Purification and Characterization of Human Coagulation Factor V *

We have purified human coagulation Factor V 6,000fold to homogeneity from citrated plasma using polyethylene glycol 6000 precipitation, adsorption of Factor V to barium citrate, DEAE-Sepharose chromatography, and gel filtration on Ultrogel AcA 34 (yield 21%). Human Factor V is a single polypeptide chain before and after disulfide bond reduction with an apparent M, = 335,000 as determined by electrophoresis on 5% acrylamide sodium dodecyl sulfate gels. Human Factor V is a glycoprotein containing 13% by weight carbohydrate and there is a high content of sialic acid (86 residues/mol) compared to the other sugars. When human Factor V is treated with thrombin, coagulation activity increases 25to 30-fold to a specific activity of 1.7 to 2.0 units/pg. Thrombin activation is accompanied by the cleavage of three bonds in the Factor V molecule. We have detected activation intermediates with apparent M, = 295,000 and 248,000 and final products with apparent M, = 150,000,121,000, and a doublet at 95,000-91,000 as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The final products of thrombin activation of human Factor V and bovine Factor V are similar, yet the intermediates observed are different. This suggests that cleavages are made at similar locations in bovine and human Factor V, but that they occur in a different sequence. When human Factor V is treated with the Factor V activator from Russell’s viper venom, it is split into two components with apparent M, = 303,000 and 95,00001,000 and is fully activated. The increase in coagulation activity observed upon treatment of human Factor V with thrombin or the Factor V activator from Russell’s viper venom seems to correlate with the generation of the doublet M, = 95,000-91,000 component.

X, (91, and high affinity platelet binding sites (10). Recently, activated bovine protein C has been shown to inactivate bovine Factor V, through limited and specific proteolysis (11,12). Rosenberg et al. (13) reported a purification of human Factor V. This product appeared as a single band on isoelectric focusing gels, but was not characterized by SDS'-polyacrylamide gel electrophoresis. After treatment with thrombin there was a 2-to %fold increase in coagulation activity and the Factor V, obtained had a specific activity of 177 units/mg. Bolhuis et al. (14) also purified human Factor V. This preparation appeared as a single band on SDS-polyacrylamide gels without 2-mercaptoethanol with an apparent M, = 300,000.
No SDS-polyacrylamide gels of samples reduced with 2-mercaptoethanol were shown. Following treatment with thrombin there was an 8-fold increase in coagulation activity, and the Factor V, had a specific activity of 120 units/mg. In this paper we report the purification of human Factor V to homogeneity from fresh frozen plasma. In addition we examine activation of human Factor V by thrombin and the Factor V-activating enzyme from Russell's viper venom. Our Factor V preparation when activated by thrombin had a specific activity of 1.7 to 2.0 units/pg, which is 10-fold higher than previous preparations.

RESULTS
Purification of Human Factor V-As was true for bovine Factor V (4), the keys to the successful purification of human Factor V were: 1) rapid purification steps performed without interruption; 2) inclusion of calcium in buffers to stabilize Factor V and 3) the use of benzamidine and diisopropylfluorophosphate to prevent proteolysis of Factor V. We have purified human Factor V approximately 6000-fold (see Table  I) with a 21% yield based on coagulation assay. Ten milligrams of homogenous Factor V were obtained from 4 liters of fresh frozen citrated plasma. The isolation of Factor V was completed within 42 h after the plasma was thawed. The unique step in our Factor V isolation is based on the finding that human Factor V is adsorbed to barium citrate in the presence but not in the absence of 11% polyethylene glycol 6000. Thus, we achieved a 33-fold purification of Factor V after this step. '' Protein concentration was estimated by absorbance at 280 nm assuming 23% = 10. The concentration of the final product human Factor V was measured assuming E% = 8.9.
