ISOLATION OF BIOTIN (VITAMIN H) FROM LIVER*
BY VINCENT DU VIGNEAUD, KLAUS HOFMANN,t AND
DONALD B. MELVILLE1
(From the Depurtment of Biochemistry, Cornell University Medical College,
New York City)
Downloaded from www.jbc.org by on August 30, 2009
AND PAUL GYORGY
(From the Babies and Childrens Hospital, and the Department of Pediatrics,
School of Medicine, Western Reserve University, Cleveland)
(Received for publication, June 6, 1941)
In the course of our work on the isolation of vitamin H, the anti-
egg white injury factor, it began to be suspected that vitamin H
might be identical with biotin, a yeast growth factor, which had
been isolated as the methyl ester from egg yolk by Kijgl and Tennis
(2), and with coenzyme R, a growth and respiration factor for
rhizobia (3). All the evidence we were able to adduce, experi-
mentally and from the literature, on distribution, stability, solu-
bility, and various other chemical and physical properties favored
the identity of vitamin H with these two factors. The similarity
between biotin and coensyme R has been pointed out by other
workers (4, 5). Our comparison of the biotin, coenzyme R, and
vitamin H activities of electrodialysate fractions of our vitamin H
preparation showed a striking parallelism in the distribution of
the three activities. On the basis of this work we suggested (6)
* A preliminary report of this work has appeared (1).
In this cooperative research reported here, the Cornell group has been
responsible for the chemical work and for the biotin assays, and the Western
Reserve group for the vitamin H assays.
The authors would like to express their appreciation to W. 0. Frohring,
president of the S. M. -4. Corporation, for his interest and genuine
cooperation.
t Dr. Hofmann’s participation in this work has been made possible by a
fellowship granted to him by the Societ,y for Chemical Industry, Basel,
Switzerland.
1 S. M. A. Corporation Fellow.
643
644 Isolation of Biotin
that vitamin H, biotin, and coenzyme R “are either identical or
indeed closely related compounds.”
Shortly after the above work was presented, we were able to
test the possible identity of biotin with vitamin H by vitamin H
assay of a solution containing 150 y of crystalline biotin methyl
ester kindly placed at our disposal by Professor Kogl. On the basis
of the ratio between the biotin activity of Kogl’s material and the
biotin activity and vitamin H activity of other preparations we had
studied we were able to predict that the biotin preparation from
Kiigl should possess a vitamin H activity of 10,000 units per mg.
(The vitamin H unit is the minimum daily amount of substance,
Downloaded from www.jbc.org by on August 30, 2009
injected for 30 days, necessary to bring about complete disappear-
ance of egg white injury symptoms in rats.) As reported (7)) this
prediction was borne out by vitamin H assays of the Kogl sample
on rats.
Since our vitamin H concentrates were prepared from liver and
since Kijgl had isolated his biotin from egg yolk, it was necessary
for us to attempt the isolation from the liver concentrates of the
substance possessing vitamin H activity and to determine whether
or not the properties of the pure substance were the same as those
reported for biotin by Kijgl and Tennis.
The starting material for the present investigation was a vitamin
H liver concentrate supplied by one of us (P. G.) which had been
prepared from the alcohol-insoluble fraction of beef liver by high
pressure hydrolysis, precipitation of the inert material with alcohol
and acetone, and precipitation of the active substance with phos-
photungstic acid, followed by decomposition of the precipitate
with barium hydroxide (8). The vitamin H activity of the result-
ing solution was 25 to 35 units per mg. of solids and 1000 units per
cc.
Each step in the fractionation of this liver concentrate given in
the present paper was followed by biotin assays by a modification
of the yeast growth method of Snell, Eakin, and Williams (9).
Comparative assays for biotin and vitamin H activities of many
of the intermediate fractions as well as the final product demon-
strated again the parallelism of the two activities.
One of the final steps in the lengthy fractionation procedure used
by Kiigl and Tijnnis was the conversion of biotin to a biologically
active “ester-base” by esterification with acidic methanol. Esteri-
du Vigneaud, Hofmann, Melville, Gyiirgy 645
fication of our crude preparation was carried out in order to obtain
the active material in a form which would be soluble in organic
solvents and which might lend itself to purification by chromato-
graphic adsorption procedures.
