PROTEINS: Structure, Function, and Genetics 31:391–405 (1998) Structural Investigation of C4b-Binding Protein by Molecular Modeling: Localization of Putative Binding Sites Bruno O. Villoutreix,1* Ylva Härdig,1 Anders Wallqvist,2 David G. Covell,3 Pablo Garcı́a de Frutos,1 and Björn Dahlbäck1 1Department of Clinical Chemistry, The Wallenberg Laboratory, University Hospital, Lund University, Malmö, Sweden 2Department of Chemistry, Rutgers University, Piscataway, New Jersey 3Frederick Cancer Research and Development Center, National Cancer Institute, Science Applications International Corporation, Frederick, Maryland ABSTRACT C4b-binding protein (C4BP) contributes to the regulation of the classical pathway of the complement system and plays an important role in blood coagulation. The main human C4BP isoform is composed of one b-chain and seven a-chains essentially built from three and eight complement control protein (CCP) modules, respectively, followed by a nonrepeat carboxy-terminal region involved in polymerization of the chains. C4BP is known to interact with heparin, C4b, complement factor I, serum amyloid P component, streptococcal Arp and Sir proteins, and factor VIII/VIIIa via its a-chains and with protein S through its b-chain. The principal aim of the present study was to localize regions of C4BP involved in the interaction with C4b, Arp, and heparin. For this purpose, a computer model of the 8 CCP modules of C4BP a-chain was constructed, taking into account data from previous electron microscopy (EM) studies. This structure was investigated in the context of known and/or new experimental data. Analysis of the a-chain model, together with monoclonal antibody studies and heparin binding experiments, suggests that a patch of positively charged residues, at the interface between the first and second CCP modules, plays an important role in the interaction between C4BP and C4b/Arp/Sir/heparin. Putative binding sites, secondary-structure prediction for the central core, and an overall reevaluation of the size of the C4BP molecule are also presented. An understanding of these intermolecular interactions should contribute to the rational design of potential therapeutic agents aiming at interfering specifically some of these protein–protein interactions. Proteins 31:391– 405, 1998. r 1998 Wiley-Liss, Inc. r 1998 WILEY-LISS, INC. Key words: complement control protein; protein modeling; blood coagulation; C4b-binding protein INTRODUCTION Human C4b-binding protein (C4BP) is a highmolecular-weight glycoprotein (570 kDa) composed, for the main isoform, of seven identical a-chains (549 residues each with three potential glycosylation sites) and a single b-chain (235 residues with five potential glycosylation sites).1,2 Like many other complement regulators (e.g., factor H), C4BP contains several 60 amino acid long repeating modules (59 repeats for human C4BP) termed complement control protein (CCP) modules, also called short consensus repeats (SCRs) or Sushi domains.3,4 The C4BP a- and b-chains are made of eight and three CCP modules, respectively, further continued by a nonrepeat carboxy-terminal region of about 60 residues involved in polymerization of the chains. C4BP has been studied by EM and X-ray scattering and has been shown to have a spider-like organization.5–7 C4BP was first identified as a regulator of the complement system8,9 and has been shown to be an acute phase protein.10,11 Upon C4b binding, C4BP interferes with the formation and decay of the classical pathway C3-convertase (C4b2a) and functions as a cofactor to factor I in the proteolytic degradation of C4b. In addition to C4b, C4BP is known to bind two Grant sponsor: The Swedish Medical Research Council; Grant numbers: 07143 and 11793; Grant sponsor: The Alfred Österlund Trust; Grant sponsor: The Albert Påhlsson Trust; Grant sponsor: The Göran Gustafsson Trust; Grant sponsor: The King Gustav V and Queen Victoria Trust; Grant sponsor: Ax:Son Johnsons Trust; Grant sponsor: The University Hospital, Malmö; Grant sponsor: The Louis Jeantet Foundation of Medicine. *Correspondence to: Bruno O. Villoutreix, Department of Clinical Chemistry, University Hospital, Lund University, S205 02 Malmö, Sweden. E-mail: email@example.com. lu.se Received 19 September 1997; Accepted 22 December 1997 392 B.O. VILLOUTREIX ET AL. plasma proteins: vitamin K-dependent protein S (PS), an anticoagulant cofactor to activated protein C (APC),2,12 and serum amyloid P component (SAP), a protein found in all types of amyloid deposits, including those present in Alzheimer disease.13–15 The PS–C4BP interaction is of significant importance in coagulation because about 70% of PS is noncovalently bound to C4BP and when complexed, PS loses its APC cofactor activity. PS deficiencies have been reported in patients suffering from thrombosis, and in this situation only the concentration of free PS can be linked unambiguously to thrombotic diseases.16 It is possible that C4BP binds to another coagulation plasma protein, factor VIII/VIIIa.17 It was recently shown that C4BP interacts with proteins Arp and Sir, two Ig-binding cell surface proteins from the Gram-positive bacterium Streptococcus pyogenes, both belonging to the antiphagocytic M protein family.18 Such pathogens can cause suppurative and inflammatory infections of the skin and throat and postinfection sequelae such as acute glomerulonephritis. The C4BP–Arp or –Sir interaction has been suggested to increase virulence by protection from complement attack.19 Finally, the negatively charged heparin molecule binds to C4BP and it was shown that this interaction competes with the binding of C4b but not of PS.20,21 Several studies have aimed at localizing the binding sites for C4BP ligands. PS, via its sex hormone binding globulin domain (SHBG),22 is the only known protein that binds to the C4BP b-chain23,24 and this interaction involves only the most amino-terminal CCP module of the b-chain.25 The C4b-binding site was first suggested to be in the middle of the a-chain (6th and/or 7th CCP).26,27 However, more recent studies have shown the importance of the Nterminal part (CCPs 1 to 3).28–30 The C4b- and Arp/Sir-binding sites are partially overlapping and involve CCP modules 1 to 3 of the a-chain.29 SAP has been shown to bind to the central core fragment of C4BP.15 The C4BP core fragment is obtained after digestion of the molecule by chymotrypsin, which cleaves the a-chain within CCP7, thereby liberating most of the chain (CCPs 1 to 7) from the remaining part of the protein.23 In contrast, the b-chain, also digested by chymostrypsin,23 is not involved in the C4BP–SAP interaction. The stoichiometry of the SAP–C4BP and PS–C4BP interactions are 1:1, whereas each a-chain, most likely, contains binding sites for C4b, Arp/Sir, and heparin. However, at least in the case of the C4BP–C4b interaction, only two to four C4b molecules bind to C4BP, presumably due to steric hindrance.31 Only limited information concerning the tertiary structure of complement control proteins is available. However, the three-dimensional (3D) structure of several CCP modules has been determined by NMR spectroscopy.32–34 These modules share important structural similarity, with local differences, mainly within loops, confirming that the consensus sequence forming their hydrophobic core is of major importance in conferring topological similarities. In consequence, these NMR structures are valid templates for