Kinetic Analysis of the Interactions between Vaccinia Virus Complement Control Protein and Human Complement Proteins C3b and C4b

Abstract

The vaccinia virus complement control protein (VCP) is an immune evasion protein of vaccinia virus. Previously, VCP has been shown to bind and support inactivation of host complement proteins C3b and C4b and to protect the vaccinia virions from antibody-dependent complement-enhanced neutralization. However, the molecular mechanisms involved in the interaction of VCP with its target proteins C3b and C4b have not yet been elucidated. We have utilized surface plasmon resonance technology to study the interaction of VCP with C3b and C4b. We measured the kinetics of binding of the viral protein to its target proteins and compared it with human complement regulators factor H and sCR1, assessed the influence of immobilization of ligand on the binding kinetics, examined the effect of ionic contacts on these interactions, and sublocalized the binding site on C3b and C4b. Our results indicate that (i) the orientation of the ligand is important for accurate determination of the binding constants, as well as the mechanism of binding; (ii) in contrast to factor H and sCR1, the binding of VCP to C3b and C4b follows a simple 1:1 binding model and does not involve multiple-site interactions as predicted earlier; (iii) VCP has a 4.6-fold higher affinity for C4b than that for C3b, which is also reflected in its factor I cofactor activity; (iv) ionic interactions are important for VCP-C3b and VCP-C4b complex formation; (v) VCP does not bind simultaneously to C3b and C4b; and (vi) the binding site of VCP on C3b and C4b is located in the C3dg and C4c regions, respectively.

The complement system is an ancient and integral component of the host's immunological defense, evolved to combat against all of the invading pathogens, including viruses (9, 27, 38). Recognition of viruses by either of the three complement activation pathways (classical, lectin, or alternative) leads to the formation of C3 convertases, which cleave the central complement component C3 into an anaphylatoxic peptide C3a and an opsonic fragment C3b. Once the newly formed C3b molecules are deposited on the viral surfaces, they are considered as non-self by the complement system and result in further activation of the pathways and neutralization of viruses through various mechanisms (5, 9). However, during the long coexistence of viruses with the host immune system, certain viruses have developed ingenious strategies to evade the complement system. Important examples include the poxviruses, herpesviruses, retroviruses, paramyxoviruses, and picornaviruses (2, 13, 31, 33, 37, 46, 52).

Vaccinia virus, a cytoplasmic double-stranded DNA virus, is the most extensively characterized member of the Orthopoxvirus genus (18, 36). It is known to have developed two distinct mechanisms to subvert the host complement system: (i) it encodes a 27-kDa secretory protein, the vaccinia virus complement control protein (VCP), which is homologous to human complement control proteins and is a potent inhibitor of complement (23, 32, 42), and (ii) it incorporates the host complement control proteins (CD46, CD55, and CD59) into the outer envelope of extracellular enveloped virus (EEV), which renders resistance to EEV against complement (53).

VCP is one of the first documented viral immune evasion proteins. It is encoded by the C21L gene and is secreted by the cells infected with vaccinia virus. The primary structure of VCP consists of 263 amino acids, with a 19-amino-acid signal peptide sequence. It consists of four tandem copies of short consensus repeats (SCRs) or complement control protein (CCP) modules (24), a characteristic structure of host CCPs, and bears 26 to 38% sequence similarity to human CCPs. The nuclear magnetic resonance structure of pairs of VCP modules and the crystal structure of the entire VCP molecule have been determined (15, 39). The structure shows that each SCR folds into a compact 6β-strand structure, which is similar to the SCRs of human CCPs. Further, it also showed that the molecule has an extended structure from SCR 1 to 3 and a turn between SCR 1 to 3 and SCR 4. The most striking feature revealed in the crystal structure was the charge distribution of the four SCR domains: the SCR domains 1 and 4 carry a positive field around them, whereas the middle two domains are predominantly surrounded by a negative field (39).

Functional studies have shown that VCP is a potent inhibitor of complement activation (23, 32, 42). It is known to bind to complement proteins C3b and C4b and inhibit both the classical and the alternative pathways of complement activation (32, 42). Studies on the mechanism of complement inactivation by VCP have revealed that it acts as a cofactor for factor I-mediated cleavages of C3b and C4b (42), inhibits the formation of the classical/lectin and alternative pathway C3 convertases (32), and accelerates the decay of these C3 convertases into their subunits (32). Interestingly, VCP also contains two putative heparin-binding sites; the first site overlaps between SCR 1 and SCR 2, and the second site is located in SCR 4 (39). The physiological significance of these heparin-binding sites with respect to complement activation is not clear at present. In addition to in vitro studies on the mechanism of complement inhibition, the role of VCP has also been examined in vivo in experimental animals. It has been shown that recombinant vaccinia virus that does not express VCP is attenuated in vivo (17).

Although it is clear that VCP inhibits complement activation by interacting with complement proteins C3b and C4b, the nature of these interactions and the binding mechanisms involved are not well understood. Earlier studies have shown that VCP requires all four SCRs for binding to C3b (41, 50), and the crystal structure data suggest the possibility of multiple binding sites for C3b and C4b in VCP (39). Thus, whether VCP binds to C3b at a single or at multiple sites is not clear. Furthermore, there is a general belief that the binding of factor I cofactors to C3b and C4b causes conformational changes in these molecules and facilitates the binding of factor I (6, 51). Whether VCP-C3b/C4b complexes undergo structural reorientation is not known.

