Replacing the Factor VIII C1 Domain with a Second C2 Domain Reduces Factor VIII Stability and Affinity for Factor IXa*

Background: FVIII possesses a duplicated C domain designated C1 and C2. Results: Replacing the C1 domain with C2 reduces FVIII stability and the affinity of FVIIIa for FIXa. Conclusion: The C1 domain likely contributes to FIXa binding and forms a stable interface with the A3 domain. Significance: The FVIII C1 domain is critical to FVIII structure and function. Factor VIII (FVIII) consists of a heavy chain (A1(a1)A2(a2)B domains) and light chain ((a3)A3C1C2 domains). To gain insights into a role of the FVIII C domains, we eliminated the C1 domain by replacing it with the homologous C2 domain. FVIII stability of the mutant (FVIIIC2C2) as measured by thermal decay at 55 °C of FVIII activity was markedly reduced (∼11-fold), whereas the decay rate of FVIIIa due to A2 subunit dissociation was similar to WT FVIIIa. The binding affinity of FVIIIC2C2 for phospholipid membranes as measured by fluorescence resonance energy transfer was modestly lower (∼2.8-fold) than that for WT FVIII. Among several anti-FVIII antibodies tested (anti-C1 (GMA8011), anti-C2 (ESH4 and ESH8), and anti-A3 (2D2) antibody), only ESH4 inhibited membrane binding of both WT FVIII and FVIIIC2C2. FVIIIa cofactor activity measured in the presence of each of the above antibodies was examined by FXa generation assays. The activity of WT FVIIIa was inhibited by both GMA8011 and ESH4, whereas the activity of FVIIIC2C2 was inhibited by both the anti-C2 antibodies, ESH4 and ESH8. Interestingly, factor IXa (FIXa) binding affinity for WT FVIIIa was significantly reduced in the presence of GMA8011 (∼10-fold), whereas the anti-C2 antibodies reduced FIXa binding affinity of FVIIIC2C2 variant (∼4-fold). Together, the reduced stability plus impaired FIXa interaction of FVIIIC2C2 suggest that the C1 domain resides in close proximity to FIXa in the FXase complex and contributes a critical role to FVIII structure and function.

where the lowercase a designates short (ϳ30 -40-residue) segments rich in acidic residues (see Ref. 1 for review). FVIII is activated by thrombin-or FXa-catalyzed cleavages at the a1A2, a2B, and a3A3 junctions. The resulting product, FVIIIa, is a heterotrimer composed of subunits designated A1, A2, and A3C1C2. FVIIIa functions as a cofactor for the serine protease FIXa in the conversion of zymogen FX to the serine protease, FXa (see Ref. 1

for review).
Binding of FVIIIa to the phospholipid vesicle (PLV) surface is essential for cofactor function and maximal FXase activity (2). This binding requires negative charge provided by stereospecific phosphatidyl-L-serine (2,3). A number of studies suggest that both FVIII C1 and C2 domains participate in phospholipid membrane binding (4 -9). In addition, the intermediate resolution x-ray structures of FVIII (10,11) show that the C1 and C2 domains are aligned such that both domains may interact with the PLV surface. Indeed, the presence of both C1 and C2 domains appears required for optimal membrane interaction (12).
FVIII C1 and C2 domains are composed of ␤-barrel structure (10,11,13) and are ϳ66% homologous (39.7% identity). In the current study, we generated an FVIII mutant, FVIII C2C2 , where the C1 domain is replaced by the C2 domain. Experiments were performed to evaluate stability parameters as well as membrane binding and functional properties of this variant as a cofactor for FIXa. Results from this study suggest that reductions in stability and cofactor function result from alterations in FVIII interdomain interactions and reduced affinity for FIXa. These results support an essential role for the C1 domain in FVIII structure and intermolecular interactions. maleimide) (Invitrogen), ␣-thrombin, FVIIa, FIXa␤, FX, and FXa (Enzyme Research Laboratories, South Bend, IN), hirudin (DiaPharma, West Chester, OH), the chromogenic FXa substrate, Pefachrome Xa (Pefa-5523, CH 3 OCO-D-CHA-Gly-Arg-pNA⅐AcOH; Centerchem Inc. Norwalk CT), and enhanced chemifluorescence reagent (GE Healthcare) were purchased from the indicated vendors.
