Species Specificity of Vaccinia Virus Complement Control Protein for the Bovine Classical Pathway Is Governed Primarily by Direct Interaction of Its Acidic Residues with Factor I

ABSTRACT

Poxviruses display species tropism—variola virus is a human-specific virus, while vaccinia virus causes repeated outbreaks in dairy cattle. Consistent with this, variola virus complement regulator SPICE (smallpox inhibitor of complement enzymes) exhibits selectivity in inhibiting the human alternative complement pathway and vaccinia virus complement regulator VCP (vaccinia virus complement control protein) displays selectivity in inhibiting the bovine alternative complement pathway. In the present study, we examined the species specificity of VCP and SPICE for the classical pathway (CP). We observed that VCP is ∼43-fold superior to SPICE in inhibiting bovine CP. Further, functional assays revealed that increased inhibitory activity of VCP for bovine CP is solely due to its enhanced cofactor activity, with no effect on decay of bovine CP C3-convertase. To probe the structural basis of this specificity, we utilized single- and multi-amino-acid substitution mutants wherein 1 or more of the 11 variant VCP residues were substituted in the SPICE template. Examination of these mutants for their ability to inhibit bovine CP revealed that E108, E120, and E144 are primarily responsible for imparting the specificity and contribute to the enhanced cofactor activity of VCP. Binding and functional assays suggested that these residues interact with bovine factor I but not with bovine C4(H₂O) (a moiety conformationally similar to C4b). Mapping of these residues onto the modeled structure of bovine C4b-VCP-bovine factor I supported the mutagenesis data. Taken together, our data help explain why the vaccine strain of vaccinia virus was able to gain a foothold in domesticated animals.

IMPORTANCE Vaccinia virus was used for smallpox vaccination. The vaccine-derived virus is now circulating and causing outbreaks in dairy cattle in India and Brazil. However, the reason for this tropism is unknown. It is well recognized that the virus is susceptible to neutralization by the complement classical pathway (CP). Because the virus encodes a soluble complement regulator, VCP, we examined whether this protein displays selectivity in targeting bovine CP. Our data show that it does exhibit selectivity in inhibiting the bovine CP and that this is primarily determined by its amino acids E108, E120, and E144, which interact with bovine serine protease factor I to inactivate bovine C4b—one of the two subunits of CP C3-convertase. Of note, the variola complement regulator SPICE contains positively charged residues at these positions. Thus, these variant residues in VCP help enhance its potency against the bovine CP and thereby the fitness of the virus in cattle.

INTRODUCTION

Poxviruses are among the most successful pathogens (1). It is believed that one of the key reasons for their success is their ability to subvert both innate and adaptive immune barriers of the host (2, 3). The most notable member of this family is variola virus, the causative agent of smallpox, which killed hundreds of millions of people before its successful eradication in 1977 owing in part to mass vaccination by vaccinia virus under the aegis of the World Health Organization (4). Intriguingly, though variola viruses and vaccinia viruses are closely related, they differ in host tropism (5). Variola virus is a strictly human-specific virus, while vaccinia virus not only infects humans but is feral in India (6, 7) and Brazil (8) and is known to cause repeated outbreaks in dairy cattle (7, 9), though its natural host is unknown (5, 10).

The complement system represents one of the major innate immune responses to viruses, targeting cell-free virus and virus-infected cells (11, 12) and augmenting virus-specific B- and T-cell responses (13–16). It is thus expected that viruses encode proteins to subvert the complement attack. In unison with this, both variola and vaccinia viruses have been shown to encode complement regulators, named SPICE (smallpox inhibitor of complement enzymes) and VCP (vaccinia virus complement control protein [CCP]), respectively, which are mimics of the host regulator of complement activation (RCA) proteins (17). Dissection of the mechanism of their action revealed that, like the human RCA proteins, they target the C3-cleaving C3 convertase enzymes by binding and decaying these enzymes (termed decay-accelerating activity [DAA]) as well as by inactivating the subunits of these enzymes with the help of a plasma protease factor I (FI) (termed cofactor activity [CFA]).

The most intriguing feature of SPICE and VCP is that they display species specificity; i.e., SPICE preferentially inhibits human complement (18, 19), while VCP preferentially inhibits bovine complement (20). In particular, SPICE was shown to be 100-fold and 6-fold more potent than VCP in inactivating human C3b and C4b—the components of alternative and classical pathway (CP) C3-convertases (18). Similarly, VCP was shown to be 12-fold better than SPICE in inactivating bovine C3b and 36-fold more potent in decaying the bovine alternative pathway C3-convertase (20); its effect on bovine C4b and the CP C3-convertase, however, was not studied.

Both SPICE and VCP are secretory proteins but have the ability to bind to the infected cells via glycosaminoglycans (21). In addition, VCP has been shown to bind to the infected cells via the viral protein A56 (22). They are entirely formed by four compact domains, termed complement control protein (CCP) domains, linked by four amino acid linkers, thus resembling a “beads-on-a-string” structure (23, 24). Interestingly, they differ only in 11 amino acids which are scattered in domains 2, 3, and 4 (18, 25). Efforts to identify the amino acids that provide a functional advantage to SPICE with respect to human complement showed that two residues (namely, K108 and K120) are critical for enhancing its ability to inactivate human C3b (26, 27). Later, substitution of each of the 11 variant amino acids of SPICE in VCP revealed that substitution of four residues (H98Y/S103Y/E108K/E120K) was enough to make VCP as potent as SPICE against human C3b and C4b (19). Likewise, substitution of three residues of VCP in SPICE was shown to convert SPICE to a state of potency equal to that of VCP against bovine C3b and the bovine alternative pathway (20).

Vaccinia virus has two infectious forms—the intracellular mature virion (MV) and the extracellular enveloped virion (EV). Both the forms have been shown to activate the complement system, resulting in their neutralization; however, this requires the presence of virus-specific antibodies (28–31), indicating that these forms are susceptible to neutralization only by the CP. Thus, it implies that effective inhibition of the bovine CP by VCP is necessary to protect the virus from complement-mediated inactivation in cattle. However, whether VCP inhibits the bovine CP is not known.

In the present study, we thus asked whether VCP is capable of inhibiting the bovine CP and whether it displays selectivity in inhibiting this pathway. In addition, we also determined the mechanism of inactivation of bovine CP by VCP and which adaptive mutations of VCP partake in this function. Our data demonstrate that VCP is effective in inhibiting the bovine CP in a species-specific manner and that the three acidic residues, which interact with bovine factor I, are responsible for this function.

(This work was done in partial fulfillment of the requirements for a Ph.D. thesis by J.K. at the S. P. Pune University, Pune, India.)

RESULTS

VCP displays selectivity for inhibition of the bovine classical complement pathway.

