The Sequence Glu 1811 –Lys 1818 of Human Blood Coagulation Factor VIII Comprises a Binding Site for Activated Factor IX*

In previous studies we have shown that the interac- tion between factors IXa and VIII involves the light chain of factor VIII and that this interaction is inhibited by the monoclonal antibody CLB-CAg A against the factor VIII region Gln 1778 –Asp 1840 (Lenting, P. J., Donath, M. J. S. H., van Mourik, J. A., and Mertens, K. (1994) J. Biol. Chem. 269, 7150–7155). Employing distinct recombinant factor VIII fragments, we now have localized the epitope of this antibody more precisely between the A3 domain residues Glu 1801 and Met 1823 . Hydropathy analysis indicated that this region is part of a major hydrophilic exosite within the A3 domain. The interaction of factor IXa with this exosite was studied by em- ploying overlapping synthetic peptides encompassing the factor VIII region Tyr 1786 –Ala 1834 . Factor IXa binding was found to be particularly efficient to peptides corre- sponding to the factor VIII sequences Lys 1804 –Lys 1818 and Glu 1811 –Gln 1820 . The same peptides proved effective in binding antibody CLB-CAg A. Further analysis re- vealed that peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 interfere with binding of factor IXa to immobilized fac- tor VIII light chain ( K i (cid:39) 0.2 m M and 0.3 m M , respective-ly). Moreover, these peptides inhibit factor X activation by factor IXa in the presence of factor VIIIa ( K i

Human blood coagulation factor VIII (FVIII) 1 is an essential protein of the hemostatic system, which is evident from the severe bleeding disorder hemophilia A that is associated with FVIII deficiency or dysfunction (1). FVIII is synthesized as a single chain polypeptide containing a number of discrete do-mains arranged in the sequence A1-A2-B-A3-C1-C2 (2,3). Examination of its primary structure reveals that FVIII shares considerable homology with the plasma proteins factor V (FV) and ceruloplasmin (4 -6). Whereas ceruloplasmin comprises a triple A domain structure (A1-A2-A3), FV displays the same domain structure (7,8). In contrast to FV, FVIII predominantly circulates as a heterodimeric protein, consisting of a Me 2ϩlinked light and heavy chain (9 -11). The heavy chain contains the A1-A2-B domains and is heterogeneous (M r 90,000 -200,000) due to limited proteolysis at a number of positions within the B domain. The light chain of FVIII (M r 80,000) comprises the domains A3-C1-C2 (10,12).
In the intrinsic pathway of blood coagulation, FVIII functions as a nonenzymatic cofactor in the factor X (FX)-activating complex (13). Within this complex, the serine protease factor IXa␤ (FIXa) activates FX in the presence of calcium ions, phospholipids, and activated FVIII. In order to play its role in the generation of FXa, FVIII has to be activated (14,15). Activation is achieved by limited proteolysis in both the FVIII heavy and light chain by FXa or thrombin (12), which results in the formation of a heterotrimeric product, FVIIIa (16,17). The relatively labile FVIIIa heterotrimer is known to be stabilized by the enzyme FIXa in the presence of phospholipids (18). In addition, it has been reported that the phospholipid-FIXa complex enhances the reassociation of isolated FVIIIa subunits into the FVIIIa heterotrimer (19), indicating that FVIIIa is capable of directly interacting with FIXa.
Several studies have been performed in order to characterize the assembly of the FIX⅐FVIII complex in more detail (19 -22). The FVIII heavy chain regions Ser 558 -Gln 565 and Arg 698 -Ser 710 have been recognized to represent FIXa interactive sites (22,23). Previously, we have shown that FVIII light chain comprises an exosite that binds FIXa with high affinity (21). In the same study, we found that the FIXa-FVIII light chain interaction was inhibited by the anti-FVIII antibody CLB-CAg A, which is known to bind to the FVIII A3 domain region Gln 1778 -Asp 1840 (24). In the present study, we addressed the possibility that this region is involved in the assembly of the enzyme-cofactor complex. Therefore, we first located the binding site of antibody CLB-CAg A in more detail. Subsequently, a series of synthetic peptides was employed in order to define the FVIII region involved in FIXa binding. This approach allowed us to identify the FVIII light chain region Glu 1811 -Lys 1818 as being involved in FIXa binding and in the assembly of the FX-activating FIXa⅐FVIIIa complex.