After adding sodium chloride to a final concentration of 0.1 M, the Factor V eluted from barium citrate was applied to a DEAE-Sepharose column (Fig. 1). Fractions from the trailing edge of the activity peak were found to contain degraded Factor V, so these fractions were omitted from the pool. This step resulted in a 16-fold purification with a 40% recovery of activity. Bolhuis et al. (14) obtained poor recoveries of Factor V activity on Sephadex G-200, but found that acrylamideagarose-based resins (Ultrogel) allowed good recoveries of activity. We used Ultrogel AcA 34 and found that Factor V eluted in the void volume of the column. The high concentration of benzamidine made it impossible to determine protein concentration by absorbance measurements. To determine the protein elution prof'iie for this step, we equilibrated an AcA 34 column with buffer containing 1 mM benzamidine and applied a Factor V sample eluted from DEAE-Sepharose corresponding to about one-fourth the amount shown in Table  I (Fig. 2). Although chromatography in 1 mM benzamidine increased the amount of degraded factor when examined by SDS-gel electrophoresis, it did illustrate the separation of Factor V from contaminants. The AcA 34 pool from a column run with 10 mM benzamidine was concentrated to a 2-ml volume and stored at -70°C. Purified Factor V stored in this manner was stable for at least 3 months as judged by total activity and SDS-gel electrophoresis. Loss of Factor V activity due to adsorption is a problem at concentrations less than 200 pg/ml. This can be prevented by the inclusion of 5 mg/ml of bovine serum albumin in samples.
Analysis of Human Factor V by SDS-Polyacrylamide Gel Electrophoresis-Human Factor V is a single-chain polypeptide with an apparent M , = 335,000 as determined by electrophoresis on 5% acrylamide-SDS gels before and after disulfide bond reduction (see Fig, 4, zero time point gel). In addition to staining with Coomassie blue, Factor V stains with PAS reagent, indicating that it is a glycoprotein (data not shown). The apparent molecular weight for human Factor V derived from SDS-gel electrophoresis is an estimate since Factor V is a glycoprotein (20). However, this molecular weight is similar to the values reported for bovine Factor V. Amino Acid and Carbohydrate Composition-The amino acid and carbohydrate composition of our human Factor V preparation is compared with the composition for bovine Factor V reported by Nesheim et aZ. in Table 11. The compositions are similar except that human Factor V has a higher half-cystine and phenylalanine content and lower proline content compared to bovine Factor V. Using the method of Harris and Teller (28), we estimate that the amino acid sequence of human Factor V is 65 to 87% homologous with the bovine protein.  This study. Composition expressed as residues/335,000 g of glycoprotein (residues/290,000 g of amino acids).
Data taken from Nesheim et al. (4). Composition expressed as residues/329,000 g of glycoprotein (residues/290,000 g of amino acids).
' Determined as cysteic acid according to Hirs (21). Determined by the method of Edelhoch (22). Determined by the method of Liu and Chang (34).
Determined from amino acid analysis.
* Determined by the method of Warren (25).
Both human and bovine Factor V are glycoproteins containing 13% and 12% carbohydrat,e, respectively. Detailed carbohydrate composition data is not available for bovine Factor V. We find that human Factor V has an unusual found only 20 mol of N-acetylglucosamine/335,000 g of human Factor V. It is possible that part of this difference is due to incomplete hydrolysis of N-acetylglucosamine on our part since our hydrolysis conditions were milder than those employed by Nesheim et al. (4).
Extinction Coefficient-We determined the concentration of Factor V using the synthetic boundary method of Babul and Stellwagen (26). We assumed a refractive increment of 4.1 fringes/mg of glycoprotein/ml. The results obtained when amino acid and carbohydrate composition data were used to calculate the Factor V concentration were in good agreement with the concentration measured in the ultracentrifuge. The extinction coefficient, E&&, was calculated to be 8.9. Activation of Factor V by Thrombin-When purified human Factor V was incubated with catalytic amounts of thrombin, there was a 25-to 30-fold increase in activity as measured by coagulation assay. In several experiments the specific activity of thrombin-activated Factor V, ranged from 1.7 to 2.0 units/pg. Concomitant with the increase in clotting activity was the disappearance of the M, = 335,000 component and the appearance of several new lower molecular weight components on SDS gels. In order to study the mechanism of activation of Factor V by thrombin, we slowed the reaction by lowering the temperature to 4°C. When human Factor V is incubated with 2 units of human thrombin/ml, there is a gradual increase in clotting activity (Fig. 3). After a 15-min incubation with thrombin at 4'C Factor V, activity had increased 5.5-fold. At 15 min the reaction temperature was raised to 37°C and the activation process was rapidly completed. In this experiment Factor V activity increased 29-fold; the fully activated Factor V had a specific activity of 1. ,000 component also appears as a doublet. All of the bands which stained with Coomassie blue also stained with PAS reagent. In addition another band appeared in gels that were stained with PAS reagent that was not visible in Coomassie blue-stained gels. This band appeared to be a final product of Factor V activation. It had an apparent M, = 190,000 (actually assumed to be 150,000, see below) on 5% acrylamide-SDS gels. These results indicate that all of the fragments which result from thrombin activation of Factor V contain carbohydrate.
