tumour involving the pleura and peritoneum. J. Biol. Chem., 133: 83-91,
autocrine fashion induces both random and directional motility (45).                                     1940.
However, one major difference compared to previously described                                     20. Wagner, J. C.. Munday, D. E., and Haringlon. J. S. Histochemical demonstration of
                                                                                                         hyaluronic acid in pleura! mesotheliomas. J. Pathol. Bacterio!., 84: 73—78.1962.
AMF is the inability of the M, 67,000 fractions to induce cell trans-                              21. McCarthy. J. B.. Palm. S. L., and Furcht, L. T. Migration by haptotaxis of a Schwann
location on uncoated or gelatin-coated filters. Melanoma and prostatic                                  cell tumor line to the basement membrane glycoprotein laminin. J. Cell Biol., 97:
carcinoma AMF can be detected on gelatin-coated filters (14, 46) and                                     772-777, 1983.
                                                                                                   22. McCarthy, J. B.. and Furchl, L. T. Laminin and fibronectin promote the haptotactic
breast carcinoma AMF is equipotent on uncoated, gelatin-coated, or                                                                                             C
                                                                                                         migration of B16 mouse melanoma cells in vitro. ¡. ell Biol., 98: 1474-1480. 1984.
ECM-coated filters (47). HT-1080 fibrosarcoma-derived AMF also                                     23. Taraboletti, G., Roberts. D. D.. and Liotta. L. A. Thrombospondin-induced        tumour
induces cell translocation on uncoated substrata (48).                                                  cell migration: haptotaxis and chemotaxis are mediated by different molecular do
                                                                                                        mains. J. Cell Biol., 105: 2409-2415, 1987.
   An intriguing possibility based on the present observations is col                              24. Aznavoorian. S.. Stracke. M. L., Krutzsch, H., Schiffmann, E., and Liotta. L. A.
laboration of the M, 67,000 material and ECM components. Thus,                                          Signal transduction for chemotaxis and haptotaxis by matrix molecules in tumour
                                                                                                        cells. J. Cell Biol., 110: 1427-1438, 1990.
type IV collagen was required for motile response of MM cells to the
                                                                                                   25. Pauli, B. U., Schwarz, D. E., Thonar. E. J-M.. and Kuetlner, K. E. Tumour invasion
motility-stimulating substance present in the Mr 67,000 FPLC frac                                       and host extracellular matrix. Cancer Mestasis Rev., 2: 129-152, 1983.
tions. This observation is similar to the finding that certain growth                              26. Carter, S. B. Principles of cell motility: the direction of cell movement and cancer
                                                                                                        invasion. Nature (Lond.), 208: 1183-1187, 1960.
factors such as Activin A and basic fibroblast growth factor act as
                                                                                                   27. Carter. S. B. Haptotaxis and the mechanism of cell motilily. Nature (Lond.), 213:
mitogens for mouse embryonal carcinoma cells provided that these are                                    256-260, 1967.
cultured on fibronectin or laminin-coated substrata (49). No mitogenic                             28. Scarpa, S., Modesti, A., and Triche. T. J. Extracellular matrix synthesis by undiffer-
                                                                                                        entiated childhood tumour cell lines. Am. J. Pathol., 729: 74-85, 1987.
effect was observed in the absence of ECM components.
                                                                                                   29. Alitalo, K.. Vaheri, A.. Krieg. T.. and Timpl. R. Biosynthesis of type IV procollagen
   In conclusion, cultured MM cells synthesize soluble and insoluble                                    and of other basement membrane proteins by human tumour cell line. Eur. J. Bio
ECM components and a A/r 67,000 autocrine motility factor-like sub                                      chem., 109: 247-255, 1980.
stance. These secreted substances induce a motile response in the                                  30. Barsky, S. H., Layfield. L.. Varki. N.. and Bhuta, S. Two tumours with basement
                                                                                                        membrane-producing potential. Cancer (Phila.). 61: 1798-1806, 1988.
producer cells, indicating a possible autocrine motility-stimulating                               31. Stenman, S., and Vaheri. A. Fibronectin in human solid tumours. Int. J. Cancer. 27:
loop. These observations may be of importance for the highly inva                                       427^05, 1981.