Materials-Protein
A-Sepharose CL-4B was from Pharmacia Biotech Inc. Microtiter plates (Immulon) were from Dynatech (Plockingen, Germany). The in vitro transcription and translation kits employing the SP6-expression system as well as the plasmid pSP64 were from Promega. Restriction enzymes were obtained from Life Technologies, Inc. Goat anti-mouse antibodies, rabbit anti-mouse antibodies, and human serum albumin (HSA) were from the Central Laboratory of the Netherlands Red Cross Blood Transfusion Service (Amsterdam, The Netherlands).
Antibodies and Other Proteins-The monoclonal anti-FVIII antibodies CLB-CAg 12, CLB-CAg 69, and CLB-CAg A have been described previously (24,25). The murine anti-FIX antibody CLB-FIX 14 (isotype IgG 1 ) was obtained as outlined previously, employing a screening strategy based on binding to immobilized FIX in the absence of calcium ions (21). Monoclonal antibodies were purified from culture medium employing Protein A-Sepharose CL-4B as recommended by the manufacturer. Polyclonal antibodies against human FIX were obtained as described (21). Antibodies were conjugated with horseradish peroxidase as described (26). Human FVIII light chain was purified as outlined previously (21). Human FIXa was prepared from immunopurified FIX as described (27).
Construction of Recombinant FVIII Fragments-Plasmid pSP/F8 -80K1 (24) was used as template for the construction of truncated FVIII fragments employing the polymerase chain reaction. The DNA fragments were made by using the sense primer 5Ј-ACG ATT TAG GTG ACA CTA TAG-3Ј (containing a part of the SP-6 promotor) in combination with the respective antisense primers 5Ј-TTA GGA TCC TCA CAT ATG ATG TTG CAC TTT-3Ј, 5Ј-TTA GGA TCC TCA TTC ATT AGG CTT GAC AAA-3Ј, 5Ј-TTA GGA TCC TCA TTC TGC TCC TTG CCT CTG-3Ј, and 5Ј-TTA GGA TCC TCA AGA AAT AAG GCT AGA ATA-3Ј. Polymerase chain reaction amplifications of these primer combinations yielded the FVIII DNA sequences between base pairs 4893 and 5676, 5640, 5610, and 5580, respectively. The antisense primers enclosed a BamHI restriction site (underlined) and a stop-codon (boldface type). The sense primer was designed to contain a HindIII restriction site at the 5Ј terminus after polymerase chain reaction. After digestion of the polymerase chain reaction products and plasmid pSP64 with HindIII and BamHI, the products were ligated into the plasmid. The identities of the constructs were verified by restriction mapping and dideoxy termination sequencing. The constructs encode for the FVIII fragments Asp 1562 -Met 1823 , Asp 1562 -Glu 1811 , Asp 1562 -Glu 1801 , and Asp 1562 -Ser 1791 .
Immunoprecipitation Studies-The FVIII fragments were in vitro transcribed and translated employing the SP6 expression system as described previously (24). Translation was performed in the presence of [ 35 S]methionine to obtain radiolabeled FVIII fragments, which were analyzed by 12% (w/v) SDS-polyacrylamide gel electrophoresis and subsequent autoradiography. All constructs appeared to encode for single polypeptides (not shown). The fragments were diluted 100-fold in 25 mM Tris (pH 8.0), 5 mM NaCl, 2 mM EDTA, 0.5% (w/v) dideoxycholate, 0.5% (v/v) Triton X-100 (immunoprecipitation buffer). The fragments were then incubated with monoclonal antibody CLB-CAg A or CLB-CAg 69 (5 g/ml) for 1 h at room temperature. Subsequently, rabbit anti-mouse antibodies immobilized onto Protein A-Sepharose were added, and the incubation was continued for another hour at room temperature. After washing twice with immunoprecipitation buffer and once with 0.1 M NaCl, 0.1% (v/v) Triton X-100, 50 mM Tris (pH 8.0), the beads were boiled for 5 min in 8% (w/v) SDS, 40% (v/v) glycerol, 0.04% (w/v) bromphenol blue, 20 mM dithiothreitol, 0.25 M Tris (pH 6.8). Finally, the supernatants were subjected to 12% (w/v) SDS-polyacrylamide gel electrophoresis, and polypeptides were visualized by autoradiography.