comparative modeling studies.35 CCP modules are characterized by the conservation of several key residues, among which four cysteines disulfidebridged in a 1–3, 2–4 fashion. The consensus residues form a compact hydrophobic core that is surrounded by five b-strands.33 The study of C4BP using X-ray crystallography or NMR spectroscopy has not been possible because of the size of the molecule, the presence of several isoforms, and heterogeneous glycosylation. Therefore molecular modeling methods seem very appropriate to study the 3D structure of this protein. In the present investigation, comparative model building of the eight CCP modules from the C4BP a-chain was carried out, taking into account electron micrograph data with the goal of identifying potential binding sites, and with a special emphasis on the Arp/C4b/ heparin–C4BP interactions. For this purpose, interactive structural analysis of the C4BP a-chain model, together with computation of some of its electrostatic properties and theoretical enumeration of possible binding sites, was undertaken. The interaction between Arp/C4b/heparin and C4BP was studied, experimentally guided by the analysis of the model structure. The human C4BP a-chain model was also analyzed in conjunction with previously reported interspecies multiple sequence alignment.36,37 These results allow us to propose molecular mechanisms for some of the C4BP interactions. MATERIALS AND METHODS Homology Modeling The modeling of CCP modules from the C4BP b-chain has been previously reported.35,38 The same approach was applied for building the CCPs from the a-chain and is briefly described below. From the 20 CCP modules of factor H, the 3D structures of CCPs 5, 16, and a 15–16 pair have been investigated by NMR spectroscopy.32–34 The coordinates of these modules were obtained from the Protein Data Bank (PDB).39,40 A Silicon Graphics Indigo2 R10000 workstation was employed for the construction of the models, together with the molecular modeling programs InsightII, Biopolymer, Homology, Discover, and Delphi (Biosym-MSI, San Diego, CA, USA). Model building The sequence of each CCP module of the C4BP a-chain was aligned to the factor H sequence (repeat 15th or 16th) according to Figure 1. The eight CCP modules were first built independently. The NMR coordinates from either H15 or H16 were used as initial templates. The deletions in C4BP relative to the reference proteins were refined by 200 cycles of energy minimization, keeping the rest of the struc- INVESTIGATION OF C4BP BY MOLECULAR MODELING 393 Fig. 1. Sequence alignment of the eight CCP modules from the a-chain of human C4BP to the 15th or 16th factor H modules and secondary-structure prediction for the nonrepeat central core region. The dashed lines indicate the disulfide bridging pattern of the C4BP a-chain CCP modules. The sixth CCP module has an extra disulfide shown by a solid black line (see text). Cysteine residues belonging to the CCP modules are numbered above the sequences. The ‘$‘ symbol depicts consensus sequence for N-glycosylation. C4BP modules were built using the presented alignment, i.e., H15 was used as template for C4BP CCPs 1, 5, and 6, while H16 was used for C4BP CCPs 2, 3, 4, 7, and 8. Black lines above and below the C4BP nonrepeat sequences correspond to the PHD secondary structure prediction for the a- and b-chains, respectively, and indicate a-helices. ture fixed. The insertion regions were built from a search41 among some of the high-resolution protein structures present at the PDB. The minor steric clashes and bond strains due to the introduction of the new sidechains and loop building were regularized by a short energy minimization. Then, the first and second CCP modules of the C4BP a-chain were superimposed onto the H15–16 pair. H15 of a second H15–16 pair was superimposed on the H16 component of the first pair. The third CCP of the a-chain of C4BP was then superimposed on the H16 structure of the second H15–16 pair. The process was repeated until the a-chain was built. The model was then briefly energy minimized. Fig. 2. Validation of the model structure. a (left): Ribbon representation of the C4BP aCCP1 model structure. Richardson rendering for the theoretical model of the first module of the a-chain. The disulfide bridges are in blue, while some key residues conserved in the CCP modules (hydrophobic core) are shown in magenta. The secondarystructure elements (b-sheets, in yellow) and other structural features (turns, in green) were assigned according to the NMR template. The remaining part of the structure is in white. b (above): Profile plot for the entire a-chain model. The compatibility score for residues in a 21-amino acid sliding window is presented. The horizontal black line represents the 0 value. Such profile indicates that the protein is well folded. Figure 3. INVESTIGATION OF C4BP BY MOLECULAR MODELING 395 Fig. 4. C4BP a-chain model structure. a (left): Solid ribbon representation of the a-chain. Overall representation of the model structure with the cysteine residues shown in blue and asparagines, part of a consensus sequence for N-glycosylation, in magenta. These asparagine sidechains are accessible to the solvent in this model. The two chymotrypsin cleavage sites are highlighted in red. b (right): The a-chain CCP1 and CCP2 model structures. Some residues accessible to the solvent and forming patches (hydrophobic, polar, positively charged) of potential impor- tance for intermolecular interactions are presented (see text). The NMR structure of a dodecamere heparin molecule65 was docked manually onto the positive cluster in order to probe further the possible C4BP–heparin interaction. K30, strictly conserved in the sequences,37 could also be involved in the interaction with heparin. The distance between K30 and K79 is about 40 Å and the length of the heparin molecule is about 50 Å. The insertion region when compared to the NMR template (between C106 and C122) is painted in green for orientation. Fig. 3. Schematic diagram of human C4BP. Model structures for the a- and b-chains were organized according to electron micrograph images.5,6 The previously reported model for the three CCP modules of the b-chain35 was rebuilt with the intermodule angle stretched in a similar fashion as one of the a-chain. The interaction between human C4BP and PS has been recently narrowed to the first CCP module of the C4BP b-chain25 and to the PS SHBG-like region.22,62 The PS binding site, by means of synthetic peptides competition assays, was suggested to involve a linear epitope (residues 31 to 45) on the first CCP of the C4BP b-chain.63 Indeed, a cluster of solvent-exposed hydrophobic residues and two lysines at the surface of the first CCP module are likely to play important roles in the PS–C4BP interaction.35 Monoclonal antibody 6F6, shown to interact with a linear epitope involving C4BP b-chain residues 51–65, is also known to inhibit the binding of PS to C4BP.64 Binding sites at the surface of the a-chain discussed in this article are presented. The distances reported are in Angstroms. 