Stoichiometry of the Porcine Factor VIII-von Willebrand Factor Association*

Factor VI11 and von Willebrand factor (vWF) are glycoproteins that form a tightly bound complex in plasma. The interaction of porcine factor VIII with porcine vWF was studied by analytical velocity sedi- mentation. A single -240-kDa species of factor VI11 was isolated for use in the analysis. In contrast, when analyzed by agarose/sodium dodecyl sulfate-polyacryl-amide gel electrophoresis, vWF consisted of a popula- tion of >10 multimers derived from a 270-kDa monomer. A single boundary (8&w = 7.2 s) was observed during velocity sedimentation of factor VI11 at 260,000 X g. A single boundary also was observed for vWF (weight-average 8;o.w = 21 S) at 42,000 x g. Under condition of excess factor VIII, the weight-average s;~,,,, of the factor VIII-vWF complex was 40 S at 42,000 X g. At 260,000 X g, the factor VIII-vWF complex had sedimented completely, leaving only free factor VIII. The height of the plateau region of the factor VIII sedimentation velocity curve at 260,000 X g was studied as a function of several starting concen- trations of vWF. The experiments were done under conditions in which the effect of radial dilution was negligible so that the plateau height was a measure of the concentration of free factor VIII. The plateau height decreased linearly as the concentration of vWF was increased, indicating that the association was essentially irreversible under the conditions used. A stoi- chiometry of 1.2 vWF monomers/factor

Stoichiometry of the Porcine Factor VIII-von Willebrand Factor Association* (Received for publication, July 27, 1987) Pete LollarS and Carlo G. Parker From the Departments of Medicine and Biochemistry, University of Vermont, Burlington, Vermont 05405 Factor VI11 and von Willebrand factor (vWF) are glycoproteins that form a tightly bound complex in plasma. The interaction of porcine factor VIII with porcine vWF was studied by analytical velocity sedimentation. A single -240-kDa species of factor VI11 was isolated for use in the analysis. In contrast, when analyzed by agarose/sodium dodecyl sulfate-polyacrylamide gel electrophoresis, vWF consisted of a population of >10 multimers derived from a 270-kDa monomer. A single boundary (8&w = 7.2 s) was observed during velocity sedimentation of factor VI11 at 260,000 X g. A single boundary also was observed for vWF (weight-average 8;o.w = 21 S) at 42,000 x g. Under condition of excess factor VIII, the weight-average s;~,,,, of the factor VIII-vWF complex was 40 S at 42,000 X g. At 260,000 X g, the factor VIII-vWF complex had sedimented completely, leaving only free factor VIII. The height of the plateau region of the factor VIII sedimentation velocity curve at 260,000 X g was studied as a function of several starting concentrations of vWF. The experiments were done under conditions in which the effect of radial dilution was negligible so that the plateau height was a measure of the concentration of free factor VIII. The plateau height decreased linearly as the concentration of vWF was increased, indicating that the association was essentially irreversible under the conditions used. A stoichiometry of 1.2 vWF monomers/factor VI11 molecule was calculated from the slope of the line. Assuming one factor VIII-binding site/vWF monomer, these results indicate that all factor VIII-binding sites are accessible in the vWF multimer.
Factor VI11 and von Willebrand factor (vWF)~ are glycoproteins that are necessary for normal hemostasis. Their corresponding deficiency states are known as hemophilia A and von Willebrand's disease, respectively. When activated by proteolytic cleavage, factor VI11 is a cofactor for factor IXa in the activation of factor X in the intrinsic pathway of blood * This work was supported by United States Public Health Service Clinical Investigator Award HL-01538 (to P. L.), by Grant-in-Aid 85-119 from the American Heart Association with funds contributed in part by the Vermont Affiliate, and by Vermont Specialized Center for Research in Thrombosis Grant HL-35058. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $To whom correspondence should be addressed: University of Vermont, Given Medical Bldg., Burlington, VT 05405.