Binding Assays-Synthetic peptides were immobilized (0.8 nmol/well added) to microtiter wells in a volume of 100 l, and remaining binding sites were blocked with 2% (w/v) HSA in 0.1% (v/v) Tween 20, 0.1 M NaCl, 25 mM Tris (pH 7.4). After washing, the immobilized peptides were incubated with antibody CLB-CAg A (625 nM) or FIXa (50 nM) in the same buffer for 1 h at 37°C in a 100-l volume. After washing, bound antibody CLB-CAg A was probed by incubating with peroxidase-conjugated goat anti-mouse antibodies for 15 min at room temperature and detected by peroxidase hydrolysis of the substrate 3,3Ј,5,5Ј-tetramethylbenzidine (Sigma). Bound FIXa was probed by employing the peroxidase-conjugated anti-FIX antibody CLB-FIX 14 (5 g/ml). Binding of FIXa to immobilized FVIII light chain and calculation of binding parameters were performed as described (21). The affinities of FIX and FIXa for peptides in solution were determined employing a previously described method (30). Briefly, FIX or FIXa (30 nM) was incubated with various concentrations of peptide (0.2-0.6 mM) in a buffer containing 2% (w/v) HSA, 0.1% (v/v) Tween 20, 0.1 M NaCl, 25 mM Tris (pH 7.4). The mixtures were incubated for 16 h at room temperature in order to reach equilibrium. Samples were subsequently incubated with immobilized peptide to allow noncomplexed FIX or FIXa to bind to the immobilized peptide. FIX or FIXa bound to the immobilized peptide was quantified employing the peroxidase-conjugated antibody CLB-FIX 14. The dissociation constants for the interaction with peptides in solution could be calculated as described (30).
FX Activation-FXa formation was determined as described (31). FX (0.2 M) was activated in 3 mM CaCl 2 , 0.1 M NaCl, 0.2 mg/ml HSA, 0.05 M Tris (pH 7.4) at 37°C by FIXa (0.7 nM) in the presence of phospholipids (0.1 mM) and FVIIIa (0.4 nM). FVIII was preactivated in the same buffer for 5 min by thrombin prior to the addition. FX activation experiments in the absence of FVIIIa were performed employing a FIXa concentration of 30 nM. FXa formation was quantified employing the chromogenic substrate S-2222 (Chromogenix AB, Mölndal, Sweden). An active site titrated FXa preparation was used as a reference to convert absorbance values into molar FXa concentrations. (21). Since this antibody is known to bind to the FVIII region Gln 1778 -Asp 1840 (24), the same region may be involved in the interaction with FIXa. To address this possibility, we first located the binding site for this antibody in more detail. Constructs comprising DNA sequences encoding for the FVIII fragments Asp 1562 -Met 1823 and carboxyl-terminal truncations thereof ( Fig. 1) were in vitro transcribed and translated, the latter of which in the presence of [ 35 S]methionine. As determined by SDS-polyacrylamide gel electrophoresis, each radiolabeled polypeptide migrated as a single band with the expected M r between 32,000 and 36,000 (not shown). These polypeptides were examined for their binding to antibodies CLB-CAg A and CLB-CAg 69 in immunoprecipitation studies. In these studies, antibody CLB-CAg 69 served as a control that should bind to all four polypeptides, as its epitope is known to encompass residues Lys 1673 -Arg 1689 (24). Indeed, the antibody effectively bound the various polypeptides to the same extent ( Fig. 1, lanes 1-4). With respect to antibody CLB-CAg A, the largest polypeptide (Asp 1562 -Met 1823 ) was readily recognized by this antibody, whereas only a faint band was observed for the polypeptide Asp 1562 -Glu 1811 (Fig. 1, lanes 7 and 8). In contrast, the two smaller polypeptides comprising the sequences Asp 1562 -Glu 1801 and Asp 1562 -Ser 1791 did not bind to antibody CLB-CAg A (Fig. 1, lanes 5 and 6). From these results it appears that residues between Glu 1801 and Met 1823 are of particular importance for binding of antibody CLB-CAg A to the FVIII polypeptide Asp 1562 -Met 1823 .