In the present study, we have utilized surface plasmon resonance (SPR) technology to decipher the mechanism of molecular recognition between VCP and its target proteins C3b and C4b and compared it to human CCPs. Our data show that, unlike host CCPs, interactions of VCP with C3b and C4b follow a simple 1:1 binding mechanism and that these complex formations do not involve conformational changes or multiple-site binding. These interactions, however, are highly dependent on ionic strength, a feature very similar to human CCP-C3b/C4b interactions. We also present data on the sublocalization of VCP binding sites on C3b and C4b. Our data show that these sites are located in the C3dg and C4c regions of C3b and C4b, respectively. A significant body of knowledge suggests that C3dg acts as a “bridge” between innate and acquired immunity (7, 29, 43). Thus, it is possible that, apart from inhibiting the complement system, VCP may also inhibit the acquired immunity by blocking the C3dg-CD21 interaction. To date, VCP is the only viral protein known to interact with C3dg.

MATERIALS AND METHODS

Expression and purification of recombinant VCP.

The vaccinia virus complement control protein cloned in Pichia pastoris was expressed and purified as described earlier with minor modifications (42). In brief, a single colony of recombinant P. pastoris expressing VCP was inoculated in 10 ml of BMGY medium (100 mM potassium phosphate [pH 6.0], 10 g of yeast extract/liter, 20 g of peptone/liter, 13.4 g of yeast nitrogen base/liter, 0.4 mg of biotin/liter, and 1% glycerol) and incubated overnight at 30°C in a shaking incubator. This inoculum was added to 1 liter of BMGY, followed by incubation for 48 h at 30°C with shaking. The cells were centrifuged, resuspended in 100 ml of BMMY (BMGY containing 0.5% methanol but without 1% glycerol), followed by incubation at 30°C for an additional 24 h with vigorous shaking. After incubation, cells were pelleted and the supernatant containing VCP was collected for purification.

The culture supernatant was sequentially precipitated with 20 and 60% ammonium sulfate at 0°C. The 60% pellet was suspended and dialyzed against phosphate-buffered saline (PBS; 10 mM sodium phosphate [pH 7.4] and 145 mM NaCl) and loaded onto a DEAE-Sephacel column (Sigma, St. Louis, Mo.) preequilibrated with 10 mM sodium phosphate buffer (pH 7.4) containing 250 mM NaCl. The flowthrough was collected, and buffer exchange was performed by using PD-10 desalting columns (Amersham Pharmacia Biotech, Uppsala, Sweden) preequilibrated with 5 mM sodium acetate (pH 4.0). The sample was then loaded onto Mono S 5/5 column (Amersham Pharmacia Biotech). The bound proteins were eluted with a linear salt gradient of 0 to 1.0 M NaCl. The eluted fractions were analyzed by sodium dodecyl sulfate-11% polyacrylamide gel electrophoresis (SDS-11% PAGE), and the fractions containing VCP were pooled, dialyzed into PBS, and concentrated by ultrafiltration.

Complement proteins and cleavage products.

The human complement proteins C3, factor H and factor I, were kindly provided by Michael K. Pangburn (University of Texas Health Center, Tyler, Tex.) and the recombinant human soluble CR1 (sCR1) was a generous gift of Henry Marsh (AVANT Immunotherapeutics, Inc., Needham, Mass.). C3b was generated by limited trypsin cleavage and purified on a Mono Q 5/5 column (Amersham Pharmacia Biotech) (42, 44). C4b was purified as follows. Twenty parts of human plasma were mixed with one part inhibitor solution (1 M KH₂PO₄, 0.2 M Na₄EDTA, 0.2 M benzamidine, and 1 mM phenylmethylsulfonyl fluoride) and precipitated first with 4.5% polyethylene glycol and then with 12% polyethylene glycol at 0°C. The 12% pellet was dissolved in 3.2 mM sodium phosphate (pH 7.4) containing 6.4 mM EDTA, 31.8 mM ɛ amino caproic acid, 6.4 mM benzamidine hydrochloride, and 0.02% sodium azide and then loaded onto a Source Q column (1.2 by 9.5 cm; Amersham Pharmacia Biotech). The bound proteins were eluted with a linear gradient of 0 to 0.5 M NaCl. Fractions were spiked with 1 mM PEFA-block (Roche, Mannheim, Germany), and C4b-containing fractions (generated during purification), identified by SDS-PAGE and immunodiffusion, were pooled and loaded onto a Mono Q 5/5 column (Pharmacia) in 10 mM sodium phosphate (pH 7.9). Bound proteins were eluted with a linear salt gradient of 0 to 0.5 M NaCl and subjected to SDS-PAGE analysis. Homogeneous C4b fractions were pooled and dialyzed into PBS. Experiments were also performed with C4b purchased from Calbiochem (La Jolla, Calif.).

The cleavage products of C3b and C4b were generated as described below. iC3b was generated by incubating 1 mg of trypsin-generated C3b with 166 μg of factor H and 16 μg of factor I in 1.2 ml of PBS (pH 7.4) at 37°C for 3 h; C3c and C3dg were generated by incubating 1 mg of trypsin-generated C3b with 166 μg of sCR1 and 16 μg of factor I in 1.1 ml of PBS (pH 7.4) at 37°C for 3 h; and C4c and C4d were generated by incubating 1 mg of C4b with 25 μg of sCR1 and 8 μg of factor I in 1.0 ml of PBS (pH 7.4) at 37°C for 2 h. The cleavage products of C3b, as well as of C4b, were purified on a Mono Q column as previously described (21, 42). The purity of all of the proteins exceeded 95%, as determined by SDS-PAGE and immunodiffusion analysis.

Site-specific biotinylation of C3b and C4b.