von Willebrand factor circulates as a population of multimers (12). The monomeric subunit is synthesized in endothelial cells (13) and megakaryocytes (14) as a single polypeptide chain with a molecular mass of 270 kDa (15). Multimers are assembled in the cell of origin by linking subunits via disulfide bonds (16). The size range of porcine vWF has been estimated to be 1-20 MDa by agarose/SDS gel electrophoresis (12).
von Willebrand factor participates in platelet-vessel wall adhesion and perhaps platelet aggregation through interactions with collagen, non-collagen vessel wall components, and platelet surface glycoproteins (17). Additionally, vWF binds factor VI11 to form a tight, noncovalently linked complex (18). This interaction prolongs the half-life of factor VI11 in plasma (7).
The binding capacity of native, multimeric vWF for factor VI11 is unknown. We have isolated a single -240-kDa species of porcine factor VIII and determined its extinction coefficient by far-UV absorbance spectroscopy. Additionally, the extinction coefficient of porcine vWF has been determined by dry weight measurement and differential refractometry (with equivalent results). Using these rigorously defined preparations, the factor VIII-vWF interaction has been analyzed by velocity sedimentation. The strength of the interaction coupled with the large difference in size between factor VIII and vWF or the factor VIII-vWF complex makes it possible to distinguish free factor VI11 from factor VI11 in complex. We have used this method to measure the stoichiometry of the interaction of factor VI11 with multimeric vWF.

EXPERIMENTAL PROCEDURES
Electrophoresis-Discontinuous SDS-PAGE was done using the buffer system of Laemmli (19). Samples containing 1% (w/v) SDS with or without 1-2% (v/v) B-mercaptoethanol were heated for 2-5 min in a heating block maintained at 100 "C. Factor VI11 was analyzed by electrophoresis of nonreduced samples in 7% acrylamide gels. Reduced vWF was analyzed by electrophoresis in 5% acrylamide gels. Myosin (200 kDa), phosphorylase b (97 kDa), bovine serum albumin (68 kDa), ovalbumin (43 kDa), and chymotrypsinogen (26 kDa) (Bethesda Research Laboratories) were used as standards. Additionally, human fibrinogen (340 kDa), purified according to the method of Straughn and Wagner (20), was used as a standard in gels containing vWF. Agarose/SDS-PAGE of vWF was done as described by Chopek et al. (15). Serum from a patient with Waldenstrom's macroglobulinemia was used to identify immunoglobulin M as a molecular mass marker (1 MDa) (12). Factor VI11 was detected by silver staining (21). vWF was detected by staining with Coomassie Blue R-50.
Factor VIII and uon Willebrand Factor Assays-Factor VI11 and thrombin-activated factor VIII were measured by one-and two-stage assays, respectively, using activated partial thromboplastin and hu-man factor VIII-deficient plasma (George King Biomedical) (4). One unit of porcine factor VI11 was defined as the amount of factor VI11 in 1 ml of citrated normal porcine plasma.
Porcine von Willebrand factor was measured in an assay based on its ability to agglutinate fixed human platelets in the presence of ristocetin (Bio/Data Corp. Hatboro, PA) (22). One unit of porcine von Willebrand factor was defined as the amount in 1 ml of citrated normal porcine plasma.
(Porton Products) was dissolved by adding 4 ml of 0.1 M MES, 0.1 M Isolation of Proteins-Commercial porcine factor VI11 concentrate NaCl, 0.255 M MgCl,, pH 6.0, to each of 10 bottles of concentrate at room temperature to dissociate factor VI11 from vWF. Each bottle contained approximately 8 mg of total protein and 60 units of factor VIII. Porcine factor VI11 was isolated from commercial concentrate by murine monoclonal anti-porcine factor VIII-Sepharose chromatography? The hybridoma cell line producing the antibody was generously provided by Dr. D. N. Fass (Mayo Clinic, Rochester, MN).