Glycoproteins have been reported to give anomalously high apparent molecular weight values when subjected to electrophoresis in SDS on polyacrylamide gels (20). Segrest and Jackson found that this anomalous behavior decreases when increasing concentrations of acrylamide are used. We examined the migration of the products of thrombin-activated Factor V. on 5, 7.5, 10, and 12.5% acrylamide-SDS gels. No change in apparent molecular weight was noted for the M , = 121,000 and M , = 95,000-91,000 components when the acrylamide concentration was increased. However, the M , = 190,OOO component reached a minimum apparent M , = 150,000 as the acrylamide concentration was increased. In addition, no new low molecular weight Coomassie blue-or PAS reagentstaining activation components were identified on the higher percentage of acrylamide gels. When Factor V is activated by thrombin a t 22"C, the results obtained are identical with those shown in Fig. 4 except that both the M , = 295,000 and M , = 248,000 components are seen 1 min after thrombin is added.
When Factor V from the trailing edge of the activity peak on DEAE-Sepharose was further purified on AcA 34 and analyzed by SDS-polyacrylamide gel electrophoresis, three components were seen. The first had an apparent M , = 335,000 and corresponded to intact Factor V. The other components had apparent M , = 248,000 and 150,000. When this preparation was treated with thrombin, all three high molecular weight components disappeared and the products and specific activity were identical with fully activated Factor V,. The M , = 150,000 component present here stained well with Coomassie blue, distinguishing it from the one detected during activation of intact Factor V with thrombin, which only stained with PAS reagent. The M , = 248,000 component may correspond to the M , = 248,000 component seen during thrombin activation of intact Factor V. The Coomassie blue-staining M , = 150,000 component is not seen during activation of intact Factor V. These additional components may represent Factor V cleaved during the purification process by a protease other than thrombin.
Activation of '251-labeled Factor V by Thrombin-In order to more finny establish that the M , = 150,000 PAS-positive component was a final product of Factor V activation, and to look for additional components that do not stain with Coomassie blue or PAS reagent, we treated "'I-labeled Factor V with thrombin. Fig. 5 shows the radiochromatogram of a 7.5% SDS-polyacrylamide gel of thrombin-activated '"I-labeled Factor V,. The peak in gel slice 6 corresponds to the M, = 150,000 component. The peak in slice 9 corresponds to the M , = 121,000 component. The peak in slice 13 represents the M , = 95,000-91,000 doublet which is not resolved in the radiochromatogram. The peaks in slices 21 and 24 were present in the '251-labeled Factor V preparation before treatment with thrombin and probably represent highly radiolabeled contaminants. Activation time courses of l"I-labeled Factor V incubated with thrombin showed patterns identical with those shown in Fig. 4  Activation of Factor V by V-CP-V-CP is an enzyme isolated from Russell's viper venom which activates Factor V (7). When Factor V was incubated with 10 pg/ml of V-CP at 37OC, Factor V activity increased 26-fold within 10 min to a specific activity of 1.8 units/pg. Coagulation activity was not further increased by the addition of 2 units/ml of thrombin. In the experiment shown in Fig. 6 300,000 when electrophoresed on 5% polyacrylamide gels, but these workers did not show samples reduced with 2-mercaptoethanol. Using their method we have found that the high molecular weight component seen on SDS-polyacrylamide gels without 2-mercaptoethanol disappears upon reduction and is therefore unrelated to intact Factor V." Our human Factor V appears as a single high molecular weight polypeptide chain on SDS-polyacrylamide gels in the presence and absence of 2-mercaptoethanol. Comparing the specific activities of thrombin-activated Factor V. , we estimate that the previously reported human Factor V preparations were less than 10% pure. Our preparation is activated 25-to 30-fold when treated with thrombin. Previous human Factor V preparations were activated only 2-to 3-fold (13) and 8-fold (14), suggesting that they also contained Factor V. . Human and bovine Factor V have similar mobilities upon electrophoresis on 5% polyacrylamide-SDS gels (data not shown). Nesheim et al. (4) found the molecular weight of their purified bovine Factor V preparation to be 330,000 as determined by sedimentation equilibrium and sedimentation velocity studies. Using this value for the molecular weight of our bovine Factor V preparation, we estimate an apparent M , = 335,000 for human Factor V. Bartlett et al. (30) ,reported molecular weights between 800,000 and 1,000,000 for purified bovine Factor V and partially purified human Factor V. However, the gel filtration method employed by these workers assumes that Factor V is a globular protein. the component reported by Esmon (5) and Nesheim et al. (4), and an l2 subunit which is a disulfide-linked dimer with M , = 400,000. We do not find any component resembling the latter in our human Factor V preparation. Human Factor V is a glycoprotein containing 13% carbohydrate. There is an unusually large amount of sialic acid present in human Factor V. This suggests the possibility of a novel, highly sialated oligosaccharide structure(s) (29). Whether sialic acid residues have any importance in the function of Factor V remains to be determined.