                                                                                                   32. Asplund. T., Versnel. M. A.. Laurent, T. C.. and Heldin. P. Human mesothelioma cells
sive, motile behavior of MM cells in vivo.
                                                                                                        produce factors that stimulate the production of hyaluronan by mesothelial cells and
                                                                                                        fibroblasts. Cancer Res.. 5.?: 388-392. 1993.
REFERENCES                                                                                         33. Laemmli, U. K. Cleavage of structural proteins during assembly of the head of
                                                                                                        bacteriophage T4. Nature (Lond.), 227: 680-685, 1970.
 1. Animan, K. A. Clinical presentation and natural history of benign and malignant                34. Sundqvist. K-G., and Otteskog, P. A labile complex at the lymphocyte surface. The
    mesothelioma. Semin. Oncol., K: 313-320, 1981.                                                      effect of the substratum on its expression. J. Immunol., 138: 825-832, 1987.
 2. Suzuki, Y. Pathology of human malignant mesothelioma. Semin. Oncol.. 8: 268-282,               35. Bradford, M. A rapid and sensitive method for the quantilation of microgram quan
     1980.                                                                                              tities of protein utilising the principle of protein-dye binding. Anal. Biochem., 72:
 3. Semb. G. Diffuse malignant pleural mesothelioma: a clinicopathological study of 10                  248-254, 1976.
    fatal cases. Acta Chir. Scand.. 126: 78-91, 1963.                                              36. Kohn, E. C.. Francis, E. A., Liotta. L. A., and Schiffmann. E. Heterogeneity of the
 4. Hillerdahl. (i. Malignant mesothelioma 1982: review of 4710 published cases. Br. J.                 motility responses in malignant tumor cells: a biological basis for the diversity and
    Dis. Chest. 77: 321-343. 1983.                                                                      homing of metastatic cells. Int. J. Cancer. 46: 287-292. 1990.
 5. Siri. A.. Carnemolla. B.. Raffanti. S., Castellani. P.. Balzano. E., and Zardi, L.             37. Ruoslahti. E.. and Pierschbacher, M. D. New perspectives in cell adhesion: ROD and
    Fibronectin concentrations in pleura! effusions of patients with malignant and non-                 integrins. Science (Washington DC), 238: 491^497, 1987.
    malignanl diseases. Cancer Lett., 22: 1-9, 1984.
                                                                                                   38. Stracke, M. L., Krutzsch, H. C.. Unsworth, E. J., Õrestad, A., Cioce, V, Schiffmann.
 6. Roboz, J.. Greaves. J., Silides, D.. Chachinian, A. P.. and Holland. J. F. Hyaluronic               E., and Liotta. L. A. Identification, purification, and partial sequence analysis of
    acid content of effusions as a diagnostic aid for malignant mesothelioma. Cancer.                   autotaxin. a novel motility-stimulating protein. J. Biol. Chem.. 267: 2524-2529. 1992.
    Res., 45: 1850-1854. 1985.
                                                                                                   39. Beck. K.. Hunter. I., and Engel, J. Structure and function of laminin: anatomy of a
 7. Orr. W.. Varani. J.. Gondck. M. D.. Ward, P. A., and Mundy, G. R. Chemotactic                       mullidomain glycoprolein. FASEB J., 4: 148-160, 1990.
    responses of tumour cells to products of resorhing bone. Science (Washington DC).              40. Hynes. R. O. Fibronectins. pp. 7-24. New York: Springer-Verlag, 1990.
    203: 176-179, 1979.
                                                                                                   41. Hynes, R. O. Fibronectins. pp. 176-199. New York: Springer-Verlag, 1990.
 8. Hujanen. E. S.. and Terranova, V. P. Migration of tumour cells to organ-derived
    chemoattractants. Cancer Res., 45: 3517-3521, 1985.                                            42. Liolta. L. A., Rao, C. N.. and Barsky, S. H. Tumor invasion and the extracellular
                                                                                                        matrix. Lab. Invest.. 49: 636-649. 1983.