396 B.O. VILLOUTREIX ET AL. Theoretical Enumeration of Potential Binding Sites The method for identifying potential binding sites is described in Young et al.,42 but a brief summary will be presented here. Geometry A simplified model consisting of only the Ca coordinates was used to represent the geometry of each residue.42–46 The lattice model of Ca atoms for the protein studied here fit the molecule coordinates within 1-Å RMS deviation.47 Lattice positions exterior to the molecular surface can be examined for their interactions with residues of the target molecule. This method can be thought as similar to calculating molecular surfaces based on their accessibility to a water molecule probe.48 The probable binding strength of each exterior position is determined by simply counting the number and type of target Ca positions within 7.5 Å of any exterior points. The amino acid composition of the defined clusters thus scores the relative strength for which a ligand might bind at that site. Scoring binding strength The binding strength for each exterior position is the sum of scores for its constituent residues: N fcluster 5 S i51 ei (1) where N is the number of neighboring (d , 7.5 Å) residues. Parameters for each residue type were based on assigning a hydrophobic value* determined from residue-based contact energies as calculated by Miyazawa and Jernigan.49 These contact energies can be understood in terms of the hydrophobic– hydrophilic designations of amino acids and the pairings that contribute most to protein stability.44,46,49 Hydrophobicity assignments using this procedure show a strong correlation with the TanfordNozaki scale50 and others, as shown by Cornette et al.51 Determination of scores for all surface regions can be completed in less than one CPU minute on a typical Silicon Graphics Workstation. Electrostatic Potential Electrostatic potential calculations were carried out for the C4BP a-chain model structure. The Delphi package, which solves the Poisson-Boltzmann equation by a finite difference method,52,53 was used for the calculation of the 3D distribution of the electrostatic potential at physiological ionic strength and pH. The atomic coordinates of the model structure were mapped into a 3D grid with a resolution of about one grid point/2 Å. The grid was chosen to leave a 10-Å border between the protein and the grid edge. The dielectric constant was set to 9 for the protein interior and 80 for the surrounding solvent. Atomic radii definitions were taken from the Delphi default parameters with the radii of hydrogen atoms set to zero. A standard set of formal charges was assigned to the titratable residues (e.g., Arg, 10.5 e at the Nh1 and Nh2 nucleus). The N-terminal amino group was given a charge of 11 e while the Cterminal carboxy group was considered neutral. The resulting 3D isopotential contours were analyzed interactively within InsightII. Secondary-Structure Prediction for Central Core Nonrepeat Region Secondary structure prediction was performed with the Profile-fed neural network systems from HeiDelberg (PHD)54,55 for the nonrepeat region of the human C4BP a- and b-chains. This method was selected since it has an average cross-validated accuracy of around 72% on a set of 250 sequences with known structure. The entire human sequence of the mature chains was provided as query input. Experimental Design Proteins C4BP5 and C456 were purified from human plasma according to the procedure described in the respective references. C4(H2O) was used instead of C4b and was prepared by repeated freezing and thawing of purified C4.57,58 Protein Arp was a kind gift from Dr. G. Lindahl. C4BP was radiolabeled using Iodobeads (Pierce Chemical Co., Rockford, IL, USA). The monoclonal antibody against C4BP (mAb104) used in this study has been described previously.30 Inhibition experiments Microtiterplates were coated with C4(H2O) or Arp, 50 µl/well at concentrations of 10 µg/ml in coating buffer. Increasing amounts of standard heparin, low-molecular-weight heparin, or mAb104 in dilution buffer were then added, together with a trace amount of 125I-labeled C4BP (20,000 cpm/well). The final volume was 50 µl/well. After an overnight incubation, the plates were washed and the amount of bound C4BP detected using a g-counter. RESULTS Comparative Model Building Overall validation of the C4BP model structure *The values of ei are: F 25.12, M 24.91, I 24.88, L 24.65, W 24.36, V 24.17, C 24.00, Y 23.24, A 22.82, H 22.75, G 22.34, T 22.30, P 22.22, R 22.18, S 22.07, Q 21.98, E 21.94, N 21.90, D 21.81, K 21.50. All eight individual CCP models from the C4BP a-chain were screened interactively and their 3D structure as well as the distribution of the hydropho- INVESTIGATION OF C4BP BY MOLECULAR MODELING Fig. 5. Theoretical enumeration of potential binding sites. Clusters of residues identified by theoretical mean as potential binding site area as reported in Table I are mapped in a simplified manner onto the C4BP model structure. The Ca atom of the potentially N-glycosylated asparagines is shown in a CPK representation. bic, polar, and charged residues were found in accordance with the NMR templates. The distribution of polar and apolar atoms of a protein is a wellestablished way to assess the quality of a 3D structure.59 The compact hydrophobic core of the CCP modules34 is found in all eight C4BP individual models, indicating that, despite a relatively low sequence identity between the NMR template and C4BP, our structural prediction is accurate. As an example, some key sidechains contributing to this hydrophobic core are shown for the first CCP of the 397 Fig. 6. Electrostatic surface potential of the a-chain model. The model structure is shown with the same orientation as the one used for Figure 5. The electrostatic isosurfaces at a level of -0.6 (red) and 10.6 (blue) kcal/mol/e are shown. A positive region, at the interface between aCCP1 and aCCP2, that play a role in the interaction between C4BP and C4b/Arp/heparin is highlighted. Other positively charged residues of possible interest are also listed (see text). a-chain (Fig. 2a). The accuracy of a protein structure can also be assessed by inspection of its 3D profile.60 When the compatibility scores of a misfolded protein are plotted as a function of the amino acid sequence, the overall profile is often below 0.1 and can dip several times below zero.60 We have run a similar calculation for the entire C4BP a-chain model (Fig. 2b). The profile is always above zero and the selfcompatibility scores are essentially above 0.1, supporting the quality of our model structure. The backbone angle values were investigated using ProStat (Biosym-MSI) and were found compatible with those of experimentally determined structures. 