Successful establishment of viral infection in the host is contingent upon the ability of the virus to subvert the antiviral immune responses. Keeping this in consideration as well as that vaccinia virus was shown to infect various domestic animals during mass smallpox vaccination (usually cattle [32–35]) and that the infectious forms of vaccinia virus are sensitive to the CP-mediated neutralization, we hypothesized that VCP ought to be better suited to inhibition of the CP of domestic animals, particularly that of cattle. We thus measured the inhibitory activity of VCP against the CP of various animal species by employing the hemolytic assay. As expected, VCP showed greater inhibition of the CP of animals such as calf, buffalo, goat, and cat than of that of human (Fig. 1). More importantly, the concentration of VCP required to inhibit calf and buffalo CP was the lowest of those required for inhibition of the CP of the examined species (Fig. 1), which is consistent with its repeated outbreaks seen in cow (36) and buffalo (7).

FIG 1.

Relative levels of inhibition of the classical complement pathway of various species by VCP and SPICE. Inhibition of the classical complement pathway in nonprimate and primate species by VCP (filled circle) and SPICE (open circle) was assessed by the hemolytic assay. Activation of the complement was triggered by antibody-coated sheep erythrocytes (EA). IC₅₀s for VCP were as follows: calf, 10.6 nM; buffalo, 7.5 nM; dog, 195.5 nM; cat, 102 nM; goat, 25.2 nM; monkey, 47.5 nM; human, 173.8 nM. IC₅₀s for SPICE were as follows: calf, 468 nM; buffalo, 312 nM; dog, 1,928 nM; cat, 838 nM; goat, 115 nM; monkey, 21.1 nM; human, 112.2 nM.

Next, we performed inhibition of the CP of various animals with SPICE. The results showed that SPICE is most effective against the CP of monkey and human and is relatively less effective against the CP of calf and buffalo. Consequently, the maximum relative difference in the inhibitory activity of VCP and SPICE was observed against calf (43-fold) and buffalo (45-fold) CP; there was, however, little (≤2-fold) or no difference in the inhibitory activities of VCP and SPICE against monkey and human CP (Fig. 1).

Acidic residues in the central domains of VCP are primarily responsible for its enhanced inhibition of the bovine classical complement pathway.

It was clear from the results described above that VCP displays selectivity in inhibiting the bovine CP compared to SPICE. Because VCP and SPICE differ only in 11 amino acids, we next determined which of the variant residues of VCP contribute to its enhanced activity against the bovine CP. We thus utilized 11 single-amino-acid mutants of SPICE, wherein each of the variant amino acids of VCP was substituted at the corresponding position (Fig. 2), and examined their ability to inhibit the CP-mediated hemolytic activity of calf sera. A significant increase in the inhibitory activity was observed in 5 of the 11 mutants, namely, those with mutations K108E, K120E, N178D, N144E, and T214K, compared to the SPICE results (Fig. 3 and Table 1). These results thus indicated that acidic residues located in CCP domains 2 and 3 are primarily responsible for the selectivity of VCP toward bovine CP. Notably, the maximum increase in the activity was observed with the substitution at position 108 (K108E), which was also responsible for the maximum gain in the inhibitory activity toward the bovine alternative pathway (20).

FIG 2.

VCP model showing 11 amino acid variations compared to SPICE and SDS-PAGE analysis of the single- and multiple-amino-acid substitution mutants of SPICE. (A) Surface representation of the front and back faces of a homology model of VCP built using the recently solved SPICE crystal structure (PDB ID: 5fob). The 11 amino acid differences in VCP compared to SPICE are marked and colored in red. (B) Amino acid sequence comparison of VCP and SPICE showing 11 amino acid differences; the disulfide linkages are represented according to the SPICE crystal structure. The variant residues in VCP are highlighted in red and marked by arrows. The numbers denote their positions in the mature protein. (C) Analysis of purified VCP, SPICE, and the substitution mutants on 10% SDS-PAGE. The proteins were run under reducing conditions and subjected to Coomassie blue staining for visualization. MW, molecular weight; SPICE-double, K108E/K120E; SPICE-triple 1, K108E/K120E/N144E; SPICE-triple 2, K108E/K120E/N178D; SPICE-tetra, K108E/K120E/L131S/N144E; SPICE-penta, H77Q/K108E/K120E/L131S/N144E.

FIG 3.

Inhibition of the bovine classical pathway of complement by VCP, SPICE, and the single-amino-acid substitution mutants of SPICE. Antibody-sensitized sheep erythrocytes (EA) were mixed with various concentrations of VCP, SPICE, or the indicated single-amino-acid substitution mutant in GVB⁺⁺ buffer. The reaction mixtures were then mixed with calf sera as a source of complement and incubated at 37°C for 1 h. The cell lysis was measured at 405 nm, and data were normalized and plotted. (Upper panel) Mutants which showed ≥3-fold increase in activity compared to SPICE. (Lower panel) Mutants which showed <3-fold increase in activity compared to SPICE. Data shown in the graphs are representative of results of one of the three experiments summarized in Table 1.

TABLE 1.

Summary of complement regulatory activities of various VCP mutants

Wild-type/mutant protein	IC50 for CP lysis (μM)b	Relative CP lysis activitya	Time (min) for 50% cleavage of bovine C4(H2O) α-chainb	Relative cofactor activitya
SPICE	0.517 ± 0.09	1.0	409 ± 42.8	1
H77Q	0.180 ± 0.04	2.9	135 ± 33.8	3.0
Y98H	0.206 ± 0.03	2.5	742 + 136	0.55
Y103S	0.219 ± 0.06	2.4	884 ± 85.1	0.46
K108E	0.015 ± 0.003	34.4	61.0 ± 13.5	6.7
K120E	0.074 ± 0.01	7.0	46.0 ± 6.9	8.9
L131S	0.180 ± 0.01	2.9	71.3 ± 21.3	5.7
N144E	0.137 ± 0.006	3.8	90.0 ± 16.7	4.5
N178D	0.099 ± 0.02	5.2	182 ± 9.1	2.2
L193S	0.183 ± 0.06	2.8	216 ± 21.6	1.9
T214K	0.150 ± 0.01	3.4	181 + 15.3	2.3
Q236K	0.198 + 0.01	2.6	210 ± 20.7	1.9
SPICE-Double (K108E/K120E)	0.014 ± 0.002	36.9	11.2 ± 1.2	36.6
SPICE-Triple1 (K108E/K120E/N144E)	0.011 ± 0.001	47.0	6.5 ± 1.8	62.9
SPICE-Triple2 (K108E/K120E/N178D)	0.025 ± 0.001	20.7	12.5 ± 3.5	32.7
SPICE-Tetra (K108E/K120/L131S/N144E)	0.01 ± 0.001	51.7	3.8 + 0.3	107
SPICE-Penta (H77Q/K108E/K120E/L131S/N144E)	0.018 ± 0.001	28.7	3.53 ± 0.05	116
VCP	0.012 ± 0.002	43.0	8.33 ± 1.5	49.0

^aData represent relative activity compared to SPICE. Boldface indicates the mutants and data with a ≥3-fold difference in activity, which was considered significant.

