Vitamin K and the Biosynthesis of Prothrombin II. STRUCTURAL COMPARISON OF NORMAL AND DICOUMAROL-INDUCED BOVINE PROTHROMBIN*

SUMMARY Highly purified dicoumarol-induced bovine prothrombin, which does not bind calcium ions and has no prothrombin activity, has been structurally compared with normal prothrombin. Quantitative amino acid and carbohydrate analysis gave identical results for both prothrombins, as did analysis of the NHz-terminal and the COOH-terminal amino acids and molecular weight determination with the sodium dodecyl sulfate gel electrophoretic technique. Peptide maps of tryptic peptides prepared from the reduced and aminoethylated normal and dicoumarol-induced prothrombin were identical. These results suggest that the difference in properties between the two prothrombins are caused by a minor structural difference or a conformational difference. Ouchterlony immunodiffusion analysis gave a reaction of complete immunological identity between the two prothrombins, whereas the quantitative immunoprecipitation technique indicated antigenic difference between them. Furthermore, it was found that normal prothrombin has calcium ion-de-pendent antigenic determinants. The sedimentation coeffi-cient, Stokes molecular radius, the titration curves for the tyrosine phenolic groups, and the fluorescence emission spectra were identical, which corroborates


JOHAN STENFLO
(Received for publication, April 5,1972) From the Department of Clinical Chemistry, University of Lund, Malmti General Hospital , Malm6, #weden SUMMARY Highly purified dicoumarol-induced bovine prothrombin, which does not bind calcium ions and has no prothrombin activity, has been structurally compared with normal prothrombin.
Quantitative amino acid and carbohydrate analysis gave identical results for both prothrombins, as did analysis of the NHz-terminal and the COOH-terminal amino acids and molecular weight determination with the sodium dodecyl sulfate gel electrophoretic technique. Peptide maps of tryptic peptides prepared from the reduced and aminoethylated normal and dicoumarol-induced prothrombin were identical.
These results suggest that the difference in properties between the two prothrombins are caused by a minor structural difference or a conformational difference. Ouchterlony immunodiffusion analysis gave a reaction of complete immunological identity between the two prothrombins, whereas the quantitative immunoprecipitation technique indicated antigenic difference between them.
Furthermore, it was found that normal prothrombin has calcium ion-dependent antigenic determinants.
The sedimentation coefficient, Stokes molecular radius, the titration curves for the tyrosine phenolic groups, and the fluorescence emission spectra were identical, which corroborates that the difference between the normal and the dicoumarol-induced prothrombin does not engage the entire molecule.
The results obtained by polyacrylamide gel electrophoresis in 8 M urea may suggest that the difference includes an anomalous pairing of half-cystine residues.
The identification and purification of a dicoumarol-induced prothrombin from bovine plasma was described (1) in a previous communication.
It had the same main antigenic determinants as normal prothrombin but lacked prothrombin activity.
Further, in contrast with normal prothrombin, it was not adsorbed to barium citrate, and in the presence of calcium ions its electrophoretic mobility was higher than that of normal prothrombin.
Dicoumarol is an antagonist of vitamin EC. Knowledge of the structural difference between this dicoumarol-induced and normal prothrombin is therefore desirable, especially since it might elucidate the role of vitamin K in the bioproduction of extracellular proteins.
In this investigation t'he dicoumarol-induced prothrombin is characterized and its struct,ure is compared with that of normal prothrombin.
A preliminary report of parts of this work has been published earlier (2).

Materials
The normal and the dicoumarol-induced prothrombin used in this study were purified in the way described earlier (1). Urea solutions were passed through a mixed bed ion exchange resin prior to USC. Guanidine hydrochloride was obtained from Mann ("ultrapure") and used without further purification. Acrylamide and N ,N'-methylene bisacrylamide were recrystallized as already described (1). Iodoacetic acid was recrystallized from diethylether. Dithiothreitol was obtained from Nutritional Biochemicals, ethyleneimine from Fluka and 5,5'-dithiobis(2. nitrobenzoic acid) from British Drug House Ltd. l-Ethyl-3- (3dimethyl-aminopropyl) carbodiimide, glycine methylester, and anhydrous hydrazine were obtained from Pierce. Phenylisothiocyanate (Eastman), pyridine, and triethylamine were purified as described by Sjiiquist (3). Trypsin treated with t.osylpheaylalanylchloromethyl ketone was obtained from Worthington. Ampholytes (Ampholine) were from LKB Produkter AB, Stockholm, Sweden.