398 B.O. VILLOUTREIX ET AL. The relative orientation of the modules, however, is difficult to predict, since the NMR data for the H15–16 pair indicate flexibility between the two repeats.32 A variable intermodule angle is further supported after observations of the electron micrographs obtained for factor H61 and C4BP,6 where some regions of the molecules appear twisted. However, it is not known if these bends are specific to these proteins or induced by experimental procedures. We assumed that the overall orientation of one CCP module relative to the next would be similar to the one observed for the H15–16 NMR pair but with the intermodule angle stretched in order to match the overall length of the a-chain as deduced from previous EM studies.5,6 One CCP module has an ellipsoid shape, with a long axis of about 32 Å, thus eight CCP modules should have an approximate theoretical length of 250 Å, to which about 10–15 Å could be added due to the presence of the intermodule linker regions. Thus one would expect a theoretical length of about 260–270 Å for an extended tentacle of eight CCPs. The a-chain homology model proposed here has an extended conformation with a length of about 260 Å (Fig. 3). An overall length of 330 Å between the N-terminus of the a-chain and the center of the central core has been reported.5 The diameter of the core has been estimated to be around 60 Å . Structural Analysis and Theoretical Enumeration of Potential Binding Sites The interactive structural analysis of the a-chain (Fig. 4a,b) models, in conjunction with investigation of the a-chain sequences (rabbit,36 rat,37 bovine,66,67 human,68 mouse69), as described by Hillarp et al.,37 is reported below. Potential binding sites on the achains are listed in Table I and are graphically represented in Figure 5. CCP1 (E residues 1 to 63) Three patches of exposed hydrophobic residues are observed and are generally well conserved in the C4BP sequences from different species. The first patch involves residues P4, A12, P13, M14, I16, and L18. The second displays F10, L34, Y37, and Y62. These residues could shield another hydrophobic patch located on aCCP2 involving F84, V108, V113, and the C65–C106 disulfide. The third hydrophobic patch contains V55, Y56, F59, I61, and aCCP2 F84 (Fig. 4b). This surface forms part of a theoretically identified binding site (cluster h, Table I and Fig. 5), together with four polar residues: S42, T43, T45, and T47. The latter residues are included in an area presenting numerous short and polar sidechains, generally conserved in the sequences (residues T25, T27, T28, S40, S50, T58). Charged residues are unevenly distributed on this module, with D15, E20, D51, E53, R22, and K24 at the N-terminal pole, K30 in the middle, and R39 and H41 at the C-terminal pole. Among them, E20, R22, K24, K30, and R39 are generally conserved in the interspecies sequence analysis, or replaced by a long polar residue or an amino acid of opposite charge.37 a CCP2 (E residues 64 to 125) At the surface of aCCP2 a patch of positively charged residues well conserved within the different species 36,37 can be observed. This involves K63, R64, R66, H67, and K7g with contributions from aCCP1 R39 and possibly H41 and K30 (Fig. 4b).Only E70 and D81 somewhat counterbalance these positive charges. This area is part of a theoretically identified binding site (cluster b, Table I and Figs. 5 and 6). Close to this positive surface, a loop presenting with a short insertion when compared to the NMR template forms also part of cluster b (Table I and Fig. 5). Several short and polar sidechains, conserved in the species, are also noted on one face of aCCP2. The residues involved are T80, S83, S86, S91, S101, T102, and T103 (Fig. 4b). a CCP3 (E residues 126 to 189) An insertion comprising residues 174–178 with respect to the template is found at the interface with aCCP2 (Fig. 1). A consensus sequence for N-glycosylation is present at residue N173 (Figs. 1 and 4a). This residue is solvent-accessible in the model and is located at the interface with aCCP2 in a region rich in polar and charged residues (aCCP2 R72, N73, Q121, E123, and aCCP3 E172, E174). One face of this module displays several charged residues (K126, K128, D132, R134, R137, H138, E141, E142, D156), most often conserved in the C4BP sequences. Another region, at the interface with aCCP2, presents several solvent-accessible hydrophobic residues, aCCP2 F97, L98, I99, P120, V125, and aCCP3 K126 (carbon sidechain), F144, A146, Y147, F149. This last set of residues is part of a potential binding site (Table I, cluster d, and Fig. 5). A patch of Ser/Thr residues is also noted on aCCP3 (conserved in the species) and involves S139, S150, T152, S154, S166, S168, and S182. a CCP4 (E residues 190 to 249) One face presents several charged residues not well conserved between the species (R192, K193, D195, E200, D236, K238), with possible contribution from aCCP3 R158 (Table I, cluster c, and Fig. 5). A face with several hydrophobic sidechains, which tends to be conserved in the sequences, was also 399 INVESTIGATION OF C4BP BY MOLECULAR MODELING TABLE I. Theoretical Enumeration of Potential Binding Sites for the a-Chain of Human C4BP C4BP a-chain clusters a b c d e f g h i Residues E405, C406, D407, Y410, I411, L412, V413, Q415, A416, P430, Q431, and C432. G36, Y62, K63, R64, C65, R66, F84, V108, Q109, D110, G112, V113, and G114. I189, T190, C191, R192, K193, P194, D195, V196, S197, H198, G199, and E200. L98, I99, P120, Q121, C122, E123, I124, V125, and Y147. C351, H352, E353, C374, G375, D376, I377, C378, Y422, and S423. R134, N135, D156, P157, R158, F159, I189, T190, A235, and S237. S154, C155, L161, L162, G163, H164, A165, and C186. S42, T43, Q44, T45, L46, T47, V55, Y56, N57, and T58. F159, S160, K188, I189, T190, I208, Y209, and N210. Location Comments On aCCP7, close to aCCP8 This area could interact with SAP The second best ranking score was mainly on aCCP2 and at the interface with aCCP1 On aCCP4 with contribution of the linker region between the third and fourth repeats At the aCCP2–aCCP3 interface This region is nearby and part of a patch of positively charged residues Interface between aCCP6 and 7 Possibly nearby residues F97; 144 and 149 also play a role This forms a small groove at the surface of the molecule Interface between aCCP3 and 4 This region is close from the third best scoring cluster (c) On aCCP3 Nearby the cluster f On aCCP1 Nearby the second best scoring cluster (b) This area is nearby clusters f, g, and c Interface between aCCP3 and 4 observed. This surface involves mainly M201, F205, P207, I208, V216, K218 (carbon sidechain), V224, L225, and V230. However, these residues are not all close in space but are distributed over one entire face of the module. a CCP5 (E residues 250 to 315) This module is not present in mouse C4BP.37,68 The only striking solvent-exposed patches involve a set of charged residues and, on the opposite side of the repeat, five threonines. These include residues E263, K270, E271, D272, R281, R283, D293, E294, T291, T292, T296, T297, and T307. Overall, these residues tend to be conserved in the C4BP sequences. a CCP6 (E residues 316 to 376) This module is not present in mouse C4BP.37,68 aCCP6 has most likely an extra disulfide bridge involving C316 with C339 (Figs. 1 and 4a), which was modeled by assuming the bridging pattern suggested by Norman et al.34 An interesting surface involves charged residues E326, H330, R331, K332, R334, H338, D345, H352, and E353. A small set of Ser/Thr residues is also noted nearby this patch. This surface contains T328, S348, S350, T354, S355, and S358. These two patches are partially conserved in the sequences. The only hydrophobic patch noted at the surface of this module contains residues V340, Y341, F342, and Y343 at the interface with aCCP5. a CCP7 (E residues 377 to 434) Two patches of charged residues are noted here, one involving residues K399, E400, E401, and K417, and on another face K383, H386, H388, K390, E405, D407, K408, and K433, further continued by aCCP8 R437 and R481. A solvent-accessible hydrophobic patch was found on this module and involves mainly I403, C406, I411, L412, V413, and A416, further continued on aCCP8 L435 and V454 (Table I, cluster a, and Fig. 5). Chymotrypsin cleaves this module at residues Tyr395 and Trp42523 (Fig. 4a). a CCP8 (E residues 436 to 490) Two N-glycosylation sites are predicted at N458 and N480 (Figs. 1 and 4a). N480 is somewhat more shielded from the solvent than N458 but is still accessible. N480 is at the interface with aCCP7 and surrounded by five positively charged residues (aCCP8 R437, R481, and aCCP7 H386, K408, K433, further extended by K383, H388, K390). N458 has also several polar and charged residues in its direct vicinity (K450, E455, E457, T460, S474, T476). One patch of charged amino acids is noted and involves R437, K438, E440, R445, D449, D451, and D464. These charged residues tend to be conserved. One solvent-accessible hydrophobic patch is found and involves residues V469, V470, P472, Y484, and P485, which are close from three charged amino acids generally conserved in the sequences, E486, K489, and E491. 