^bData represent means ± SD of results of three independent experiments.

VCP is known to regulate the human CP by targeting the C3-convertase by two mechanisms (termed decay-accelerating activity [CP-DAA] and cofactor activity [CFA]). During CP-DAA, VCP inactivates CP C3-convertase C4b2a by irreversibly displacing its C2a protease subunit from C4b, while during CFA, it inactivates C4b by supporting its cleavage by factor I (37–39). We thus next sought to determine whether the enhanced activity of VCP toward the bovine CP is due to enhancement in one or both of these activities. In addition, we also looked into the variant residues of VCP that contribute to the enhanced CP-DAA and/or C4b CFA.

To measure the relative levels of CP-DAA of VCP and SPICE, we assembled the bovine CP C3-convertase on antibody-sensitized sheep erythrocytes (EA) by incubating these cells with cobra venom factor (CVF)-treated bovine sera (as a source of C142). As expected, the assembled CP C3-convertase displayed spontaneous decay under physiological conditions (Fig. 4A). However, neither VCP nor SPICE could significantly accelerate the decay of the bovine CP C3-convertase (Fig. 4B). Accordingly, none of the single-amino-acid mutants displayed significant decay of the CP C3-convertase (Fig. 4B). These results clearly reflected that the enhanced activity of VCP toward the bovine CP ought to be due to its greater CFA toward bovine C4b. Measurement of the relative levels of CFA of VCP and SPICE against bovine C4(H₂O), which is conformationally similar to C4b (40), revealed that this indeed is true: VCP displayed a 49-fold increase in activity compared to SPICE (Fig. 5 and Table 1). Next, examination of the contribution of each of the variant residues in the augmentation of the CFA of VCP suggested that five substitutions (H77Q, K108E, K120E, L131S, and N144E) contributed to this increase. The order of activity was K120E > K108E > L131S > N144E > H77Q. It should be noted that 3 of the 5 substitutions were negative-charge substitutions, suggesting yet again that, primarily, the presence of acidic residues at the specified positions of CCP domains 2 and 3 dictates the specificity of VCP for the bovine CP.

FIG 4.

Measurement of the bovine classical pathway C3-convertatse decay-accelerating activity of VCP, SPICE, and the single-amino-acid substitution mutants of SPICE. (A) Measurement of natural decay of bovine classical pathway C3-convertase (C4b2a). The bovine classical pathway C3-convertase was assembled on antibody-sensitized sheep erythrocytes (EA) as described in Materials and Methods and allowed to decay at 22°C for the indicated time intervals. The remaining convertase activity was determined by incubating the cells for 30 min at 37°C with guinea pig sera containing 40 mM EDTA. (B) The CP C3-convertase was assembled on antibody-coated sheep erythrocytes and allowed to decay in the presence of three different concentrations of the regulators. Data obtained were normalized by considering the lysis in the absence of an inhibitor to represent 100% lysis. Results represent means ± SD of results of three independent experiments.

FIG 5.

Amino acid variations in domains 2 and 3 of VCP compared to SPICE are responsible for its enhanced cofactor activity against bovine complement C4. (A) Comparison of factor I cofactor activity of VCP, SPICE, and the single-amino-acid substitution mutants of SPICE against bovine C4. The cofactor activities of the regulators were compared by incubating these regulators (denoted by arrowheads) with bovine (Bo) C4(H₂O) and bovine factor I and measuring the cleavage of the α-chain of bovine C4(H₂O). (B) Relative cofactor activities of VCP, SPICE, and the mutants. The intensity of α-chain that remained at various time points after incubation was quantitated by densitometric analysis and is represented graphically against time. Data shown in the graphs are representative of results of one of the three experiments summarized in Table 1.

The data described above suggested that the CFA of VCP is the major factor skewing its specificity for the bovine CP. It is well known that VCP imparts this activity owing to its interaction with C4b and factor I. Thus, we measured the binding abilities of VCP, SPICE, and the mutants with bovine C4(H₂O) to determine whether the enhanced CFA of VCP against bovine C4(H₂O) is a result of its superior binding to this molecule. For binding measurements, we employed a surface plasmon resonance (SPR)-based assay described earlier (41). In this assay, we oriented the C4(H₂O) on Sensor Chip SA (SA chip) in its physiological orientation and flowed VCP, SPICE, or the mutants over it. VCP displayed moderately increased binding to bovine C4(H₂O) compared to SPICE, but the mutants that showed increased CFA compared to SPICE (H77Q, K108E, K120E, L131S, and N144E) showed binding either similar to or lower than that seen with SPICE (Fig. 6). These results therefore suggest that the increased CFA of VCP against C4(H₂O) is not due to enhancement of its binding to C4(H₂O).

FIG 6.

SPR-based binding analysis of bovine C4(H₂O) with VCP, SPICE, and the single-amino-acid substitution mutants of SPICE. Sensogram overlays representing the interaction of VCP, SPICE, and the single-amino-acid substitution mutants are shown. Biotinylated bovine C4(H₂O) was immobilized on a SA sensor chip, and a 20 nM concentration of VCP, SPICE, or the mutant was injected over the chip to measure binding. (Upper panel) Interaction of SPICE single-amino-acid mutants that exhibited ≥3-fold increases in cofactor activity compared to SPICE (Fig. 5B). (Lower panel) Interaction of SPICE single-amino-acid mutants that showed less than <3-fold increases in cofactor activity compared to SPICE (Fig. 5B).

The specificity of VCP for the bovine classical complement pathway is dictated by the acidic residues at positions 108, 120, and 144.

Our exercise described above identified five determinants (H77Q, K108E, K120E, L131S, and N144E) that contribute to the enhanced CFA of VCP against bovine C4(H₂O) compared to SPICE. Next, in order to identify the major determinants capable of directing the specificity of VCP for the bovine CP, we generated and tested the inhibitory activities of multiresidue mutants of SPICE. Hence, we constructed a SPICE double mutant (SPICE-Double [K108E/K120E]), two SPICE triple mutants (SPICE-Triple 1 [K108E/K120E/N144E] and SPICE-Triple 2 [K108E/K120E/N178D]), a SPICE tetramutant (SPICE-Tetra [K108E/K120E/N144E/L131S]), and a SPICE pentamutant (SPICE-Penta [K108E/K120E/N144E/L131S/H77Q]) and measured their ability to inhibit the bovine CP-mediated lysis of erythrocytes and to impart CFA against C4(H₂O).