Interaction between Recombinant FVIII Fragments and Antibody CLB-CAg A-Binding of FIXa to FVIII light chain is inhibited by the monoclonal anti-FVIII antibody CLB-CAg A
Interaction between Synthetic Peptides and Antibody CLB-CAg A or FIXa-A hydropathy analysis of the FVIII A3 domain primary structure was performed in order to determine the hydrophilicity of the region Glu 1801 -Met 1823 . As can be seen in Fig. 2, this region is part of a markedly hydrophilic exosite encompassing the residues Arg 1781 -Asp 1842 , which indicates that the region Glu 1801 -Met 1823 may be exposed at the exterior of the FVIII light chain molecule. The possibility that the hydrophilic exosite comprises a FIXa binding region was addressed by employing a series of overlapping peptides that encompass the FVIII region between Tyr 1789 and Ala 1834 (Fig.  2). The interaction between these peptides and FIXa was as-sessed in binding studies employing immobilized peptides. FIXa displayed particular effective binding to peptide Lys 1804 -Lys 1818 and, to a lesser extent, to peptide Glu 1811 -Gln 1820 (Fig. 3). The same series of peptides were used to examine antibody CLB-CAg A binding (Fig. 3). Antibody CLB-CAg A displayed a similar pattern of specificity for these peptides as FIXa (Fig. 3), as most effective binding was observed for peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 . Collectively, these data indicate that the A3 domain comprises a hydrophilic exosite that contributes to FIXa binding.

Effect of Synthetic Peptides on FIXa Binding to FVIII Light
Chain-In order to investigate the interaction between FIXa and various synthetic peptides in solution, the effect of synthetic peptides on binding of FIXa to immobilized FVIII light chain was determined. In these experiments not all of the peptides of Fig. 2 could be evaluated, as the solubility of peptides Tyr 1786 -Glu 1801 , Tyr 1815 -Ala 1834 , and His 1822 -Ala 1834 was limited to concentrations that are below the concentrations required for competition studies in solution. As shown in Fig. 4A, the peptide comprising the sequence Gly 1799 -Lys 1813 was incapable of interfering with binding of FIXa to immobilized FVIII light chain. In contrast, peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 were found to inhibit binding of FIXa to immobilized FVIII light chain in a dose-dependent manner (Fig. 4A). By analyzing these data in a model for competitive inhibition (21), the inhibition constants were calculated to be 0.19 Ϯ 0.01 mM (mean Ϯ S.D.) and 0.27 Ϯ 0.02 mM for peptide Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 , respectively. Thus, both peptides effectively compete with binding of FVIII light chain to FIXa. Therefore, these peptides should also interfere with FVIII-dependent activation of FX by FIXa.
Effect of Synthetic Peptides on FX Activation-The effect of peptides Glu 1811 -Gln 1820 , Lys 1804 -Lys 1818 , and Gly 1799 -Lys 1813 on FX activation by FIXa in the presence of phospholipids, calcium ions, and FVIII was determined. In the presence of peptide Gly 1799 -Lys 1813 little, if any, inhibition was observed (Fig. 4B). In contrast, FX activation was effectively inhibited in the presence of peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 (Fig. 4B). Neither of the peptides inhibited FX activation in the absence of FVIII (not shown). Thus, peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 indeed interfere with a FVIII-dependent step within the process of FX activation. The mechanism of inhibition was addressed by studying the effect of the inhibitory peptides on the kinetic parameters K m and V max . In experiments employing peptide Lys 1804 -Lys 1818 , the apparent K m remained unchanged, whereas the apparent V max was dependent on the concentration of peptide (Fig. 5). Thus, peptide Lys 1804 -Lys 1818 inhibits FVIII-dependent FX activation by FIXa by a noncompetitive mechanism. The same results were obtained employing the peptide Glu 1811 -Gln 1820 (not shown). Using the normal model of noncompetitive inhibition, the inhibition constant was found to be 0.23 Ϯ 0.05 mM and 0.25 Ϯ 0.02 mM for peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 , respectively. These values are similar to the inhibition constant found for the inhibition of the FIXa-FVIII light chain interaction (see Fig. 4A). Apparently, both peptide Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 inhibit FVIII-dependent FX activation by FIXa by interfering in the association between FIXa and FVIII light chain.