400 B.O. VILLOUTREIX ET AL. Fig. 7. Inhibition experiments. a: Increasing concentrations of heparin was allowed to compete with 125I C4BP for binding to immobilized C4(H2O) (U), protein Arp (N), and mAb104 (M). 100% binding was estimated in the absence of fluid-phase competitor. b: Increasing concentrations of mAb 104 was allowed to compete with 125I C4BP for binding to immobilized Arp. The data were obtained after three different experiments. Electrostatic Potential and Heparin Effects on C4BP Interactions tration of heparin was required to displace the binding of Arp and mAb 104 as compared to C4(H2O). Similar results were obtained with lowmolecular-weight heparin (data not shown). The ability of mAb 104 to displace the binding of C4BP to immobilized Arp was tested in a similar assay (Fig. 7). It was found that this antibody was an efficient inhibitor of the Arp–C4BP interaction. Since the epitope for mAb 104 is located either on the first aCCP or at the interface between aCCP1 and aCCP2,30 the above data suggests that C4b, Arp, and mAb 104 are interacting in part with the positive cluster found on the first and second CCP modules. Three other positive patches that could be potential binding sites for heparin were found in the a-chain. One is located on aCCP5 and involves K270, R281, and R283. A site on aCCP6 could involve six charged residues, H330, R331, K332, R334, H338, and H352. The last most likely site for heparin binding is at the interface between aCCP7 and aCCP8 and could involve eight positively charged residues, K383, H386, K390, K408, K433, R437, and R481 (Fig. 6). The overall electrostatic potential of the a-chain is mosaic, but some modules tend to be covered either by positive or negative surface. The most striking surfaces are reported below. The interface between the first and the second repeats is mainly positive at the energy level computed (Fig. 6), as well as at -1 or -2 kcal/mol/e (data not shown). This region corresponds to the conserved patch of positively charged residues discussed above (Fig. 4b). Another electropositive region of possible interest is at the interface between aCCP7 and aCCP8 and could be of importance for interaction with the mainly electronegative b-chain35 or with other molecules such as SAP. The negative surfaces are mainly at the N-terminal pole of aCCP3, at the interface between aCCP5 and aCCP6, and at the C-terminal pole of aCCP8. This last region on aCCP8 could be of importance for interactions with the nonrepeat segment of C4BP or with SAP. Areas presenting a richer content in positively charged residues could be involved in specific or nonspecific heparin binding. One site, well conserved in the C4BP sequences (6 to 8 positively charged residues), was discussed above and is located at the interface between aCCP1 and aCCP2 (Figs. 4b and 6). A series of experiments were designed to evaluate the role of this positively charged cluster in the C4BP interaction with heparin and with Arp/C4b. Increasing concentration of heparin was added, together with a trace amount of 125I-labeled C4BP, to microtiter plates coated with C4(H2O), Arp, or mAb 104 (a monoclonal antibody known to compete with the C4BP–C4b interaction). As can be seen in Figure 7, heparin was able to interfere with all three interactions, although a higher concen- Secondary-Structure Prediction for Central Core Nonrepeat Region The C4BP b- and a-chains contained only bstrands, as expected, for the region encompassing the three and eight CCP modules, respectively. The nonrepeat areas were found essentially helical (74% for the b-chain and 53% for the a-chain) (Fig. 1). This last point will be analyzed in the Discussion section. DISCUSSION C4BP is a complex and important regulatory protein of the complement cascade. In addition, its role INVESTIGATION OF C4BP BY MOLECULAR MODELING in the coagulation system is certainly of importance since it is known that patients with low free PS level are at higher risk of thromboembolic disease.16,70 The potential role of C4BP in some bacterial infectious processes has been pointed out recently.18,19 It is known that many pathogenic microorganisms have evolved mechanisms to avoid complement attack, in some cases by expressing a protein with functional similarity to a complement regulatory molecule or by acquiring such a protein from the host.71 It has also been reported that C4BP exacerbates the inflammatory response to Escherichia coli infusions in baboons.72 For these reasons, the study of the 3D structure and mechanisms of actions of C4BP are of importance. In the present investigation we have developed a model for C4BP using as template previously reported NMR structures of CCP modules from factor H. Comparative model building73,74 is the best method presently available to predict the 3D structure of a protein75 and has already been used successfully to investigate CCP modules.35,38,76–79 These modules are found in several families of proteins and seem to be well adapted for protein–protein interaction. The interface area between these modules is nicely designed to ‘hook‘ approaching molecules. Intermolecular associations involve generally interactions between the surface-accessible portions of molecules and seem to be mainly determined by the details of geometric and chemical complementarity.45 Recently, Young et al.42 have described a method for finding regions on protein surfaces where a residue-based hydrophobicity potential is high. These regions generally coincide with binding sites of ligands. In addition, the graphical analysis of the three-dimensional distribution of the electrostatic potential can help in pointing toward regions of functional importance.80–82 Analysis of a-Chain Our investigation aimed at the localization of residues possibly involved in heparin, C4b, and Arp/Sir binding. Several studies support the importance of the amino-terminal part (CCPs 1 to 3) for the binding of C4b, including EM analysis,6,30 binding assays with chimeric constructs,28–30 and sequence comparisons.36,37 A recent report indicates that the C4b- and Arp/Sir-binding sites are partially overlapping18 and involve CCP modules 1 to 3 of the a-chain.29 Furthermore, it was shown that the interactions C4BP–C4b31 and C4BP–Arp29 are ionic strength-dependent. These suggest that electrostatic forces could play a role in the complex formation, since at this concentration salts should essentially shield charges. This hypothesis is further supported by our study, which shows that heparin 401 competes with the C4BP–Arp interaction, highlighting the importance of positively charged residues. It is known that heparin competes also with the C4BP– C4b interaction.20 Monoclonal antibody 10430 and mAb G11H1129 seem to interact between aCCP1 and aCCP2. These two antibodies