Three heterodimers of porcine factor VI11 having apparent molecular masses by SDS-PAGE of 160 + 76, 130 + 76, and 82 + 76 kDa have been described previously (4,23). The 76-kDa chain represents a common light chain, whereas the three heaviest chains have a common NHz-terminal sequence (2) and are derived from variable proteolysis of the middle region, or B domain (24), of the parent singlechain factor VI11 molecule. Two of the four lots of commercial concentrate that have been examined have lacked the 130-kDa heavy chain species. The 160 + 76 (or -240-kDa) heterodimer was isolated from a lot lacking the 130-kDa species by cation-exchange HPLC using an HR 5/5 Mono S column (Pharmacia LKB Biotechnology Inc.)? The average of two preparations yielded a specific activity of 1100 units/mg in a one-stage assay for factor VIII. Factor VI11 was activated greater than 50-fold by thrombin in a two-stage assay. Factor VI11 was stored in the HPLC elution buffer, 0.4 M NaCl, 10 mM histidine C1, 5 mM CaClZ, 0.01% Tween 80, pH 6.0 (Buffer A + 0.4 M NaCl), at 4 'C and used within 24 h of preparation.
vWF also was isolated from commercial porcine factor VI11 concentrate. Greater than 95% of the protein did not adsorb to antifactor VIII-Sepharose during the isolation of factor VI11 and was concentrated from 40 to 20 ml by ultrafiltration through an Amicon PM-10 membrane. vWF was isolated by Sephacryl S-lo00 (Pharmacia LKB Biotechnology Inc.) (2.5 X 85-cm column) gel permeation chromatography and concentrated using conditions described previously for human vWF (25). The final preparation, approximately 1 mg/ml in 2 ml of 0.15 M NaCl, 0.02 M Tris-C1, pH 7.4 (TBS), was stored at 4 "C. The final product had a specific activity of 120 units/ mg and was used within 48 b of preparation. On SDS-PAGE using reducing conditions, greater than 90% of the preparation migrated as a single band at -250 kDa. On composite agarose/SDS-PAGE using nonreducing conditions, approximately 10 bands, corresponding to vWF multimers, were identified. Additionally, the largest forms were not resolved into discrete bands. The smallest band had an apparent molecular mass of approximately 1 MDa since it migrated slightly slower than immunoglobulin M.
Extinction Coefficients-The extinction coefficient at 280 nm of the -240-kDa factor VI11 heterodimer was determined by far-UV spectroscopy as described by van Iersal et al. (26). Samples underwent buffer exchange into 0.1 M sodium phosphate, pH 6.5, by chromatography on a Sephadex G-25 superfine column (Pharmacia LKB Biotechnology Inc.) before analysis. Determinations were done in duplicate from samples from a single preparation. The extinction coefficient at 280 nm of vWF was determined in duplicate on two preparations by: 1) differential refractometry using a double-sector synthetic boundary cell in a Beckman Model E analytical ultracentrifuge (27); or 2) dry weight measurement (28). The absorbance of vWF at 280 nm was corrected for light scattering by plotting In (turbidity) against In (wavelength) from 320 to 400 nm at 10-nm intervals, followed by extrapolation to obtain the turbidity at 280 nm (29). The slope of the line was 3.0 for two separate preparations, indicating significant departure from the fourth power Rayleigh scattering law, presumably because the diameter of the scattering particles is larger than the wavelength of the incident light (30). The following values (E:'&) were obtained: -240-kDa factor VI11 heterodimer, 0.88, vWF (dry weight), 0.810; and vWF (refractometry), 0.838. An average value for vWF of 0.824 was used for subsequent cdculations.
Analytical Ultracentrifugation-Velocity sedimentation profiles were obtained using a Beckman Model E analytical ultracentrifuge equipped with a photoelectric scanner, cylindrical lens, and mirrored Lollar P., Parker C. G., and Tracy R. P., BIood, in press. optics. Carriage travel and absorbance data were captured by a microcomputer using an ISAAC data acquisition system (Cyborg). Calculations were done using software developed by Dr. M. N. Blackburn, (Louisiana State University, Shreveport, LA). Measurements were made at 280 nm on 0.25-ml samples at 19-21 "C in double-sector cells that contained sapphire windows. Buffers used were Buffer A + 0.3 M NaCl and TBS for factor VIII and vWF, respectively. Sedimentation coefficients were calculated from linear least-squares regression analysis of the natural logarithm of the equivalent boundary position, i, versus time. For samples containing only factor VIII, i was determined by midpoint analysis. Because of its known heterogeneity, i of samples containing vWF was determined both by midpoint and second moment analysis (31). The latter calculation yields the weight-average sedimentation coefficient.