C3b and C4b contain a free −SH group, which is generated as a result of hydrolysis of the thioester bond present in these proteins (28). The free −SH groups in these proteins were labeled by using EZ-Link PEO-maleimide-activated biotin (Pierce, Rockford, Ill.) (45). Briefly, 0.5 mg of C3b or C4b in PBS was dialyzed into 0.1 M sodium phosphate (pH 6.0) containing 5 mM EDTA for 16 h at 4°C. EZ-Link PEO-maleimide-activated biotin was dissolved in 0.1 M sodium phosphate-150 mM NaCl-1 mM EDTA (pH 7.2) at a concentration of 10 mM and then mixed and incubated with C3b or C4b at a 1:70 molar ratio for 30 min at room temperature. Free biotin was removed by passing the reaction mixture through PD-10 column, followed by dialysis for 48 h at 4°C. The C3b (iC3b and C3dg) and C4b (C4d) fragments were obtained by proteolytic cleavages of the same stock of biotinylated C3b and C4b as described above. Monitoring of the biotinylation reactions, cleavages, and purifications was performed by SDS-PAGE, followed by Western blotting with avidin-horseradish peroxidase (HRP) conjugate.

Biotinylation of C3c and C4c.

Biotinylation of C3c and C4c was performed by using EZ-Link-Sulfo-NHS-Biotin (Pierce). In brief, 0.2 mg of C3c and C4c were dialyzed against PBS (pH 7.2). EZ-Link-Sulfo-NHS-Biotin was dissolved in PBS (pH 7.2) at a concentration of 2 mM, added to C3c or C4c at a molar ratio of 1:18, and incubated on ice for 2 h. The reaction mixtures were then extensively dialyzed against PBS (pH 7.4) at 4°C, and biotinylation was confirmed by Western blot analysis with avidin-HRP conjugate.

Western blot analysis.

Site-specific biotinylation of C3b, C4b, and their fragments (iC3b, C3dg, and C4d) was confirmed by Western blotting. Labeled proteins were separated on SDS-PAGE and electrotransferred onto polyvinylidene difluoride membrane (Bio-Rad, Hercules, Calif.). The membranes were blocked overnight at 4°C with constant rocking with 5% ECL blocking reagent (Amersham Biosciences, Buckinghamshire, United Kingdom). The incorporation of biotin was determined by probing the blots with 1:5,000 diluted avidin-HRP conjugate (Bio-Rad) for 2 h at room temperature and a wash with Tris-buffered saline (20 mM Tris, 150 mM NaCl [pH 7.5]) containing 0.05% Tween 20. Labeled proteins were detected by using a SuperSignal West Pico chemiluminescent kit (Pierce).

SPR measurements.

The kinetics of binding of VCP to C3b, C4b, and their fragments were determined on the SPR-based biosensor BIACORE 2000 (Biacore AB, Uppsala, Sweden). All of the experiments were performed in PBS-T (10 mM sodium phosphate and 145 mM NaCl [pH 7.4] containing 0.05% Tween 20) at 25°C unless mentioned otherwise. The addition of 0.05% Tween 20 blocked the nonspecific adsorption of analytes to the sensor chips. In each experiment, ligands were coupled either to Fc-2 or Fc-4, and Fc-1 or Fc-3 served as control flow cells. Experiments were performed either by immobilizing VCP or by immobilizing the target proteins onto the sensor chips; VCP (2000 Rus) was immobilized onto a CM5 chip by using amine-coupling chemistry, while biotinylated-C3b (2300 Rus) and -C4b (2200 Rus) were immobilized onto streptavidin chips (Sensor Chip SA; Biacore AB). Binding was measured at 50 μl/min to avoid the mass transport effect. Binding was measured for 120 s, and dissociation was monitored for an additional 180 s. The sensor chips were regenerated with 30-s pulses of 0.2 M sodium carbonate (pH 9.5). Sensograms obtained for the control flow cell were subtracted from the data for the flow cell immobilized with a ligand. The SPR data were analyzed by BIAevaluation software version 3.2 with global fitting. When data did not fit to a simple 1:1 Langmuir binding model it was evaluated by linear transformation analysis as previously described (44, 45, 55) by plotting dRU/dt Vs RU, where RU is the relative response of the biosensor at time t. The apparent equilibrium dissociation constant (K_D) was calculated from the equation K_D = k_d/k_a, where k_d is the dissociation rate constant and k_a is the association rate constant.

Measurement of factor I cofactor activity of VCP for C3b and C4b.

A quantitative analysis of factor I cofactor activity of VCP for C3b and C4b was determined in 10 mM phosphate buffer (pH 7.4) containing 145 mM NaCl. In these assays, 2.7 μg of C3b or 2.9 μg of C4b was mixed with 100 ng of factor I and various concentrations of VCP and then incubated at 37°C for 4 h in a total volume of 15 μl. The reactions were stopped by adding sample buffer containing dithiothreitol and then electrophoresed on an 8% SDS-PAGE gel for determining C3b cleavages and a 9% SDS-PAGE gel for C4b cleavages. The cleavage products were visualized by staining the gel with Coomassie blue. The gels were scanned for densitometric analysis by using ChemiDoc XRS system (Bio-Rad, Segrate, Italy) to calculate the percentage of α′ chain. The data obtained was normalized by considering 100% α′ chain to be equal to the α′-chain intensity obtained in the absence of factor I (control). The binding data were fit by using nonlinear regression analysis (GraFit; Erithacus Software, London, United Kingdom), and a four-parameter logistic analysis was performed to identify the best-fit concentration of VCP causing 50% α′-chain cleavage.

RESULTS

Kinetic analysis of binding of VCP to C3b and C4b.

Although previous studies have identified host proteins C3b and C4b as the target proteins for VCP, the kinetic model of binding involved in the bimolecular interaction of VCP with its target proteins is still not clear.