By adsorption of the crude ester from chloroform solution on
aluminum oxide and by elution with acetone, the potency of the
material could be raised from 25 to 50 vitamin H units per mg. to
1000 to 2000 vitamin H units per mg. A second adsorption on
aluminum oxide and elution yielded material of a potency of 3000
to 6000 vitamin H units per mg. The eluate was evaporated and
the residue taken up in chloroform. The solution was extracted
Downloaded from www.jbc.org by on August 30, 2009
with dilute HCl, and the extract was then evaporated to dryness.
FIG. 1. Biotin methyl ester. X 150
This preparation was esterified and the ester hydrochloride which
was formed was converted into the free ester. Crystals were
obtained which melted at 154-158”. By repeated crystallizations
from a mixture of methanol and ether a product of constant melting
point and biological activity was obtained which melted sharply
in the Kofler micro melting point apparatus at 166-167” (uncor-
rected). Sublimation in ~~CUO followed by crystallization from a
mixture of methanol and ether did not change the melting point,
crystalline form, or biological activity. The substance crystal-
lized, as shown in Fig. 1, in long, thin, plate-like needles from the
methanol-ether mixture. From a chloroform-petroleum ether
mixture it crystallized in needles as described by K6gl and TGnnis.
The biological activity and melting point of the pure biotin
methyl ester were confirmed by several separate isolations.
646 Isolation of Biotin
Using the yeast growth method, we have compared the biotin
activity of our purified crystalline material with the biotin activity
of the vitamin H preparation (34 vitamin H units per mg.) that
we had used as an arbitrary standard in our previous communica-
tions (6, 7). Expressed in terms of vitamin H units, the various
preparations of purified product that we have made have all con-
sistently yielded, by the yeast growth method, the high value of
27,000 (&lo per cent) vitamin H units per mg. Direct vitamin
H assays of our crystals, made on rats by the curative method,
were in agreement with this high potency. From 5 liters of the
liver concentrate (8) we have been able to isolate 70 mg. of pure
Downloaded from www.jbc.org by on August 30, 2009
biotin methyl ester, which corresponds to a yield of 38 per cent
based on our starting material.
The analytical values we obtained for the pure crystalline com-
pound, as shown in the experimental part, agree most closely with
the empirical formula of CnH,,03N$. Molecular weight deter-
minations by the Rast method with various solvents were not
entirely satisfactory. However, all the results obtained with bor-
neol indicated t,hat the empirical formula as derived from the
analytical data on the basis of 1 sulfur atom in the molecule is
unquestionably correct. Analyses of several crystalline deriva-
tives and degradation products, which will be described in a later
publication, likewise support this formula. This formula is in
agreement with that given by K6gl (lo), although it should be
pointed out that K6gl gave no analytical data. The melting
point for the preparation from which K6gl derived this formula
was not given; so one must therefore assumethat his previously
published melting point of 148” still applies. It is to be noted that
the melt’ing point of the pure biotin methyl ester obtained by us is
18-19” higher than the value reported by Kiigl and Tijnnis, and
no explanation for this is as yet apparent. The similarity, how-
cvcr, in solubility, in crystalline form, and in composition leads
us to believe that the compound isolated by us from liver is iden-
tical with that isolated by Kcgl and Tiinnis from egg yolk.
As WChave stated, the solution of crystalline biotin methyl ester
supplied us by K6gl showed a potency of 10,000 vitamin H units
per mg., as compared with the value of 27,000 vitamin H units
per mg. found for our preparations both by comparative assays
du Vigneaud, Hofmann, Melville, Gyiirgy 647
for yeast growth and for anti-egg white injury activities.1 How-
ever, we do not know the melting point or purity of this particular
preparation of Kogl. We therefore do not wish to stress this
difference in degree of potency. It is within possibility that the
Kijgl sample may have suffered loss of potency from the time it
was made up until it was assayed here. We can therefore simply
report the activity of it as we found it. Direct comparison of the
crystalline compounds is, of course, desirable and we hope that
future circumstances will permit such comparison.
EXPERIMENTAL
Downloaded from www.jbc.org by on August 30, 2009
Method of Assay-Biotin determinations were carried out by a
yeast growth method which is essentially that of Snell, Eakin,
and Williams (9), except that Fleischmann Strain 139 of Saccharo-
myces cerevisiae was used as the test organism and turbidity meas-
urements of yeast growth were made in the Klett-Summerson
photoelectric calorimeter, with a blue filter (Klett No. 42). This
instrument is equipped with a logarithmic scale, so that the tur-
bidity readings observed are a linear function of the yeast growth.