The association of factor VI11 with vWF in the ultracentrifuge was studied as follows. Factor VIII in Buffer A + 0.4 M NaCl was mixed with vWF in TBS at a ratio (v/v) of 1.5:1.25 for 20 min at room temperature (final NaCl concentration, 0.3 M; final pH, 6.5). After the sample was loaded, the rotor speed was increased to 24,000 rpm (42,000 X g). Scans were taken at 4-min intervals within 2 min of reaching the target speed. This rotor speed allowed observation of the sedimenting factor VIII-vWF complex boundary. After five scans were obtained, the rotor speed was increased to 60,000 rpm (260,000 X g), followed by scans a t 4-min intervals to observe the boundary of free (unbound) factor VIII. The height of the plateau region is directly related to the total concentration of sedimenting protein. Therefore, the height obtained from the first scan (zero-time scan) was used as a measure of the free factor VIII. The concentration of free factor VI11 was determined using the plateau height of the zero-time scan of a known concentration of factor VI11 in the absence of vWF. Since the height of the plateau region decreases with time due to radial dilution (31), the zero-time scan was started within 2 min after the target speed was reached to minimize this effect. Additionally, the sedimentation of vWF in the absence of factor VI11 was done at 260,000 X g to obtain the base line due to nonsedimenting absorbance in the vWF preparation. This base-line correction, which was small, was also subtracted from the plateau height before calculating the concentration of free factor VIII.

RESULTS AND DISCUSSION
Velocity Sedimentation of Factor VIII and vWF-We isolated the largest heterodimer of porcine factor VI11 (-240 kDa) by cation-exchange nondenaturing HPLC in sufficient quantities for study by analytical ultracentrifugation (Fig. 1). This represents the first reported isolation procedure for a homogeneous preparation of factor VIII. The lack of heterogeneity simplified the analysis of the interaction of factor VI11 with vWF. A single boundary was observed during velocity sedimentation of factor VIII (Fig. 2), which is consistent with a single sedimenting species. The sedimentation coefficient obtained from linear regression analysis of the plot shown in Fig. 2 (inset) was 7.2 S. The same value (within 0.2 S) was obtained when the concentration of factor VI11 was decreased 4-fold, indicating that there was no concentration dependence of the sedimentation coefficient under the conditions used in this study.
Despite its known heterogeneity, only a single boundary was seen during velocity sedimentation of vWF (Fig. 3). This presumably represents overlapping boundaries due to the size distribution of multimers over regularly spaced intervals (12). Heterogeneity is also indicated from the comparison of the sedimentation coefficients calculated by either midpoint and second moment analysis. The two methods gave different values (Table I), which is consistent with heterogeneity (31). The sedimentation coefficient of vWF was not significantly different at a 4-fold lower concentration, indicating that it has no concentration dependence under the conditions used in this study (not shown). Therefore, the value in the Table I represents the weight-average s&, of the sample. vWF was purified by gel permeation chromatography. Agarose/SDS-PAGE of various fractions from this step showed that the distribution of the population of multimers shifted to larger 97. multimers as the elution volume decreased; as described previously (32). Since a fraction of vWF was not pooled (-20% by assay) to avoid lower molecular mass contaminants, the weight-average sedimentation coefficient obtained from this preparation may be higher than that of circulating vWF. Previous velocity sedimentation comparing purified and plasma porcine vWF in sucrose density gradients yielded similar results; the estimated sedimentation coefficient is 25-32 S (33). The porcine vWF used in this study has similar properties since: 1) more than 10 multimers with apparent molecular mass >1 MDa were observed after electrophoresis in agarose/SDS-PAGE (not shown); and 2) the weight-average sedimentation coefficient of the vWF was 21 S (Table I)  The -240-kDa factor VI11 heterodimer was mixed with vWF as described under "Experimental Procedures," and the resulting solution was analyzed by ultracentrifugation at 42,000 x g. The results for one concentration of vWF are shown in Fig. 4. The absorbance near the meniscus (-6.5 cm) is greater than zero due to free factor VIII. The weight-average sedimentation Coefficient, calculated from second moment analysis after subtracting the base line due to free factor VIII, was 39 S. By midpoint analysis, the sedimentation coefficient was 48 S, again indicating heterogeneity of the sample (Table I).