When human Factor V is activated by thrombin, we observe two components that are activation intermediates and three components that are final products. In Table I11 these components, identified by their apparent molecular weights, are compared with those components observed during the activation of bovine Factor V by thrombin. The correspondence between bovine and human peptides seems likely but has not yet been rigorously established.
There are three bonds in the Factor V molecule cleaved by thrombin, one of which is also cleaved by V-CP. At 4OC, thrombin forms the final product with M , =  The components seen in partially degraded human Factor V are similar to the two intermediates seen during the activation of bovine Factor V by thrombin. The reason that different intermediates are observed when human and bovine Factor V are activated with thrombin may be explained most simply by a different order of bond cleavages ( 7 ) . It is clear that the apparent molecular weights used to designate the components in this activation scheme may not represent the true values. This can be expected because of their large size and the fact that they are heavily glycosylated. The true molecular weights of Factor V and the components of Factor V, wiU require the isolation and characterization of the intermediate and end products of Factor V activation. We do not know whether the molecular weights of the various human components are actually larger than the corresponding bovine products as appears from comparison of data from different laboratories.
In contrast to thrombin, V-CP produces products of apparent M , = 95,000-91,OOO and M , = 303,000. The fact that Factor V which has been treated with V-CP and then thrombin gives a gel pattern identical with Factor V which has been treated with thrombin alone suggests that this bond is also cleaved by thrombin. Cleavage of the V-CP-sensitive bond results in complete activation of Factor V. When Factor V is treated with thrombin as in Fig. 3, most of the Factor V activity does not appear until the M , = 95,000-91,000 component is visualized. It therefore seems probable that cleavage of the Factor V to molecule to produce the M, = 95,000-91,000 component by either thrombin or V-CP is the necessary event required for the development of Factor V coagulation activity. The development of small amounts of Factor V activity during the 4°C incubation in Fig. 4 might be explained by the formation of small amounts of the M , = 95,000-91,000 component which stain too faintly to be seen. This conclusion is different from those made by workers studying the activation of bovine Factor V by thrombin, who concluded that the appearance of activity correlated with the appearance of the bovine M , = 94,000 (6) or M , = 110,000 (5) component or the disappearance of the bovine M , = 150,000 component (33). After this work was completed we received a manuscript by Dahlback (35) that reports the successful purification of human Factor V by a method different from that reported here. This preparation was essentially identical with ours although it was obtained in a much lower yield. After treating human Factor V with thrombin, Dahlback also observed activation intermediates that were different from those occurring during activation of bovine Factor V in agreement with our results.

Acknowledgments-We wish to thank Owen Bates and Ralph
Bradshaw for assistance in the amino acid analysis, Jacques Baenziger for assistance in the carbohydrate analysis, Bruce Barshop for assistance in running the model E ultracentrifuge, Craig M. Jackson for providing the Factor V-activating enzyme from Russell's viper venom and bovine Factor V, and Bjorn Dahlback for providing a preprint of his manuscript. Helpful discussions with George Braze and Craig M. Jackson are also gratefully acknowledged.
Addendum-We have recently measured high affinity binding of human Factor V, to platelets. Sodium dodecyl sulfate-gel electrophoresis of platelet pellets discloses that the M , = 121,000 and M , = 95,000-91,OOO components bind to platelets, but that the M , = 150,000 component does not. This indicates that the M , = 150,000 component is not really a component of Factor V. as we imply in our paper but rather that it is an "activation peptide."