Construction, Expression, and Purification of WT and Variant FVIII-WT FVIII and variants (FVIII C2C2 ) with C1 residues 2022-2168 replaced with C2 residues 2175-2325 were constructed as B-domainless FVIII, lacking residues Gln 744 -Ser 1637 in the B-domain (14) (see Fig. 1A). Recombinant WT and variant FVIII forms were stably expressed in baby hamster kidney cells and purified as described previously (15). Protein yields for the variants ranged from Ͼ10 to ϳ100 g from two 750-cm 2 culture flasks, with purity from ϳ85% to Ͼ95% as judged by SDS-PAGE. The primary contaminant in the FVIII preparations was albumin. FVIII concentration was measured using an enzyme-linked immunosorbent assay (ELISA), and FVIII activity was determined by one-stage clotting and twostage chromogenic FXa generation assays described below.
ELISA-A sandwich ELISA was performed as described previously (16) using purified recombinant FVIII (Kogenate, Bayer Corp.) as a standard. FVIII capture used the anti-C2 monoclonal antibody (GMA8003, Green Mountain Antibody), and the anti-A2 monoclonal antibody (R8B12, Green Mountain Antibody) was employed for FVIII detection following its biotinylation.
SDS-PAGE and Western Blotting-FVIII proteins (0.34 g) were subjected to electrophoresis under reducing conditions (0.1 M dithiothreitol) using 10% polyacrylamide gels at constant voltage (150 V). Proteins were transferred to a polyvinylidene fluoride membrane and probed with an anti-A3 monoclonal antibody (2D2), and protein bands were visualized by chemifluorescence (570 nm) using a Storm 860 PhosphorImager (GE Healthcare).
Dot Blotting-Up to 0.25 pmol of FVIII proteins in 100 l of 20 mM HEPES, 0.1 M NaCl, 0.01% Tween 20, pH 7.2, were transferred to a PVDF membrane using a microfiltration blotting device (Bio-Rad). Proteins were probed with 2D2 or an anti-C1 antibody (GMA8011) and detected by the previously described method (17).
One-stage Clotting Assay-One-stage clotting assays were performed using substrate plasma chemically depleted of FVIII according a method as described previously (18) and assayed using a Diagnostica Stago clotting instrument.
FVIII Thermal Decay-WT and FVIII C2C2 (4 nM) in HEPES buffer were incubated at 55°C, aliquots were removed at the indicated time points, and activity was determined using the FXa generation assay.
FVIIIa Activity Decay-WT and FVIII C2C2 (1.5 nM) in HEPES buffer in the presence or absence of 20 M PLV were activated using 20 nM thrombin for 1 min at 23°C. Reactions were immediately quenched by hirudin (10 units/ml) to inactivate thrombin, aliquots removed at the indicated times, and activity was determined using the FXa generation assay.
FVIII Activity in the Presence of FVIII Antibody-WT and FVIII C2C2 (1 nM) in HEPES buffer in the presence or absence of 20 M PLV were activated using 20 nM thrombin for 1 min at 23°C followed by adding hirudin (10 units/ml) to inactivate thrombin. The reactions were incubated with 300 nM FVIII antibodies (GMA8011, ESH4, ESH8, or 2D2) for 2 min, and the activity was determined using the FXa generation assay.
Phospholipid Vesicle Preparation-Phospholipid vesicles containing OR (OR-PLV) were prepared by mixing 10 mg of PC:PE:PS and 0.6 mg of OR in 1 ml of chloroform and processed as described (22). This method yielded a concentration of 16.0 mM PC:PE:PS and 0.31 mM OR. OR concentration was determined by absorbance at 564 nm (molar extinction coefficient ϭ 95,400). The number of OR molecules per unit of phospholipid area (Å 2 ) was estimated to be 2.7 ϫ 10 Ϫ4 OR molecules/Å 2 based on the criterion that each phospholipid occupies an area of 70 Å 2 (22).