Sephadex G-100 was obtained from Pharmacia.

Methods
Reduction and Carboxymethylation-Reduction and carboxymethylation was carried out as described by Merino and Snell (4) with the following modifications.
The reduction was carried out in 6 M guanidine hydrochloride for 3 hours at 37" and carboxymethylation was allowed to proceed for 2 hours before the sample was dialyzed at 4", first against 0.1 31 SH4HC03 and then thoroughly against distilled deionized water.
The preciyit,ated material was lyophilized.
Amino Acid Composition-Lyophilized, salt-free samples of normal and dicoumarol-induced prothrombin (1 to 2 mg) were hydrolyzed in 6 N HCl in sealed, evacuated Pyrex tubes at 110" for 24 and 72 hours (5). The analyses were performed \\-ith the 5 AH automatic amino acid analyzer.
Norleucine was used as internal standard.
Half-cystine was determined as cysteic acid after performic acid oxidation (7). The tryptophan content was estimated with the technique of Bence and Schmid (8). The amide content was determined with the method of Hoare and Koshland (9): activation of the free carboxyl groups with 1.ethyl-3.(3-dimethylaminopropyl) carbodiimide and reaction with glycine met,hylester resulted in the coupling of glycine to the free carboxyl groups.
The reaction was performed with the protein (5 mg) in 5 $1 guanidine hydrochloride at 25" in a Radiometer pH-stat with the pH maintained at 4.75 with 0.4 M HCl. The consumption of acid had virtually ceased after 60 min. After extensive dialysis against 1 mM HCl and lyophilization the amount of glycine incorporated was determined by amino acid analysis.
After washing the reaction mixture with 4 x 1 ml of benzcne-~ethylene chloride-water (3: 1:4, v/v/v, upper phase) the sample was flushed with N? and then lyophilized.
Cyclization and release of the NH%-terminal amino acid as the phenylthiohydantoin derivative was performed with 400 ~1 of HCl-saturated acetic acid-water (2:1, v/v) at 40" for 2.5 hours. The sample was then taken almost to dryness by evaporation in vccc~o. After addition of 500 ~1 of distilled water the phenylthiohydantoin amino acids were extracted with 3 X 1 ml of ethylacetate.
The combined extracts were taken to dryness by evaporation in vucuo and the residue dissolved in 90% acetic acid. The phenylthiohydantoin amino acids were identified by thin layer chromatography on precoated silica gel plates with fluorescent indicator (Merck). System V of Jeppsson and Sjiiquist (12) was used for the chromatography.
The spots were eluted with 95y0 ethanol and the ultraviolet spectra of the phenylthiohgdantoin amino acids were recorded.
The COOH-terminal amino acid of normal and dicoumarolinduced prothrombin was determined with hydrazinolysis as described by Fraenkel-Conrat and Chun Ming Tsung (13). The react.ion was allowed to proceed for 16 hours at 60" with hydrazine sulfate as a catalyst (14). The dried hydrazinolysate was dissolved in wa,ter and the neutral and acidic amino acids were separated from the basic amino acids and from the amino acid hydrazides on a column of Amberlite IRC-50.
After lyophilization free amino acids were determined with the amino acid analyzer.
Carbohydrate $nalyses-For hexosamine determination the prothrombin was hydrolysed for 6 hours in 4 51 HCl at 100" in sealed, evacuated Pyrex tubes (15). Hexosamine was determined on the short column of the amino acid analyzer with cY-amino-fi-guanido-propionic acid as internal standard. Standard solutions of glucosamine were subjected to the same treatment.. Sialic acid was determined after hydrolysis with 0.1 M HzS04 at 80" for 1 hour with the thiobarbituric acid assay (16). Gas chromatographic analysis of the carbohydrates present in one sample of both the normal and the dicoumarol-induced pro-thrombin was kindly performed by Dr. Hans Bennich at the Wallenberg Laboratory, University of Uppsala, using the method of Clamp et al. (17).