The experiment was repeated at several concentrations of vWF with the factor VI11 concentration held constant. The calculated sedimentation coefficients did not vary appreciably with concentration (Table I). This indicates that the experiments were done under conditions of excess factor VI11 such that vWF was completely saturated.
Stoichiometry of the Factor VIII-vWF Interaction-After each experiment shown in the Table I, the relative centrifugal field was increased to 260,000 X g. Because of its size, the factor VIII-vWF complex had sedimented completely by the time the rotor had reached the target speed, which allowed the analysis of free (unbound) factor VIII. The plateau height of the sedimentation velocity curve yields the concentration of free factor VI11 when calculated as described under "Experimental Procedures." The plateau height decreased as the vWF concentration increased due to the incorporation of factor VI11 into the complex, but there was no detectable change in the sedimentation coefficient of factor VIII. The  . 4. Velocity sedimentation of a mixture of -240-kDa factor VIII and vWF. Factor VI11 and vWF were mixed and subjected to ultracentrifugation at 42,000 X g as described under "Experimental Procedures." The final concentrations of factor VI11 and vWF were 0.09 and 0.119 mg/ml, respectively. Other details are as described for Fig. 3. free factor VIII concentration as a function of the nominal concentration of vWF is plotted in Fig. 5 . A linear relationship is evident as the concentration of v W F is increased to 0.117 mg/ml, indicating that the reaction is essentially irreversible under the conditions used. This is supported by the fact that no boundary was observed at 0.14 mg/ml vWF. Therefore, stoichiometry of the interaction was determined from the slope of the line (34). The calculation was done by averaging the results of two experiments using different preparations of factor VI11 and vWF. Converting to molar ratios using a molecular mass of 240 and 270 kDa for factor VI11 and vWF, respectively, a value of 1.2 vWF monomers/factor VI11 molecule was obtained. This indicates that the stoichiometry of the interaction approaches 1:l for completely saturated vWF. Equivalently, the mass of the completely saturated vWF is 1.9 times that of factor VIII-free vWF.
The concentrations of factor VI11 and vWF that are required for the calculation must be known accurately. This was done for factor VIII using far-UV spectroscopy. Because of its large size, solutions of vWF are turbid. Solution turbidity increases with decreasing wavelength, which precluded the use of this spectroscopic technique. Instead, determinations were done by both differential refractometry and dry weight determination with similar results. The extinction coefficient at 280 nm was then calculated after correcting for turbidity.
It is clear from inspecting the data in Table I that the factor VIII-vWF complex has significantly different hydrodynamic properties than free vWF. The sedimentation coefficient depends on several factors including the molecular mass and frictional coefficient of the sedimenting species. The latter depends on shape and degree of hydration. At a fixed frictional coefficient, the sedimentation coefficient, s, of a single species varies as s a m2I3, where m is the molecular mass (31). For heterogeneous substances such as vWF and the factor VIII-vWF complex, it is not strictly valid to calculate the sedimentation coefficient given an average molecular weight. Nevertheless, it is interesting to do the calculation as a rough estimate. Given a sedimentation coefficient of 21 S and a 1:l stoichiometry for the interaction of factor VI11 with the vWF monomer, the calculated sedimentation coefficient for the complex is 32 S. This is lower than the observed value of 40

FIG. 5.