inhibit the contact C4BP–C4b. Moreover, the mAb 104–C4BP interaction is also displaced by heparin (Fig. 7). Taken together, these data suggest that an important site of interaction for C4b involves at least the positive cluster at the interface between aCCP1 and aCCP2 (Fig. 4b). However, since a higher concentration of heparin was necessary to displace the C4BP–Arp interaction when compared to C4BP–C4b, it is expected that the binding sites are not entirely equivalent. This is supported by the fact that the positive cluster between CCPs 1 and 2 is very conserved in C4BP sequences, while only C4BP from human (or closely related primates) interacts with Arp.29 We have compared the sequences of C4BP CCPs 1 and 2 with the ones of other C4b-binding proteins, complement receptor 1 (CR1), and decay accelerating factor (DAF). There is an important amino acid conservation in the proposed C4b-binding site. Interestingly, C4BP, CR1, and DAF possess a positively charged cluster at the interface between two CCPs. This cluster is located between CCPs 1 and 2 for C4BP and CR1, and between CCPs 2 and 3 for DAF. In addition, several residues forming human C4BP cluster b (Table I) are conserved also in CR1 and DAF. These residues are in the C4BP, numbering G36, K63, C65, R66, F84, I108 (or V), and V113.* Furthermore, in CR1, residues G35, R64, and N65 (C4BP G36, R66, and H67, respectively) have been shown to be important for C4b recognition by sitedirected mutagenesis.83 Also, C4BP R66 and H67 have been found to be important in Arp and C4b binding by mutagenesis.29 A model for the four CCPs of DAF also suggests that a positive surface, together with a hydrophobic patch mainly at the interface between CCPs 2 and 3, is of importance for the interaction with the C3 convertase.78 All these data support that cluster b is a binding site for C4b and Arp. The sequence expected to be responsible for the interaction with C4BP on the Arp protein contains several conserved negatively charged residues.19 This Arp sequence could thus interact with the positive cluster found between C4BP aCCP1 and aCCP2. C4b residues 738 to 826 have been suggested to play a role in the interaction with C4BP.27 The region *The sequences compared were human, orangutan, mouse, and guinea pig DAF; human, rabbit, rat, mouse, and bovine C4BP; human, chimpanzee, and hamadryas baboon CR1; and the mouse CRRY and CRY proteins. 402 B.O. VILLOUTREIX ET AL. encompassing C4b residues 740–757 contains a stretch of negatively charged residues (740EILQEEDLIDEDDIPVRS757)27 that would seem well adapted for the interaction with the aCCP1–aCCP2 positive cluster. Theoretical analysis of the a-chain also suggests that CCP3 and CCP4 contain potential binding sites (Fig. 5). These regions could play some role in the recognition of factor I, C4b, or Arp. Interestingly, no binding sites were found on CCP 5 and CCP6, but only at the interface between CCP 6 and CCP7 (Fig. 5). The fifth and sixth repeats are indeed missing in the mouse sequence,69 suggesting that these two modules may not be important for the function of the molecule. This is further supported because a recombinant C4BP molecule with the three amino-terminal CCPs from mouse did bind human C4b.28 Monoclonal antibody 7F227 has been reported to interact with aCCP6 and to inhibit the C4b–C4BP interaction. It seems likely that CCP6 does not contain the binding site for C4b and that the competition observed is due to steric hindrance. The top ranking score on the a-chain (Fig. 5 and Table I) could be involved in an interaction with SAP since this cluster (cluster a) is nearby the central core (Figs. 3 and 4) and two potential glycosylation sites. It is known that deglycosylation lowers the affinity for the C4BP– SAP interaction.15 Basic Structure and Central Core of C4BP The predominant form of human C4BP consists of seven identical a-chains and a single b-chain linked together by disulfide bonds within the C-terminal nonrepeat region. Covalent bonds between the achains, however, are not necessary for association since the cysteine residues in this region of the molecule are not present in mouse C4BP, which also lacks the b-chain.69,84 This indicates that noncovalent forces play an important role in the polymerization of the chains. Secondary-structure prediction for the nonrepeat regions suggested this area to be predominantly helical (Fig. 1). The central core was shown via EM experiments to have a ring-like structure with an inner diameter of 13 Å and an outer diameter of 60 Å.5 Building part of the nonrepeat core as a-helices allows for the formation of a structure presenting mainly one face with hydrophobic sidechains and another with hydrophilic amino acids (Fig. 8). Taken together, these data suggest that the central core nonrepeat region could be arranged as a barrel with the hydrophobic face of the helices packing against each other and the hydrophilic surfaces having contact with the solvent. Such hydrophobic surfaces could explain why polymerization of the chains is independent of disulfide bridging. Fig. 8. Amphipathic a-helix for regions of the nonrepeat central core. Several segments from the nonrepeat central core area are predicted to adopt an a-helical conformation (see Fig. 1). The region of residues 515 to 533 from the a-chain and of residues 200 to 210 from the b-chain were built by assuming a standard helical geometry. It can be seen clearly that one face of the helix displays essentially hydrophobic residues while the other contains essentially hydrophilic amino acids. Additional information is needed, however, to build the entire central core region. CONCLUSIONS The lack of 3D information for the nonrepeat core and a-chain of C4BP prompted us to undertake the present molecular modeling study. Most of the putative binding sites described here are in agreement with the experimental data on record, validating the overall accuracy of our model structure. A region involved in the interaction of Arp/C4b/heparin with C4BP, presenting positively charged residues, should be located at the interface between the first and second CCP modules of the a-chain. Heparin-like molecules on the surface of cells could play a role in the functioning of C4BP, such as local downregulation of the complement system. Polysulfated compounds and chemically modified peptides could be used to inhibit undesired macromolecular interactions. Our results will be useful for further characterization of these protein assemblages and for investigation of the central core region. Site-directed mutagenesis is presently undergoing in our laboratory to assess the role of the predicted binding sites. ACKNOWLEDGMENTS We are grateful to Dr. G. Lindahl for providing Arp protein. REFERENCES 1. Hillarp, A., Hessing, M., Dahlbäck, B. Protein S binding in relation to the subunit composition of human C4b-binding protein. FEBS Lett. 259:53–56, 1989. 2. Dahlbäck, B. Protein S and C4b-binding protein: Compo- INVESTIGATION OF C4BP BY MOLECULAR MODELING 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. nents involved in the regulation of the protein C anticoagulant system. Thromb. Haemost. 