Each of the Three Binding Sites on Complement Factor H Interacts with a Distinct Site on C3b*

Factor H (fH) restricts activation of the alternative pathway of complement at the level of C3, both in the fluid phase and on self-structures, but allows the activation to proceed on foreign structures. To study the interactions between fH and C3b we used surface plasmon resonance analysis (Biacore®) and eight recombinantly expressed fH constructs containing fragments of the 20 short consensus repeat domains (SCRs) of fH. We analyzed the binding of these constructs to C3b and its cleavage products C3c and C3d. Three binding sites for C3b were found on fH. Site 1 was localized to the five amino-terminal SCRs (SCR1–5), and its reciprocal binding site on C3b was found to be lost upon the cleavage of C3b to C3c and C3d. Site 2 on fH was localized by exclu-sion probably within or near SCRs 12–14 (fragment SCR8–20 bound to C3b, C3c, and C3d; SCR8–11 did not bind to C3b at all; and SCR15–20 bound only to

The main functions of the complement system, a set of plasma and membrane-bound proteins, are to protect the human body against invading organisms, to remove debris from plasma and tissues, and to enhance cell-mediated immune responses. The system can be activated through three major pathways: the alternative, the classical, and the lectin pathway. The activation of the alternative pathway (AP) 1 is initiated spontaneously and independently of antibodies or specific target structures needed for the classical and the lectin path-ways. Consequently, AP attacks biological structures that are not specifically protected against activation (1,2).
AP is continuously activated at a low rate in human plasma. Plasma protein C3, the key component of AP, undergoes slow, spontaneous hydrolysis to C3(H 2 O) (3). This provides a subunit for the initial C3 convertase of AP, C3(H 2 O)Bb, which cleaves fluid phase C3, generating metastable C3b that can attach covalently to hydroxyl or amine groups on surfaces. After the binding of factor B (fB) and the cleavage of C3b-bound fB by factor D (fD), the actual C3 convertase of AP, C3bBb, is formed. This leads to a progressively increasing deposition of C3b through the formation of new metastable C3b molecules and new C3 convertases (4,5). This amplification cascade leads to an effective opsonization of foreign structures with C3b and to the simultaneous formation of complement membrane attack complexes through the propagation of the terminal complement cascade (6,7).
AP activation is regulated in its early step, at the level of C3b and C3bBb, by plasma protein factor H (fH) and by three membrane-bound molecules (CD35, i.e. complement receptor 1; CD46, i.e. membrane cofactor protein; and CD55, i.e. decayaccelerating factor) (8). fH regulates AP activation by competing with fB for binding to C3b, by enhancing the dissociation of the C3bBb complex, and by acting as a cofactor for factor I, which leads to the irreversible proteolytic inactivation of C3b (9 -11). fH is a crucial regulator of AP in plasma because it limits the number of functionally active C3b molecules. In addition, fH is practically the only regulator involved in the protection of self-structures that lack the membrane-bound regulators (e.g. basement membranes in kidney glomeruli) (12)(13)(14). This is exemplified by rare cases of fH deficiency in human beings (15)(16)(17)(18) and pigs (19,20) that lead to complement-dependent kidney damage. The ability of AP to discriminate between activating and nonactivating structures depends on the differential binding of fH to C3b on different types of surfaces. The affinity of fH is relatively high for C3b molecules deposited on self-structures ("nonactivators"), whereas C3b on foreign targets ("activators") has a relatively low affinity for fH, a fact that leads to opsonization and subsequent target damage (5,21,22). Because the non-activating surfaces usually contain negatively charged surface structures (sialic acids and glycosaminoglycans), interactions between fH and polyanions have been intensively characterized. fH has been found to contain two or three binding sites for heparin and/or sialic acid on SCR7, SCR20, and possibly SCR12-14 (23)(24)(25)(26)(27)(28). It has been suggested that these heparin-binding sites have a role in the ability of AP to discriminate between activators and nonactivators; however, the physiological role of these sites is not yet fully understood.