Phospholipid Binding of FVIII as Measured by Fluorescence Resonance Energy Transfer (FRET)-Titration of PyMPO maleimide-labeled WT FVIII or FVIII C2C2 was performed according to the methods described previously (8,23). Briefly, 25 nM FVIII proteins (with or without PyMPO-labeling) in 20 mM HEPES, 0.1 M NaCl, 0.01% Tween 20, 0.01% BSA, 5 mM CaCl 2, pH 7.2, containing 300 M PC vesicles were titrated by adding 0 -60 M PLV or OR-PLV. Three separate titrations were performed including one where labeled FVIII was titrated with PLV without OR (sample-0), a labeled FVIII was titrated with PLV containing OR (sample-1), and an unlabeled FVIII was titrated with PLV with OR (sample-2). After the addition of PLV, samples were incubated for 10 min prior to determining emission fluorescence (540 -546 nm; bandwidth 16 nm) by exciting at 417 nm (bandwidth: 2 nm) using an Aminco-Bowman Series 2 luminescence spectrometer (Thermo Spectronic). Actual fluorescence after quenching by OR (F) was calculated by subtracting sample-2 fluorescence from sample-1 fluorescence. Relative fluorescence (F/F 0 ), the ratio of F to control sample-0 fluorescence (F 0 ), was plotted against phospholipid concentration.
FIXa Binding Affinity-FVIII (0.5 nM) in HEPES buffer containing 20 M PSPCPE was activated by 20 nM thrombin for 1 min and immediately reacted with hirudin (10 units/ml), and the reaction was incubated in the absence or presence of each antibody (300 nM) for 2 min. Samples were then reacted with the indicated concentration of FIXa, and activity was measured by FXa generation assay.
Michaelis-Menten Kinetics-Thrombin-activated FVIII as described above (0.5 nM) in HEPES buffer containing 20 M PSPCPE was incubated with 40 nM FIXa, and FXa generation was initiated by adding the indicated concentrations of FX. Data were fitted to the Michaelis-Menten equation by nonlinear least squares regression, and parameter values were obtained.
FVIIIa Inactivation by Activated Protein C (APC) or FXa-APC-or FXa-mediated inactivation of FVIIIa was performed as described previously (24,25). Briefly, 150 nM FVIII was activated with 30 nM thrombin for 10 min at 37°C. After mixing with 10 units/ml hirudin and 100 M PLV, inactivation reactions were initiated by adding 3 nM APC or 5 nM FXa. Aliquots were removed at the indicated times, and residual FVIIIa activity was measured by a one-stage clotting assay.
Data Analysis-Values for FVIIIa activity decay as a function of time were fitted to a single exponential decay curve nonlinear least squares regression using the equation where A is residual FVIIIa activity (nM/min/nM FVIII), A 0 is the initial activity, k is the apparent rate constant, and t is the time after FVIII activation when thrombin was quenched.
For FVIII-PLV binding kinetics, we used the following equation where F/F 0 is relative fluorescence, A is the concentration of FVIII (25 nM), X is the concentration of phospholipid, K d is a dissociation constant, n is a ratio of binding stoichiometry (phospholipid:FVIII), and Q max is the maximum quenching value. The value of n (ϭ 100) was estimated as described previously (8).
FIXa-FVIII binding affinity used the following equation where A is initial velocity (nM/min/nM FVIII), X is the concentration of FIXa in nM, K d is the dissociation constant, B is the FVIIIa concentration, and V max is the maximum activity at saturation. We utilized a second order polynomial equation as employed previously for an unbiased estimation of the initial reaction rate (24).