Peptide LYapping-Complete reduction and aminoethylation of normal and dicoumarol-induced prothrombin were carried out essentially according to Slobin and Singer (18). A 0.25 to 0.5% solution of the protein in 0.2 M Tris-HCl, 0.01 M EDTA, 7 RI guanidine hydrochloride was made 0.1 M in dithiothreitol. The vial was flushed with X2, stoppered, and incubated at room temperature.
After 2 hours an equal volume of 3 M Tris-HCl, 0.01 nf EDTA, 7 M in guanidine hydrochloride pH 8.0 was added. With the solution still under Nz, ethyleneimine was added in five equal portions at 5-min intervals to a IO-fold molar excess over dithiothreitol.
One hour after the first addition of ethyleneimine 1 volume of ice cold, distilled and deionized water was added. The solution was dialyzed against distilled, deionized water and lyophilized.
The reduced, aminoethylated protein was suspended to a concentration of 5 mg per ml in 0.1 M NHhHCOs pH 8.5 containing phenol red as a pH indicator.
Digestion was carried out with tosylphenylalanylchloromethyl ketone-treated trypsin at 37" for 8 to 14 hours at an enzyme to substrate ratio of 1:50 (w/w). At the end of the digestion when no core remained, the samples were frozen and lyophilized.
The lyophilized digest was dissolved in distilled water. About 1.5 mg was applied to a Whatman No. 3M,M paper. High voltage electrophoresis was performed at 1.5 kv for 3% hours in pyridine-acetic acid-water (100:10:890, v/v/v) at pH 6.1 on an apparatus with a water cooled plate (19). In some experiments a guide strip was stained with ninhydrin to locate the neutral peptides which were cut out and sewn to another Whatman No. 3RIM paper. Electrophoresis was then run at 2.5 kv for 45 min in formic acid-acetic acid-water (10:29: 193, v/v/v) at pH 1.8 in a tank under Warsol (20). After electrophoresis the papers were dried by fanning, sewn to new sheets of Whatman No. 3MM paper, and subjected to ascending chromatography in pyridineisoamyl alcohol-water (35:35:30, v/v/v) for 14 to 16 hours (21). The air-dried papers were dipped in 0.5% ninhydrin in acetone and developed at room temperature overnight. Alternatively they were stained for sulfur with platinic iodide followed by a basic ninhydrin stain (22).
Cyanogen Bromide Treatment-Cyanogen bromide degradation was performed with the method described by Steers et al. (23). The protein (5 mg) was dissolved in 90% formic acid. After dilution with water to 70% formic acid cyanogen bromide (250fold molar excess relative to methionine) was added. The reaction was allowed to proceed for 20 to 24 hours at room temperature, after which the sample was diluted with water (1O:l) and the proteins recovered by lyophilization.
Polyacrylamide Gel Electrophoresis and Isoelectric Focusing-Polyacrylamide gel electrophoresis in urea was carried out either at pH 8.9 in 8 M urea using 5% or 7% acrylamide gel (24) or at pH 2.7 in 9 M urea using 5% acrylamide gel (25). The electrophoresis was performed at 1.5 ma per tube for about 4 hours. The gels were stained with 0.05 y0 Amido black B (Merck) in 7 y0 acetic acid and destained electrophoretically.
Polyacrylamide gel electrophoresis in sodium dodecyl sulfate was carried out as described by Weber and Osborn (26) using 40% of the standard concentration of methylenebisacrylamide.
Thyroglobulin, phosphorylase a, y-globulin, H chain and horse liver alcohol dehydrogenase were used as molecular weight markers using the molecular weights given by Weber and Osborn (26).
Isoelectric focusing was carried out in thin layers of poly-acrylamide gel essentially as described by Awedh et al. (27). The composition of the solutions was identical to that used by Spencer and King (28) except that the urea concentration in the gel was increased to 7.5 RI. A mixture of equal volumes of pH 3 to 5 and pH 5 to 7 -1mpholine was used with the addition of 10 y0 of this volume of pH 3 to 10 *Ampholine.
The carbon anode and cathode were wetted with 20y0 phosphoric acid and 20% ethylenediamine, respectively, immediately prior to the run. Electrofocusing was performed at room temperature for about 16 hours. The voltage was increased slowly to 400 volts so that the current did not exceed 6 ma. The gels were stained in a mixture of 200 ml of 5% trichloroacetic acid-57, sulfosalicylic acid, 2 ml of 1 y0 Coomassie blue and 40 ml of methanol (28).