Although each of the five multi-amino-acid substitution mutants of SPICE described above demonstrated a substantial increase in its ability to inhibit the bovine CP-mediated lysis of erythrocytes as well as to inactivate C4(H₂O), the major shift in activities was achieved by changing lysines to glutamates at position 108 and position 120 (K108E/K120E) (Fig. 7A and B) (Table 1). Further, substitution of a third glutamate at position 144 generated a mutant (K108E/K120E/N144E) capable of inhibiting CP-mediated lysis and inactivating C4(H₂O) at a level equal to that of VCP, indicating that it is primarily the glutamates at positions 108, 120, and 144 that skew the specificity of VCP for the bovine CP (Fig. 7A and B; Table 1). Of note, though further substitution of serine at position 131 in the triple mutant (K108E/K120E/L131S//N144E) showed no increase in the CP inhibitory activity of the mutant, it showed a considerable increase in the ability to inactivate C4(H₂O) (Fig. 7A and B; Table 1), suggesting the importance of S131 in the CFA.

FIG 7.

Negatively charged variant residues of VCP are principally responsible for its specificity for the bovine classical complement pathway. (A) Inhibition of the bovine classical complement pathway by VCP, SPICE, and SPICE multi-amino-acid mutants. The activity was measured using a hemolytic assay. Data shown in the graphs are representative of results of one of the three experiments summarized in Table 1. (B) The cofactor activity of VCP, SPICE, and SPICE multi-amino-acid substitution mutants against bovine (Bo) C4(H₂O). Data shown in the graphs are representative of results of one of the three experiments summarized in Table 1. (C) SPR-based binding analysis of VCP, SPICE, and the SPICE multi-amino-acid substitution mutants with respect to bovine C4(H₂O). Sensogram overlay graphs were grouped on the basis of cofactor activity. The left panel shows binding of SPICE mutants that showed reduced cofactor activity toward bovine C4(H₂O) compared to VCP. The right panel shows binding of SPICE mutants that showed cofactor activity toward C4(H₂O) equal to or greater than than VCP. Key: conc., concentration; SPICE-double, K108E/K120E; SPICE-triple 1, K108E/K120E/N144E; SPICE-triple 2, K108E/K120E/N178D; SPICE-tetra, K108E/K120E/L131S/N144E; SPICE-penta, H77Q/K108E/K120E/L131S/N144E.

Our binding analysis of single-amino-acid mutants with bovine C4(H₂O) indicated that the gain in the specificity of the mutants for the bovine CP was not due to their increased binding to C4(H₂O) (Fig. 6). In order to substantiate this view, we also measured the binding of multi-amino-acid substitution mutants with the C4(H₂O). As expected, mutant SPICE Triple 1 (K108E/K120E/N144E) that displayed activity identical to that of VCP did not show any increase in binding to C4(H₂O). If anything, its binding was lower than that seen with SPICE (Fig. 7C). Similarly decreased binding was also observed for the SPICE-Double mutant (K108E/K120E) that showed a maximum switch in the specificity (Fig. 7C). The SPICE-Triple 2 mutant (K108E/K120E/N178D), however, displayed a moderate increase in binding, suggesting the likely role of D178 in binding to C4(H₂O).

The species-selective cofactor activity of VCP is primarily dictated by its interaction with factor I.

It was apparent from the data obtained as described above that the enhanced CFA of VCP toward bovine C4(H₂O) cannot be attributed to its interaction with C4(H₂O). Therefore, to resolve whether it can be attributed to its enhanced interaction with factor I, we performed analyses of the CFA of VCP, SPICE, and the mutants against C4(H₂O) using bovine C4(H₂O) and human factor I.

Incubation of bovine C4(H₂O) and bovine factor I with VCP, SPICE, or the mutants demonstrated that VCP is highly efficient in supporting the inactivation of C4(H₂O) compared to SPICE and that substitutions of three glutamates at positions 108, 120, and 144 are responsible for this increased specificity of VCP (Fig. 8A). Importantly, replacement of bovine factor I with human factor I in the assay described above resulted in drastic abrogation of the CFA of VCP as well as of the mutants. Here, the activity of VCP as well as that of all three mutants decreased more than 40-fold (Fig. 8B). Thus, it is the interaction of factor I that determines the species-selective CFA of VCP.

FIG 8.

Species specificity of VCP for the bovine classical pathway is governed by its interaction with bovine factor I. The cofactor activity of VCP, SPICE, and the multi-amino-acid mutants of SPICE was measured by incubating each of them with bovine (Bo) C4(H₂O) and bovine (Bo) or human (Hu) factor I for the indicated time periods. The data are represented graphically as means of results of two independent experiments. (A) VCP, SPICE, or the indicated mutant was incubated with bovine C4(H₂O) and bovine factor I. (B) VCP, SPICE, or the indicated mutant was incubated with bovine C4(H₂O) and human factor I. Key: SPICE-triple 1, K108E/K120E/N144E; SPICE-tetra, K108E/K120E/L131S/N144E; SPICE-penta, H77Q/K108E/K120E/L131S/N144E.

Mapping of factor I interacting residues of VCP onto the modeled structure of bovine C4b:VCP:bovine factor I and bovine C4b:VCP:human factor I complexes.

Our CFA data suggested possible interactions of VCP residues E108, E120, and E144 with bovine factor I (Fig. 8). To get a detailed view of the interactions of these residues with bovine factor I and to explain why human factor I does not efficiently interact with VCP, we generated geometrically, stereochemically, and energetically sound models of the bovine C4b:VCP:bovine factor I and bovine C4b:VCP:human factor I complexes. In brief, we first modeled the structures of bovine C4b, VCP, and bovine factor I on experimentally determined structures of human C4b, SPICE, and human factor I, respectively. Next, to model the trimolecular complexes, we first modeled the bovine C4b:VCP complex using the human C3b:SPICE structure (PDB identifier [ID] 5fob) and then placed bovine or human factor I onto this complex as achieved earlier for modeling the human C3b:factor H:factor I complex (42) and human C3b:Kapo:factor I complex (43). Both the models were subjected to energy minimization and validated for geometrical and stereochemical properties using PROCHECK (44) (for details and the structural adjustments that we had to perform to avoid clashes, see Materials and Methods). Our models indicated that specific interactions are responsible for the experimentally observed behavior. The negatively charged glutamates at positions 108, 120, and 144 in VCP were found to directly interact with the positively charged residues of bovine factor I, but such interactions were absent in human factor I. (i) E108 of VCP was found to make an ionic interaction with K108 of bovine factor I, but no such interaction was seen with human factor I. (ii) E120 of VCP showed an ionic interaction with K92 of bovine factor I. At a similar position, human factor I also had K306, albeit with its side chain pointing in the opposite direction. Another residue seen in the vicinity of E120 was L307, which again does not seem to be involved in any favorable interactions with VCP. (iii) E144 of VCP showed a direct interaction with the oppositely charged R425 of bovine factor I. In addition, R425 of bovine factor I also showed weak interactions with S131 of VCP. In human factor I, however, no such interactions were observed (Fig. 9).