where [FVIIIa] is the FVIIIa concentration in nM, t is the time in minutes, A is the initial concentration in nM of FVIIIa and B is the slope at time 0. Rates of FVIIIa inactivation were calculated by dividing the absolute value of B by the concentration of APC or FXa. Computation for nonlinear least squares regression analysis was performed using a standard curve-fitting algorithm (Gauss-Newton algorithm using the method of Levenberg-Marquardt). Fig. 1. The C1 and C2 domains fold into nearly identical ␤-barrel structures (10,11,13). The corresponding region within the disulfide bridge of the C1 domain (residues Cys 2021 -Cys 2169 ) was replaced by C2 domain residues contained within the disulfide bridge bordered by Cys 2184 -Cys 2326 . Thus we replaced residues 2022-2168 with the C2 sequence (2175-2325) to generate FVIII C2C2 . Expressed and purified FVIII proteins were subjected to Western blotting analysis using the A3 domain-specific, 2D2 antibody to verify the molecular weight of LC. The LC of FVIII C2C2 migrated to the same position as WT FVIII Fig. 1B (left panel). The C1 domain-specific antibody, GMA8011, did not work well with Western blotting, likely due to denaturation of the FVIII proteins; however, this reagent was employed to assess the native FVIII in a dot blotting format. Dilutions of the WT and FVIII C2C2 proteins were spotted in the membrane and blotted with either the 2D2 or the GMA8011 antibodies (Fig. 1B, right). The 2D2 antibody showed dose-dependent reactions with both FVIII forms, whereas the C1 domain-specific GMA8011 antibody only recognized the FVIII WT in a dose-dependent fash- , proteins were transferred to PVDF membrane. Right panel, the indicated amounts of purified WT and FVIII C2C2 proteins were transferred to PVDF membrane using a dot-blot apparatus and probed with 2D2 or GMA8011 antibody, and FVIII proteins were visualized by chemifluorescence as described under "Experimental Procedures."

FVIII C2C2 Sequence Structure, Western Blotting/Dot Blotting Analyses, and Cofactor Activity-The domain construction of the FVIII C2C2 variant is shown in
Factor VIII with C1 Replaced by C2 OCTOBER 25, 2013 • VOLUME 288 • NUMBER 43 ion. Together, these data indicate that the LC of FVIII C2C2 contains the duplicated C2 domain. FVIII C2C2 possessed a low but appreciable specific activity as measured by a one-stage clotting assay and by a two-stage FXa generation assay (ϳ16 and ϳ35% of the WT FVIII value, respectively) ( Table 1).
Stability of FVIII Variants-FVIII thermal decay at 55°C as measured by FXa generation assay can be used to monitor the stability of intersubunit interaction (6). Fig. 2A shows results of the thermal decay at 55°C for WT FVIII and FVIII C2C2 . Results obtained for WT FVIII showed ϳ80 and ϳ40% activity remaining after 5 and 18 min, respectively. However, the FVIII C2C2 variant appeared significantly more labile with activity decaying to ϳ20% of the initial level in 5 min ( Fig. 2A). The estimated decay rate for FVIII C2C2 was 11.2-fold higher as compared with WT FVIII (Table 2). Although the interaction of FVIII and PLV at this elevated temperature is unclear, in the presence of PLV, the decay rate of WT FVIII was reduced by ϳ2-fold, whereas the decay rate of FVIII C2C2 was essentially unchanged (data not shown).
FVIIIa activity decay is governed by the dissociation of A2 subunit (26). Fig. 2B shows the results of FVIIIa activity decay for WT FVIII and the FVIII C2C2 variant. In the absence of PLV, FVIIIa activity of both WT FVIII and FVIII C2C2 decayed similarly, showing 40 -50% activity at 4 min. In the presence of PLV, more FVIIIa activity (ϳ70%) was remaining after a 4-min incubation (Fig. 2B). However, FVIII C2C2 activity was similarly reduced, showing ϳ40% activity after 4 min. In the absence of PLV, the FVIIIa decay rate of FVIII C2C2 was 1.5-fold greater than the WT value ( Table 2). In the presence of PLV, the WT FVIIIa decay rate was reduced by ϳ2-fold as compared with the    value measured in the absence of PLV, whereas the decay rate for FVIIIa C2C2 was independent of the presence of PLV.
FVIIIa Activity Inhibition by FVIII Antibodies-We examined FVIIIa cofactor activity in the presence of a panel of anti-FVIII monoclonal antibodies directed against domains in the FVIII LC (Fig. 3). ESH8 is known to bind a region in the FVIII C2 domain that overlaps a thrombin-binding site (27). Because we were interested in the effects of the antibodies on FVIIIa activity, we first activated FVIII with thrombin prior to reaction with the antibodies. However, due to the rapid decay of FVIIIa activity by A2 subunit dissociation, we limited the antibody incubation time to 2 min. Activity was measured by FXa generation assay. Under these conditions, the anti-A3 antibody (2D2) did not inhibit cofactor activity of either WT FVIII or FVIII C2C2 .