Interaction of FVIII Light Chain or Synthetic Peptides with Uncleaved and Cleaved Forms of FIX-The FIX zymogen is dissimilar to the fully activated FIXa in that it does not activate FX (13,27). To examine whether FIX and FIXa also differ in cofactor binding, the interaction with FVIII light chain and, more specifically, with peptide Lys 1804 -Lys 1818 was addressed. In line with our previous finding (27), FIX was less efficient than FIXa in binding FVIII light chain (Fig. 6). In contrast, when FIX and FIXa were compared for binding to the immobilized peptide Lys 1804 -Lys 1818 , this peptide bound the FIX zymogen and the FIXa enzyme to the same extent (Fig. 6, inset). The same observation was made employing peptide Glu 1811 -Gln 1820 (not shown). However, studies employing immobilized peptides do not provide quantitative binding parameters. Therefore, the interaction between peptide Lys 1804 -Lys 1818 and FIX or FIXa was further studied in solution under equilibrium conditions (see "Experimental Procedures"). These experiments revealed an apparent dissociation constant of 0.20 Ϯ 0.02 mM (mean Ϯ S.D.) for the peptide-FIXa interaction. This value is similar to the inhibition constants found for the inhibition of the FIXa-FVIII light chain interaction and FVIII-dependent activation of FX (see Fig. 4). The same value (0.23 Ϯ 0.02 mM) was also found for the peptide-FIX interaction, demonstrating that indeed peptide Lys 1804 -Lys 1818 is equally effective in binding FIX or FIXa. Thus, peptide Lys 1804 -Lys 1818 does not distinguish between uncleaved and cleaved forms of FIX, whereas FVIII light chain does. This finding was confirmed by using FVIII light chain as competitor for the interaction between immobilized peptide Lys 1804 -Lys 1818 and FIX or FIXa. Binding of FIXa to the immobilized peptide was readily inhibited in the presence of FVIII light chain (Fig. 7). In contrast, FVIII light chain proved to be inefficient in interfering with binding of FIX to the immobilized peptide. These data demonstrate that peptide Lys 1804 -Lys 1818 and FVIII light chain compete for binding to FIXa but not to the FIX zymogen.

DISCUSSION
During the process of FX activation, the enzyme FIXa assembles with the nonenzymatic cofactor FVIIIa into a lipidbound complex. In previous studies we have shown that FVIII light chain contains a site that binds FIXa with high affinity (K d Ϸ 15 nM) and that FIXa binding is inhibited by the FVIII light chain-directed antibody CLB-CAg A (21). Here we show that this antibody is directed against an extensive hydrophilic exosite within the A3 domain (Figs. 2 and 3). Because such hydrophilic regions are likely to be exposed at the exterior of the protein (32), we addressed the possibility that this exosite comprises a FIXa binding site. Indeed, FIXa binds to synthetic peptides that consist of FVIII sequences that are part of the hydrophilic exosite Arg 1781 -Asp 1842 (Fig. 3). Competition studies demonstrated that peptides corresponding to the exosite regions Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 effectively inhibit binding of FIXa to immobilized FVIII light chain (Fig. 4A). The same peptides also interfere with FVIII-dependent activation of FX by FIXa (Fig. 4B). Inhibition of FX activation is noncompetitive (Fig. 5), which strongly suggests that the peptides inhibit the enzyme FIXa by binding at a site distinct from the substrate binding pocket. Collectively, our data demonstrate that peptides consisting of the FVIII amino acid residues Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 represent a FIXa binding site. It is of importance to note that the K i for the binary FIXa-FVIII light chain interaction is similar to the K i found for FX activation in the complete FX activating complex, thus including the entire FVIIIa heterotrimer (Figs. 4 and 5). Assembly of the functional FIXa⅐FVIIIa complex apparently is directly related to binding of FIXa to the FVIII A3 domain exosite. In this respect it should be mentioned that FIXa binding is not an exclusive property of the FVIII A3 domain. FIXa recognition sites have been identified within the FVIII A2 domain regions Ser 558 -Gln 565 (22) and Arg 698 -Ser 710 (23). As synthetic peptides corresponding to these A2 domain regions also interfere with FVIIIa cofactor function, it seems reasonable to assume that both FVIII heavy chain and light chain regions participate in FIXa-FVIIIa complex formation.
As peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 proved more efficient in their interaction with FIXa than the other peptides tested (Figs. 3 and 4), we propose that the minimal requirements for FIXa binding are met by the overlapping residues Glu 1811 -Lys 1818 . This region, including its direct environment (residues Gly 1799 -Gln 1820 ), is strikingly rich in basic Lys residues, which are located at positions 1804, 1808, 1813, and 1818 (Fig. 8). These Lys residues appear to be unique for the FVIII A3 domain, as they are not only lacking in the FVIII A1-and A2 domains but also in the A3 domains of the structurally related proteins FV and ceruloplasmin (Fig. 8). The same Lys residues are conserved in the FVIII A3 domain of a rodent species (Fig. 8), which would be compatible with the involvement of these residues in a FVIII A3 domain-specific event such as FIXa binding. However, peptide Gly 1799 -Lys 1813 with Lys residues at 1804, 1808, and 1813 proved considerably less efficient in its interaction with FIXa than peptide Glu 1811 -Gln 1820 with Lys residues at 1813 and 1818 (Figs. 3 and 4). Apparently, the presence of the Lys residues alone is not sufficient for FIXa binding. It should be mentioned that peptide Glu 1811 -Gln 1820 contains a triplet of aromatic residues (Tyr 1815 -Phe 1816 -Trp 1817 ), which is also present in the inhibitory peptide Lys 1804 -Lys 1818 but lacking in the noninhibitory peptide Gly 1799 -Lys 1813 . Because part of this sequence is conserved in other A domains (Fig. 8), it is unclear how these residues may be involved in a FVIII A3 domain-specific function. To what extent individual amino acids in the FIXa binding region contribute to FIXa⅐FVIII light chain complex formation, therefore, remains to be investigated. It seems of interest to note that mutations at positions Ser 1784 , Leu 1789 , Met 1823 , Pro 1825 , and Thr 1826 have been determined to be associated with moderately severe hemophilia A (33). As these mutations are in close proximity to the FIXa binding region, it is tempting to speculate that the bleeding tendency that is associated with these mutations is due to a suboptimal assembly of the FIXa⅐FVIII light chain complex.
Recently, we demonstrated that uncleaved FVIII light chain is similar to FVIII light chain derivatives that have been cleaved by the activators thrombin or FXa in that they display similar affinity for FIXa (31). Apparently, the FIXa recognition site is fully exposed in the intact FVIII light chain. In agreement with previous observations, we have found that FVIII light chain is more efficient in binding the fully activated FIXa than the uncleaved FIX zymogen (Figs. 6 and 7; Ref. 27). These Various concentrations of Glu-Gly-Arg chloromethyl ketone-treated FIXa (q) or FIX (E) were incubated with immobilized FVIII light chain (0.7 pmol/well) as described under "Experimental Procedures." Association between FVIII light chain and FIX or FIXa was assessed as described (21). Inset, various concentrations of Glu-Gly-Arg chloromethyl ketone-treated FIXa (q) or FIX (E) were incubated with immobilized peptide Lys 1804 -Lys 1818 (0.8 nmol/well added) as described under "Experimental Procedures." Binding was detected employing the peroxidase-labeled anti-FIX antibody CLB-FIX 14. Absorbance was measured at 450 nm using 540 nm as reference. data indicate that FVIII light chain displays preferential binding to the enzyme FIXa rather than to the nonactivated FIX zymogen. In this regard, FVIII light chain seems similar to FVa and thrombomodulin, because these cofactors are more efficient in binding to their respective enzymes than to the uncleaved proenzymes (34,35). Surprisingly, this seems to be untrue for the FIXa-binding peptides Lys 1804 -Lys 1818 and Glu 1811 -Gln 1820 , because these peptides do not distinguish between the enzyme FIXa and the FIX zymogen (Figs. 6 and 7). Several possibilities may be considered that may explain these observations. First, it is possible that the relative size of the FVIII light chain prevents binding to the intact FIX zymogen, while this restriction is overcome by limited proteolysis of the zymogen at its activation sites Arg 145 or Arg 180 (27). It should be noted here that cleavage at Arg 145 is sufficient for full exposure of the FVIII light chain binding site, whereas cleavage at Arg 180 results in a suboptimal exposure (27). Alternatively, the conformation of the FIXa-binding motif in synthetic peptides may differ from its conformation in the complete FVIII light chain. This may be due to other portions of the light chain that provide the region Glu 1811 -Lys 1818 its specificity for binding to the activated form of FIX. In this respect it is of importance to note that the Asn residue at position 1810 is a potential site for N-linked glycosylation in FVIII (2,3). As this site is located adjacent to the FIX-binding motif Glu 1811 -Lys 1818 , it seems conceivable that glycosylation of this site contributes to the specificity for binding of FIXa to its binding sequence Glu 1811 -Lys 1818 . FIG. 8. Comparison of the FVIII A3 domain region Gly 1799 -Gln 1820 with corresponding portions of other A domains. The primary structure of the A3 domain of human FVIII (2, 3) is compared with the corresponding regions of the A3 domains of murine FVIII (mFVIII) (36), human FV (hFV) (4,5), and ceruloplasmin (hCer.) (6) and with both A1 and A2 domains of human FVIII (hFVIII) (2,3). The sequences are aligned as described (36,37). Amino acids that are identical to the corresponding residues in the FVIII A3 domain are boxed.