Ezfinction Coegicienfs---The concentrations of the purified prothrombins were measured from the absorbance at 280 nm. The extinction coefficients (E::m) at 280 nm of the purified proteins were determined in the following way. After filtration through a column of Sephadex G-25 fine in 0.05 M phosphate buffer pH 7.5, the protein-containing fractions were pooled and the absorbance at 280 nm measured in a Zeiss PM Q II spectrophotometer.
n'itrogen determinations were carried out on the same samples using a micro Kjeldahl technique.' From these data the ext,inction coefficients were calculated using the values for nitrogen content (see below) obtained from the amino acid and carbohydrate analyses.
Determination of Sedimentation Coejqicient and Stokes Radii-Ultracentrifugal analyses2 were performed at 20" in 0.05 M phosphate buffer, 0.5 JI in SaCl pH 7.0 in a Spinco model E analytical ultracentrifuge equipped with the schlieren optical system. Sedimentation runs were made at 59,780 rpm in 12 mm 4" single sector cells. B direct comparison between normal and dicoumarol-induced prot'hrombin was performed in wedge cells. All calculations were corrected to water at 20" and zero protein concentration as described by Schachmann (29).
Stokes radius was determined with the method of Laurent and Killander (30). Analytical gel chromatography was carried out at 4" on a column (1.20 x 104 cm) of Sephadex G-100 equilibrated with 0.04 M Tris-HCl, 0.5 h* NaCl and 2 mM EDT-4 pH 7.4. The column was equipped with flow adapters and eluted upward at a flow rate of 2.9 ml per cm2 per hour with a Persplex peristaltic pump (LKB, Stockholm, Sweden). The column effluent was continuously monitored at 280 nm with a Uvicord (LKB, Stockholm, Sweden).
All analyses were performed in duplicate.
K,, was measured as described by Karlsson el al. (31). Stokes molecular radius was calculated from the formula given by Laurent and Killander (30). KW = prL(r,+r,? where K,, is the volume available for the protein in the gel, rs is Stokes radius, L and rT are constants characteristic of the gel. L was determined from the K,, of human albumin (rs = 35.5 A) and r, was assumed to be 6.5 A (30, 32).

Immunochemical
J1ethods--Double immunodiffusion was carried out as described by Ouchterlony (33). Quantitative precipit,ation of normal and dicoumarol-induced prothrombin 1 These analyses were performed at the Centrala Analyslaboratoriet, Department of Chemistry, Uppsala University. 2 The ultracentrifugal analyses were kindly performed by Dr. U. B. Hansson.
was performed with an antiserum raised against normal bovine prothrombin (1). The antiserum was diluted 1:3 and dialyzed against 0.05 M Tris-HCl buffer, 5 mM in ED'TB, pH 7.4. Various amounts of the antigens dialyzed against the same buffer were added to 0.5 ml of the antiserum and the final volumes made up to 1.0 ml with the buffer. The tubes were incubated at 37" for 1 hour and then at 4" for 16 hours. The precipitates were collected by centrifugation, washed twice with the above buffer, and then dried.
One-tenth milliliter of 0.1 1~ NaOH was added and the samples warmed until the precipitates dissolved, after which 1.2 ml of 6 M guanidine hydrochloride in 0.4 JI Tris-HCl buffer, 5 mM in EDTX pH 7.4 was added. After careful mixing the absorbancies at 280 nm were recorded.
In some experiments the same Tris buffers were used, but containing 5 rnlr calcium chloride instead of EDT\.
Specfrophotometric Titration of Tyosine Phenolic, Groups-The two prothrombins were dialyzed against 0.05 hf Tris-HCl, 0.15 M NaCl pH 8.0. Titrations were made in l-cm quartz cells in a Zeiss PMQ II spectrophotometer using protein concentrations of 0.4 to 0.5 mg per ml. Small additions of NaOH were made with a Hamilton microliter syringe followed by thorough mixing. After each addition the pH of the sample was measured and the absorbance at 295 nm, recorded.
The absorbance values were corrected for the small changes in volume.
The molar absorbance increase at 295 nm, was assumed to be 2700 per phenolic group titrated (34, 35).