To explain the ability of AP to discriminate between self-and nonself-structures, it is essential to characterize the interaction sites between fH and surface-bound C3b. Two sites on C3b are known to be involved in the binding of fH, one on the proteolytic fragment C3d (C3 residues 1178 -1252) (29) and another at the amino terminus of the ␣ chain (C3 residues 727-768) (30 -32). It has been suggested that a total of three C3b-binding sites exist on fH. One has been located to the amino-terminal tryptic fragment (SCR1-6a) (33,34) and further mapped by functional assays to . The other two binding sites have not yet been well characterized. Results obtained by using either several anti-fH mAbs (38) or deletion mutants lacking 5-10 SCR domains (39) showed that there is more than one binding site on fH for C3b. One of the two new sites was localized to , and it has been suggested that the other new site lies within SCRs 6 -10 (39). So far the locations of the reciprocal binding sites on C3b have not been addressed or identified.
Because of the central role of AP activation, we analyzed the interactions between C3b and fH. We employed the surface plasmon resonance method for a detailed analysis of the binding sites. This assay monitors real-time changes in the refractive index in the fluid phase close (ϳ300 nm) to the sensor chip surface (40,41). Surface plasmon resonance has been successfully used for the detection and characterization of interactions between two label-free biomolecules (42,43).
Our specific goals in this study were (i) to verify the existence of the suggested three binding sites for C3b on fH, (ii) to map the location of the binding sites within the fH protein, and (iii) to identify the reciprocal binding sites on C3b. Our results show that fH has at least three binding sites for C3b. Site 1 consists of SCRs 1-5, site 2 is probably within or near SCRs 12-14, and site 3 is within SCRs 19 -20. We also localize the corresponding interaction sites to distinct domains of the C3 molecule. fH site 1 binds specifically to intact C3b and is destroyed by the cleavage of C3b to C3c and C3d, site 2 binds to C3c, and site 3 interacts with the C3d region. Multiple reciprocal binding sites explain the complexity of the interactions between these two central complement components. Furthermore, multiple reciprocal binding sites suggest mechanisms that might explain why the C3b-fH interaction is uniquely sensitive to the nature of the surface to which C3b is bound.

EXPERIMENTAL PROCEDURES
Reagents and Complement Components-Trypsin treated with tosylamide phenylethyl chloromethyl ketone and soybean trypsin-chymotrypsin inhibitor were obtained from Sigma. Soybean trypsin-chymotrypsin inhibitor was used at the ratio 1:3 (w/w) of trypsin to the inhibitor. C3, fB, fD, and fH were purified from human plasma by the methods previously described (44,45). Purity of the proteins was examined by SDS-PAGE and was found to be Ͼ 90%.
Purification of C3 Fragments-C3b was prepared by purified C3, factor B, factor D, and NiCl 2 as described earlier (45,46). C3c was isolated using factor I and its cofactors fH and soluble complement receptor 1 (CR1). Briefly, C3b (1 mg, 1.0 mg/ml) was mixed with factor H (10 g), factor I (1.0 g), and soluble CR1 (5.0 g). The mixture was incubated for 8 h at 37°C and 15 h at 22°C with continuous slow shaking. The supernatant was subjected to anion exchange chromatography on a Mono-Q column (Amersham Pharmacia Biotech), and the fractions containing C3c were pooled (50 -400 mM NaCl gradient). Thereafter, the C3c preparation was subjected to affinity chromatography using previously described C3d-specific mAb 4C2 (45) to eliminate all the C3b fragments that contained parts of the C3d or C3dg regions. Analysis of the C3c preparation by SDS-PAGE under reducing conditions showed three bands with apparent molecular masses of 75, 27, and 40 kDa, as expected. No reactivity of the C3c preparation was seen by using anti-C3d mAb 4C2 in the Western blot analysis.