FIG 9.

Mapping of high-propensity interactions of functionally important residues of VCP onto the modeled structure of the C4b-VCP-factor I complex. (A) The model of the trimolecular complex was built as described in Materials and Methods. Individual chains of C4b have been colored differently and are depicted as a cartoon. The colors green, cyan, and pink represent the beta, alpha, and gamma chains of bovine C4b. The model of bovine factor I is also shown as cartoon (yellow), while the model of VCP is shown in surface representation. (B and C) Potential interactions (present within a 5-Å radius) of the functionally important residues of VCP (in cartoon; salmon) with bovine and human factor I (in cartoon; yellow). The functionally important residues are shown as sticks and colored by atom type in the theme of the original protein color. The left panels (B) depict the interaction of VCP with bovine factor I, while the right panels (C) depict the interaction of VCP with human factor I. (i) The upper left panel shows a direct chain ionic interaction between oppositely charged residues of VCP (E108) and bovine factor I (K108); no such interaction was found between E108 of VCP and human factor I (upper right panel). (ii) The middle left panel shows an ionic interaction between side chains of E120 of VCP and K92 of bovine factor I; once again, such an interaction was not observed between E120 of VCP and human factor I (middle right panel). (iii) The lower left panel shows interactions of E144 and S131 of VCP with R425 of bovine factor I. The side chain of this arginine residue seems to sit in a pocket created by the main chain of E144 and side chain of S131. The main chain of E144 of VCP shows direct interaction with the oppositely charged side chain of R425 of bovine factor I; the R425 of bovine factor I also shows weak interactions with S131 of VCP.

To examine whether the negatively charged residues are conserved at positions 108, 120, and 144 in VCP of various vaccinia virus strains as well as orthologs of VCP expressed by other poxviruses that infect domestic animals, we aligned the sequences of VCP and VCP-like sequences. For selecting proper sequences for the alignment, we first constructed the phylogenetic tree of VCP-like sequences of various poxviruses using the neighbor joining method (45). Thereafter, we selected the sequences that clustered only in the VCP clade. This resulted in exclusion of B5R-like sequences as well as VCP-like sequences containing only CCP domain 2 or 3 with transmembrane regions. From the alignment, it was clear that the negatively charged residues are well conserved in positions in VCP of various strains of vaccinia viruses as well as VCP orthologs of cowpox virus, ectromelia virus, rabbitpox virus, camelpox virus, and horsepox virus (Fig. 10).

FIG 10.

Sequence alignment of poxviral complement regulators. The protein sequences of CCP domains 2 and 3 of VCP and VCP orthologs from various poxviruses were aligned using Clustal Omega (version 1.2.3). The arrows in black indicate conserved residues at positions 108, 120, and 144. Abbreviations: VAR-B, Variola virus-Bangladesh-1975; VAR-C, Variola virus-Congo-1965; VAR-G, Variola virus-Garcia-1966; VAR-I67, Variola virus-India-1967; VAR-I71, Variola virus-India-1971; CPV-FR, Cowpox virus-France 2001; CPV-BR, Cowpox virus-Brighton Red; CPV-N, Cowpox virus-Norway-1994; CPV-G90, Cowpox virus-Germany-1990; CPV-UK, Cowpox virus-UK 2000-K2984; ECT-H, Ectromelia virus isolate “Hampstead egg”; MON-Z, Monkeypox virus-Zaire-96-I-16; ECT-M, Ectromelia virus-Moscow; ECT-N, Ectromelia virus isolate “Naval 1995”; HSPV, Horsepox virus “MNR-76”; VACV-DUKE, Vaccinia virus-Duke; CMPV, Camelpox virus isolate “M-96”; CPV-GRI, Cowpox virus GRI-90; CPV-FIN, Cowpox virus-Finland-2000; ECT-I, Ectromelia virus isolate “Ishibashi-I-111”; RPV, Rabbitpox virus-Utrecht; VACV-A, Vaccinia virus Acambis; VACV-WR, Vaccinia virus Western Reserve; VACV-DRYV, Vaccinia virus Dryvax; VACV-LC, Vaccinia virus LC 16m8; CPV-A, Cowpox virus Austria 1999; VACV-GLV, Vaccinia virus-GLV-1h68.

DISCUSSION

Poxviruses thwart the host complement system by employing two strategies: (i) they encode their own soluble complement regulator (12, 46); (ii) they pirate the membrane-bound complement regulators from the host (29). Intriguingly, the sequence similarity among various poxvirus-encoded complement regulators exceeds 90%, suggesting their recent evolutionary split and low mutation rates. Further, it also implies that the mutations accumulated in the poxviral complement regulators after the split are likely to be significant with respect to their adaptation in new hosts, as these viruses have been shown to evolve in the presence of host pressure (47). Earlier, we have shown that vaccinia virus complement regulator VCP displays selectivity toward the bovine alternative complement pathway (20). Here, we show that its selectivity is also extended toward the bovine classical complement pathway, which is capable of neutralizing both infectious forms of vaccinia—the intracellular mature virus and the extracellular enveloped virus (28, 29). Notably, our data indicate that the determinants responsible for the specificity of VCP for the bovine CP are the acidic residues (E108, E120, and E144) present in CCP domains 2 and 3 of VCP that interact with bovine factor I.

Examination of the relative levels of inhibitory activity of VCP and SPICE against the classical complement pathway of various species showed that VCP is much (43-fold to 45-fold) more effective against bovine complement than SPICE; such differences, however, were not observed against primate complement (Fig. 1). Since these proteins accumulated only 11 mutations after the split, the data indicated that the selectivity of VCP for the bovine CP is driven by one or more of these variant residues and prompted us to determine which residue(s) contributes to the specificity. Analysis of the bovine CP inhibitory activity of 11 single-residue substitution mutants of SPICE indicated that it is primarily the negatively charged residues (E108, E120, D178, and E144) of VCP that determine its specificity (Fig. 3). In line with this, previous data on the specificity of VCP for the bovine alternative pathway also indicated that the Glu residues at positions 108, 120, and 144 are the key functional determinants (20). Thus, accumulation of negatively charged residues in CCP domains 2 and 3 seems to provide a functional advantage to VCP with respect to inhibiting both the classical and alternative pathways of bovine complement. It is notable that the mutant with a K108E mutation in SPICE was the most influential in CP lysis inhibition and that its activity was nearly equivalent to that of VCP, which was not reflected in its cofactor activity against C4(H₂O) (Fig. 3 and 5; Table 1). We suggest that the increased effect of the mutant seen in CP lysis is due to its combined effects on the CP and on the alternative pathway loop as this loop is known to be activated during CP activation and to contribute to lysis (48, 49). It should be noted that this mutant has the largest effect on the bovine alternative pathway (20).