Fluorescence Spectra--Fluorescence emission spect.ra were measured at ambient temperature (23 f I") in l-cm cells in an hminco-Bowman spectrophotofluoromet.er equipped with an X-Y recorder (Houston Omnigraph 2000). The excitation and emission slit widths were 1.0 and 0.5 mm, respectively.
hleasurements were made wit,11 protein concentrations of 0.05 mg per ml in 0.05 11 Tris-HCl buffer pH 8.0 which was 0.15 M in SaCl and 5 rnlr in EDTA.
Only uncorrected spectra were recorded.

Chemical Composition of Normal and Dicoumarolinduced
Prothrombin Amino Acid Composition-The amino acid composition was determined on two preparations of each of the two prothrombins. As shown in Table I, the compositions of t,he normal and the dicoumarol-induced prothrombin appeared to be identical. The half-cystine value for the dicoumarol-induced prothrombin is probably one residue too low since titrations with 5,5'-dithiobis-(2.nitrobenzoic acid) in 5.5 31 guanidine hydrochloride indicated that there was no free sulfhydryl group in either of the two prothrombins. This is corroborated by the fact that separate sulfur determinations (36) gave a value of 1.24% for both prothrombins,3 whereas judging from the amino acid analyses, the sulfur content is 1.22% and 1.17G70 for the normal and the dicoumarolinduced prothrombin.
Separate phosphorous determinations (37) gave values corresponding to less than 1 mole of phosphorus per mole of normal prothrombin and less than 2 moles per mole of dicoumarol-induced prothrombin.3 End Group Analyses-Quantitative NH*-terminal amino acid determinations with a modified Edman procedure (11) repeatedly yielded a single spot on the thin layer chromatograms with a RF value corresponding to that of the phenylthiohydantoin of alanine and with a typical ultraviolet spectrum.
The recovery was 0.6 to 0.7 mole of phenylthiohydantoin-alanine per mole of protein (uncorrected for operational losses) for both the normal and the dicoumarol-induced prothrombin. On hydrazinolysis with anhydrous hydrazine (Pierce) and with hydrazine sulfate as catalyst serine was the principal amino acid released from both prothrombins.
The yield varied between 0.65 and 1.2 residues per 72,000 g of protein if a recovery factor of 96 y. is anticipated (14). Glycine was regularly demonstrated in amounts of 0.2 to 0.3 residue per mole of protein, whereas alanine, aspartic acid, and threonine varied bet\lTeen 0.1 and 0.2 mole per mole of protein.
Identical results were obtained when the hydrazine had been refluxed on sodium hydroxide, distilled under nitrogen, and used immediately.
If unsatisfactory hydrazine was used the amount of glycine increased considerably, whereas the amount of serine released was fairly constant.
Since the hydrazides of glycine and serine are known to be very labile, control experiments on human transferrin and bovine serum albumin were performed.
The amount of serine released in these proteins was negligible.
httempts were made to check that serine is the carboxy terminal amino acid by carboxypeptidase _4 and B digestion of reduced and carboxymethylated samples in 0.2 bf Tris-HCl buffer, pH 8.0, 0.05 31 in sodium lauryl sulfate (13). Samples removed after various time intervals were analyzed on the automatic amino acid analyzer.
However, no amino Analysis-The data obtained for t,he carbohydrate analyses of the normal and dicoumarol-induced prothrombin are summarized in Table II. The values are in fair agreement with those reported earlier by Magnusson (39). There was no significant difference in the composition of the carbohydrate prosthetic group between the normal and dicoumarolinduced prothrombin.
Glucosamine analyses were performed both on the short column of the amino acid analyzer and by gas liquid chromatography.
The cause of the discrepancy between the glucosamine values obtained with the two methods was not investigated.
Sialic acid determinations with the thiobarbituric acid method gave values between 3% and 4y0 for both prothrombins. On agarose gel electrophoresis at pH 8.6 of the purified normal and dicoumarol-induced prothrombins before and after digestion with neuraminidase, the electrophoretic mobilities of both proteins were reduced to the same extent, which also indicates an identical number of sialic acid residues. Peptide Mapping-Tryptic peptide maps were prepared from the normal and the dicoumarol-induced prothrombin, reduced, and aminoethylated in 7 M guanidine hydrochloride (Fig. 1). To improve the separation of the neutral peptides they were cut out, stitched to another paper, and resolved by a second electrophoresis at pH 1.8 followed by ascending chromatography (Fig. 2). From the amino acid composition about 90 peptide spots were expected. About 66 spots were regularly detected in the peptide maps of both the normal and the dicoumarol-induced prothrombin.