66:49–58, 1991. Reid, K.B.M., Bentley, D.R., Campbell, I.D., et al. Complement system proteins which interacts with C3b or C4b: A superfamily of structurally related proteins. Immunol. Today 7:230–234, 1986. Janatova, J., Reid, K.B.M., Willis, A.C. Disulfide bonds are localized within the short consensus repeat units of complement regulatory proteins: C4b-binding protein. Biochemistry 28:4754–4761, 1989. Dahlbäck, B., Müller-Eberhard, H.J. Ultrastructure of C4b-binding protein fragments formed by limited proteolysis using chymotrypsin. J. Biol. Chem. 259:11631–11634, 1984. Dahlbäck, B., Smith, C.A., Müller-Eberhardt, H.J. Visualization of human C4b-binding protein and its complexes with vitamin K-dependent protein S and complement protein C4b. Proc. Natl. Acad. Sci. USA 80:3461–3465, 1983. Perkins, S.J., Chung, L.B., Reid, K.B.M. Unusual ultrastructure of complement component C4b-binding protein of human complement by synchrotron X-ray scattering and hydrodynamic analysis. Biochem. J. 233:799–807, 1986. Cooper, N.R. Isolation and analysis of the mechanism of action of an inactivator of C4b in normal human serum. J. Exp. Med. 141:890–903, 1975. Shiraishi, S., Stroud, R.M. Cleavage product of C4b produced by enzymes in human serum. Immunochemistry 12:935–939, 1975. Barnum, S.R., Dahlbäck, B. C4b-binding protein, a regulatory component of the classical pathway of complement, is an acute-phase protein and is elevated in systemic lupus erythematosus. Compl. Inflamm. 7:71–77, 1990. Garcı́a de Frutos, P., Mohammed Alim, R.I., Härdig, Y., Zöller, B., Dahlbäck, B. Differential regulation of a- and b-chains of C4b-binding protein during acute-phase response resulting in stable plasma levels of free anticoagulant protein S. Blood 84:815–822, 1994. Dahlbäck, B., Stenflo, J. High molecular weight complex in human plasma between vitamin K-dependent protein S and complement component C4b-binding protein. Proc. Natl. Acad. Sci. USA 78:2512–2516, 1981. de Beer, F.C., Baltz, M.L., Holford, S., Feinstein, A., Pepys, M.B. Fibronectin and C4-binding protein are selectively bound by aggregated amyloid P component. J. Exp. Med. 154:1134–1139, 1981. Schwalbe, R.A., Dahlbäck, B., Hillarp, A., Nelsestuen, G.L. Independent association of serum amyloid P component, protein S, and complement C4b with complement C4bbinding protein and subsequent association of the complex with membranes. J. Biol. Chem. 265:21749–21757, 1990. Garcı́a de Frutos, P., Härdig, Y., Dahlbäck, B. Serum amyloid P component binding to C4b-binding protein. J. Biol. Chem. 270:26950–26955, 1995. Simmonds, R.E., Zöller, B., Ireland, H., et al. Genetic and phenotypic analysis of a large (122 member) protein S deficient knindred provides an explanation for the familial coexistence of type I and III plasma phenotypes. Blood 89:4364–4370, 1997. Koppelmann, S.J., van’t Veer, C., Sixma, J.J., Bouma, B.N. Synergistic inhibition of the intrinsic factor X activation by protein S and C4b-binding protein. Blood 86:2653–2660, 1995. Thern, A., Stenberg, L., Dahlbäck, B., Lindahl, G. Igbinding surface proteins of Streptococcus pyogenes also bind human C4b-binding protein (C4BP), a regulatory component of the complement system. J. Immunol. 154:375– 386, 1995. Johnsson, E., Thern, A., Dahlbäck, B., Hedén, L.-O., Wikström, M., Lindahl, G. A highly variable region in members of the streptococcal M protein family binds the human 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 403 complement regulator C4BP. J. Immunol. 157:3021–3029, 1996. Hessing, M., Vlooswijk, R.A., Hackeng, T.M., Kanters, D., Bouma, B. The localization of heparin-binding fragments on human C4b-binding protein. J. Immunol. 144:204–208, 1990. Garcı́a de Frutos, P., Dahlbäck, B. Interaction between serum amyloid component and C4b-binding protein associated with inhibition of factor I-mediated C4b degradation. J. Immunol. 152:2430–2437, 1994. He, X., Shen, L., Malmborg, A., Smith, K.J., Dahlbäck, B., Linse, S. Binding site for C4b-binding protein in vitamin K-dependent protein S fully contained in the carboxyterminal laminin-G-type repeats. A study using recombinant factor IX-protein S chimeras and surface plasmon resonance. Biochemistry 36:3745–3754, 1997. Hillarp, A., Dahlbäck, B. The protein S-binding site localized to the central core of C4b-binding protein. J. Biol. Chem. 262:11300–11307, 1987. Hillarp, A., Dahlbäck, B. Novel subunit in C4b-binding protein required for protein S binding. J. Biol. Chem. 263:12759–12764, 1988. Härdig, Y., Dahlbäck, B. The amino-terminal module of the C4b-binding protein b-chain contains the protein Sbinding site. J. Biol. Chem. 271:20861–20864, 1996. Chung, L.P., Reid, K.B.M. Structural and functional studies on C4b-binding protein, a regulatory component of the human complement system. Biosci. Rep. 5:855–865, 1985. Hessing, M., Kanters, D., Takeya, H., et al. The region Ser333–Arg356 of the a-chain of human C4b-binding protein is involved in the binding of complement C4b. FEBS Lett. 317:228–232, 1993. Ogata, R.T., Mathias, P., Bradt, B.M., Cooper, N.J. Murine C4b-binding protein. Mapping of the ligand-binding site and the N-terminus of the pre-protein. J. Immunol. 150: 2273–2280, 1993. Accardo, P., Sánchez-Corral, P., Criado, O., Garcı́a, E., Rodrı́guez de Córdoba, S. Binding of human complement component C4b-binding protein (C4BP) to Streptococcus pyogenes involves the C4b-binding site. J. Immunol. 157: 4935–4939, 1996. Härdig, Y., Hillarp, A., Dahlbäck, B. The amino-terminal module of the C4b-binding protein a-chain is crucial for the C4b binding and Factor I-cofactor function. Biochem. J. 323:469–475, 1997. Ziccardi, R.J., Dahlbäck, B., Müller-Eberhard, H.J. Characterization of the interaction of human C4b-binding protein (C4BP) with physiological ligands. J. Biol. Chem. 259: 13674–13679, 1984. Barlow, P.N., Steinkasserer, A., Norman, D.G., et al. Solution structure of a pair of complement modules by nuclear magnetic resonance. J. Mol. Biol. 232:268–284, 1993. Barlow, P.N., Norman, D.G., Steinkasserer, A., et al. Solution structure of the fifth repeat of factor H: A second example of the complement control protein module. Biochemistry 31:3626–3634, 1992. Norman, D.G., Barlow, P.N., Baron, M., Day, A.J., Sim, R.B., Campbell, I.D. Three-dimensional structure of a complement control protein module in solution. J. Mol. Biol. 219:717–725, 1991. Villoutreix, B.O., Fernández, J.A., Teleman, O., Griffin, J.H. Comparative modeling of the three SCR domains of the b-chain of C4b-binding and evaluation of potential sites of interaction with protein S. Protein Eng. 8:1253–1258, 1995. Garcı́a de Frutos, P., Dahlbäck, B. cDNA structure of rabbit C4b-binding protein a-chain: Preserved sequence motive in complement regulatory protein modules which bind C4b. Biochim. Biophys. Acta 1261:285–289, 1995. Hillarp, A., Wiklund, H., Thern, A., Dahlbäck, B. Molecular cloning of rat C4b-binding protein a- and b-chains. J. Immunol. 158:1315–1323, 1997. 404 B.O. VILLOUTREIX ET AL. 38. Fernández, J.A., Villoutreix, B.O., Hackeng, T.M., Griffin, J.H., Bouma, B.N. Analysis of protein S C4b-binding protein interactions by homology modeling and inhibitory antibodies. Biochemistry 33:11073–11078, 1994. 39. Abola, E.E., Bernstein, F.C., Bryant, S.H., Koetzle, T.F., Weng, J. Information content, software systems, scientific applications. In: ‘Crystallographic Databases.‘ Allen, F.H., Bergerhoff, G., Sievers, R., (eds.). Bonn: Data Commission of the International Union of Crystallography, 1987:107– 132. 40. Bernstein, F.C., Koetzle, T.F., Williams, G.J.B., et al. The Protein Data Bank: a computer-based archival file for macromolecular structures. J. Mol. Biol. 112:535–542, 1977. 41. Jones, T., Thirup, S. Using known substructures in protein model building and crystallography. EMBO J. 5:819–822, 1986. 