Free factor VI11 as a function of the nominal vWF concentration. The plateau height of the sedimentation profile at 260,000 X g was converted to the concentration of free factor VI11 as described under "Results and Discussion" and plotted versus the nominal concentration of vWF. The data points correspond to the samples from the experiment described for Table I S and indicates that the complex may be influenced by a relatively smaller frictional drag than is free vWF. This could be due to tighter coiling of the vWF multimer when it binds factor VIII.
In Fig. 6, a scale model of the completely saturated factor VIII-vWF complex is proposed. The structure of vWF is based on the work of Fretto et al. (35) and Fowler et al. (32). Their ultrastructural studies indicate that the vWF monomer is a fibrillar structure 60 nm long consisting of an NH2-terminal G domain and a COOH-terminal R domain. The monomers are arranged -GR-RG-GR-RG-to form multimers. The vWF monomer is depicted as two cylinders placed end-to-end. The thick cylinder corresponds to the G domain, and the thin cylinder corresponds to the R domain. The radii of the anhydrous cylinders were calculated to be 1.6 and 0.85 nm for G and R, respectively, using published values for their molecular weights and partial specific volumes (35). The C-terminal ends of the monomer are linked to form the so-called vWF protomer (Fig. 6, h e r ) . Factor VI11 is shown as two spheres corresponding to the 76-and 166-kDa chains. Their radii were calculated to be 2.8 and 3.6 nm, respectively, assuming a partial specific volume of 0.72 mg/g. Factor VI11 binds within 273 residues of the NH2-terminal end of the 2050-residue vWF monomer (361, perhaps via the 76-kDa chain (37). In Fig. 6 (upper), a 2.2-MDavWF multimer is shown with 8 bound factor VI11 molecules. Although the vWF monomer may be rigid and elongated, electron microscopic and light scattering studies indicate that most multimers are supercoiled in solution instead of arranged linearly as in Fig. 6 (38,  39). If we assume that there is one binding site for factor VIII/vWF monomer, the results of this study are consistent with the hypothesis that all potential binding sites are accessible in native vWF.
If there is one potential binding site for factor VI11 for each vWF monomer, it is interesting to ask how saturated vWF is in vivo. The reported range of specific activities of human factor VI11 is 2300-8000 units/mg (5)(6)(7)(8)(9)(10)(11). At 1 unit/ml, this corresponds to 0.1-0.4 pg/ml. The concentration of human vWF has been estimated to be 5-10 pg/ml (18). According to these numbers, human vWF is 1-8% saturated with factor VIII. Recently, we found from the specific activity calculations and also from direct radioimmunoassay that the plasma concentration of porcine factor VI11 is 2-3 gg/ml, which is considerably higher than that for human factor VIK2 The concentration of porcine vWF, calculated from the specific activity of the purified material from this or previous (40, 41) studies, is estimated to be 4-8 pg/ml. Thus, porcine vWF may be >50% saturated with factor VIII. At sites of hemostasis, it is conceivable that the local concentration of factor VI11 increases significantly relative to vWF, thereby leading to increased saturation of vWF. It is possible that factor VI11 might regulate vWF function under these conditions.
Velocity sedimentation offers several advantages as a technique to study the interaction of factor VI11 with vWF. The factor VIII-vWF equilibrium is essentially irreversible under the conditions used in this study. With excess factor VIII, this results in the formation of only two species: factor VI11 and the completely saturated complex. Since the species differ markedly in sedimentation properties (7 S uersm 40 S), they can be differentiated easily. Thus, the technique yields quantitative information about the interaction in solution. Additionally, the use of the photoelectric scanner interfaced to a microcomputer allows the measurement of the concentration and sedimentation coefficient of less than 10 gg of factor VIII, vWF, or the factor VIII-vWF complex in approximately 1 h. This technique may be useful in the study of other aspects of the interaction of factor VI11 with vWF, such as subunit interactions, and the regulation of the interaction by proteolytic modification of factor VI11 and/or vWF.