To gain insight into the interaction of VCP with C3b and C4b, we utilized SPR technology and compared its interaction to that of human regulators factor H and sCR1 (Fig. 1). VCP was coupled to a sensor chip by using the standard amine-coupling chemistry, and C3b and C4b were flown over the chip to measure binding. VCP bound to both the analytes in a dose-dependent manner (Fig. 2). The binding data (sensograms) for these interactions could not be fitted to a simple 1:1 binding model. Linear transformation analysis showed nonlinear plots, suggesting that these bindings follow complex models. We then attempted to fit these data to complex models to investigate the nature of these interactions. The data showed a close fit to a bivalent analyte model with χ² values for VCP-C3b and VCP-C4b interactions of 0.559 and 0.353, respectively (Table 1), suggesting that this is the most likely model of interaction. There was, however, another possibility that amine coupling produced a heterogeneous ligand surface which resulted in complex binding. This was further supported by the fact that the data fit well even to the heterogeneous ligand model. The χ² value for VCP-C3b interaction was 0.297 and that for VCP-C4b interaction was 0.264.

FIG. 1.

Schematic representation of structures of VCP, factor H, and sCR1. VCP is entirely composed of four SCR or CCP modules, whereas the human complement regulators factor H and complement receptor 1 are composed of 20 and 30 SCRs, respectively. The SCR domains of each protein are numbered, and the binding domains for C3b and C4b are identified (3, 19, 20, 26, 41, 48, 50). The relative binding activity of VCP to that of factor H and sCR1 is shown on the right. These activities are based on the data presented in Table 1. , Factor H binds to C4b only in a buffer with very low ionic strength (47).

FIG. 2.

Analysis of the binding of immobilized VCP to C3b and C4b by SPR. The left panels show sensogram overlays for interactions between immobilized VCP and C3b or C4b. The concentration of analyte injected is indicated at the right of each sensogram overlay. Solid lines correspond to the global fitting of the data simultaneously. Both C3b and C4b data fit to a bivalent analyte model (A+B ↔ AB; AB + B ↔ AB₂; BIAevaluation 3.2). The right panels show linear transformations of the association phase data for the respective sensogram data shown on the left. The straight lines are linear least square fits to the data.

TABLE 1.

Kinetic and affinity data for the interactions of VCP and human complement control proteins factor H (fH) and sCR1 with C3b, C4b, and their fragments

Ligandb	Analyte	kd1(1/s)/ka1(1/Ms)	SE (kd1/ka1)	KD1c	kd2(1/s)/ka2(1/Ms)a	SE (kd2/ka2)	KD2	χ2
VCP	C3b	9.61 × 10−2/9.58 × 103	2.52 × 10−3/179	10 μM	1.03 × 10−3/1.5 × 102	1.08 × 10−5/2.75	0.69 × 10−5	0.559
VCP	C4b	0.103/3.17 × 104	2.93 × 10−3/482	3.2 μM	1.34 × 10−3/3.83 × 102	1.71 × 10−5/5.88	0.35 × 10−5	0.353
C3b	VCP	0.554/7.66 × 105	2.5 × 10−2/6.38 × 104	0.724 μM	NA	NA	NA	0.887
C4b	VCP	0.38/2.42 × 106	1.1 × 10−2/9.69 × 104	0.157 μM	NA	NA	NA	0.526
C3b	CR1	5.74 × 10−2/4.4 × 106		13 nM*
C3b	fH	5.98 × 10−2/1.1 × 106		54.4 nM*
C4b	CR1	4.17 × 10−2/3.8 × 106		11 nM*
C3dg	VCP	NA	NA	0.179 μM†	NA	NA	NA	0.111
C4c	VCP	3.93 × 10−4/2.66 × 103	4.16 × 10−5/61	0.148 μM	NA	NA	NA	0.122

^ak_a2 (1/Ms) = k_a2 (1/RUs)×100×molecular weight of the ligand. NA, not applicable.

^bVCP was immobilized on CM5 chip by using standard amine coupling chemistry. C3b, C4b, and C4dg were site specifically biotinylated by labeling their free −SH groups and immobilized on SA chips. C4c was biotinylated by labeling its amino groups and immobilized on an SA chip.

^cData were calculated by global fitting analysis (BIA evaluation 3.2); *, data were calculated by linear transformation analysis (35, 44); †, data were calculated by steady-state analysis (BIAevaluation 3.2).

To determine whether bivalent binding truly reflected the binding interactions and was not an artifact of the heterogeneous ligand surface, we designed the following experiment. We oriented C3b and C4b onto a streptavidin chip by labeling their free −SH group with biotin (Fig. 3). This approach not only provided a homogeneous ligand surface but also immobilized these proteins in their physiological orientation. In complement proteins C3 and C4, Cys⁹⁸⁸ and Cys⁹⁹¹ are involved in the formation of the internal thioester bond.

FIG. 3.