Heating of t.he peptide maps at 80" gave no additional spots. Twenty-four of the spots were positive in the platinic iodide stain, whereas 27 to 28 spot's were expected from the amino acid composition.
The peptide spot marked 1 (Fig. 1) is stained somewhat heavier in the map of the dicoumarol-induced prothrombin but is also present in the map of the normal prothrombin.
The peptide spot marked 2, which was positive in the platinic iodide stain, exhibited a somen-hat varying RF value from one run to another in peptide maps from both prothrombins. Peptide maps prepared from different preparations of the two prothrombins yielded reproducible patterns. Furbhermore no core was visible on completion of the t'ryptic digestion or at the point of application of the digested sample on the paper.
Thus, judging from these findings the two prothrombins are identical in amino acid sequence and carbohydrate composition. thrombin at pH 8.9, 8 M urea and at pH 2.7, 9 M urea are shown in Fig. 3. Mixtures of reduced and alkylated normal and dicoumarol-induced prothrombii migrated as a single band at both pH values, whereas mixtures of the two unmodified prothrombins could be completely separated into two protein bands in prolonged electrophoretic runs with the normal prothrombin in the most anodal position. At pH 2.7 no separation could be achieved between unmodified normal and dicoumarol-induced prothrombin. At present no entirely satisfactory explanation can be given for the fact that the two untreated prothrombins separated at pH 8.9, but not at pH 2.7. However, it was probably not due to charge differences between the two proteins, but possibly to more complete denaturation, and thus a larger molecular radius of the abnormal prothrombin than of the normal prothrombin in the slightly alkaline pH range. Since urea is a more efficient denaturing agent at low pH values (40), both prothrombins are probably completely denatured at pH 2.7 and consequently might have the same molecular radius and electrophoretic mobility.
To confirm that there is no charge difference between the two prothrombins they were analyzed by isoelectric focusing in polyacrylamide gel containing 7.5 M urea in a pH 3 to 7 gradient. Three bands were regularly seen in both prothrombins. However, the normal and the dicoumarol-induced prothrombins gave identical patterns (Fig. 4).
illolecular Size of Normal and Dicoumarol-induced Prothrombin -Sedimentation velocity studies of the normal and the dicoumarol-induced prothrombin in 0.05 M phosphate buffer (pH 7.0), 0.5 M in NaCl in the wedge cell revealed two single symmetrical peaks sedimenting with apparently the same velocity (Fig. 5). The value for the s$,~ of the dicoumarol-induced prothrombin extrapolated from sedimentation runs at four different protein concentrations was 4.9.
The Stokes molecular radius for the two prothrombins was determined on a calibrated column of Sephadex G-100. A buffer with high ionic strength was chosen since at lower ionic strengths Tishkoff et al. (41)  edly dependent on the protein concentration in a way that suggests that the prothrombin molecules undergo reversible associations. Duplicate determinations on the same column gave values between 40 and 41 A for both prothrombins.
To obtain evidence for the assumed identity in molecular weight for the two prothrombins electrophoresis in sodium dodecyl sulfate was carried out according to Weber and Osborn (26). Mixtures of normal and dicoumarol-induced prothrombin gave only one protein band (Fig. 6). A molecular weight of 72,000 f 1,000 (average of three determinations with four molecular weight markers) was obtained.
Zmmunochemical Properties-When the normal and the dicoumarol-induced prothrombins were analyzed by crossed immunoelectrophoresis in calcium ion containing buffer, the precipit,ate produced by the normal prothrombin was heavier stained than the precipitate produced by the dicoumarol-induced prothrombin both when unfractionated plasma samples and mixtures of the two purified proteins were analyzed (1). This finding prompted an investigation of the two proteins with the Ouchterlony immunodiffusion technique.