Each sample to be assayed was made up in a series of decimal dilu-
tions of 1 and 0.3 cc., involving some twelve dilutions over a
millionfold range. By plotting the resulting turbidity readings
against the logarithm of the concentrations used, an S-shaped
curve was obtained for each sample that was approximately
linear over a considerable portion of the middle range. Com-
parison of the interpolated concentration at which half maximum
growth occurred in the assay sample with the corresponding con-
centration obtained in the same manner with a solution of known
vitamin H activity permitted calculation of the vitamin H equiv-
alence of the sample being assayed.
Isolation Procedure, EsteriJication-500 cc. of the vitamin H
concentrate from liver mentioned in the introduction were con-
centrated to dryness in vacua and kept at 100’ for 1 hour. The
1 As mentioned in the preliminary report of this work (1) the above ratio
was likewise found for the activities of the two preparations in promoting
the growth of Rhizobium trifolii (coenzyme R activity) as determined by
Professor Dean Burk and t,he growth of Clostridium butylicum as determined
by Dr. D. W. Woolley.
Isolation of Biotin
gummy residue was then taken up in 200 cc. of absolute methanol
and was again concentrated to dryness. 200 cc. of absolute meth-
anol containing 5 per cent of dry HCl gas were then added and the
solution was refluxed for 1 hour. The methanol was removed
in vacua and the residue was reesterified in the same manner.
The solution was then concentrated to a thick syrup in vacua;
the residue was cooled to 0” and was washed into a separatory
funnel with 50 cc. of ice water. 200 cc. of ice-cold ethyl acetate
were then added, followed by 30 gm. of solid potassium bicar-
bonate, and the mixture was shaken until all of the bicarbonate
Downloaded from www.jbc.org by on August 30, 2009
was dissolved. The water layer, which was alkaline to litmus, was
then extracted three more times with 100 cc. portions of ethyl
acetate. The combined ethyl acetate extracts were washed
repeatedly with water, dried over sodium sulfate, and evaporated
to dryness in vacua. The ethyl acetate fraction contained 80 to
90 per cent of the activity of the starting material and it had a
potency of 50 to 100 vitamin H units per mg. The water residues
and washings were combined, evaporated to dryness in vacua,
and the residue was reesterified by the procedure outlined above.
First Adsorption on Aluminum Oxide-Several ester fractions
were combined and purified further by chromatographic adsorp-
tion. One typical experiment was carried out as follows: 18 gm.
of the crude ester fraction, containing 1,500,OOO vitamin H units,
were dissolved in 150 cc. of chloroform and the solution was filtered
through a column of 50 gm. of aluminum oxide (standardized
according to Brockmann). The column was washed with the
various solvents indicated in Table I, each washing being col-
lected separately and evaporated to dryness. The samples were
then assayed. The results of such an experiment are summarized
in Table I. All the fractions containing an activity of 1000 vita-
min H units per mg. or better were combined and used for the next
step in the purification.
Second Adsorption on Aluminum Oxide-2.55 gm. of this con-
centrate derived from several preparations which had been put
through the first adsorption step and which contained 2,585,OOO
vitamin H units were dissolved in 50 cc. of acetone, and the solu-
tion was filtered through a column of 100 gm. of the aluminum
oxide. The chromatograph was again fractionated by elution with
mixtures of solvents as indicated in Table II. Fractions 343-V
du Vigneaud, Hofmann, Melville, Gyijrgy 649
TABLE I
Distribution of Biotin Activity in First Chromatograph
Vitamin
Weight Activity H units
Column elated with Fraction No.
Zli3’
cc. w7. vitaminH units
Chloroform 300 339-x1 15,000 147,000 9.8
Acetone 300 339-VIII 805 1,000,OOO 1242
“ 90% + meth- 50 339-1x 21 45,000 2140
an01 10%
“ ‘I 50 339-x 76 145,000 1910
Downloaded from www.jbc.org by on August 30, 2009
iI “ 50 339-x1 256 310,000 1210
‘I ‘I 50 339-x11 102 45,000 440
Total ._..........................._..., 16,260 1,692,OOO
TABLE II
Distribution of Biotin Activity in Second Chromatograph
-
Column &ted with I kaction No. Weight Activity
cc. ml. Zamin H unit B
Acetone 200 343-I 622 l,ooo 1.6
“ 90% + meth- 130 343-11 583 Inactive 0
anol 10%
“ “ 700
10 343-111 82 8.5
‘I ‘I 1110
10 343-IV 72 80,000
I‘ ‘I 10 343-v 52 178,000 3420
“ I< 10 45 162,000
343-VI 3600
“ L‘ 10 343-VII 42 220,000 5240
“ ‘I 10 343-VIII 37 220,000 5950
“ “ 200,000
10 343-1X 30 6670
‘I “ 10 343-x 89 312,000 3500
‘I LL 10 343-X1 65 240,000 3700
CL L‘ 162,000
10 343-X11 68 2380
I‘ “ 10 343-X111 76 120,000 1580
6‘ “ 10 343-XIV 50 60,000 1200
-.