C3d was prepared from purified human C3 by a modification of the method of Eggertsen et al. (47). Briefly, C3 at a concentration of 4 mg/ml in veronal-buffered saline (VBS: 5.0 mM barbital and 146 mM NaCl, pH 7.35) was treated with tosylamide phenylethyl chloromethyl ketone-treated trypsin for 2 h at 37°C at a molar enzyme/substrate ratio of 1:60. The reaction was stopped by addition of soybean trypsin-chymotrypsin inhibitor. C3d was isolated by eluting the trypsinized C3 from a Mono-Q column (Amersham Pharmacia Biotech) with a 0.1-0.4 M NaCl gradient in 25 mM Tris-HCl, pH 8.3. In accordance with earlier findings (47), C3d was eluted in an asymmetrical peak shortly before the major peak. In SDS-PAGE analysis under reducing conditions there was one major band and two minor bands of C3d with apparent molecular masses of 30 -35 kDa. The identity of all three bands as C3d was confirmed by their reactivity with the C3d-specific mAb 4C2 in Western blot analysis.
Recombinant Fragments of fH-Cloning, expression in the baculovirus system, and purification of most of the used recombinant fragments of fH have been described earlier (27,36,50). The additional fragments representing SCR8 -11, SCR15-18, and SCR15-20 were generated using the pBSV-8His expression vector (50). Preparation of the Biacore® Instrumentation-Real-time monitored surface plasmon resonance assays were performed using a Biacore®-2000 instrument, and the data were analyzed using the BiaEvaluation 3.0 software (Biacore AB, Uppsala, Sweden). Carboxylated dextran chips (sensor chip CM5, research grade from Biacore AB) were used in all the assays. Flow cells of the CM5 chips were used either for a standard amine coupling procedure or prepared for the direct enzymatic coupling of C3b by using a standard activation-deactivation procedure without adding any protein between the steps. The activation step was performed with fresh solution containing N-hydroxysuccinimide and N-ethyl-NЈ-(dimethylaminopropyl)-carbodiimide (Biacore AB, 7-15-min injection at a flow rate of 5 l/min) and was followed by deactivation with ethanolamine-HCl (1.0 M at pH 8.5) (Biacore AB, 7-15-min injection). Hepes-buffered saline (Biagrade, Biacore AB) or 1 ⁄3 VBS was used as the flow buffer throughout. After these initial steps VBS or 1 ⁄3 VBS was used as the continuous flow buffer at 5 l/min; only degassed buffers were used.
Amine Coupling of Proteins onto the Biacore® Chip-C3b, C3c, and C3d were coupled onto the CM5 chip using the standard amine coupling procedure as recommended by the manufacturer. The proteins to be coupled were dialyzed against 10 mM acetate buffer (pH 5.0 -5.7) to achieve a negative net charge for the amine coupling. Briefly, the chip surface was activated with N-ethyl-NЈ-(dimethylaminopropyl)-carbodiimide (7-15-min injection, 5 l/min), and either purified C3b (50 g/ml, 20 l), C3c (70 g/ml, 30 l), or C3d (130 g/ml, 20 l) was injected to reach an appropriate level of coupling for the binding experiments, i.e. 5,000 -20,000 resonance units (RU). Afterward, the flow cells were deactivated as described above. Before the experiments, the flow cells were washed thoroughly with 1 ⁄3 VBS and 3 M NaCl in 10 mM acetate buffer, pH 4.6.