On the other hand, the anti-C1 antibody (GMA8011) inhibited WT FVIII activity by ϳ40% but did not inhibit FVIII C2C2 activity. Two anti-C2 antibodies were evaluated. ESH8 showed little if any inhibition of WT FVIII, whereas ESH4 showed ϳ20% inhibition. However, both ESH8 and ESH4 inhibited the FVIII C2C2 variant to levels of ϳ50 and ϳ40%, respectively, of the original activity.
Binding of WT FVIII and FVII C2C2 to PLV-To test whether the above FVIIIa activity inhibition by antibodies was due to altered PLV binding of FVIIIa, we measured PLV binding affinity in the presence of the antibodies. PLV titrations of WT FVIII and FVIII C2C2 were performed, and binding was detected by FRET using PyMPO-labeled FVIII (donor) and OR-PLV (acceptor) (Fig. 4). Relative fluorescence from PyMPO-labeled WT FVIII (Fig. 4A) as well as from FVIII C2C2 (Fig. 4B) decreased in a hyperbolic fashion as the concentration of OR-PLV was increased with saturation occurring at ϳ10 M OR-PLV. With the exception of results obtained in the presence of ESH4, no significant changes in titration profiles were observed for either WT FVIII or FVIII C2C2. However, in the presence of ESH4, the extent of quenching due to OR-PLV binding was markedly reduced for both WT FVIII and FVIII C2C2 . K d values estimated for FVIII C2C2 binding to OR-PLV indicated a modestly reduced affinity (ϳ2.8-fold) as compared with WT FVIII ( Table 3). The FVIII C2C2 variant also showed a modest reduction in the extent of quenching at saturation as expressed by relative fluorescence (F max ) of WT FVIII. In the presence of GMA8011, ESH8, or 2D2, the K d values for WT FVIII binding to OR-PLV were slightly increased (1.9 -2.8-fold), whereas a marked increase in the K d value obtained in the presence of ESH4 was observed (ϳ100-fold) that was accompanied by a significant increase (1.3-fold) in the F max value. Similarly, the K d values for FVIII C2C2 binding to OR-PLV in the presence of GMA8011, ESH8, or 2D2 did not show significant differences from the value obtained in its absence, whereas the K d value in the pres- , open diamonds; and 2D2, closed circles) were titrated with PLV containing OR, and emission at 540 -546 nm was monitored as described under "Experimental Procedures." F 0 is the fluorescence intensity of the sample titrated with unlabeled PLV. F is the corrected fluorescence intensity of the sample titrated with PLV containing OR. The acceptor density was 2.7 ϫ 10 Ϫ4 OR molecules/Å 2 . Each point represents the value averaged from three separate determinations. Data were fitted to an equilibrium binding equation by nonlinear least squares regression as described under "Experimental Procedures," and dashed (without antibody) and solid lines (with antibody) were drawn.