Site-specific biotinylation of C3b and C4b. The free −SH groups of C3b and C4b were labeled with PEO-maleimide biotin, and labeled proteins were cleaved into iC3b or C3c and C3dg and C4c and C4d and analyzed by SDS-PAGE and Western blotting as described in Materials and Methods. C3b cleavage in the presence of factors H and I results in cleavage of the α′ chain into N-terminal 68-kDa and C-terminal 43-kDa fragments; the appearance of these fragments indicates the generation of iC3b, whereas C3b cleavage in the presence of sCR1 and factor I results in cleavage of the α′ chain into N-terminal 25-kDa fragment, C3dg, and C-terminal 43-kDa fragments, which indicates the generation of C3c and C3dg. C4b cleavage in the presence of sCR1 and factor I results in cleavage of the α′ chain into N-terminal 25-kDa, C-terminal 16-kDa, and central C4d fragments; the appearance of these fragments indicates the generation of C4c and C4d. (A) Diagram showing covalent attachment of C3b and C4b to the activating surface. (B) Diagram depicting the experimental design. Site-specific biotinylated C3b and C4b were immobilized on a streptavidin chip (SA chip). (C) Coomassie blue staining (left) and Western blot (right) of biotinylated C3b and its cleavage products. Lane 1, biotinylated C3b; lane 2, biotinylated C3b cleaved with factors I and H; lane 3, biotinylated C3b cleaved with factor I and sCR1. (D) Schematic representation of C3b, C3c, and C3dg structures. Arrows indicate factor I-mediated cleavages generated in the presence of cofactors and closed balloon indicates the location of −SH group labeled with biotin. (E) Coomassie blue staining (left) and Western blot (right) of biotinylated C4b and its cleavage products. Lane 1, biotinylated C4b; lane 2, biotinylated C4b cleaved with factor I and sCR1. (F) Schematic representation of C4b, C4c, and C4d structures. Arrows indicate factor I-mediated cleavages generated in the presence of sCR1, and the closed balloon indicates the location of −SH group labeled with biotin.

During complement activation, proteolytic cleavage of C3 and C4 result in generation of C3b and C4b and exposure of their thioester bond. This bond reacts with hydroxyl or amino groups present on the activating surfaces to form an ester or an amide bond. The covalent attachment of C3b and C4b onto the activating surfaces orients them on these surfaces (Fig. 3A). During this reaction, a free −SH group is generated (Fig. 3A). We labeled this −SH group with biotin and oriented the labeled molecules on streptavidin chip (Fig. 3B). Thus, orientation of C3b and C4b in our experimental setup mimicked the physiological orientation of these proteins. The specificity of this labeling was verified by analyzing the reactivity of labeled C3b and C4b with avidin-HRP in a Western blot assay. The results depicted in Fig. 3C and E show that biotin incorporation in C3b and C4b was associated with the α′ chains, indicating thereby that biotin was incorporated at the thioester site. To further confirm this, we cleaved labeled C3b and C4b molecules with factor I in the presence of either factor H or sCR1. Biotin incorporation was observed only in fragments containing free −SH group; in C3b fragments it was associated with the 68-kDa fragment and C3dg, whereas in C4b fragments it was associated with C4d (Fig. 3C and E).

When VCP was flown over streptavidin chips immobilized with oriented C3b or C4b, it bound in a dose-dependent and saturable manner (Fig. 4). The kinetics of binding were distinctly different from those obtained when VCP was immobilized by using amine coupling chemistry. Global fitting analysis of the sensograms showed a good fit to a 1:1 binding model (χ² values < 0.9, Table 1) and linear transformation of the binding data showed a single component. Together, these data indicated that both VCP-C3b and VCP-C4b interactions follow a simple 1:1 binding model. We then compared k_a and k_d of the VCP-C3b interaction with those of the VCP-C4b interactions. There was a 3.2-fold increase in k_a and a 1.5-fold decrease in k_d for the VCP-C4b interaction compared to the VCP-C3b interaction. Together, this resulted in a 4.6-fold increase in the affinity of VCP for C4b compared to C3b (Table 1).

FIG. 4.

Analysis of binding of VCP to immobilized C3b and C4b. The left panels show overlay plots of the binding of immobilized C3b and C4b to VCP. Various concentrations of VCP were injected over a streptavidin chip immobilized with C3b or C4b. Solid lines represent the global fitting of the data to a 1:1 Langmuir binding model (A+B ↔ AB; BIAevaluation 3.2). The right panels show linear transformations of the association data for the respective sensogram data shown on the left. The straight lines are linear least-square fits to the data.

Next, we analyzed binding of oriented C3b and C4b with multivalent proteins sCR1 and factor H (Fig. 1). As expected, binding of sCR1 to C3b and C4b and factor H to C3b did not follow a 1:1 binding model (Fig. 5). Because the data did not fit to a 1:1 model, the association and dissociation rate constants were calculated by linear transformation analysis as previously described (44, 45, 55). The affinities of sCR1 and factor H for C3b were 56- and 13-fold greater, respectively, than VCP, whereas that of sCR1 for C4b was 14-fold greater than that of VCP.

FIG. 5.

Binding of sCR1 and factor H to immobilized C3b and C4b. The left panels show overlay plots for binding of sCR1, factor H to immobilized C3b, and sCR1 to immobilized C4b. Various concentrations of analytes were injected over streptavidin chips containing biotinylated C3b or C4b. Solid lines shown in the C3b-CR1 panel represent the global fitting of the data to a bivalent analyte model (A+B ↔ AB; AB + B ↔ AB₂; BIAevaluation 3.2). The middle panels show the linear transformation of the association phase data of the respective sensogram data shown on the left. The straight lines are linear least-square fits to the data. The inset shows values of k_s (determined from the slope of the fits of the association data) replotted against analyte concentration. The slope of this plot provided k_a. The right panels show the linear transformation of the highest concentration of the dissociation phase data of the respective analyte shown on the left. The slope of the fits provided the off rates (k_d).

Comparison of the factor I cofactor activity of VCP for C3b and C4b.

The data presented above indicated that VCP has a 4.6-fold-higher affinity for C4b than C3b. In order to test whether this difference in affinity is also reflected in its function, we measured relative factor I cofactor activity of VCP for C3b and C4b. Figure 6 shows a comparison of cofactor activity of VCP in factor I-mediated cleavage of C3b and C4b. In this assay, equimolar concentrations of C3b or C4b were incubated with factor I, and various concentrations of VCP in a physiological ionic strength buffer. The reaction mixtures were run on SDS-PAGE gels and cleavage of the α′ chain of C3b and C4b were quantitated by densitometric analysis. The concentration of VCP required to cleave 50% of α′ chain of C3b and C4b was 0.059 and 0.018 μM, respectively, indicating that better binding of VCP to C4b leads to better cofactor activity of VCP for C4b.