Enzymatic Coupling of C3b onto the Biacore® Chip-To exploit the inherent ability of C3b to bind to surfaces via a thiol group, C3b was coupled to the sensor chip surface by means of an enzymatic coupling procedure. Throughout the enzymatic coupling VBS-Ni 2ϩ (1 mM NiCl 2 ) was used as the continuous flow buffer (5 l/min flow rate), and the temperature of the chip was adjusted to 30°C. First, 16.7 g of freshly purified C3, 2.3 g of fB, and 0.05 g of fD were mixed in 50 l of VBS-Ni 2ϩ at 0°C and injected into a flow cell. After 2-5 min 50 l of VBS-Ni 2ϩ containing 2.3 g of fB and 0.05 g of fD was injected (fB ϩ fD injection) followed immediately by 50 l of VBS-Ni 2ϩ containing 16.7 g of C3 (C3 injection). Thereafter, fB ϩ fD and C3 injections were sequentially repeated five times. Because the whole procedure was performed using real-time monitoring, the amount of injected C3 (100 g/ml) could be adjusted so as to achieve a final increase of the resonance signal between 5,000 and 10,000 RUs during the last cycle. Usually a total of 60 -135 g of C3 was injected during the procedure. After C3b coupling the flow buffer was changed to 1 ⁄3 VBS, and the flow cells were washed for about 2 h with 1 ⁄3 VBS and 2-3 M NaCl in acetate buffer, pH 4.6, before performing the binding assays to allow fB and noncovalently bound C3b or C3 to detach from the sensor chip surface.

Expression of the New Recombinant Fragments of fH-
In addition to the previously described recombinant fragments of fH, we generated new mutant proteins spanning the middle and carboxyl-terminal parts of fH (constructs containing [15][16][17][18][15][16][17][18][19][20] as tools for the fH-C3b interaction studies. All fragments were purified by Ni 2ϩ chelate chromatography. In SDS-PAGE the apparent molecular masses of the fragments were 28 kDa for SCR8 -11 and SCR15-18 and 41 kDa for SCR15-20 (Fig. 1A). In Western blotting analysis, polyclonal anti-fH antibodies bound to all the recombinant fragments. The fragments SCR15-18 and SCR15-20 were found to form some homodimers of ϳ56 and 82 kDa, respectively (Fig. 1B).
Coupling of C3b onto the Biacore® Sensor Chip Surface-To analyze the binding of the different recombinant fragments of fH to fluid phase and surface-bound C3b in the Biacore® instrument, two methods were used to couple C3b onto the sensor chip surface. Firstly, a highly efficient enzymatic method was used, exploiting the ability of nascent C3b to generate a physiological thioester bond. A curve indicating the enzymatic coupling of C3b to hydroxyl groups on a CM5 sensor chip is shown in Fig. 2. Secondly, a standard protocol for the amine coupling of preformed C3b to the CM5 sensor chip was used. In this approach the amount of coupled C3b was found to depend on the activation efficiency and the amount of C3b injected, as expected.
Binding of the Recombinant Fragments of fH to C3b-A total of eight different recombinant mutant fragments of fH were analyzed for their C3b binding capacity by means of the surface plasmon resonance technique. A schematic representation of the fragments is given in Fig. 3. As a positive control we used fH purified from human plasma. Association of the fluid phase ligand was detected by an increase in the number of RUs as a function of time during the injection. Dissociation of the protein complex was followed after the injection. The resonance curves obtained by injecting the ligand into the C3b-coupled flow cell and into the control flow cell were superimposed. The result was considered negative if no increase in the resonance signal was observed in two separate assays using a ligand concentration higher than 80 g/ml.
One of the fH-binding sites for C3b is located in the four most amino-terminal SCR domains (51,52). We used two recombinant fragments of these amino-terminal SCRs to verify the C3b binding with the Biacore® equipment. Fragment SCR1-6 bound to C3b, but fragment SCR2-6 did not (Fig. 4).
We also analyzed the binding of five recombinant fragments representing the middle and carboxyl-terminal regions of fH. Fragments SCR8 -20, SCR15-20, and SCR19 -20 bound to C3b, whereas fragments SCR8 -11 and SCR15-18 did not (Fig.  4). Thus the carboxyl-terminal SCRs 19 -20 contain one binding site for C3b. Binding of the fH fragments was specific, because none of them bound to the control flow cells (blank channels) (Fig. 4).