TABLE 3 PLV binding parameters of WT and FVIII C2C2
PyMPO-labeled WT-FVIII or FVIII C2C2 at 25 nM in the absence or presence of 300 nM antibody was titrated with PLV containing OR, and emission at 540 -546 nm was monitored as described under "Experimental Procedures" and plotted in Fig. 4, A and B, as a function  inhibits FVIII binding to phospholipid (9,28,29). Our results also show a similar specificity of inhibition for the FVIII C2C2 variant. However, we note that the magnitude of PLV binding affinity reduction by ESH4 on FVIII C2C2 was lower as compared with WT FVIII. Furthermore, the F max value for FVIII C2C2 in the presence of ESH4 was significantly higher than that for WT FVIII. The reason(s) for these differential effects of ESH4 on WT FVIII and FVIII C2C2 is not clear, but may reflect the fact that FVIII C2C2 contains two binding sites for these antibodies. FIXa Binding Affinity-Overall, the impaired PLV binding due to the presence of ESH4 may only partially explain the reduction in FVIIIa activity inhibition by this antibody. Thus we further analyzed the effect of these antibodies on the FVIIIa-FIXa interaction. Functional affinity of WT FVIII and FVIII C2C2 for FIXa was measured by titrating (thrombin-activated) FVIIIa (0.5 nM) with the indicated concentration of FIXa and assessing activity by FXa generation assay. The reconstituted FXase activity values were plotted as a function of FIXa concentration, and results are shown in Fig. 5. FXase activity of WT FVIII and FVIII C2C2 increased to a saturable level as FIXa concentration was increased. From the fitted curves, K d values (Table 4) for FVIII C2C2 showed an 8.7-fold reduced affinity for FIXa as compared with WT FVIII. As shown in Fig. 5, GMA8011 markedly inhibited WT FVIIIa activity. This inhibition was not explained by impaired PLV binding because FXa generation reactions were run with excess (20 M) PLV in the assay. Interestingly, in the presence of GMA8011, FIXa affinity for WT FVIII was markedly reduced (10.4-fold, Table 4). Although ESH4 significantly inhibited PLV binding of WT FVIII, this antibody did not alter FIXa binding affinity. Both ESH4 and ESH8 modestly (ϳ4fold) inhibited FIXa binding of FVIII C2C2 . None of antibodies showed inhibitory effects on either FXase complex with substrate FX (data not shown).
Michaelis-Menten Kinetics and FVIIIa Inactivation by APC or FXa-The structural integrity of the FVIII C2C2 variant was assessed by additional functional experiments that included Michaelis-Menten kinetics and FVIIIa inactivation by both APC and FXa (Fig. 6). FXase activity with WT FVIII and FVIII C2C2 titrated with FX showed hyperbolic curves that were saturable (Fig. 6A). Estimated K m values showed essentially no differences for the WT FVIII and FVIII C2C2 (32.8 Ϯ 2.6 and 26.3 Ϯ 1.2 nM, respectively). Furthermore, both WT FVIII and FVIII C2C2 were inactivated by APC or FXa nearly linearly with activity reduced by ϳ50% in ϳ12 min (Fig. 6B). Estimated inactivation rates were similar for APC inactivation rates for WT FVIII and FVIII C2C2 (2.62 Ϯ 0.14 and 2.97 Ϯ 0.32 min Ϫ1 , respectively) and FXa inactivation rates for WT FVIII and  Initial velocity of FXa generation was measured as described under "Experimental Procedures" and plotted in Fig. 5

DISCUSSION
In a previous study, we generated an FVIII variant lacking the C2 domain (⌬C2-FVIII) (8) and showed that this variant retained high affinity for PLV, thus providing evidence for the significant contribution of the C1 domain in PLV binding. Attempts to produce an analogous FVIII variant lacking the C1 domain have been unsuccessful for reasons that are not fully understood. Therefore, to gain insights into the role of the C domains in a variant lacking C1, we generated an FVIII variant, FVIII C2C2 , where C1 is now replaced with a second C2 domain. This variant showed several functional defects as compared with the WT protein. The FVIII C2C2 variant retained low but appreciable cofactor activity. Similar to ⌬C2-FVIII, the thermal stability of FVIII C2C2 was dramatically reduced as compared with WT. Furthermore, whereas PLV binding affinity of FVIII C2C2 was modestly lower (2.8-fold) than WT FVIII, the FIXa binding affinity was markedly reduced (8.7-fold).
FVIII C1 and C2 domains show 66.2% sequence homology (39.7% identity). A portion of these sequences is shown in Fig. 6. Phospholipid-binding sites have been identified in both domains (4 -9). Because FVIII C2C2 contains a duplicated C2 domain, it was not surprising that the magnitude of the reduction observed in PLV binding affinity as compared with WT FVIII (Ͻ3-fold) was less than that of ⌬C2-FVIII (ϳ14-fold) (8). However, FVIII C2C2 had only half of the cofactor activity as compared with ⌬C2-FVIII as measured in an FXa generation assay where PLV is saturating. This reduced activity likely derives in part from an altered interaction with FIXa, as well as reduced protein stability as a result of weakened interdomain interactions.