FIG. 6.

Comparison of factor I cofactor activity of VCP for C3b and C4b. Equimolar concentrations of C3b or C4b and factor I were incubated in 10 mM sodium phosphate (pH 7.4) containing 145 mM NaCl with increasing concentrations of VCP at 37°C for 4 h (upper and middle panels). Cleavage products were visualized by running the samples on 8% and 9% SDS-PAGE gels for C3b and C4b, respectively, and staining with Coomassie blue. The intensities of the α′ chains of C3b and C4b were determined by densitometric analysis and are represented graphically (lower panel).

Effect of NaCl on binding of VCP to C3b and C4b.

VCP is structurally and functionally similar to host CCPs. Previous mutation analysis of charged residues of CCPs, as well as C3b, have highlighted the importance of ionic interactions in CCP-C3b/C4b complex formations (4, 6, 11, 25, 30, 40). In order to understand whether electrostatic interactions also play a critical role in binding of VCP to C3b and C4b, we measured binding at different salt concentrations. VCP immobilized on a CM5 sensor chip was allowed to bind to C3b or C4b in the presence of different NaCl concentrations. The buffer used for binding was 10 mM phosphate (pH 7.4) containing 0.05% Tween 20 and various concentrations of NaCl ranging from 30 to 400 mM. It is clear from the data that binding declined with increases in the NaCl concentration and was entirely abolished at 400 mM NaCl (Fig. 7). We have verified these data by performing experiments in the reverse orientation, wherein C3b or C4b were oriented on the sensor chip and VCP was allowed to bind at different salt concentrations (data not shown). The data confirmed that both VCP-C3b and VCP-C4b complex formations are strongly dependent on salt concentration. These data indicate that, like host CCPs, ionic interactions are also critical in the formation of VCP-C3b and VCP-C4b complexes.

FIG. 7.

Effect of NaCl concentration on the binding of VCP to C3b and C4b. Sensogram overlays for binding of immobilized VCP to C3b (125 nM) and C4b (100 nM) in the presence of various concentrations of NaCl are shown. On the left, the binding response (RU) is plotted against time. On the right, the maximum binding response obtained for each buffer condition is plotted against the NaCl concentration.

Does VCP bind simultaneously to C3b and C4b?

Because VCP interacts with C3b, as well as C4b, we sought to determine whether VCP could interact simultaneously with C3b and C4b. To address this issue, we evaluated the binding of VCP or VCP preincubated with C4b to C3b immobilized on a streptavidin chip. A positive control was formed by analyzing binding of VCP preincubated with anti-VCP monoclonal antibody (MAb) NCCS 67.2 to immobilized C3b. This MAb binds to VCP (K_D = 3.8 nM [determined by SPR analysis]) but does not inhibit the factor I cofactor activity of VCP for C3b and C4b (data not shown). In this assay, the expected response was if both C3b and C4b bound simultaneously to VCP then there would be a buildup of response in VCP-C4b complex binding compared to binding of VCP alone. Binding of VCP alone yielded a maximum response of 22 RU, whereas binding of VCP-C4b complex (VCP preincubated with C4b) yielded a response of 19 RU, demonstrating that there was no buildup in response (Fig. 8A). As expected, the binding of VCP-MAb complex (positive control) showed a buildup in response with a maximum signal of 61 RU. The MAb and C4b by themselves yielded no response. Similar results were obtained when the experiment was performed by using the converse arrangement (Fig. 8B). Thus, binding of VCP and VCP-C3b complex to immobilized C4b showed maximum responses of 39 and 40 RU, respectively, whereas VCP-MAb binding (positive control) resulted in a maximum response of 270 RU. The MAb and C3b alone did not show any binding to C4b. These data indicate that VCP does not complex simultaneously with C3b and C4b.

FIG. 8.

Simultaneous binding of C3b and C4b to VCP. (A) VCP (0.6 μM), VCP (0.6 μM) preincubated with C4b (0.56 μM), or VCP (0.6 μM) preincubated with anti-VCP MAb (0.4 μM [this antibody does not inhibit the functional activity of VCP]) at 25°C for 30 min was injected over a streptavidin chip containing biotinylated C3b, and the association and dissociation phases were monitored. (B) VCP (0.6 μM), VCP (0.6 μM) preincubated with C3b (0.6 μM), or VCP (0.6 μM) preincubated with anti-VCP MAb (0.2 μM) at 25°C for 30 min was injected over a streptavidin chip containing biotinylated C4b, and binding and dissociation were monitored.

Localization of VCP binding site on C3b and C4b.

In the present study, we also attempted to sublocalize the binding site of VCP on C3b and C4b. For this purpose, we characterized binding of VCP to physiologic fragments of C3b (C3c and C3dg) and C4b (C4c and C4d) with SPR. Biotinylated C3dg and C4d were generated by cleaving the site-specific biotinylated C3b and C4b with sCR1 and factor I (Fig. 4), whereas biotinylated C3c and C4c were generated by cleaving the unlabeled C3b and C4b with factor I and sCR1 and labeling them with EZ-Link-Sulfo-NHS-Biotin. The labeled C3dg/C4d (1100 RU/880 RU) and C3c/C4c (1300 RU/850 RU) were immobilized on Fc-2 and Fc-3 of streptavidin chips, and Fc-1 served as a control. Binding was studied by flowing VCP in PBS (pH 7.4). VCP bound to C3dg and C4c but not to C3c and C4d (Fig. 9). Binding to both C3dg and C4c was dose dependent. The K_D values for C3dg and C4c were 0.18 and 0.15 μM, respectively. Binding of VCP to C4c instead of C4d, as one may expect, is not unusual since a recent study reported that CR1 binds to C4c but not C4d (8). From the data presented here it is evident that the nature of interaction of VCP with C3b is different than that of C4b.