Like the human RCA proteins, VCP regulates human complement by targeting C3-convertases (50); this was attributed to its ability to support factor I-mediated inactivation of C3b and C4b (cofactor activity) and to accelerate the decay of the CP C3-convertase (decay-acceleration activity) (38, 39). It was thus expected that the inhibitory effect of VCP on the bovine CP would be due to its cofactor activity for C4b [here we utilized C4(H₂O)] as well as to its decay-acceleration activity for the CP C3-convertase C4b2a. However, our data indicated that its inactivation of bovine CP is solely due to its cofactor activity for C4(H₂O), with little effect, if any, on the decay of the CP C3-convertase (Fig. 4 and 5). Notably, these data differ from those showing VCP's ability to target the bovine alternative pathway C3-convertase in that it regulates alternative pathway C3-convertase by inactivating C3b as well as by accelerating the decay of the convertase (20). The likely explanation for VCP's ability to decay human CP C3-convertase, but not bovine CP C3-convertase, is presented below. According to a prevalent model of CP C3-convertase decay mediated by the RCA proteins, the decay is a result of binding of the RCA protein to both subunits of the convertase, i.e., to C4b and C2a (51, 52). Thus, the ability of VCP to efficiently decay human CP C3-convertase, but not bovine CP C3-convertase, is likely due to its lack of binding to bovine C4b or to C2a or both. A comparison of the levels of direct binding of VCP to human and bovine C4b/C4(H₂O) indicated that it binds equally well to the two proteins (data not shown). Thus, its perturbed ability to decay the bovine CP C3-convertase is likely due to its lack of binding to C2a.

Our efforts to identify the minimum residues that dictate the specificity of VCP for the bovine CP indicated that three mutations in SPICE (K108E/K120E/N144E) are enough to switch its specificity to the bovine CP and that this is due to a gain in the cofactor activity (Fig. 7). Because cofactor activity is driven by the interaction of the regulator with the target protein [here, C4(H₂O)] as well as with factor I, the mutations in SPICE described above are likely to enhance its interaction with one or both of these proteins. Data obtained here by utilizing a direct binding assay as well as a cofactor assay and a combination of bovine factor I and human factor I (Fig. 7 and 8) indicated that the charged residues at positions 108, 120, and 144 interact directly with bovine factor I and dictate the specificity. This conclusion is also supported by the earlier data on the specificity of VCP for the bovine alternative pathway wherein E108, E120, and E144 were suggested to interact with bovine factor I (20). Although C3b and C4b are structural homologs, it is not yet clear whether factor I docks onto C3b-RCA and C4b-RCA complexes in similar manners. Our data support the view that the manners in which factor I docks onto these complexes are similar.

Though biochemical data obtained clearly pointed out the involvement of E108, E120, and E144 of VCP in binding to bovine factor I, the structural basis of these interactions was not clear. Thus, to obtain a detailed view of the functionally important interactions, we mapped the critical residues onto the modeled structures of bovine C4b-VCP-bovine factor I and bovine C4b-VCP-human factor I complexes. It is pertinent that modeling of these complexes was possible solely due to the availability of the crystal structures of C3b-SPICE complex (24) as well as of C4b (53) and factor I (42). Modeling of the bovine C4b-VCP structure and examination of E108, E120, and E144 of VCP in this model showed that these residues are clearly exposed to the solvent and are located away from the binding interface. Next, to help reveal whether these residues interact with factor I in the ternary complex, we took leads from the previous work (43) and placed bovine factor I onto the bovine C4b-VCP complex. A closer look at VCP residues E108, E120, and E144 (which impart enhanced cofactor activity) showed that these residues create favorable ionic interactions with K108, K92, and R425 of factor I, respectively. Additionally, we observed that S131 of VCP also makes a contact with R425 of factor I, which explains why the SPICE-Tetra mutant (K108E/K120E/L131S/N144E) is a better cofactor than the SPICE-Triple I mutant (K108E/K120E/N144E) (Table 1). Such favorable interactions, however, were absent when human factor I was placed onto the bovine C4b-VCP complex instead of bovine factor I, which explains our in vitro data. Interestingly, of the three critical bovine factor I residues, K108 and K92 are located in the factor I-membrane attack complex (FIMAC) domain, while R425 is located in the serine protease (SP) domain. An earlier mutagenesis study on human factor I demonstrated that mutations in the FIMAC and SP domains in particular obliterate its ability to inactivate C3b and C4b, irrespective of the RCA protein used (54, 55). Our model thus supports the findings described above.

In summary, our results indicate that VCP's specificity for the bovine CP is a consequence of its enhanced C4b cofactor activity, which is shaped by the presence of three acidic residues of domains 2 and 3 that make a direct interaction with bovine factor I. Importantly, these acidic residues are conserved in VCP of various poxviruses that infect domestic animals. We therefore suggest that poxviral complement regulators are among of the mediators of poxvirus tropism. These results also add to our knowledge on how poxviruses overcome the host complement system to induce a state of viral pathogenesis.

MATERIALS AND METHODS

Reagents, complement proteins, and buffers.

Anti-sheep erythrocyte antibody was procured from ICN Biomedical Inc. (Irvine, CA). Human factor I (FI) was purchased from Complement Technology, Inc. (Tyler, TX). Bovine factor I was purified as described previously (20). Cobra venom factor (CVF) was purified from Naja naja kaouthia venom as described previously (56) with minor modifications (57). Veronal-buffered saline (VBS) contained 5 mM barbital, 145 mM NaCl, and 0.02% sodium azide (pH 7.4). Gelatin veronal buffer (GVB) was VBS containing 0.1% gelatin, GVB⁺⁺ was GVB containing 0.5 mM MgCl₂ and 0.15 mM CaCl₂, and GVBE was GVB with 10 mM EDTA. MgEGTA contained 0.1 M MgCl₂ and 0.1 M EGTA, and phosphate-buffered saline (PBS; pH 7.4) contained 10 mM sodium phosphate and 145 mM NaCl.

Bovine complement C4 purification and C4(H₂O) generation.