42. Young, L., Jernigan, R.L., Covell, D.G. A role for hydrophobicity in protein–protein interactions. Protein Sci. 3:717– 729, 1994. 43. Levitt, M. A simplified representation of protein conformations for rapid simulation of protein folding. J. Mol. Biol. 104:59–107, 1976. 44. Covell, D., Jernigan, R. Conformations of folded proteins in restricted space. Biochemistry 29:3287–3294, 1990. 45. Dill, K.A. Dominant forces in protein folding. Biochemistry 29:7133–7155, 1990. 46. Covell, D.G. Lattice model simulations of polypeptide chain folding. J. Mol. Biol. 235:1032–1043, 1994. 47. Covell, D.G. Folding protein alpha-carbon chains into compact forms by Monte Carlo methods. Proteins 14:192– 204, 1992. 48. Christenson, H.K., Claesson, P.M. Cavitation and the interaction between macroscopic surfaces. Science 239:390– 392, 1988. 49. Miyazawa, S., Jernigan, R.L. Estimation of effective interresidue contact energies from protein crystal structures: Quasi-chemical approximation. Macromolecules 18:534– 552, 1985. 50. Nozaki, Y., Tanford, C. The solubility of amino acids and two glycine peptides in aqueous ethanol and dioxane solutions. J. Biol. Chem. 246:2211–2217, 1971. 51. Cornette, J.L., Cease, K.B., Margalit, H., Spouge, J.L., Berzofsky, J.A., DeLisi, C. Hydrophobicity scales and computational techniques for detecting amphipathic structures in proteins. J. Mol. Biol. 195:659–685, 1987. 52. Gilson, M., Sharp, K., Honig, B. Calculating the electrostatic potential for molecules in solution. J. Comp. Chem. 9:327–335, 1987. 53. Klapper, I., Hagstrom, R., Fine, R., Sharp, K., Honig, B. Focusing of electric fields in the active site of Cu–Zn superoxide dismutase: Effects of ionic strength and amino acid modification. Proteins 1:47–59, 1986. 54. Rost, B. PHD: Predicting one-dimensional protein structure by profile based neural networks. Methods Enzymol. 266:525–539, 1996. 55. Rost, B., Sander, C. Prediction of protein secondary structure at better than 70% accuracy. J. Mol. Biol. 232:584– 599, 1993. 56. Andersson, M., Hansson, A., Englund, G., Dahlbäck, B. Inhibition of complement components C3 and C4 by cadralazine and its active metabolite. Eur. J. Clin. Pharmacol. 40:261–265, 1991. 57. von Zabern, I., Gigli, I. Conformational changes in complement component C4 induced by activation, treatment with amines, chaotropes, or freezing–thawing, detectable by radioiodination using lactoperoxidase. J. Immunol. 128: 1439–1442, 1982. 58. von Zabern, I., Bloom, E.L., Chu, V., Gigli, I. The fourth component of human complement treated with amines or chaotropes or frozen–thawed (C4b-like C4): Interaction with C4-binding protein and cleavage by C3b/C4b inactivator. J. Immunol. 128:1433–1438, 1982. 59. Novotny, J., Bruccoleri, R., Karplus, M. An analysis of 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. incorrectly folded protein models. J. Mol. Biol. 177:787– 818, 1984. Lüthy, R., Bowie, J.U., Eisenberg, D. Assessment of protein models with three-dimensional profiles. Nature 356:83–85, 1992. DiScipio, R.G. Ultrastructures and interactions of complement factors H and I. J. Immunol. 149:2592–2599, 1992. Villoutreix, B.O., Garcı́a de Frutos, P., Lövenklev, M., Linse, S., Fernlund, P., Dahlbäck, B. The SHBG region of the anticoagulant cofactor protein S: Secondary-structure prediction, circular dichroism and analysis of naturally occuring mutations. Proteins 29:478–491, 1997. Fernández, J.A., Griffin, J.H. A protein S binding site on C4b-binding protein involves b-chain residues 31–45. J. Biol. Chem. 269:2535–2540, 1994. Hessing, M., Kanters, D., Heijnen, H.F.G., Hackeng, T.M., Sixma, J.J., Bouma, BN. Structure function studies on human C4b-binding protein using monoclonal antibodies. Eur. J. Immunol. 21:2077–2085, 1991. Mulloy, B., Forster, M.J., Jones, C., Davies, D.B. NMR and molecular modeling studies of the solution conformation of heparin. Biochem. J. 293:849–858, 1993. Hillarp, A., Thern, A, Dahlbäck, B. Bovine C4b-binding protein: Molecular cloning of the alpha- and beta-chains provides structural background for lack of complex formation with protein S. J. Immunol. 153:4109–4199, 1994. Hillarp, A., Dahlbäck, B. Cloning of cDNA coding for the b-chain of human complement component C4b-binding protein: Sequence homology with the a-chain. Proc. Natl. Acad. Sci. USA 87:1183–1187, 1990. Chung, L.P., Bentley, D.R., Reid, K.B.M. Molecular cloning and characterization of the cDNA coding for C4b-binding protrein, a regulatory protein of the classical pathway of the human complement system. Biochem. J. 230:133–141, 1985. Kristensen, T., Ogata, R.T., Chung, L.P., Reid, K.B.M., Tack, B.F. cDNA structure of murine C4b-binding protein, a regulatory component of the serum complement system. Biochemistry 26:4668–4674, 1987. Comp, P., Esmon, C. Recurrent venous thromboembolism in patients with a partial deficiency of protein S. N. Engl. J. Med. 311:1525–1528, 1984. Cooper, N.R. Complement evasion strategies of microorganisms. Immunol. Today 12:327–331, 1991. Taylor, F.B., Dahlbäck, B, Chang, A.C., et al. Role of free protein S and C4b-binding protein in regulating the coagulant response to Escherichia coli. Blood 86:2642–2652, 1995. Blundell, T.L., Sibanda, B.L., Sternberg, M.J.E., Thornton, J.M. Knowledge-based prediction of protein structures and design of novel molecule. Nature 326:347–352, 1987. Greer, J. Comparative modeling methods: Application to the family of the mammalian serine proteases. Proteins 7: 317–334, 1990. Sánchez, R., Sali, A. Advances in comparative protein structure modeling. Curr. Opin. Struct. Biol. 7:206–214, 1997. Rossi, V., Gaboriaud, C., Lacroix, M., et al. Structure of the catalytic region of human complement protease C1s: Study by chemical cross-linking and three-dimensional homology modeling. Biochemistry 34:7311–7321, 1995. Molina, H., Perkins, S.J., Guthridge, J., Gorka, J., Kinoshita, T., Holers, V.M. Characterization of a complement receptor 2 (CDR2, CD21) ligand-binding site for C3. J. Immunol. 154:5426–5435, 1995. Kuttner-Kondo, L., Medof, M.E., Brodbeck, W., Shoham, M. Molecular modeling and mechanism of action of human decay-accelerating factor. Protein Eng. 9:1143–1149, 1996. Mumenthaler C, Schneider U, Buchholz CJ, Koller D, Braun W, Cattaneo R. A 3D model for the measles virus INVESTIGATION OF C4BP BY MOLECULAR MODELING receptor CD46 based on homology modeling, Monte Carlo simulations and hemagglutinin binding studies. Protein Sci. 6:588–597, 1997. 80. Honig, B., Nicholls, A. Classical electrostatics in biology and chemistry. Science 268:1144–1149, 1995. 81. Villoutreix, B.O., Spassov, V.Z., Atanasov, B.P., Herve, G., Ladjimi, M.M. Structural modeling and electrostatic properties of aspartate transcarbamylase from Saccharomyces cerevisiae. Proteins 19:230–243, 1994. 82. Karshikoff, A., Bode, W., Tulinsky, A., Stone, S.R. Electrostatic interactions in the association of proteins: An analy- 405 sis of thrombin–hirudin complex. Protein Sci. 1:727–735, 1992. 83. Krych, M., Hourcade, D., Atkinson, J.P. Sites within the complement C3b/C4b receptor important for the specificity of ligand binding. Proc. Natl. Acad. Sci. USA 88:4353–4357, 1991. 84. Rodrı́guez de Cordoba, S., Pérez-Blas, M., Ramos-Ruiz, P., Sánchez-Corral, P., Pardo-Manuel de Villena, F., ReyCampos J. The gene coding for the beta-chain of C4bbinding protein (C4BPB) has become a pseudogene in the mouse. Genomics 21:501–509, 1994.