The result obtained with an antiserum raised against normal bovine prothrombin was compatible with complete identity of the two prothrombins (Fig. 7). To obtain more detailed information the quantitative precipitin technique was used. Precipitation curves were prepared from both prothrombins with buffer and antisera containing either 5 mM EDTA or 5 mM CaClz (Fig. 8). With the dicoumarolinduced prothrombin there was only a slight decrease in precipitation with EDTA instead of Ca2f, presumably due to decomplementation of the antiserum (42). In contrast with these findings, the normal prothrombin exhibited far greater precipitation in the equivalence zone in the presence of Ca2+ than in the presence of EDTA.
The precipitation was also greater than that obtained with the dicoumarol-induced prothrombin.
These results indicate a Ca*-induced conformational change in the normal prothrombin. Furthermore the dicoumarol-induced prothrombin has a conformation different from that of normal prothrombin both with and without Ca2+.

Other Properties oj Normal and Dicoumarol-induced
Prothrombin-The difference between the two prothrombins in calcium ion binding, electrophoretic mobility, and immunochemical properties indicated that they had different conformations. Attempts were therefore made to corroborate this by spectrophotometric titration of tyrosine residues and by measuring fluorescence emission spectra. However, the tyrosine titration curves proved identical for both prothrombins.
The titration data indicated 18 tyrosine residues, approximately 5 of which were freely exposed to titration, whereas the remaining ones seemed to have anomalously high pK values. Cazf or 5 mM EDTA in the buffer did not influence the shape of the titration curves. Fluorescence emission spectra were recorded with excitation at 280 nm. The spectra of the normal and the dicoumarol-induced prothrombin were identical with maxima at 340 nm. Emission spectra recorded after excitation at several different wave lengths did not reveal any nonprotein chromophores.
Both normal and dicoumarol-induced prothrombin contain 7 to 8 methionine residues.
On polyacrylamide gel electrophoretic analyses in 8 M urea of cyanogen bromide fragments from the two prothrombins with intact disulfide bonds two major protein zones were visible at pH 8.9 and four at pH 2.7. The protein zones obtained with the two prothrombins seemed identical at both pH values. However, when mixtures of cyanogen bromide fragments from the normal and the dicoumarol-induced prothrombin were analyzed on the same gel, a pattern with four protein zones was repeatedly obtained indicating small differences in electrophoretic mobility between the fragments from the two prothrombins at pH 8.9. At pH 2.7 the mixture gave a pattern identical with that obtained with the individual proteins (Fig. 9).

DISCUSSION
The preparations of normal and dicoumarol-induced prothrombin used in the present study are homogenous by several criteria including electrophoresis in agarose gel and polyacrylamide gel without dissociating medium (1) and with 8 or 9 M urea at two different pH values. Yet molecular polymorphism was found in purified preparations of the two prothrombins when analyzed by isoelectric focusing in polyacrylamide gel containing 7.5 M urea. Three fractions were found in both the normal and in the dicoumarol-induced prothrombin. There was no difference between the two proteins.
Since the purification methods used for the two prothrombins are entirely different (1) this microheterogeneity is probably not a preparation artifact.
The abnormal prothrombin synthesized during dicoumarol administration as well as the normal prothrombin has a molecular weight of 72,000 as determined with the sodium dodecyl sulfate gel electrophoresis technique, which is in agreement with the sediment,a.tion enuilibrium molecular weight of 74,000 reported 8173 by Ingwall and Scheraga (43) for normal prothrombin. The results of amino acid and carbohydrate analysis reported in this paper are in agreement with values report.ed earlier (44) and identical for the normal and the dicoumarol-induced prot.hrombin. In both, 1 residue of alanine per molecule of prothrombin was found as amino terminus in agreement. with several earlier reports (45)(46)(47).
Attempts to determine the COOH-terminal amino acid with the carboxypept.idase method were not successful, as in Magnusson's (38) investigation.
Therefore, hydrazinolysis was resorted to. With this method a reasonable yield of serine was obtained from both prothrombins. This is in agreement with the proposal of Magnusson (48) that the 0 chain of thrombin which has a COOH-terminal serine constitut'es the COOH-terminal portion of the prothrombin molecule.
However, definite conclusions regarding the COOH-terminal amino acid of prothrombin must await confirmation with other methods.
Several analyses of tryptic peptide maps of the normal and the dicoumarol-induced prothrombin gave very reproducible results.
