The Light Chain of Factor VIII Comprises a Binding Site for Low Density Lipoprotein Receptor-related Protein*

In the present study, the interaction between the endocytic receptor low density lipoprotein receptor-related protein (LRP) and coagulation factor VIII (FVIII) was investigated. Using purified components, FVIII was found to bind to LRP in a reversible and dose-dependent manner (K d ≈ 60 nm). The interaction appeared to be specific because the LRP antagonist receptor-associated protein readily inhibited binding of FVIII to LRP (IC50 ≈ 1 nm). In addition, a 12-fold molar excess of the physiological carrier of FVIII,i.e. von Willebrand factor (vWF), reduced the binding of FVIII to LRP by over 90%. Cellular degradation of125I-labeled FVIII by LRP-expressing cells (≈ 8 fmol/105 cells after a 4.5-h incubation) was reduced by approximately 70% in the presence of receptor-associated protein. LRP-directed antibodies inhibited degradation to a similar extent, indicating that LRP indeed contributes to binding and transport of FVIII to the intracellular degradation pathway. Degradation of FVIII was completely inhibited by vWF. Because vWF binding by FVIII involves its light chain, LRP binding to this subunit was studied. In ligand blotting experiments, binding of FVIII light chain to LRP could be visualized. More detailed analysis revealed that FVIII light chain interacts with LRP with moderate affinity (k on≈ 5 × 104 m −1s−1; k off ≈ 2.5 × 10−3 s−1; K d ≈ 50 nm). Furthermore, experiments using recombinant FVIII C2 domain showed that this domain contributes to the interaction with LRP. In contrast, no association of FVIII heavy chain to LRP could be detected under the same experimental conditions. Collectively, our data demonstrate that in vitro LRP is able to bind FVIII at the cell surface and to mediate its transport to the intracellular degradation pathway. FVIII-LRP interaction involves the FVIII light chain, and FVIII-vWF complex formation plays a regulatory role in LRP binding. Our findings may explain the beneficial effect of vWF on thein vivo survival of FVIII.

Low density lipoprotein receptor-related protein (LRP), 1 also known as ␣2-macroglobulin receptor, is a member of the low density lipoprotein receptor family of endocytic receptors (for a review, see Refs. 1 and 2). It consists of a heavy chain and a light chain, which are associated in a noncovalent manner. The 85-kDa light chain comprises the transmembrane and cytoplasmic domains, whereas the ligand binding regions are located within the 515-kDa heavy chain (3). LRP is abundantly present in various tissues such as the liver, placenta, lung, and brain (4) and is expressed in an array of cell types: parenchymal cells, neurons and astrocytes, Leydig cells, smooth muscle cells, monocytes, and fibroblasts (4). Commonly used cell lines such as monkey kidney COS cells and Chinese hamster ovary (CHO) cells also express LRP (5,6). The function of LRP is to mediate the binding and transport of ligands from the cell surface to the endosomal degradation pathway (1,2). Binding and internalization of ligands is antagonized by a 39-kDa chaperone protein designated receptor-associated protein (RAP) (7,8). Currently, a wide spectrum of structurally and functionally unrelated ligands involved in a variety of processes such as lipoprotein metabolism, cell growth and migration, and neuronal regeneration (1, 2) has been identified. Furthermore, LRP seems to be linked to the process of blood coagulation. This is apparent from the observations that LRP recognizes thrombin/ antithrombin and factor Xa/␣2-macroglobulin complexes and the Kunitz-type inhibitor tissue factor pathway inhibitor (9 -11). In addition, LRP contributes to down-regulation of tissue factor expression at the surface of monocytes (12).