Binding of the fH Constructs to the C3b Fragments C3c and C3d-To map the reciprocal binding sites on C3b the proteolytic fragments C3c and C3d were generated. The generated C3c was found to be free of fragments containing C3d, and the C3d was found to be free of fragments containing C3c as determined by SDS-PAGE and Western blotting (Fig. 5). C3b and its cleavage products C3c and C3d were coupled to the sensor chip surface via amine coupling, and the binding of the various recombinant fH fragments was analyzed. The amounts of C3c and C3d bound to the chip surface were ϳ15.9 and ϳ12.8 ng/mm 2 , respectively. fH bound to C3b, C3c, and C3d, whereas the fH constructs containing the amino-terminal C3b-binding site (SCR1-5) bound specifically to intact C3b but not to C3c or C3d (Fig. 6). This indicates that the binding site for the fH SCRs 1-5 was lost on the C3b molecule upon its cleavage to C3c and C3d. The fH constructs containing the carboxyl-terminal C3b-binding site (SCR15-20 and SCR19 -20) bound to C3b and C3d but not to C3c. Because SCR8 -20 was found to bind to C3b, C3d, and C3c, it is evident that this fH construct contains two C3b-binding sites, one located at the carboxyl terminus and the second in the middle part of the protein. Because the fragment SCR8 -11 did not bind to C3b or to its fragments and because SCR15-20 did not bind to C3c, this additional C3bbinding site (the C3c-binding site) is likely to be located within or near the SCR domains 12-14.

DISCUSSION
In this study we have mapped three distinct binding sites for C3b on the complement regulator fH. These sites were localized to the amino-terminal SCRs 1-5 (site 1), to the middle region within or near SCRs 12-14 (site 2), and to the carboxyl-terminal SCRs 19 -20 (site 3). Importantly, we observed that each of these fH-binding sites interacted with a distinct site on C3b. Site 1 bound to intact C3b but not to its cleavage fragments C3c or C3d, site 2 bound to C3b and C3c, and site 3 bound to the C3b and C3d fragments. A schematic representation of the fH-C3b binding results is shown in Fig. 7.
The first of the two main issues of the current study was to analyze the binding sites on fH for C3b. Our results about the location of the three binding sites on fH are, in general, in agreement with previous reports. The amino-terminal binding site of fH (site 1) was originally located to a reduced and alkylated tryptic fragment spanning SCRs 1-6a (the trypsin cleavage site is in the middle of SCR 6) (33). In the same study the reduced carboxyl-terminal fragment (SCR6b-20) did not bind to C3b in an enzyme-linked immunosorbent assay. The binding of intact fH to C3b was dramatically reduced in the presence of the monoclonal antibody OX24, which was known to bind to the amino-terminal tryptic fragment. This result gave an impression that, by binding to the amino-terminal tryptic fragment, OX24 blocked the only binding site for C3b on fH (34,53). Later, the amino-terminal binding site was studied using recombinant fH fragments in an enzyme-linked immunosorbent assay, and SCR1-4 was the shortest fragment that was found to display decay-accelerating activity (37). By means of a combination of several anti-fH mAbs mapped to the recombinant fH fragments, an additional C3b-binding site was mapped on fH (38). In agreement with these results, the existence of multiple distinct binding sites was next demonstrated using deletion mutants of fH. Because fragments consisting of or representing SCR1-10 or SCR11-20 bound to fH, it was concluded that there are at least two binding sites for C3b on fH (39). By comparing the binding efficiencies of the various deletion mutants, Sharma and Pangburn (39) suggested a total of three binding sites, which were broadly located to SCRs 1-5,  6 -10, and 15-20. In the present study we confirmed the existence of three distinct binding sites for C3b on fH and mapped one of these sites more precisely. In terms of the location of the amino-terminal (site 1) and carboxyl-terminal (site 3) binding sites, our study fully agrees with the previous ones. In addition, it provides strong evidence for the existence of a binding site in the middle part of fH. The location of this middle binding site for C3b on fH (site 2) has not yet been exactly determined. However, in contrast to Sharma and Pangburn (39), we localize this site within or near SCRs 12-14, not within the SCRs 6 -10. There are at least two possible ways to explain this difference. First, on the basis of current knowledge one cannot exclude the existence of a fourth binding site for C3b on fH, for example within SCRs 7-8. Second, localization of the C3b/C3c binding region within or near SCRs 12-14 in our study was made by exclusion, on the basis of the observations that the recombinant fragments SCRs 8 -11 and 15-18 did not bind to C3c or C3b. Thus the possibility remains that, for example, a hypothetical C3b/C3c-binding site on SCRs 9 -10 would not be exposed in the fragment SCR8 -11, because this is only four SCRs long, but would be exposed in the fragment SCR8 -20.