Unlike human C4, bovine C4 is highly susceptible to plasma proteases. The purification method described below, which is a modified version of a previously described method (58), resulted in successful purification of bovine C4. Here, calf blood was collected in 136 mM trisodium citrate (10:1 [vol/vol]). Thereafter, it was kept on ice and the inhibitor solution (1 M KH₂PO₄, 1 M benzamidine hydrochloride, 0.4 M EDTA, 0.02% sodium azide, pH 7.4) (20:1 [vol/vol]) was added to the blood. It was then centrifuged at 3,000 rpm for 10 min at 4°C to collect the plasma, which was mixed with 50 mM ε-amino caproic acid (EACA; Sigma-Aldrich, St. Louis, MO) prior to freezing at −80°C. To purify C4, the frozen plasma was thawed, mixed with 2.5 mM diisopropyl phosphorofluoridate (DFP) (Chemsworth, Surat, India) and 1 mM phenylmethylsulfonyl fluoride (PMSF), and subjected to precipitation by adding 65 mM BaCl₂ (MP Biomedicals, Santa Ana, CA). The supernatant was diluted 2-fold with 10 mM Tris (pH 8.2) containing 100 mM NaCl, 10 mM EDTA, 10 mM benzimidine hydrochloride, 50 mM EACA, 0.02% sodium azide, and 2.5 mM DFP and loaded onto a Q Sepharose FF 16/10 column (GE Healthcare Life Sciences, Pittsburgh, PA) equilibrated in the buffer mentioned above. Elution of the bound proteins was performed with a linear gradient of 100 to 500 mM NaCl. The fractions collected were mixed immediately with 2.5 mM DFP and 6 mM sodium phosphate to adjust pH to 7.4. The bovine C4-containing fractions, as identified by 9% SDS-PAGE and Western blot analysis using anti-C4d monoclonal antibody (Abcam, MA), were pooled, dialyzed against 10 mM Tris (pH 8.2) containing 100 mM NaCl, and loaded onto a preequilibrated Mono Q 10/10 column (Amersham Pharmacia Biotech, Uppsala, Sweden). The bound protein fractions eluted over a linear gradient of 100 to 800 mM NaCl were spiked with 2.5 mM DFP, and the pH was adjusted to 7.4. The fractions containing C4 were pooled and dialyzed against 20 mM sodium acetate (pH 5.6) and loaded onto a preequilibrated Mono S (5/5) column (Amersham Pharmacia Biotech). The bound proteins were eluted with a linear gradient of 0 to 800 mM NaCl, and fractions were replenished with 2.5 mM DFP. The fractions containing homogeneous bovine C4 as judged by SDS-PAGE and Western blot analysis were pooled, concentrated, and dialyzed against PBS (pH 7.4) containing 0.02% sodium azide. The purity of bovine C4 was ∼85%. To generate C4(H₂O), which is a conformational equivalent of C4b, the aliquots of the purified native C4 were freeze-thawed >20 times. Each freeze/thaw cycle constituted 30 min of freezing at −80°C followed by 10 min of thawing at 37°C.

Construction, expression, and purification of single- and multi-amino-acid mutants of SPICE.

The construction of SPICE single-amino-acid mutants was described previously (20). The multi-amino-acid substitution mutants of SPICE were generated by the use of QuikChange II and QuikChange multisite-directed mutagenesis kits (Stratagene, La Jolla, CA). Earlier, our laboratory had constructed SPICE-Double (K108E/K120E) and SPICE-Triple 1 (K108E/K120E/N144E) mutants in pGEM-T Easy vector (20). These were used as a template to generate SPICE-Triple 2 (K108E/K120E/N178D) and SPICE-Tetra (K108E/K120E/KN144E/L131S) mutants, respectively. The SPICE-Penta (K108E/K120E/KN144E/L131S/H77Q) mutant was generated using the SPICE-Triple 1 (K108E/K120E/KN144E) clone. The primers utilized for generating these multiresidue mutants were the same as those used earlier (20). The authenticity of these clones was verified by automated DNA sequencing (1st Base; Axil Scientific, Singapore). All the mutants were subcloned into the pET-29b vector (Novagen) and transformed into Escherichia coli BL21 cells for expression. The expression, purification, and refolding of VCP, SPICE, and SPICE multi-amino-acid mutants were performed as described previously (19, 20). The refolded proteins were subjected to gel filtration (Superose-12 column; Amersham Pharmacia Biotech) to obtain the monodispersed population.

Classical pathway hemolytic assay.

Inhibition of the classical complement pathway activity of sera of different species by VCP, SPICE, and SPICE mutants was studied by measuring the inhibition of CP-mediated lysis of antibody-sensitized sheep erythrocytes (EA) as described previously (37). In brief, various concentrations of VCP, SPICE, or the mutants were incubated with 5 μl (10⁹/ml) EA in the presence of indicated sera in a total volume of 250 μl at 37°C for 1 h; the volume was adjusted by adding GVB⁺⁺. After incubation, the cells were spun at 3,000 rpm for 3 min and absorbance of the supernatants was read at 405 nm. The data collected were normalized by considering 100% lysis to be equal to the lysis that occurred in the absence of an inhibitor. Data were fitted using nonlinear regression analysis (Grafit; Erithacus Software, London, United Kingdom), and a four-parameter logistic analysis was performed to determine the best-fit 50% inhibitory concentration (IC₅₀) value. Activity differences of ≥3-fold were considered significant (20, 43, 51).

Factor I cofactor activity assay.

The factor I cofactor activity of SPICE, VCP, and the SPICE mutants against bovine C4(H₂O) was measured by employing a fluid-phase assay (41). Briefly, 24 μg of bovine C4(H₂O) was mixed with 8 μg of VCP, SPICE, or the indicated mutant and 1.6 μg of factor I in a total volume of 150 μl PBS (pH 7.4). The reaction mixture described above was then incubated at 37°C, and 15 μl of the reaction volume was taken out at the indicated time intervals and mixed with 5× SDS-PAGE sample buffer containing dithiothreitol (DTT). The cleavage products were resolved on 9% SDS-PAGE. The percentage of C4(H₂O) cleaved was determined by densitometric analysis (Quantity One software; Bio-Rad, Hercules, CA) of the remaining α-chain of C4(H₂O), which was normalized to the β-chain (loading control). Data are presented as the time required for 50% cleavage of the α-chain of C4(H₂O), which was calculated by plotting the percentage of the α-chain of C4(H₂O) that remained against time. Activity differences of ≥3-fold were considered significant (20, 43, 51).

Bovine CP C3-convertase decay acceleration assay.

The decay acceleration activity of VCP, SPICE, and the SPICE mutants against the bovine CP C3-convertase was assessed by assembling the C3-convertase on antibody-coated sheep erythrocytes (EA) using cobra venom factor (CVF)-treated bovine sera, as this cobra protein spares CP proteins C1, C4, and C2 but activates and depletes C3 and the later components (59). First, bovine sera was treated with CVF (10 μg/ml) in the presence of 5 mM MgCl₂ and kept at 37°C for 5 min. A 150-μl volume of this serum was thereafter mixed with 30 μl of EA (10⁹/ml) and 120 μl of GVB⁺⁺ and incubated for 2 min at 30°C to form the C3-convertase. The formation of the enzyme was stopped by adding 300 μl ice-cold GVB containing 40 mM EDTA. For examining the decay-accelerating activity of the viral regulators, 10 μl (0.5 × 10⁹/ml) of the C3-convertase coated EA was incubated with different concentrations of regulators in a total volume of 25 μl GVB⁺⁺ and allowed to decay for 4 min at 22°C. These reaction mixtures were then mixed with 225 μl of 1:100-diluted guinea pig serum in GVB containing 40 mM EDTA and incubated for 30 min at 37°C. The reaction mixtures were centrifuged, and the percentage of lysis was determined by measuring the absorbance at 405 nm. Data were normalized by considering 100% C3-convertase activity to be equal to the lysis that occurred in the absence of any regulator. Data show means ± standard deviations (SD) of results of three independent experiments. The concentration of inhibitor required to inhibit 50% of enzyme activity (IC₅₀) was calculated by plotting the C3-convertase activity against the inhibitor concentration. Activity differences of ≥3-fold were considered significant (20, 43, 51).