Coagulation factor VIII (FVIII) is the precursor of its activated derivative, which stimulates factor IXa-mediated activation of factor X (for recent reviews, see Refs. 13 and 14). The fact that deficiency or dysfunction of FVIII is associated with severe bleeding tendencies demonstrates that this cofactor is indispensable for appropriate hemostasis. FVIII comprises a domain structure (A1-A2-B-A3-C1-C2) (15) and circulates in plasma predominantly as a heterodimeric protein consisting of a metal ion-linked light and heavy chain (16,17). The heavy chain (90 -220 kDa) contains the A1-A2-B domains and is heterogenous as a result of limited proteolysis within the B domain. The light chain (80 kDa) consists of the A3-C1-C2 domains. The amino-and carboxyl-terminal ends of FVIII light chain together comprise the binding site for von Willebrand factor (vWF) (18,19), the physiological carrier protein of FVIII (20). The FVIII precursor is converted into its activated derivative upon limited proteolysis by thrombin (21,22). Activated FVIII consists of the A2 domain, which is noncovalently associated with the metal-ion linked A1/A3-C1-C2 dimer, whereas the B domain and the amino-terminal part of the A3 domain have been removed (23). Due to the release of the aminoterminal part, high affinity binding to vWF is lost (18). vWF prevents the FVIII precursor from binding to components of the factor X-activating complex (14,24,25). Furthermore, the halflife of FVIII is considerably reduced in the absence of vWF (26,27), indicating that vWF prevents FVIII from premature clearance. However, the mechanism by which FVIII is removed from the circulation has remained unidentified.
In the present study, we investigated the possibility that FVIII is recognized by the multifunctional receptor LRP. To this end, the interaction between LRP and FVIII or its constituent subunits has been addressed using purified components. In addition, cellular degradation of FVIII has been studied. It is demonstrated that LRP recognizes FVIII as a ligand, and that binding involves the light chain of FVIII. Furthermore, both LRP-mediated binding and degradation of FVIII are downregulated by vWF. Our data are in support of a mechanism in which LRP contributes to binding and internalization of FVIII.
Proteins-FVIII light chain, thrombin-cleaved FVIII light chain, and FVIII heavy chain were prepared as described previously (25,28). The integrity of the isolated subunits was assessed in reconstitution experiments as described previously (28). As expected, isolated subunits could effectively be reassembled into biologically active heterodimers (data not shown). Plasma-derived FVIII heterodimer was isolated as described previously (25). FVIII was labeled with Na 125 I (Amersham Pharmacia Biotech) using the IODO-GEN method (Pierce) as described previously (29), except that FVIII was stored in 150 mM NaCl, 2 mM CaCl 2 , 0.005% (v/v) Tween 20, and 20 mM Hepes (pH 7.4) at Ϫ20°C. Specific radioactivity was 4.2 (Ϯ 0.7) ϫ 10 3 cpm/fmol FVIII (mean Ϯ range; n ϭ 2). Recombinant FVIII C2 domain (residues 2172-2332) was obtained using the baculovirus expression system as described previously (30) and purified by immunoaffinity chromatography as described for FVIII light chain (25). Purified recombinant wild-type vWF was kindly provided by Prof. H. P. Schwarz (Baxter-Immuno, Vienna, Austria). Purified LRP (31) and serum containing polyclonal antibodies against LRP were a generous gift from Dr. S. K. Moestrup (University of Aarhus, Aarhus, Denmark). Total IgG was purified from the serum using protein A-Sepharose CL-4B as recommended by the manufacturer. A plasmid encoding glutathione S-transferase fused to RAP (GST-RAP) was kindly provided by Dr. J. Kuiper (Leiden University, Leiden, The Netherlands) and used for expression of GST-RAP in Escherichia coli DH5␣ as described previously (7). GST-RAP was purified using glutathione-Sepharose 4B as recommended by the manufacturer. Because the GST tag does not interfere with the binding properties of RAP (7), GST-RAP was used throughout the present study. Purified anti-FVIII antibodies CLB-CAg 69, CLB-CAg 117, and CLB-CAg A have been described previously (25). Anti-FVIII antibodies ESH4 (interferes with vWF binding) and ESH8 (promotes vWF binding to thrombin-cleaved FVIII) were obtained from American Diagnostica. Human albumin was from the division of products of CLB (Amsterdam, The Netherlands). Bovine albumin (fraction V) was from Merck.
Protein Concentrations-Protein was quantified by the method of Bradford (32), using human albumin as a standard. FVIII activity was assayed using Coatest FVIII (Chromogenix AB, Mölndal, Sweden). Pooled plasma, which was calibrated against World Health Organization standard 91-666, was used as a standard. The amount of FVIII present in 1 ml of human plasma (1 unit/ml) was assumed to correspond to 0.4 nM (28).