Total 1913 1,955,700
to 343-X111 were combined and used for the next step. Frac-
tions lower in activity were likewise combined and purified further
by chromatographic adsorption.
650 Isolation of Biotin
Extraction and ReesteriJication-The combined Fractions 343-V
to 343-X111 (504 mg. containing 1,814,OOO vitamin H units) were
dissolved in 10 cc. of chloroform and the resulting solution was
extracted fifteen times with 10 cc. portions of 3 N HCl. The
combined HCl extracts were evaporated to dryness in vacua and
the dry residue (474 mg. containing 1,812,600 vitamin H units)
was refluxed for 1 hour with 50 cc. of methanol containing 5 per
cent of HCl gas. The methanol was then removed in vacua and
the residue dissolved in a few cc. of ice water. Saturated potas-
sium bicarbonate was then added to the solution until it was alka-
line to litmus and the solution was extracted with three 50 cc.
Downloaded from www.jbc.org by on August 30, 2009
portions of ice-cold ethyl acetate. The combined ethyl acetate
extracts were washed with water, dried over sodium sulfate, and
concentrated in vacua. During the concentration crystals of
biotin methyl ester began to separate. The crystals were collected
and were washed with ethyl acetate. The yield of crude material
which melted at 155-160” was 35 mg. Purification by two crystal-
lizations from a mixture of methanol and ether yielded a material
that melted at 166-167”. The material was sublimed at 140-
150’=’ in a high vacuum (10v6 mm.) and recrystallized from meth-
anol and ether. 25 mg. of colorless elongated plates were obtained
which melted sharply at 166-167’.
The compound is very soluble in ethyl alcohol, methyl alcohol,
chloroform, and acetone, and is sparingly soluble in ether and
ethyl acetate. It possesses a specific rotation of [a]E2 = +57” for
a 1 per cent solution in chloroform. The compound has the fol-
lowing composition.
CJMhN~S.
Calculated. C 51.14, H 7.02, N 10.84, S 12.42, OCH3 12.0
Found. “ 51.40, “ 7.19, “ 10.84, “ 12.16, “ 11.5
The authors wish to express their appreciation to Miss Eleanor
Hague for carrying out the numerous biotin assays, to Dr. J. R.
Rachele for the microanalyses, and to Miss Catharine S. Rose for
her assistance in the vitamin H assays.
SUMMARY
A procedure for the isolation of the methyl ester of biotin (vita-
min H) from liver has been described. The compound after
du Vigneaud, Hofmann, Melville, Gyijrgy 651
repeated crystallization and sublimation possessed constant
biological activity and a constant melting point af 166-167”.
The analytical data agreed most closely with the empirical for-
mula, CIIHI~O~N&.
BIBLIOGRAPRY
1. Gyargy, P., Rose, C. S., Hofmann, K., Melville, D. B., and du Vigneaud,
V., Science, 92, 609 (1940).
2. K6g1, F., and Tijnnis, B., 2. physiol. Chem., 242,43 (1936).
3. Allison, F. E., Hoover, S. R., and Burk, D., Science, 78,217 (1933).
4. West, P. M., and Wilson, P. W., Xcience, 89, 607 (1939).
Downloaded from www.jbc.org by on August 30, 2009
5. Nilsson, R., Bjiilfve, G., and Burstrijm, D., Nuturwissenschuften, 27,
389 (1939).
6. GyGrgy, P., Melville, D. B., Burk, D., and du Vigneaud, V., Science, 91,
243 (1940).
7. du Vigneaud, V., Melville, D. B., Gyijrgy, P., and Rose, C. S., Science,
92, 62 (1940).
8. Gyijrgy, P., Kuhn, R., and Lederer, E., J. Biol. Chem., 131, 745 (1939).
9. Snell, E. E., Eakin, R. E., and Williams, R. J., J. Am. Chem. Sot., 62,
175 (1940).
10. Kiigl, F., Chem. and Ind., 67,49 (1938).