FIG. 9.

Binding of VCP to immobilized C3b and C4b fragments. The left panels show sensograms for interactions between VCP and C3b or C4b fragments. VCP (1.3 μM) was injected over a streptavidin chip immobilized either with C3b fragments (C3c-biotin and C3dg-biotin) or with C4b fragments (C4c-biotin and C4d-biotin). The right panels show overlay plots for interactions between immobilized C3dg or C4c and VCP. The small arrow in the C3dg-VCP panel indicates the time point used for evaluating steady-state affinity data. Solid lines shown in the C4c-VCP panel represent the global fitting of the data to a 1:1 Langmuir binding model (A + B ↔ AB; BIAevaluation 3.2).

DISCUSSION

Previous studies have established VCP as a virulence determinant of vaccinia virus. Using a skin lesion model it has been shown that vaccinia virus mutants that do not express VCP are attenuated in rabbits and guinea pigs (17). It is believed that this in vivo effect is due to enhanced complement-mediated neutralization of the mutant virus as well as the generation of specific inflammatory response at the site of infection due to the lack of VCP-mediated inactivation of the complement system. Although it is clear that VCP inactivates complement by interacting with complement proteins C3b and C4b, the nature of mechanisms involved is not clear. SPR analyses make it possible to probe the mechanism of protein-protein interaction; therefore, in the present study we have utilized this technology to unravel the molecular mechanisms underlying the interaction between VCP and its target proteins C3b and C4b.

Our real-time kinetic data obtained from SPR measurements of binding of C3b and C4b to VCP immobilized by using amine-coupling chemistry indicate that these interactions do not follow a 1:1 binding model. Linear transformation of the data showed that the reactions are multistep (Fig. 2). Such reactions could reflect the presence of multiple binding sites with different affinities, a conformational change, or a more complex model. The multistep binding model gains some support from the previous studies. (i) Based on the earlier mutagenesis data on membrane cofactor protein and the crystal structure of VCP, it has been proposed that a number of sites on VCP might interact with C3b and C4b (39). (ii) A study on factor H-C3b interaction suggested that binding of factor H to C3b causes a conformational change in C3b (51). Thus, there is a possibility that VCP and/or the target molecules (C3b and C4b) undergo structural reorientation upon binding.

The multistep binding, however, could simply be due to heterogeneity in the analyte and/or the surface under test. Both C3b and C4b are known to form dimers in solution. Such dimers are formed either due to thioester linkage formation between the α′ chains and/or the disulfide bond formation between the free −SH groups of C3b and C4b (54). We have ruled out the first possibility by purifying C3b and C4b over a gel filtration column. Prior to SPR experiments, C3b and C4b were passed over Superose 12 column (Amersham Pharmacia Biotech; two columns linked in a series), and the absence of dimers was confirmed by SDS-PAGE. Amine coupling method, however, is known to produce heterogeneity in the ligand surface due to differential attachment of ligands. Thus, to rule out this possibility, we oriented C3b and C4b onto streptavidin chips by labeling Cys⁹⁸⁸ and Cys⁹⁹¹, respectively (Fig. 3), which are involved in the thioester bond formation. This strategy not only allowed us to generate a homogeneous ligand surface but also simulated the in vivo orientation of these proteins (Fig. 3A and B); C3b and C4b bound to the streptavidin chips simulated C3b and C4b deposition on the activating surface, and VCP in solution mimicked binding of soluble VCP to deposited C3b and C4b. The data shown in Fig. 4 clearly demonstrate that the binding reactions between VCP and C3b, as well as C4b, follow a 1:1 binding model. Our data, therefore, indicate that these reactions do not involve multiple site interactions or conformational changes and are less complex than previously thought. The simple 1:1 interaction seems to be a general phenomenon between viral homologs of complement control proteins and C3b/C4b since a similar binding mechanism was observed between kaposica (the Kaposi’s sarcoma-associated herpesvirus homolog of CCP) and C3b/C4b (J. Mullick and A. Sahu, unpublished observation). How complement regulators (e.g., CR1, factor H, MCP, and C4BP) function as factor I cofactors in the degradation of C3b and C4b is not clear at present. It has been suggested, at least in the case of factor H, that binding of this molecule induces a conformational change in C3b that facilitates the binding of factor I (51). It is clear from our data that VCP does not induce conformational change either in C3b or in C4b; thus, it seems that structural reorientation in C3b and C4b is not necessary for factor I cofactor activities of viral homologs of CCPs.

VCP has previously been shown to inhibit both the classical and the alternative pathways of the complement system (32, 42). In a comparative analysis it was shown that VCP was 34-fold more active in inhibiting the classical pathway than the alternative pathway (42). It was suggested that the greater effect of VCP on the classical pathway was due to its dual effect on C3b and C4b. It is, however, possible that the greater effect of VCP on the classical pathway could be in part due to higher affinity of VCP for C4b. The comparative affinity data of VCP for C3b and C4b presented in Table 1 indicate that its affinity for C4b is 4.6-fold better than that of C3b. Furthermore, its higher affinity for C4b is also reflected in its factor I cofactor activity; VCP showed 3.3-fold greater factor I cofactor activity for C4b than C3b (Fig. 6). Based on these data we suggest that the greater activity of VCP against the classical pathway is due to its dual effect on C3b and C4b, as well as to its higher affinity for C4b.