The second of the two main issues of the current study was to analyze the binding sites on C3b for fH. Previously, two binding sites for fH on C3b have been described. The first evidence for a second site on C3b interacting with fH was obtained by inhibiting the binding of fH to C3b and C3d with a peptide spanning the residues 1187-1214 in the C3d part of C3b. This peptide inhibited the binding of fH to C3d totally but inhibited the fH-C3b interaction only partially (29). Based on these experiments and on the finding that fH binds to C3dg and to its carboxyl-terminal cleavage fragment (spanning the residues 1178 -1252 of C3), one fH-binding site is known to be contained in the C3d part of C3b (29). Until now, existing data about the other binding sites for fH on C3b have been interpreted assuming that there are only two binding sites for fH on C3b. In one study fH binding to C3c was not observed (29), but subsequently C3c was found to inhibit fH from binding to C3b (30). However, the simultaneous inhibition by C3c and C3dg did not totally abolish the binding of fH to C3b (30), a result that is in agreement with the finding in the current report (there are three sites for interaction between C3b and fH).
It has been previously shown that one of the binding sites on C3b for fH is located in the amino terminus of the C3b ␣Ј chain. This is supported by the following two results. First, in peptide inhibition studies a peptide spanning the residues 726 -768, and antibodies raised against this peptide, were found to inhibit fH from binding to C3b (30). Second, in mutagenesis studies the cofactor activity of fH was reduced by deletion of the residues 727-768 (32), and partial inhibition of several fH functions on C3b was achieved by site-directed mutagenesis of 730D731E and 736E737E (31). In addition, antibodies against C3b and/or C3c have been shown to inhibit fH binding to C3b (reviewed in Ref. 54). The amino-terminal binding site on the C3b ␣Ј chain for fH is part of the C3c fragment and not included in the C3d part of C3b. Because we observed that the carboxylterminal site on fH for C3b (SCRs 19 -20) bound only to the C3b and C3d fragments, it is probable that the binding site on fH corresponding to the site on the amino terminus of the C3b ␣Ј chain is either the fH site 1 on SCRs 1-4 or the site 2 on SCRs 8 -15.
Our results suggest that the binding sites on C3b corresponding to the amino-terminal and the carboxyl-terminal binding sites of fH are accessible both when C3b has been bound on the sensor chip surface through the enzymatic cascade (surface-bound conformation) and when C3b has been bound through amine coupling (preformed fluid phase conformation). Thus it is probable that the surface expression of these binding sites is independent of the coupling method and of the sensor chip surface. The interaction between fH and C3b seems to be unique in many ways. It is rare that two molecules of this size have multiple points of interaction, i.e. at least three different and independent binding sites, and that this interaction is influenced by the surface to which the other protein has become covalently bound. The identification of three distinct sites for interaction between C3b and fH may help us understand the recognition mechanism of AP, i.e. why fH restricts AP activation more efficiently on the nonactivating surfaces than on the activating surfaces. It is possible that not all C3b sites are exposed on activator surfaces or that additional sites on fH interact directly with the surface. Although Biacore® is a powerful tool in analyzing interactions between soluble molecules, it is not suitable for analysis of the interaction between fH and cell-surface-bound C3b. In addition, kinetic analyses using the purified proteins in Biacore® do not help to evaluate the role and importance of each of the three binding sites in target discrimination by fH and C3b. Thus, further studies are needed to determine how AP discrimination operates at the molecular level.
In conclusion, this work describes three distinct reciprocal binding sites between fH and C3b and provides information that is important for further functional studies and for the possible development of therapeutic C regulators.