Biotinylation of bovine C4(H₂O).

C4(H₂O), which is a conformational equivalent of C4b, was biotinylated at its free –SH group using EZ-Link PEO-maleimide-activated biotin (Pierce, Rockford, IL) as described previously (41). Briefly, 0.5 ml (1.5 mg/ml) of bovine C4(H₂O) was dialyzed against 0.1 M sodium phosphate (pH 6.5) containing 5 mM EDTA for 16 h at 4°C. Next, freshly prepared 20 mM PEO-maleimide biotin solution (in 0.1 M sodium phosphate [pH 7.2] containing 150 mM NaCl and 1 mM EDTA) was mixed with bovine C4(H₂O) at a 1:75 molar ratio for 30 min at room temperature. This reaction mix was then extensively dialyzed in PBS (pH 7.4) containing 0.02% sodium azide at 4°C to remove free biotin. To verify the site-specific biotinylation, C4(H₂O) was incubated with 2 μg VCP and 0.2 μg factor I in 15 μl PBS and the reaction was subjected to Western blot analysis using avidin-horseradish peroxidase (HRP) conjugate (Bio-Rad) and a Pierce ECL Plus kit (Thermo Scientific, Waltham, MA). As expected, biotinylation was observed only in the C4d region, which contains the free –SH group (data not shown).

SPR measurements.

Binding of VCP, SPICE, and the SPICE mutants to bovine C4(H₂O) was analyzed by the use of a surface plasmon resonance (SPR)-based biosensor Biacore 2000 system (Biacore, Uppsala, Sweden) as described previously (41). Here, the test flow cell (FC-2) was formed by immobilizing the site-specifically biotinylated bovine C4(H₂O) (∼5,827 response units [RU]) on a streptavidin chip (Sensor Chip SA; Biacore); a biotinylated bovine serum albumin (BSA)-immobilized (Sigma-Aldrich) flow cell (Fc-1) served as the control flow cell. To measure binding, 20 nmol of each of the regulators was flowed over the chip at a 50 μl/min flow rate at 25°C in PBS containing 0.05% Tween 20. Association and dissociation of the regulators were measured for 120 s. The sensor chip was regenerated by two 30-s pulses of 0.2 M sodium carbonate (pH 9.5). The specific binding of the viral regulators/mutants to bovine C4(H₂O) was determined by subtracting the control flow cell data from the test flow cell data.

Molecular modeling of bovine C4b-VCP-bovine factor I trimolecular complex.

To generate a model representing the ternary complex of bovine C4b:VCP:bovine FI, we first generated homology models of each of the components of the complex; these were then employed to generate the ternary complex as previously described (43). For generation of homology models of bovine C4b, we utilized the crystal structure of human C4b (PDB ID: 5jtw) (53) as the template. Briefly, the structure was modeled as different chains—A (β), B (α), and C (γ)—present in the 5jtw template structure. Hence, alignment was performed between a NCBI reference sequence (NP_001159957.1) and a crystal structure-derived sequence using ClustalW (60) and modeling was performed using Modeler 9.15 (61, 62). Overall, 50 different models were made for each of the chains, and the best one was chosen using Discrete Optimized Protein Energy (DOPE) scoring. Next, VCP was modeled using the crystal structure of SPICE (PDB ID: 5fob) (24). Here, the 11 variant amino acids of SPICE compared to VCP were changed to the corresponding VCP amino acids. Specifically, care was taken to perform such replacements in a rotamer conformation similar to that in the original crystal structure. Lastly, for generation of the homology model of bovine factor I, a reference sequence (NP_001033185.1) was aligned with the sequence of human factor I structure (PDB ID: 2xrc) (42), and 50 different models were generated using Modeler 9.15. The best model was chosen based on the lowest DOPE score. In all the models generated, the regions/loops for which the structure was missing either was not modeled or was removed postmodeling.

For generating the model of bovine C4b:VCP:bovine factor I complex, we first modeled the bovine C4b:VCP complex using the human C3b:SPICE structure (PDB ID: 5fob) (24). Briefly, bovine C4b modeled as described above was superposed on the human C3b:SPICE structure to obtain a model for the bovine C4b:SPICE complex. Later, SPICE was replaced by the VCP model (created as described above) to obtain the bovine C4b-VCP complex model. Next, to obtain a model of the ternary complex, bovine factor I was added to this model as achieved previously for modeling the C3b:factor H:factor I complex (42) and the C3b:Kapo:factor I complex (43). Of note, building of the model for the ternary complex required tilting of the C345C domain to avoid clashes between factor I and the C345C domain (43). In addition, the N-terminal region of the bovine factor I model was also reoriented to avoid stearic clashes in the bovine C4b:VCP:bovine factor I complex model (43). In the final model, one of the loops of factor I (residues G110 to R112) was seen to pass through a loop of VCP (residues L55 to G57), which is the limitation of our model. To explain the functional differences that we observed in our in vitro data, we also placed human factor I in the model for the bovine C4b:VCP:bovine factor I complex. Both the final models were subjected to energy minimization using YASARA (63). All the models were checked for geometrical and stereochemical properties using PROCHECK (44). In our trimolecular complex model of bovine C4b:VCP:bovine factor I, among a total of 2,274 residues, 87.2%, 10.8%, 1%, and 1% were located within the core, additional allowed, generously allowed, and disallowed regions of the Ramachandran plot, respectively. Similarly, in our bovine C4b:VCP:human factor I model, of the 2,277 residues, the Ramachandran plot showed 87.3%, 10.6%, 1.1%, and 1% in the core, additional allowed, generously allowed, and disallowed regions, respectively. None of the disallowed residues in these models were a part of the interacting domain or region that was of interest to us (data not shown). Residue numbers of VCP are presented according to the sequence of the mature protein and that of bovine C4, and human and bovine factor I GenBank accession numbers are given below.

Accession number(s).

Human and bovine factor I sequences are available in GenBank under accession numbers NP_001159957.1 (complement C4 precursor [Bos taurus]), GI:1335054 (unnamed protein product, partial [Homo sapiens]), and NP_001033185.1 (complement factor I precursor [Bos taurus]).

Data analysis.

Functional activity comparisons were performed by measuring the fold change. As described earlier (20, 43, 51), activity differences of ≥3-fold were considered significant. The fold changes reported represent averages of results of at least three independent experiments.