Surface Plasmon Resonance Analysis-Binding studies were performed using a BIAcore TM 2000 biosensor system, based on surface plasmon resonance (SPR) technology. SPR analysis was performed essentially as described previously (33). LRP was immobilized onto a CM5-sensor chip, using the amine coupling kit as prescribed by the supplier, at indicated densities. A control channel was routinely acti-vated and blocked in the absence of protein. Binding to coated channels was corrected for binding to noncoated channels (less than 5% of binding to coated channels). SPR analysis was assessed in 150 mM NaCl, 2 mM CaCl 2 , 0.005% (v/v) Tween 20, and 20 mM Hepes (pH 7.4) at 25°C with a flow rate of 5 l/min or 20 l/min, where appropriate. Regeneration of the sensor chip surface was performed by incubating with 100 mM H 3 PO 4 for 2 min at a flow rate of 5 l/min.
Data Analysis-For analysis of the association and dissociation curves in the obtained sensorgrams, BIAevaluation software (Biacore AB) was used. Interaction constants were determined by performing nonlinear fitting of data corrected for bulk refractive index changes according to a two-site model, using equations described previously (33). Data fitting to a one-site model proved inappropriate, as judged from residual plots and statistical parameters (data not shown).
Competition Experiments-Purified LRP (125 fmol/well) was adsorbed onto microtiter wells in 50 mM NaHCO 3 (pH 9.5) in a volume of 50 l for 16 h at 4°C. LRP-coated and noncoated wells were blocked with 1% (w/v) gelatin in 150 mM NaCl, 2 mM CaCl 2 , 0.01% (v/v) Tween 20, and 20 mM Hepes (pH 7.4) in a volume of 100 l for 2 h at 37°C. After washing, FVIII (40 nM) was added in the same buffer to LRPcoated and noncoated wells in the presence (0 -1 M) of GST-RAP in a volume of 50 l and incubated for 3 h at 37°C. Bound FVIII was quantified by incubating for 15 min at 37°C with peroxidase-labeled antibody CLB-CAg 117.
Cellular Degradation Experiments-Cellular degradation of FVIII was examined essentially as described elsewhere (34). CHO cells (ATCC CCL-61) were grown to 80 -95% confluence in 24-well plates in DMEM:F12 medium supplemented with 10% (v/v) fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Before incubation, cells were washed extensively with DMEM:F12 medium. 125 I-labeled FVIII (20 nM) was then added in a volume of 200 l in DMEM:F12 medium containing 1% (w/v) bovine albumin and 5 mM CaCl 2 . In some experiments, 125 I-labeled FVIII was added in the presence of vWF (500 nM), GST-RAP (1 M), or protein A-Sepharose purified polyclonal antibodies directed against LRP (0.9 mg/ml). After a 35-min incubation at 37°C, cells were washed three times with 500 l of DMEM:F12 to remove nonbound material. Subsequently, incubation was allowed to proceed for another 4.5 h at 37°C in a volume of 200 l of DMEM:F12 medium containing 1% (w/v) bovine albumin and 5 mM CaCl 2 , in the presence of freshly added competitor where appropriate. Then, 100 l samples were drawn to determine the amount of degraded material. Degraded material is defined as radioactivity that is soluble in 10% trichloroacetic acid (35). In all experiments, a control was included in which the amount of degradation was assessed in the absence of cells.
Ligand Blotting-Ligand blotting experiments were performed essentially as described elsewhere (35). Briefly, 1 g of purified LRP was subjected to SDS-polyacrylamide gel electrophoresis on a 5% gel under nonreducing conditions and blotted onto nitrocellulose filters. Blots were blocked with 1% (w/v) non-fat milk in 100 mM NaCl, 0.01% (v/v) Tween 20, and 20 mM Hepes (pH 7.4). The blots were subsequently incubated with FVIII light chain (50 nM) in the absence or presence of GST-RAP (125 nM) in the same buffer containing 5 mM CaCl 2 for 16 h at room temperature. After washing the blots, bound FVIII light chain was detected using the peroxidase-labeled anti-FVIII light chain-directed antibody CLB-CAg 69. Binding was visualized using 3,3-diaminobenzidine tablets (Ken-En-Tec, Copenhagen, Denmark).

RESULTS
FVIII Binding to Immobilized LRP-The interaction between LRP and FVIII was first investigated by SPR analysis using purified components. As shown in Fig. 1, an increase in response was observed as FVIII (80 nM) was passed over immobilized LRP (8 and 25 fmol/mm 2 ), demonstrating that FVIII associates with LRP. Binding appeared to be dose-dependent in that the highest response was detected in case of the highest density of LRP (Fig. 1). Upon replacement of FVIII solution with buffer, the response started to decline gradually, indicating that FVIII dissociates from immobilized LRP. The interaction between LRP and FVIII was studied in more detail by assessment of the association and dissociation rate constants k on and k off . The results are summarized in Table I. Binding of FVIII heterodimer to LRP could be described using a two-site model. The affinity constants (K d ) derived from k on and k off values are 59 and 65 nM, respectively. These data indicate that FVIII is able to bind to LRP with moderate affinity in a revers-ible and dose-dependent manner.
Effect of GST-RAP and vWF on the FVIII-LRP Interaction-The interaction between FVIII and LRP was further analyzed in competition experiments using the FVIII carrier protein vWF and the LRP antagonist RAP. With regard to RAP, its effect was tested in an immunosorbent assay by incubating immobilized LRP (125 fmol/well) with mixtures of FVIII (40 nM) and various concentrations of GST-RAP (0 -1 M). Residual FVIII binding was subsequently determined using an anti-FVIII-directed antibody. The amount of FVIII bound decreased when concentrations of GST-RAP were increased ( Fig. 2A). Half maximum binding was found at a concentration of approximately 1 nM GST-RAP, showing that RAP efficiently inhibits the binding of FVIII to LRP. The effect of vWF on the interaction between LRP and FVIII was studied by SPR analysis. Whereas FVIII (40 nM) bound efficiently to LRP (16 fmol/mm 2 ) in the absence of vWF, association with LRP was inhibited dose-dependently in the presence of vWF (10 -500 nM) (Fig. 2B). The presence of a 12-fold excess of vWF resulted in over 90% inhibition, indicating that vWF interferes with complex formation between FVIII and LRP. These data demonstrate that the interaction between FVIII and LRP may be inhibited at the level of LRP using RAP or at the level of FVIII using vWF.
Cellular Degradation of FVIII-To address the contribution of LRP to the transport of FVIII to the endosomal degradation pathway, cellular degradation of FVIII was examined in experiments using CHO cells, which express LRP constitutively (6). It appeared that 125 I-labeled FVIII was rapidly degraded by CHO cells (Ϸ 8 fmol/10 5 cells after 4.5 h), and degradation did not occur in the presence of vWF (Fig. 3). The addition of GST-RAP (1 M) inhibited degradation of 125 I-labeled FVIII by approximately 65% (Fig. 3). Moreover, a similar extent of inhibition was observed using polyclonal antibodies directed against LRP (Fig. 3). This demonstrates that LRP contributes to the cellular uptake and subsequent degradation of FVIII.
Binding of FVIII Subunits to LRP-To identify FVIII regions involved in binding of LRP, we first examined the interaction of LRP with the isolated heavy and light chain of FVIII. Using 100 nM FVIII heavy chain, no association of this subunit to LRP (25 fmol/mm 2 ) could be detected (Fig. 4A). In contrast, in the presence of FVIII light chain (100 nM), a clear increase in response was found (Fig. 4A), indicating that FVIII light chain displays binding to LRP. Binding of FVIII light chain to LRP was further investigated in ligand blotting experiments. When nitrocellulose filters containing purified LRP (1 g) were incubated with FVIII light chain (50 nM), a clear band could be visualized (Fig. 4B, lane I), whereas only a faint band was observed when incubated in the presence of FVIII light chain (50 nM) and RAP (125 nM) (Fig. 4B, lane II). Thus, these data strongly suggest that FVIII light chain comprises a site that is recognized by LRP.
Effect of Thrombin Cleavage on FVIII Light Chain Binding to LRP-Because both vWF and LRP bind to FVIII light chain (18, Fig. 4), we tested the possibility that LRP and vWF share similar sites within this part of the FVIII molecule. FVIII comprises two sites that are involved in vWF binding, one of which is located between residues 1649 and 1689 at the aminoterminal part of the light chain (18,36). Thrombin-cleaved FVIII light chain, which lacks this particular sequence, was therefore compared with intact light chain for binding to LRP by SPR analysis to reveal the kinetic parameters k on and k off . As for FVIII heterodimer, binding of both intact and thrombincleaved light chain to LRP could appropriately be described using a two-site model. Similar k on and k off values were ob- Response is indicated in Resonance Units (RU) and is corrected for nonspecific binding, which was less than 5% relative to binding to LRP-coated channels. tained for both intact and thrombin-cleaved FVIII light chain, indicating that these proteins are similar in their interaction with LRP. Apparently, binding to LRP is independent of the amino-terminal part of FVIII light chain.
Interaction between LRP and the FVIII C2 Domain-Apart from its amino-terminal part, the carboxyl-terminal C2 domain of FVIII light chain also comprises a vWF binding site (19). Because vWF binding is inhibited by the C2 domain-directed antibody ESH4, the effect of this antibody on LRP binding was tested. As shown in Fig. 5A, ESH4 interfered with FVIII light chain binding to LRP. Inhibition appeared to be specific because ESH4 was unable to affect binding of tissue-type plasminogen activator/plasminogen activator inhibitor-1 complexes to LRP (data not shown). Moreover, other FVIII light chain directed antibodies (CLB-CAg A and CLB-CAg 69) were unable to interfere with LRP binding (data not shown). Thus, these data suggest that the FVIII C2 domain contributes to the interaction with LRP. This was further investigated using recombinant C2 domain. However, even at high concentrations of this 17.5-kDa fragment (500 nM), only a modest association with LRP was observed (Fig. 5B, line I). The conformation of the C2 domain may be affected by residues elsewhere in FVIII light chain or by the C2 domain-directed antibody ESH8 (37). Binding of the C2 domain to LRP was therefore addressed in the presence of this antibody. When complexes of C2 domain and antibody ESH8 were applied to LRP, a pronounced, dosedependent increase in response could be observed (Fig. 5B, lines II-IV), indicating that ESH8 promotes binding of the isolated C2 domain to LRP. Binding was dependent on exposure of the LRP binding site in the C2 domain because C2 domain-ESH8 complexes or C2 domain alone did not associate to LRP in the presence of antibody ESH4 (Fig. 5B, lines V and  VI). Collectively, these data strongly suggest that a LRP binding site is present within the C2 domain of FVIII light chain. DISCUSSION The surface of cells comprises numerous receptors that contribute to the binding and internalization of plasma proteins. Among these receptors is LRP, a multifunctional, endocytic receptor that is involved in the transport of a wide spectrum of ligands from the cell surface to the endosomal degradation pathway. In the present study, evidence is provided that LRP recognizes coagulation procofactor FVIII as a ligand. First, in a system using purified components, FVIII proved to bind to LRP in a manner that is reversible and dose-dependent (Fig. 1). Furthermore, binding could be inhibited efficiently in the presence of the FVIII carrier protein vWF or the LRP antagonist RAP (Fig. 2). Finally, both RAP-and anti-LRP-directed antibodies interfered with cellular degradation of 125 I-labeled FVIII (Fig. 3). It is of interest to note that inhibition of LRP does not fully prevent FVIII degradation. Whether or not re-sidual degradation involves a receptor-mediated process is currently under investigation. Irrespective thereof, our data indicate that the transport of FVIII from the cellular surface to the intracellular degradation pathway involves LRP. Because FVIII is structurally and functionally unrelated to the ligands of LRP that have been described thus far, it therefore provides a novel member of the already extensive range of established ligands for LRP.
The affinity by which FVIII heterodimer binds to LRP was found to be approximately 60 nM (Table I). This is in the same range as reported for some of the other LRP ligands, such as hepatic lipase (52 nM) (38), ␤-amyloid precursor protein (80 nM) (39), two-chain urokinase (60 nM) (40), and plasminogen activator inhibitor-1 (35 nM) (33). The data obtained for FVIII binding to LRP are in agreement with a two-site binding model (Table I), indicating that multiple regions contribute to the interaction with LRP. FVIII light chain proved similar to the FVIII heterodimer in that LRP binding involves two classes of binding sites (referred to in Table I as 1 and 2). Furthermore, both proteins display similar affinity with regard to the class 1 binding site (Table I). Therefore, it seems conceivable that FVIII light chain serves an important role in the interaction between FVIII and LRP. It is noteworthy that the class 1 association and dissociation rate constants for FVIII heterodimer differ from that for FVIII light chain by 25-fold (Table  I), suggesting that exposure of the LRP binding site in FVIII light chain is distinct from that in the FVIII heterodimer. The same may be true for the class 2 binding site because its   Fig. 1, except that a flow rate of 20 l/min was used. The concentrations tested were 60 -120 nM for FVIII, 150 -250 nM for intact FVIII light chain, and 50 -150 nM for thrombin-cleaved light chain. The data obtained for all concentrations tested were analyzed to calculate association rate constants (k on ) and dissociation rate constants (k off ) as described previously using a two-site binding model (33). Each class of binding sites is referred to as 1 and 2, respectively. Affinity constants (K d ) were inferred from the ratio k off :k on . Data are based on two to six measurements using five different concentrations for each measurement. Data represent the average (ϮS.D. affinity for LRP is 2-3-fold lower in FVIII light chain than in FVIII heterodimer. One explanation for this observation may be that the exposure of LRP binding sites within FVIII light chain depends on the presence of FVIII heavy chain. A similar mechanism has been reported previously for the interaction between FVIII light chain and vWF. The affinity for vWF increases 10-fold when FVIII light chain is associated with FVIII heavy chain (19). It has been reported that the A2 domain isolated from thrombin-activated FVIII has the potential to associate with LRP (41). Our observation that FVIII heavy chain does not associate with LRP may result from a different experimental approach. Alternatively, it cannot be excluded that the LRP binding site within the FVIII A2 domain is exposed in a suboptimal manner when this domain is linked to the A1 and B domains. This opens the possibility that the binding site for LRP within the A2 domain requires proteolysis of the FVIII heavy chain by thrombin for its exposure.
A positive identification of the LRP binding site on the FVIII light chain was achieved using the monoclonal antibody ESH4 (Fig. 5), which has previously been described to be directed against the FVIII C2 domain (42). This suggests that the C2 domain contributes to the interaction with LRP. Indeed, purified recombinant C2 domain displayed modest binding to LRP (Fig. 5B). Surprisingly, the binding of C2 domain to LRP was markedly increased in the presence of antibody ESH8 (Fig. 5B), in a manner that was more pronounced than that observed for other antibodies (data not shown). It has been well established that antibody ESH8 is able to change the conformation of the C2 domain in thrombin-cleaved FVIII light chain, resulting in altered affinities for vWF and phospholipids (37,43). It seems conceivable therefore that this antibody provokes a similar event in the isolated C2 domain, which then results in a more optimal exposure of the LRP binding site. The presence of a binding site for LRP within the C2 domain may explain the inhibitory effect of vWF (Fig. 2B), which is known to bind to the C2 domain of FVIII (19). However, whether vWF and LRP share a common binding site within the C2 domain or whether vWF interferes with binding by sterical hindrance remains to be determined.
In plasma, FVIII is in a dynamic equilibrium with vWF in which approximately 95% of the FVIII molecules have been calculated to be in complex with vWF (44). In complex with vWF, cellular uptake and degradation of FVIII is almost fully Similarly, complexes of C2 domain (400 nM) with ESH4 (500 nM) and complexes of the C2 domain (400 nM) with both antibodies (500 nM each) were incubated with LRP (lines V and VI, respectively). Complexes were allowed to form for 45 min before SPR analysis. suppressed (Fig. 3), indicating that the contribution of LRP to the cellular uptake of FVIII-vWF complexes in vivo is limited. However, in the absence of vWF, FVIII is degraded efficiently in a process that involves LRP (Fig. 3). In this view, our findings may provide an explanation for the beneficial effect of vWF that has been reported with regard to the expression of recombinant FVIII in CHO cells (17,45). vWF may well contribute to the accumulation of FVIII in medium by interfering with LRPmediated uptake of FVIII. Our findings may also be of relevance to the in vivo survival of FVIII in the absence of vWF. The physiological significance of complex formation between FVIII and vWF is particularly apparent in patients having von Willebrand disease type 3, who have no detectable vWF protein. Not only do these patients have a secondary deficiency of FVIII, but they also have a considerably reduced half-life of intravenously administered FVIII (26,27). It is tempting to speculate that the decreased levels of FVIII in these patients are associated with increased binding and internalization of FVIII by LRP due to the absence of vWF.