That the number of peptides predicted from the amino acid composition was not obtained might be due to some homologous amino acid sequences in the thrombin and the nonthrombin parts of the molecule. This explanation is supported by the finding that on activation prothrombin gives rise not only to one enzyme with esterase activity, i.e. thrombin, and that the non-thrombin part of the molecule also exhibits esterase activity (49). Peptide maps of both the normal and the dicoumarol-induced prothrombin were identical with respect to all the major peptide spots. The results hithert.0 presented indicate that the differences in properties between the two prothrombins are caused by any minor structural difference, e.g. in disulfide pairing or in prosthetic group, or by conformational differences.
The finding t,hat the two prothrombins hare the same main antigenic determinants and that the sedimentat.ion coefficients, the Stokes molecular radius, the tyrosine titration curves, and the fluorescence emission spectra were identical indicates that there is no gross difference involving the entire molecule.
The quantitative precipitin curves furthermore suggest that there is a conformational difference between normal prothrombin with and without calcium ions which is in agreement with the observation that the autoactivation of prothrombin is enhanced by calcium ions. : No such calcium ion-dependent conformation was observed in the dicoumarol-induced prothrombin. Similar calcium-dependent antigenic determinants have recently been identified with the quantitative precipitin technique in glutamic acid containing synthetic polyamino acids such as (L-G~@'-L-Ala%-TyrlO), by Maurer et al. (50). The differences in electrophoretic mobi1it.y of the two prothrombins in 8 M urea at pH 8.9 when the disulfide bonds are intact seems to rule out a differential noncovalent binding of small ligands as a cause of the difference in calcium ion binding and immunochemical properties between the two proteins. The cyanogen bromide degradation of the two prothrombins with intact disulfide bonds resulted in fragments with slightly different mobility at pH 8.9, in agreement with the result obtained with the intact proteins.
The unsuccessful attempts to distinguish between the two prothrombins when they are reduced and alkylated, whereas electrophoretic separation could be achieved by polyacrylamide gel electrophoresis in 8 JZ urea when the disulfide bonds are intact may be compatible with the idea that there is a different pairing of disulfide bonds in the two prothrombins.
Since the concept of stable conformational variants might have a bearing on the present results attempts to reoxidize the normal prothrombin from random coil have been undertaken as suggested by Epstein and Schechter (51). However, so far such attempts have yielded only high molecular weight prothrombin aggregates despite a wide variation of the reosidation conditions with regard to pH, protein concentration, mercaptoethanol concentration, and calcium ion concentration.
Similarly, attempts to normalize the abnormal prothrombin by incubation with mercaptoethanol at a physiological pH have been unsuccessful. 4 Using the immunofluorcscence technique Barnhart and Anderson (52) could not identify any precursor prothrombin in the liver of dogs \\-ith the same antigenic determinants as prothrom bin. However, in the rat it has been suggested that a precursor of prothrombin is accumulated in the liver in vitamin K deficiency Tvhich UIIOIL administration of the vitamin is completed and released to the circulation (53~56). This has also been suggested by Shah and Suttie (57) who demonstrated that, the prothrombin formed in vitamin K deficient rats when radioactive amino acids and the vitamin are administered after cyclohesimide administration contains no radioactivity. If the dicoumarolLinducrd prothrombin has any relat.ion to a prothrombin precursor cannot be evaluated at present.
The fact that the normal and the dicoumarol-induced prothrombin arc identical in carbohydrate composition and have identical peptide maps is incompatible wit,h the recent suggestion that dicoumarol ad\Tersely affects the carbohydrate attachment to the polypeptide chain (58,59). Furthermore, the demonstration that the abnormal prothrombin does not possess prothrombin activity (1) is not in accord with this hypothesis, since it, has been demonstrat'ed that removal of a substantial amount of the carbohydrate from normal prothrombin has no effect on the biological activity of the protein or on its adsorption to barium salts (60).
Further experimental work on the covalent structure and the conformation of the normal and the dicoumarol-induced prothrombin is necessary before any hypothesis on the mode of action of vitamin K can be arrived at. However, the fact that vitamin K can function as an oxidation-reduction mediator in the cell and that dicoumarol inhibits the enzyme vitamin K reductase (61) or DT diaphorase (EC 1.6.99.2) (62) makes it tempting to speculate that there is an abnormal pairing of disulfide bridges in the dicoumarol-induced prothrombin which is also consistent with the experimental results so far obtained.