There is now a consensus that ionic interactions form an essential component of the binding interface between CCPs (factor H, CR1, C4BP, and MCP) and C3b/C4b. This premise is based on the following two observations: (i) these interactions show ionic strength dependence and (ii) extensive mutagenesis data show that the negatively charged residues on C3b and C4b and positively charged residues on CCPs are important for CCP-C3b/C4b interactions (4, 6, 11, 25, 30, 40). Because VCP is structurally and functionally similar to CCPs, we sought to determine whether VCP utilizes a similar mechanism to stabilize the binding with C3b and C4b. The data obtained here clearly show that binding of VCP to C3b and C4b is highly dependent on ionic interactions (Fig. 7). Thus, our data suggest that long-range electrostatic forces and ion pairing are critical in the formation of stable VCP-C3b/C4b complexes. The crystal structure of the entire VCP molecule revealed that the SCR domains 1 and 4 are highly positively charged, whereas the middle two SCR domains are surrounded by a net negative field (39). Based on our data and the premise that positively charged residues of CCPs play a predominant role in ionic interactions, it could be speculated that SCRs 1 and 4 of VCP play a significant role in ionic interactions. However, a previous study on the monkey poxvirus homolog of VCP, which is devoid of a large portion of SCR 4, showed that this protein contains complement inhibitory activity (49). Therefore, it is likely that ionic interactions in VCP occur mainly through SCR 1.

SPR analyses allow measure of bi- and multimolecular complex formation and thus serve as a powerful tool for examining surface topography. Therefore, in the present study, we also sought to determine whether VCP could bind simultaneously to C3b and C4b. We measured binding of VCP or VCP preincubated with C4b to C3b-biotin immobilized on a streptavidin chip. We observed that preincubation of VCP with C3b did not result in an increase in mass on the chip (Fig. 8A). These results were also validated by using a converse arrangement wherein VCP or VCP preincubated with C3b was allowed to bind to immobilized C4b (Fig. 8B). These data indicate that C3b and C4b binding to VCP are not independent events; thus, VCP may not be able to execute the dual functions of inactivating C3b and C4b together. Whether the lack of simultaneous binding of C3b and C4b to VCP is due to competition for the same binding site or sterically hindered access to these sites due to close proximity is not clear and requires further investigation.

Our data on the localization of VCP binding site on C3b and C4b has revealed that these sites are located in the C3dg and C4c regions, respectively (Fig. 9). It should be pointed out here that, to date, VCP is the only viral protein known to interact with C3dg (38). The other C3-interacting viral proteins, which have been studied for their interaction with C3 fragments, are glycoprotein C of herpes simplex virus types 1 and 2, and these are known to bind to C3c (21, 31). There is now a large body of evidence implicating the importance of interaction between C3dg (a proteolytically cleaved fragment of C3) with its receptor CR2 (CD21), which is present on B cells and follicular dendritic cells, in the modulation of antibody-mediated humoral response (7, 29, 43). For example, mice deficient either in C3 or in C4 protein showed impaired antibody response to suboptimal doses of T-dependent antigen, and their responses were characterized by reduced number and size of germinal centers (10, 12). Similarly, mice containing disrupted CR2 locus showed impaired antibody response to T-dependent antigens (1, 34). In addition, the administration of anti-CR2 antibody or the soluble form of CR2 suppressed the in vivo immune response (14, 16). Whether binding of VCP to C3dg inhibits its interaction with CR2 is not clear at present. However, such interaction would be of great benefit to the virus because then VCP would not only inhibit the innate (complement activation) but would also inhibit the acquired immune response (generation of specific antibody production).

A central question in the biology of vaccinia virus is, what role does complement play in controlling its infection and how does VCP help to subvert it? Previous in vitro studies have clearly shown that infectivity of both the infectious forms of vaccinia virus, i.e., intracellular immature virus (17, 53), as well as EEV (53), was destroyed by complement when anti-vaccinia virus antibodies were present during the assay. It is important to point out here that although the incorporation of host complement regulators confers resistance to EEV against the alternative pathway, this form is still susceptible to complement-mediated neutralization by the classical pathway (53). Thus, for successful propagation of the virus, it would be advantageous for the virus to encode a complement regulatory protein that would effectively inhibit the classical pathway. The data obtained thus far make it clear that VCP effectively inhibits the classical pathway-mediated inactivation of both intracellular immature virus and EEV (17, 53). It is noteworthy that this viral protein is a more effective regulator of the classical pathway than even the host complement regulators factor H and C4-binding protein (22, 42). The data presented here indicate that the greater effect of VCP against the classical pathway is due to its higher affinity for C4b than C3b (Fig. 4 and Table 1).

How VCP regulates complement-mediated inactivation of vaccinia virus in vivo is not clear at present. A prior study had shown that VCP deletion mutant produced smaller skin lesions than the wild-type virus in rabbits (17). Interestingly, lesions produced by mutant and wild-type viruses were similar for the first few days but reduced in size after day 5 in the mutant virus, and this timing correlated well with the appearance of anti-vaccinia virus antibody. It is therefore likely that VCP protects vaccinia virus from the classical pathway of complement during the late phase of infection. In the present study we have shown that VCP interacts with C3dg fragment of C3 (Fig. 9). Thus, as discussed above, it is also possible that VCP-mediated inhibition of specific antibody generation leads to protection of vaccinia virus from the classical pathway of complement activation. Together, these observations suggest that structural determinants of VCP important in interacting with C3b and C4b may act as potential targets for future therapies and drug development. The initial belief that multiple-site interactions exist between VCP and C3b and C4b (39) suggested that such an endeavor would need blocking of multiple interactions and thus would not be a realistic approach. However, our data suggest that these interactions follow a simple one-site model and therefore targeting VCP would be feasible.