 9. Rosen, E. M.. Meromsky, L., Setter. E.. Vinter. D. W.. and Goldberg. 1. D. Smooth
                                                                                                   43. AI-Adnani, M. S., Kimme, J. A., and McGee. J. O. Inappropriate production of
    muscle-derived factor stimulates mobility of human tumour cells. Invasion and Me-
    taslasis, 10: 49-64, 1990.                                                                          collagen and prolyl hydroxyproline by human breast cancer cells in vivo. Br. J.
                                                                                                        Cancer, 31: 653-660, 1975.
10. Stracke, M. L., Kohn, E. C., Aznavoorian, S. A., Wilson, L. L., Salomon, D.,
    Krutzsch, H. O. Liotta, L. A., and Schiffmann. E. Insulin-like growth factors stimu            44. Barsky, S. H., Rao, C. N.. Grotendorst, G. R., and Liotta, L. A. Increased content of
    late chemotaxis in human melanoma cells. Biochem. Biophys. Res. Commun., 153:                       type IV collagen in desmoplasia of human breast carcinoma. Am. J. Pathol., 108:
     1076-1083. 1988.                                                                                   276-292, 1982.
11. Ruff. M.. Schiffmann. E.. Terranova. V.. and Perl. C. B. Neuropeptides are chemoat             45. Stracke, M. L.. Guirguis, R., Liotta. L. A., and Schiffmann. E. Pertussis toxin inhibits
    tractants for human tumour cells and monocytes: a possible mechanism for metastasis.                stimulated motility independently of the adenylate cyclase pathway in human mela
    Clin. Immunol. Immunopathol.. 37: 387-396, 1985.                                                    noma cells. Biochem. Biophys. Res. Commun., 146: 339-345, 1987.
12. Hayashi. H.. Yoshida, K., Ozaki, T.. and Ushijima. K. Chemotactic factor associated            46. Evans, C. P., Walsh, D. S.. and Kohn. E. C. An autocrine motility factor secreted by
    with invasion of cancer cells. Nature (Lond.), 226: 174-175. 1970.                                  Dunning R-3327 rat prostatic adenocarcinoma cell subtype AT 2.1. Int. J. Cancer, 49:
13. Biirk. R. R. Factor from a transformed cell line that affects cell migration. Proc. Nati.           109-113, 1991.
    Acad. Sci. USA, 70: 369-372, 1973.                                                             47. Guirguis. R., Margulies. I., Taraboletti. G., Schiffmann. E.. and Liotta, L. A. Cyto-
14. Liotta, L. A., Mandler, R., Murano, G.. Kalz, D. A.. Gordon, R. K.. Chiang, P. K.. and              kine-induced pseudopodial protrusion is coupled to tumor cell migration. Nature
    Schiffmann, E. Tumour cell autocrine motility factor. Proc. Nati. Acad. Sci. USA, 83:               (Lond.), 329: 261-263. 1987.
    3302-3306, 1986.                                                                               48. Watanabe, H.. Canni, P., Hogan, V., Raz. T.. Silletti. V.. Nabi. I., and Raz, A.
15. Rennard. S. I.. Jaurand, M. C, Bignon, J., Kawanami, O., Ferrans. V. J.. Davidson, J..              Purification of human tumor cell autocrine motility factor and molecular cloning of
    and Crystal, R. G. Role of pleural mesothelial cells in the production of the subme-                its receptor. J. Biol. Chem., 266: 13442-13448. 1991.
    sothelial connective tissue matrix of lung. Am. Rev. Respir. Dis.. 130: 267-274. 1984.         49. Schubert. D.. and Kimura. H. Substratum-growth factor collaborations are required
16. Behbehani, A. M.. Hunter. W. 1.. Chapman. A. L.. and Lin, F. Studies of human                       for the mitogenic activities of Activin and FGF on embryolan carcinoma cells. J. Cell.
    mesothelioma. Hum. Pathol.. 13: 862-866, 1982.                                                      Biol., 114: 841-846, 1991.


Shared By: