Role of CCP2 of the C4b-binding protein b-chain in protein S binding evaluated by mutagenesis and monoclonal antibodies Joanna H. Webb 1 , Bruno O. Villoutreix 2 , Bjo¨ rn Dahlba¨ck 1 and Anna M. Blom 1 1 Division of Clinical Chemistry, Department of Laboratory Medicine, Lund University, Sweden; 2 INSERM U428, University of Paris V, France Complement regulator C4b-binding protein (C4BP) and the anticoagulant vitamin K-dependent protein S form a high affinity complex in human plasma. C4BP is composed of seven a-chains and a unique b-chain, each chain comprising repeating complement control protein (CCP) modules. The binding site for protein S mainly involves the first of the three b-chain CCPs (CCP1). However, recently it has been suggested that CCP2 of the b-chain also contributes to the binding of protein S. To elucidate the structural background for the involvement of CCP2 in the protein S binding, several recombinant b-chain CCP1-2 variants having mutations in CCP2 were expressed and tested for protein S binding. Mutations were chosen based on analysis of a homology model of the b-chain and included R60A/R101A, D66A, L105A, F114A/I116A and H108A. All mutant pro- teins bound equally well as recombinant wild type to pro- tein S. Several monoclonal antibodies against the b-chain CCP2 were raised and their influence on protein S binding characterized. Taken together, the results suggest that the role of CCP2 in protein S binding is to orient and stabilize CCP1 rather than to be directly part of the binding site. Keywords: binding site; C4BP; complement; protein S; structure-function relationship. C4b-binding protein (C4BP) is an important regulator of the classical pathway of complement. C4BP also affects the regulation of the coagulation system, as it binds protein S, which serves as a cofactor to the anticoagulant activated protein C [1]. C4BP and protein S form a high-affinity, noncovalent 1 : 1 complex, the interaction being greatly enhanced by calcium [2]. Only free protein S, which accounts for approximately 30% of the total protein S in plasma, can act as a cofactor to activated protein C [3], whereas the functions of C4BP remain unperturbed when C4BP is in complex with protein S [4]. We have recently demonstrated that protein S can serve to localize C4BP to the surface of apoptotic cells, C4BP retaining its ability to bind complement protein C4b when attached to the apoptotic cells surface [5]. C4BP has an octopus-like structure being composed of seven elongated a-chains and one shorter b-chain, the chains being held together by hydrophobic forces that are stabilized by disulphide bridges in the central core [6]. Each chain comprises several complement control protein (CCP) domains, a CCP domain being approximately 60 residues long containing two disulphide bridges and a central antiparallel b-sheet [7]. It is the first of three CCPs (CCP1) of the unique b-chain that contains the protein S binding site [8–11]. In this CCP, a large hydrophobic patch is essential for binding of protein S [12]. It has also been shown that CCP2 has a moderate influence (approximately fivefold) on the interac- tion between C4BP and protein S [13,14]. To investigate the structural contribution of CCP2 to the C4BP–protein S interaction, we have expressed several b-chain variants carrying point mutations in CCP2 in a prokaryotic expres- sion system and tested their ability to bind protein S. The mutations introduced, R60A/R101A, D66A, L105A, F114A/I116A and H108A, were chosen based on a homology-based computer generated 3D-structure of the C4BP b-chain [15]. In addition, we have raised and characterized monoclonal antibodies against b-chain CCP1-2 and tested their influence on the C4BP–protein S interaction. None of the mutants affected the interaction and taken together the results suggest that the role of CCP2 in the binding of protein S is to stabilize and orient CCP1 rather than to provide binding sites for protein S. Materials and methods Cloning procedure Cloning of wild-type C4BP b-chain CCP1-2 has been described previously [12]. This construct was then used as a template and the mutations introduced using the Quik- Change site-directed mutagenesis kit (Stratagene). Sense primers used for mutagenesis were as follows (with template used in parenthesis): R60A (wild-type b-chain CCP1-2) 5¢- ACTGAGTGCGCCTTGGGCCACTGT-3¢, R60A/R101A (R60A) 5¢-GGCAGCAATGCGAGCCAGTGTCTA-3¢, D66A (wild-type b-chain CCP1-2) 5¢-CACTGTCCTGCTC CTGTGCTG-3¢, L105A (wild-type b-chain CCP1-2) 5¢-AG CCAGTGTGCAGAGGACCAC-3¢, F114A/I116A (wildtype b-chain CCP1-2) 5¢-GCACCTCCCGCTCCCGCCTGCA Correspondence to B. Dahlba ¨ ck, Division of Clinical Chemistry, Department of Laboratory Medicine, Lund University, University Hospital Malmo ¨ , S-205 02 Sweden. Fax: + 46 40 337044, Tel.: + 46 40 331501, E-mail: Bjorn.Dahlback@klkemi.mas.lu.se Abbreviations: C4BP, C4b-binding protein; CCP, complement control protein; MoAb, monoclonal antibody; tPA, modified plasminogen activator. (Received 14 October 2002, accepted 14 November 2002) Eur. J. Biochem. 270, 93–100 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03365.x AAAGT-3¢, H108A (wild-type b-chain CCP1-2) 5¢-TGTCT AGAGGACGCCACCTGGGCA-3¢. The various b–chain constructs were then transformed into Escherichia coli DH5a bacteria and mutations were confirmed using an automated DNA sequencing (Perkin-Elmer). Expression and purification of recombinant proteins Recombinant proteins were expressed and purified essen- tially as described before [12], with the exception of an additional gel filtration purification step. Briefly, E. coli strain BL21(DE3) transformed with the cDNA coding for the recombinant proteins were induced with isopropyl thio- b- D -galactoside to start expression of the proteins. Follow- ing induction, the bacteria were sonicated and centrifuged and the bacterial pellet was dissolved in a buffer containing guanidine-HCl and reduced glutathione. The sample was again sonicated and centrifuged, and the supernatant was applied to a nickel-nitrilotriacetic acid Superflow column (QIAGEN). Proteins were eluted from the column with a buffer containing 100 m M EDTA. Fractions were chosen and pooled based on measurement of absorbance at 280 nm. Dithiothreitol (100 m M ) was added to the pooled sample. Following reduction for 2 h at 4 °C, the sample was diluted in a buffer containing 3 m M cysteine and 0.3 m M cystine and refolding of the protein was accom- plished by extensive dialysis against the same buffer. Free cysteine residues were then blocked with iodoacetamide, followed by dialysis against a buffer containing 10% (v/v) glycerol. After dialysis the proteins were applied to a MonoQ column (Amersham Pharmacia Biotech), fractions were pooled after analysis of silver staining after SDS/ PAGE. Finally, all recombinant proteins were applied on a gel filtration column (Superose 12 HR 10/30, Amersham Pharmacia Biotech), previously equilibrated with 50 m M Tris/HCl, 150 m M NaCl (NaCl/Tris), pH 8.0. The flow rate used was 0.5 mL per minute and 0.5 mL fractions were collected. Fractions were chosen and pooled based on analysis of silver staining after SDS/PAGE and stored at )70 °C until further use. Concentrations were determined by measuring the absorbance at 280 nm, using an extinc- tion coefficient (1%, 1 cm) of 10 for all recombinant proteins. Plasma purified proteins C4BP and protein S were purified from human plasma, as described before [16,17]. Protein concentration was deter- mined by measuring absorbance at 280 nm. Extinction coefficients (1%, 1 cm) used were 14.1 (C4BP) and 9.5 (protein S). Protein S was labelled with 125 Iusingthe chloramine T method. Monoclonal antibodies Monoclonal antibodies (MoAb) 15 and 44 were raised using a standard procedure, as described previously [18]. The antigen used to immunize the mice was 20 lg per mouse per injection of recombinant wild-type C4BP b-chain CCP1-2 [12]. Antibodies selected for subcloning were chosen by ELISA. Twelve antibodies were purified using protein A and protein G coupled columns (5 mL, Hi-trap, Amersham Pharmacia Biotech). Antibody-containing cell medium was applied to the columns, equilibrated with 20 m M NaPO 4 pH 7.0. The columns were then washed in the same buffer. Bound antibodies were eluted with 0.1 M glycine-HCl, pH 2.7 with 1 mL fractions being collected in tubes containing 50 lL Tris/HCl, pH 9.0. Fractions were pooled after measurement of absorbance at 280 nm and dialyzed against NaCl/Tris pH 8.0 (50 m M Tris/HCl, 150 m M NaCl). Following dialysis, the absorbance was again measured and the concentration of antibody was calculated using an extinction coefficient (1%, 1 cm) of 12.5. Anti- bodies were stored at 4 °C after addition of 0.02% NaN 3 . Antibodies were tested using dot blot for their ability to bind plasma purified C4BP. Two microlitres of serial dilutions of C4BP were dotted on a nitrocellulose membrane. After incubation with the purified antibodies (at 5 lgÆmL )1 ) detection of binding was performed with a secondary rabbit anti-mouse Ig (Dakopatt) conjugated with alkaline phos- phatase and the membrane was developed. Two antibodies showed clear recognition of plasma purified C4BP, MoAb 15 and 44. We therefore decided to proceed with more detailed analysis of these two antibodies. All mono- clonal antibodies were tested for binding to three different constructs. Two of the constructs were recombinant C4BP a-chains where CCP1 or 2 had been replaced with their b-chain counterpart (characterized in [9]). The third was the recombinant molecule consisting of b-chain CCP1 and 2 (as used for immunization of the mice). Each construct was dotted on a membrane at equal concentration. After incubation with the purified antibodies (at 5 lgÆmL )1 ) detection of binding was performed with a secondary rabbit anti-mouse serum (Dakopatts) conjugated with alkaline phosphatase and the membrane was developed. Electrophoretic and blotting techniques Recombinant proteins. Recombinant proteins were separ- ated on 15% SDS/PAGE, under reducing and nonreducing conditions (approximately 1 lg) for the silver staining and under nonreducing conditions for the radioligand blot (approximately 1.5 lg). For the radioligand blot, the proteins were transferred by electroblotting from the gel to a poly(vinylidene difluoride)-membrane. The membrane was then incubated for one hour at room temperature in a quenching solution, composed of washing buffer [50 m M Tris/HCl pH 8.0, 150 m M NaCl, 0.5% (w/v) Tween 20] supplemented with 3% fish gelatin. Then the buffer was changed to washing buffer with 2 m M CaCl 2 and trace amounts of 125 I-labeled protein S and the membrane incubated for two hours at 4 °C. The membrane was further washed in washing buffer, dried and exposed in a cassette. Finally the membrane was scanned using a Phosphor-Imager (Molecular Dynamics). Western blot. Western blot was used to test the binding of MoAb 44 to recombinant wild-type b-chain and plasma purified C4BP under both reducing and nonreducing conditions. Varying amounts of protein were run on SDS/ PAGE (15% for recombinant wild-type b-chain, 5% for unreduced plasma purified C4BP and 10% for plasma purified reduced C4BP). The proteins were transferred from the gel to a poly(vinylidene difluoride)-membrane. The 94 J. H. Webb et al.(Eur. J. Biochem. 270) Ó FEBS 2003 membrane was then incubated for one hour at room temperature in quenching solution. Following this, the membrane was incubated with MoAb 44 or 15 (5 lgÆmL )1 ) inthesamebuffer.Themembranewasthenwashedin washing buffer before being incubated with the secondary antibody (biotinylated goat anti-mouse, 2 lgÆmL )1 )for 30 min. Following this, the membrane was washed again, and incubated for 30 min with horseradish peroxidase coupled Vectastain reagent (Vector Laboratories Inc., Ca, USA) prepared according to the manufacturer’s recom- mendations. Finally, the membrane was washed and developed. Protein S binding assays Direct binding assay. Microtiter plates were coated with 50 lL of protein (wild-type or mutant recombinant b-chain) at 2 lgÆmL )1 in 75 m M Na-carbonate, pH 9.6, at 4 °C overnight. The plates were then washed three times in washing buffer with 2 m M CaCl 2 and quenched for 1 h. The plates were washed as above, and protein S added at increasing concentrations (0–240 n M , final concentration), plus trace amounts of 125 I-labeled protein S, at room temperature for three hours. Finally, the plates were washed five times in washing buffer and bound radioactivity determined in a c-counter. Competition assay. Microtiter plates were coated with 50 lL of plasma purified C4BP at 10 lgÆmL )1 in 75 m M NaCO 3 ,pH9.6,at4°C overnight. The plates were then washed three times in washing buffer and quenched in washing buffer with 3% fish gelatin for 1 h, and washed as before. Increasing amounts of plasma purified C4BP, or recombinant b-chain (wild-type and mutants), plus trace amounts of 125 I-labeled protein S were added overnight at 4 °C. The next day the plates were washed five times in the same buffer as before and bound radioactivity was measured in a c-counter. Monoclonal antibody binding assays Test of binding site for MoAbs. Microtiter plates were coated overnight with 50 lL antigen (recombinant b-chain CCP1-2 1 lgÆmL )1 , plasma purified C4BP or C4BP– protein S complex 10 lgÆmL )1 )in75m M NaCO 3 ,pH9.6, at 4 °C. Plates were washed three times in washing buffer and blocked for 1 h at room temperature with NaCl/Tris pH 7.5, 1% (w/v) BSA. Following blocking, serial dilutions of MoAb 15 or 44 (0–20 lgÆmL )1 ) were added to the immobilized C4BP for one hour at 37 °C. Plates were again washed three times, and incubated with secondary antibody (goat anti-mouse serum coupled with horse radish peroxi- dase) for 1 h at 37 °C. Plates were then washed three times before being developed and absorbance was read at 490 nm in BioTek plate reader. Test if protein S can bind recombinant C4BP b-chain in the presence of MoAbs. Microtiter plates were coated overnight with 50 lLofMoAb15or44at10lgÆmL )1 in 75 m M NaCO 3 ,pH9.6,at4°C. Plates were washed and blocked as above. Serial dilutions of wild-type recombinant C4BP b-chain (0–500 n M ) were added together with trace amounts of 125 I-labeled protein S overnight at 4 °C. The next day, the plates were washed five times in the same buffer as before and bound radioactivity was measured in a c-counter. Results Selection of amino acids in CCP2 to be mutated In this study, we were interested in investigating which regions within CCP2 of the C4BP b-chain are involved in protein S binding [13,14]. The mutagenesis strategy was chosen based on the 3D-homology model of the C4BP b-chain [15] (Fig. 1). A series of b-chain variants was created, including R60A/R101A, D66A, L105A, F114A/I116A and H108A. These changes were based on the theoretical analysis and were expected to be structurally well tolerated. This was also confirmed as recombinant mutant proteins were shown to bind protein S with high affinity, which requires a b-chain molecule with structural integrity. The purified recombinant proteins were separated on SDS/PAGE under reducing conditions (Fig. 2A). In addition, recombinant proteins with the F114A/I116A and H108A mutations introduced were separated on SDS/PAGE under nonreducing conditions (Fig. 2B). All recombinant proteins looked very similar to wild-type, except the variant carrying the F114A/I116A mutation, which was present in more than one form. Mutations in CCP2 had no effect on binding of protein S The b-chain constructs were tested in three different assays for their ability to bind protein S. In the radioligand blotting assay, proteins were separated by SDS/PAGE, blotted to a poly(vinylidene difluoride)-membrane, and allowed to bind 125 I-labeled protein S in solution (Fig. 2C). All variants yielded binding patterns similar to that of wild-type b-chain. In the binding assay, the recombinant proteins were immobilized in microtiter plates and tested for binding of 125 I-labeled protein S (Fig. 3A). The binding curves were similar, with one exception. Mutant F114A/I116A showed a slightly higher binding as compared to wild-type, with an apparent twofold higher affinity at the most. The b-chain variants were also tested for protein S binding using a competition assay. Immobilized plasma purified C4BP and fluid phase competitor (plasma purified C4BP or recombinant wild-type or mutant b-chain) were allowed to compete for binding of 125 I-labeled protein S (Fig. 3B,C and Table 1). The apparent slightly lower efficiency of the F114A/I116A variant (around threefold) to compete for protein S binding was not statistically significant. The results of the assay are dependent on the concentration of correctly folded protein. On the silver stained unreduced SDS/PAGE gel (Fig. 2B) it is apparent that the F114A/I116A mutant is present in more than one form. According to the radioligand blot (Fig. 2C), only one of the F114A/I116A forms binds protein S. The results suggest that the correctly folded F114A/I116A variant binds protein S at least equally as well as the wild-type variant. The apparent slightly decreased efficiency of the H108A variant (around 1.5 times) in the competition assay was not statistically significant. Ó FEBS 2003 The role of CCP2 of the C4BP b-chain in binding of protein S (Eur. J. Biochem. 270)95 Monoclonal antibodies General. To further elucidate the influence of CCP2 on protein S binding, we made the effort to raise monoclonal antibodies against CCP1-2 with the aim to use them in structure–function analysis. Two of 12 antibodies, MoAb 15 and MoAb 44, recognized native C4BP purified from plasma on a dot blot and were therefore selected for detailed analysis. The remaining ten antibodies only recognized the recombinant b-chain implying that access to their epitopes was sterically hindered in the fully assembled C4BP molecule. All antibodies were IgG, MoAb 15 belonged to IgG1 and MoAb 44 belonged to IgG2a class. The antibodies were tested on dot blot against the previously characterized a–b-chain hybrids [9]. All the antibodies reacted with constructs containing CCP2 and MoAb 44 in addition gave a weak reaction against the hybrids only containing CCP1 (results not shown). The conclusion was that the epitopes of all antibodies were mainly located in CCP2. Western blot. MoAb 15 and MoAb 44 were tested for their ability to recognize the recombinant wild-type b-chain and plasma purified C4BP with Western blotting, under reducing and nonreducing conditions. Both MoAbs recog- nized the reduced b-chain of plasma purified C4BP and the reduced recombinant CCP1-2. They also reacted with the unreduced b-chain in purified C4BP from plasma and with unreduced recombinant b-chain CCP1-2 (Fig. 4, only MoAb 44 shown). Binding assay. In an attempt to define the epitopes for the MoAb 15 and MoAb 44, they were tested for binding of the different recombinant b-chain CCP1-2 constructs. Two of the b-chain variants, D66A and R60A/R101A, showed weaker binding to both MoAbs (Fig. 5A,B). These two variants bound protein S equally as well as the wild-type b-chain, suggesting correct folding. It seems plausible to conclude that the epitopes of both MoAb 15 and 44 involve amino acids at mutated positions. In binding to plasma purified C4BP that was immobilized on microtiter plates, the two antibodies demonstrated a difference because MoAb 15 bound whereas MoAb 44 did not (Fig. 5C). However, for binding of MoAb 15 to occur, the plates had to be coated with high concentrations of C4BP (10 lgÆmL )1 ). When lower C4BP concentrations were used, binding of MoAb 15 only reached approxi- mately 10% (at highest concentration of MoAb 15) of the binding seen to recombinant wild-type b-chain CCP1-2 (results not shown). This suggests that the multiple, long a-chains to various extents sterically block the epitopes for MoAb 15 (partial blockage) and MoAb 44 (complete blockage) on the b-chain in immobilized C4BP. Influence of protein S binding of monoclonal antibod- ies. To investigate whether protein S and MoAb 15/ MoAb 44 had overlapping binding sites on the b chain, the following four different binding assays using microtiter plates were used. (a) First, the effect of increasing amounts of fluid phase protein S on the binding of the MoAbs to immobilized recombinant b-chain was tested. The binding of both MoAb 15 and MoAb 44 to the b-chain was Fig. 1. Schematic outline of C4BP (A) and model of C4BP b-chain CCP1-2 (B). (A) C4BP consists of seven identical a-chains and one b-chain. Each chain is made up repeating complement control protein (CCP) domains. The a-chains have eight such CCP domains, the b-chain has three. All chains are held together by disulphide bridges involving the nonrepeat carboxy-terminal regions. (B) A structural model is shown as a ribbon diagram (adapted from [15]). Residues expected to be glycosylated are shown in orange (N47, N54, N81 and N100). The glycans are presented in orange but to illustrate the fact that we neither know their exact size or orientation, they are not attached to the protein. These glycans were positioned in a way con- sistent with the covalent attachment to the molecule. Residues R60, D66, R101, L105, H108, F114 and I116 were mutated in the present study while key residues involved in protein S binding on CCP1 are coloured blue (residues I33, V31, V18 and I16) [12]. 96 J. H. Webb et al.(Eur. J. Biochem. 270) Ó FEBS 2003 unaffected even by the highest concentration of protein S used (240 n M , results not shown). (b) In the second analysis, free C4BP and C4BP–protein S complexes were immobilized and the binding of MoAb 15 tested (Moab 44 did not work in this assay system as described above). MoAb 15 was found to bind equally well to both free and complexed C4BP (Fig. 5D). (c) In the third assay, the antibodies were immobilized and recombinant b-chain added together with 125 I-labeled protein S. The 125 I-labeled protein S–b-chain complex bound to MoAb 15 but not to MoAb 44 (Fig. 6). In the MoAb 15 variant, the decreased binding of 125 I-labeled protein S observed at higher concentrations of b-chain suggested that free b chain competes with the b-chain-protein S complex for binding to the antibody (Fig. 6). (d) In the fourth assay, the effect of MoAb 44 on the binding of protein S to immobilized b-chain was investigated. In this assay protein S was found to bind equally well in the absence and presence of MoAb 44. Thus, from assays a and d it was concluded that the immobilized b-chain was able to bind MoAb 44 and protein S simultaneously. In contrast, immobilized MoAb 44 could not bind the protein S–b-chain complex (assay c). These results suggest that protein S and the two MoAbs bind to different binding sites on the b-chain but that under certain experimental conditions there was sterical hindrance between protein S and MoAb 44. Discussion The binding site for protein S on C4BP is predominantly contained in CCP1 of the b-chain [9]. A key binding surface for protein S involves residues I16, V18, V31 and I33 on CCP1 of the b-chain [12]. CCP2 has been shown to have a small positive influence on the C4BP–protein S interaction because a recombinant construct containing CCP1 but lacking CCP2 had approximately five times lower affinity for protein S than a construct containing both CCP1 and 2 [13]. In that study, chimeras between CCP domains from the C4BP b-chain and the N-terminal of a modified plasminogen activator (tPA) were used. The authors concluded that the function of CCP2 was not simply to serve as a spacer, which could be performed by any CCP domain, but that CCP2 either induced a specific conform- ational change in CCP1 or that it directly participated in the binding of protein S. In an earlier study using recombinant a–b-chain hybrids consisting of b-chainCCP1,1+2,and 1 + 2 + 3, we failed to observe this specific effect of CCP2 [9]. To address the discrepancy between the two studies, a recombinant construct consisting of CCP1 from the C4BP b-chain and CCP2 from the C4BP a-chain was fused to the modified tPA [14]. The results of this exercise again suggested b-chain CCP2 to have a specific positive influence. To further evaluate the role of b-chain CCP2 in the interaction with protein S we decided to introduce amino acid changes into putative binding sites on b-chain CCP2. The mutations were chosen based on a homology model of the C4BP b-chain [15]. We have previously shown that a solvent exposed hydrophobic patch on b-chain CCP1 was crucial for binding protein S, and that electrostatic forces played only a minor role [12,19]. The b-chain is heavily glycosylated [20] but the glycans seem to have no role in binding of protein S because prokaryotic recombinant b-chain constructs bind equally well to protein S as plasma purified C4BP. Thus, with the structural information gained from our model and experimental data, and general knowledge of protein–protein interactions, we could select areas potentially involved in the binding of protein S. We were interested in mutating residues close to the CCP1– CCP2 interface and amino acids of hydrophobic nature. Yet, it was also important to assess the role of charged or polar residues, and of amino acids located away from the intermodule region. In the present situation, three param- eters could not be well defined; the exact angle between the CCP modules (assuming there is such an angle between CCP modules), which Asn residue is glycosylated and the exact orientation of the glycans relative to the protein core (suchachaincouldbeflexible). Fig. 2. SDS/PAGE analysis of recombinant C4BP b-chain CCP1-2. (A) Indicated recombinant proteins (wild-type and mutants), approximately 0.5 lg per well, were separated on SDS/PAGE under reducing conditions. Proteins were visualized using silver staining. (B) Selected recombinant proteins (wild-type and mutants), approximately 1 lg per well, were separated on SDS/PAGE under nonreducing conditions. Proteins were visualized using silver staining. (C) Recombinant proteins (wild-type and mutants), approximately 1.5 lg per well, were separated on SDS/PAGE under nonreducing conditions. The proteins were then transferred to a poly(vinylidene difluoride)-membrane and incubated with 125 I-labeled protein S. Bound protein S was detected using a PhosphorImager. Ó FEBS 2003 The role of CCP2 of the C4BP b-chain in binding of protein S (Eur. J. Biochem. 270)97 Changes introduced in the present study involved residues R60A/R101A, D66A, L105A, F114A/I116 A, and H108A. As it is not possible to mutate all the residues of the CCP2 to investigate the protein S–C4BP interaction, we reasoned as follows. (a) Residues R60 and R101 point away from the protein core and cover one side of the CCP2. Although these residues are positively charged, such amino acids are often part of recognition site and the side-chain of Arg is for a large part hydrophobic and aromatic. (b) Residue D66 is fully exposed, and its substitution would probe another face of the CCP2. (c) Residue L105 is exposed, hydrophobic and somewhat on another face of the CCP2 as compared to the other residues where mutations have been introduced. Moreover, L105 is close to CCP1, thus further suggesting a potential role in binding of protein S. (d) Residues F114 and I116 are both solvent exposed. (e) Residue H108 is also exposed, possibly next to the protein S binding site located on CCP1. In all situations, mutations to Ala should be structurally well tolerated, as the substituted amino acids are solvent exposed and apparently not making important contacts with the remaining part of the molecule. We suggest that with these substitutions, taken together with the potentially glycosylated Asn and the epitopes for the monoclonal antibodies now reported, we cover most of the CCP2 surface that could be accessible to protein S during the interaction with C4BP. Using three different assays, we show that the mutated recombinant b- chain bound protein S equally well as the wild-type b-chain. Table 1. Comparison of the average negative log M concentration required of each recombinant construct to reduce binding of protein S to immobilized C4BP to 50% as compared to binding in the absence of fluid phase competitor. Recombinant construct Average negative log M concentration required for 50% of 125 -I protein S binding SD Wild-type 8.1 ± 0.2 H108A 7.9 ± 0.1 F114A/I116A 7.8 ± 0.2 Fig. 4. Western blot analysis of plasma purified C4BP using MoAb 44. Three micrograms of plasma purified C4BP were separated on 10% (reducing conditions) or 5% (nonreducing conditions) SDS/PAGE. The proteins were then silver stained or transferred to a poly(vinylid- ene difluoride)-membrane and subjected to Western blot using MoAb 44. (A) Lane 1 and 2 show the Western blot, whereas lane 3 and 4 show the corresponding silver stained gels of two different C4BP preparations (lanes 1 and 3, 2 and 4, respectively). The a-andb-chains are indicated with arrows. (B) Western blot of C4BP (unreduced sample) using MoAb 44. Fig. 3. Binding assay analysis of interaction between recombinant b-chain constructs and protein S. (A) Direct binding assay. Increasing amounts of cold protein S mixed with trace amounts of 125 I-labeled protein S, were added to immobilized recombinant b-chain (wild-type or mutants). Binding is expressed as per cent of the maximum binding seen in each experiment. Each point represents the mean value (± SD) from three different experiments, performed in duplicate. d, Wild- type; h, R60A/R101A; n,D66A;s, L105A; j, H108A; m, F114A/ I116A. (B and C). Competition assay. Fluid phase plasma purified C4BP or recombinant b-chain (wild-type or mutants) were allowed to compete with immobilized C4BP for binding to 125 I-labeled protein S. Binding to the immobilized C4BP is expressed as per cent of the binding seen in the absence of fluid phase inhibitor. Each point rep- resents the mean value (± SD) from at least three different experi- ments, performed in duplicate (with the exception of R60A/R101A that was performed twice in duplicate). (B) d, Wild-type; h, H108A; m, F114A/I116A; s, L105A. (C) d, Wild type; j, R60A/R101A; n, D66A; s, plasma purified C4BP. 98 J. H. Webb et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Taken together, the data on the mutations and monoclonal antibodies (see below) did not support the concept that CCP2 directly contacts protein S. Instead, CCP2 may rather contribute to the stability and/or proper orientation of the CCP1. We have raised two monoclonal antibodies, MoAb 15 and 44, against the recombinant wild-type C4BP b-chain CCP1-2. These antibodies specifically recognized both the recombinant b-chain CCP1-2 and the full-length b-chain from plasma purified C4BP using Western blotting under reducing and nonreducing circumstances. The antibodies displayed similar behaviour in Western blotting, but in the binding assay, MoAb 15 showed much higher affinity for immobilized plasma C4BP. To probe the binding site for the MoAbs a binding assay was used, where the abilities of the different recombinant b-chains (wild-type and mutants) to bind the antibody were compared. For both antibodies, the binding of D66A and R60A/R101A was greatly weakened as compared to wild- type suggesting the binding epitope for the monoclonal antibodies to be located in CCP2, involving these residues and some of the surrounding regions. Most probably residues 60 and 66 are more important for binding than residue 101, as this is located next to a potential carbohy- drate (i.e. the glycan could interfere with MoAb binding). The epitope for MoAb 44 seems to be located closer to the CCP1–CCP2 interface (but outside and/or partially over- lapping the protein S binding site), as this monoclonal weakly recognized a recombinant construct carrying only CCP1 from the b-chain. This also matches the fact that this MoAb bound less to b -chain proteins having residues R60A/R101A mutated. The binding of MoAb 15 to the b-chain was unaffected by the binding of protein S whereas the situation was more complicated with MoAb 44. Under certain experimental conditions, protein S and MoAb 44 could not bind simultaneously whereas no competition between the two b-chain ligands were seen under other conditions. Thus, immobilized MoAb 44 could not bind the protein S–b-chain complex whereas both MoAb 44 and protein S could bind simultaneously to immobilized b-chain. This suggests that the epitope for MoAb 44, possibly involving both CCP1 and CCP2, is closer to the protein S binding site than the epitope for MoAb 15. These results point in the direction that the binding site for protein S is localized on CCP1 and not on CCP2 of the b-chain. The monoclonals seem to cover at least two faces of the CCP2. As protein S does not impede binding of Fig. 5. Binding of MoAb 15 and 44 to C4BP. Microtiter plates were coated with either C4BP or the different recombinant b-chain variants and MoAb 15 or 44 were then added in solution. Binding of the antibody was detected with horse radish peroxidase-coupled secondary antibody. Binding is expressed as per cent of maximum binding seen (with highest concentration of antibody). Each point represents the mean value (± SD) from two to three different experiments, performed in duplicate (except D., which was performed once in duplicate). Bars represent S.D. values. (A) Recombinant proteins tested with MoAb 44; d, wild type b-chain; h, R60A/R101A; n, D66A; s, L105A; j, H108A; m, F114A/I116A. (B) Recombinant proteins tested with MoAb 15: d, wild-type b-chain; h, R60A/R101A; n,D66A.(C)MoAb15:h recombinant wild-type b-chain; j plasma purified C4BP. (D) MoAb 15: h, plasma purified C4BP without protein S; j plasma purified C4BP in complex with protein S. Fig. 6. Ability of MoAbs to bind CCP1-2-protein S complexes. Increasing amounts of b-chain CCP1-2 were added together with a trace amount of 125 I-protein S to microtiter wells with immobilized monoclonal antibodies. Each point represents two different experi- ments performed in duplicate. r,MoAb15;h,MoAb44. Ó FEBS 2003 The role of CCP2 of the C4BP b-chain in binding of protein S (Eur. J. Biochem. 270)99 MoAb 15 (which did not seem to bind to CCP1), and we found no effect of any of the introduced mutations, we cannot really define any accessible area or important zone that could be involved in protein S binding. Although it may be argued that some other regions remain unexplored on CCP2, we have with our set of mutations and with the likely identification of epitopes covered most of the surface of CCP2. In addition, as several sides of CCP2 appear to be covered by glycans (which are not involved in protein S binding), we consider that there is no clear free space left for the direct interaction with protein S. Thus, the role of CCP2 should be to stabilize and orient CCP1 rather than be a part of the binding site for protein S. Our study, together with previous reports, provides information useful in the theor- etical prediction of the structure of the protein S–C4BP complex. Acknowledgements We are most grateful to Ulla Persson for raising the monoclonal antibodies. Ylva Lindroth for assistance in developing the binding assay, and to Ylva Ha ¨ rdig who made the eukaryotic a/b-chain chimeras. This work was supported by grants from the Swedish Research Council, the Network for Inflammation Research funded by the Swedish Foundation for Strategic Research, the Tore Nilson’s Trust, the Greta and Johan Kock’s Trust, the O ¨ sterlunds Trust, the Craaford Trust, the Royal Physiographic Society in Lund, the Clas Groschinsky’s Trust, the King Gustav V’s 80th Anniversary Founda- tion, the Thelma Zoe ´ gas Trust, the Nanna Svartz Trust, the Magnus Bergvalls Trust, the Louis Jeantet Foundation of Medicine, and research funds from the University Hospital, Malmo ¨ . References 1. Dahlba ¨ ck, B. (1995) The protein C anticoagulant system: inherited defectsasbasisforvenousthrombosis.Thromb. Res. 77, 1–43. 2. Dahlba ¨ ck, B., Frohm, B. & Nelsestuen, G. (1990) High affinity interaction between C4b-binding protein and vitamin K-depen- dent protein S in the presence of calcium. Suggestion of a third component in blood regulating the interaction. J. Biol. Chem. 265, 16082–16087. 3. Dahlba ¨ ck, B. (1986) Inhibition of protein Ca cofactor function of human and bovine protein S by C4b-binding protein. J. Biol. Chem. 261, 12022–12027. 4. Schwalbe, R., Dahlba ¨ ck, B., Hillarp, A. & Nelsestuen, G. (1990) Assembly of protein S and C4b-binding protein on membranes. J. Biol. Chem. 265, 16074–16081. 5. Webb, J.H., Blom, A.M. & Dahlba ¨ ck, B. (2002) Vitamin K-dependent protein S localizing complement regulator C4b- binding protein to the surface of apoptotic cells. J. Immunol. 169, 2580–2586. 6. Kask, L., Hillarp, A., Ramesh, B., Dahlba ¨ ck,B.&Blom,A.M. (2002) Structural requirements for the intra-cellular subunit polymerization of the complement inhibitor C4b-binding protein. Biochemistry 41, 9349–9357. 7. Norman, D.G., Barlow, P.N., Baron, M., Day, A.J., Sim, R.B. & Campbell, I.D. (1991) Three-dimensional structure of a comple- ment control protein module in solution. J. Mol. Biol. 219, 717– 725. 8. Hillarp, A. & Dahlba ¨ ck, B. (1988) Novel subunit in C4b-binding protein required for protein S binding. J. Biol. Chem. 263, 12759– 12764. 9. Ha ¨ rdig,Y.&Dahlba ¨ ck, B. (1996) The amino-terminal module of the C4b-binding protein b-chain contains the protein S-binding site. J. Biol. Chem. 271, 20861–20867. 10. Ferna ´ ndez, J.A., Villoutreix, B.O., Hackeng, T.M., Griffin, J.H. & Bouma, B.N. (1994) Analysis of protein S C4b–binding protein interactions by homology modeling and inhibitory antibodies. Biochemistry 33, 11073–11078. 11. Ferna ´ ndez, J.A. & Griffin, J.H. (1994) A protein S binding site on C4b-binding protein involves beta chain residues 31–45. J. Biol. Chem. 269, 2535–2540. 12. Webb, J.H., Villoutreix, B.O., Dahlback, B. & Blom, A.M. (2001) Localization of a hydrophobic binding site for anticoagulant protein S on the beta-chain of complement regulator C4b-binding protein. J. Biol. Chem. 276, 4330–4337. 13. van de Poel, R.H., Meijers, J.C. & Bouma, B.N. (1999) Interaction between protein S and complement C4b-binding protein (C4BP). Affinity studies using chimeras containing c4bp beta-chain short consensus repeats. J. Biol. Chem. 274, 15144–15150. 14. van de Poel, R.H., Meijers, J.C., Dahlba ¨ ck,B.&Bouma,B.N. (1999) C4b-binding protein (C4BP) beta-chain Short Consensus Repeat-2 specifically contributes to the interaction of C4BP with protein S. Blood Cells Mol. Dis. 25, 279–286. 15. Villoutreix, B.O., Ferna ´ ndez, J.A., Teleman, O. & Griffin, J.H. (1995) Comparative modeling of the three CP modules of the beta- chain of C4BP and evaluation of potential sites of interaction with protein S. Protein Eng. 8, 1253–1258. 16. Dahlba ¨ ck, B. (1983) Purification of human C4b-binding protein and formation of its complex with vitamin K-dependent protein S. Biochem. J. 209, 847–856. 17. Dahlba ¨ ck, B. (1983) Purification of human vitamin K-dependent protein S and its limited proteolysis by thrombin. Biochem. J. 209, 837–846. 18. Brown, M.A., Stenberg, L.M., Persson, U. & Stenflo, J. (2000) Identification and purification of vitamin K-dependent proteins and peptides with monoclonal antibodies specific for gamma-carboxyglutamyl (Gla) residues. J. Biol. Chem. 275, 19795–19802. 19. Blom, A.M., Covell, D.G., Wallqvist, A., Dahlba ¨ ck,B.& Villoutreix, B.O. (1998) The C4b-binding protein–protein S interaction is hydrophobic in nature. Biochim. Biophys. Acta 1388, 181–189. 20. Hillarp, A. & Dahlba ¨ ck, B. (1990) Cloning of cDNA coding for the beta chain of human complement component C4b-binding protein: sequence homology with the alpha chain. Proc. Natl Acad. Sci. USA 87, 1183–1187. 100 J. H. Webb et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . Role of CCP2 of the C4b -binding protein b-chain in protein S binding evaluated by mutagenesis and monoclonal antibodies Joanna H. Webb 1 ,. less to b -chain proteins having residues R60A/R101A mutated. The binding of MoAb 15 to the b-chain was unaffected by the binding of protein S whereas the