A number of studies indicate that FVIII stability as measured by thermal decay experiments depends heavily on the strength of FVIII interdomain interactions at A1-A2 or A2-A3 (30 -32) and A1-C2 (33) interfaces. In addition, we recently reported that noncovalent interaction at the intrasubunit interface formed by A3 and C1 but not at the interface formed by C1 and C2 contributed to FVIII stability (34). In the FVIII C2C2 variant, the interdomain interaction of A1 and C2 is maintained, however, the interaction originally between A3 and C1 is now  Factor VIII with C1 Replaced by C2 OCTOBER 25, 2013 • VOLUME 288 • NUMBER 43 between A3 and C2. Inasmuch as mutations that strengthen or weaken the A3-C1 interaction resulted in increased or decreased FVIII stability, respectively (34), we speculate that the non-native A3-C2 domain interaction in the variant is weaker than the native A3-C1 interaction, and this makes a significant contribution to the observed lability and potentially reduced specific activity of the FVIII C2C2 variant. This contention is supported by comparison of the A3 domain-interactive sequence in the C1 domain, Thr 2114 -Thr 2220 , corresponding to C2 domain residues Leu 2273 -Asn 2277 in FVIII C2C2 , which shows a low level 28.6% homology (14.3% identity, see Fig. 7).
In a recent study using fluorescence resonance energy transfer, we estimated distance values between multiple sites in FVIIIa and the phospholipid membrane surface (23). Results from these distance calculations indicated that the molecular orientation of FVIIIa bound the phospholipid membrane with a tilt angle of 30 -50°rather than standing upright. This orientation combined with results from mutagenesis of selected A3 domain residues (Arg 1719 and Arg 1721 ) provided evidence that the A3 domain also interacted with the phospholipid membrane. The A3 domain possesses a prominent FIXa-interactive site (35), and a high affinity interaction has been observed for the A3C1C2 subunit of FVIIIa (35,36). The observation that the FVIIIa C2C2 variant showed a 10-fold reduced affinity for FIXa suggests a contribution of the C1 domain to this interaction.
Using the x-ray crystal structure of FVIII to model the FVIII-FIXa complex on PLV (11) suggests that the FVIII C1 domain resides on the membrane adjacent to FIXa. This result is consistent with the effects we observed with the anti-C1 antibody, GMA8011. Although this antibody did not inhibit the WT FVIII interaction with PLV, the antibody did inhibit membrane-dependent FVIIIa cofactor activity without affecting the K m for FX. Thus these results suggest that the antibody blocked the FVIIIa-FIXa interaction. In the case of FVIII C2C2 , the first C2 domain that replaces C1 would be adjacent to FIXa. This positioning explains the capacity of the two anti-C2 antibodies to reduce the affinity for FIXa of the variant FVIIIa while showing no effect on the affinity of FIXa for WT FVIIIa.
Recently, Albert et al. (37) reported that the ESH8 epitope was restricted to within a relatively short sequence (Ser 2265 -Val 2280 ) in the C2 domain. Based upon sequence alignments with C1 (Fig. 7), this region corresponds to C1 residues Ser 2106 -Leu 2123 , and these residues are thought to be in close proximity to an FIXa-interactive site based upon the FVIII-FIXa binding model (11). Interestingly, this C1 sequence region also contains an A3-interactive region (Thr 2114 -Thr 2120 ). Therefore, the observed ESH8-dependent inhibition of cofactor activity may be combined with an inhibition of C2-A3 interaction in the variant. Furthermore, although the epitope for ESH4 is less well defined and maps to within C2 domain residues Thr 2303 -Tyr 2332 (29), the corresponding region in C1 also appears to be in close proximity to FIXa in the modeled structure.
Results from the current study cannot distinguish whether the altered interaction of the FVIII C2C2 variant with FIXa resulted from a direct interaction with FIXa following replacement of C1 with C2 or whether the reduced affinity was due to indirect effects caused by altered interdomain interactions between the A3-C domains. However, results from this study indicate that the FVIII C1 domain is likely located near FIXa in the FXase complex. Further studies are required to determine the existence of direct interaction between the FVIII C1 domain and FIXa.