Mutational analysis of Kaposica reveals that bridging of MG2 and CUB domains of target protein is crucial for the cofactor activity of RCA proteins

Significance

The complement system discriminates self from nonself targets solely owing to the presence of complement regulators on the host cells, which primarily belong to the regulators of complement activation (RCA) protein family. These regulators control complement activation on the host cell surface by two mechanisms termed “cofactor activity” (CFA) and “decay-accelerating activity.” Here, we have identified the critical structural determinants of an RCA protein responsible for imparting the CFA and show that these determinants bridge MG2 (macroglobulin-2) and CUB (complement C1r-C1s, Uegf, Bmp1) domains of C3b and interact with factor I. As a proof of principle, we show that incorporation of putative membrane cofactor protein (CD46) regions responsible for the bridging and factor I interaction in decay acceleration factor (CD55) results in gain of function. We, thus, define the molecular events that govern CFA.

Abstract

The complement system has evolved to annul pathogens, but its improper regulation is linked with diseases. Efficient regulation of the system is primarily provided by a family of proteins termed regulators of complement activation (RCA). The knowledge of precise structural determinants of RCA proteins critical for imparting the regulatory activities and the molecular events underlying the regulatory processes, nonetheless, is still limited. Here, we have dissected the structural requirements of RCA proteins that are crucial for one of their two regulatory activities, the cofactor activity (CFA), by using the Kaposi’s sarcoma-associated herpesvirus RCA homolog Kaposica as a model protein. We have scanned the entire Kaposica molecule by sequential mutagenesis using swapping and site-directed mutagenesis, which identified residues critical for its interaction with C3b and factor I. Mapping of these residues onto the modeled structure of C3b–Kaposica–factor I complex supported the mutagenesis data. Furthermore, the model suggested that the C3b-interacting residues bridge the CUB (complement C1r-C1s, Uegf, Bmp1) and MG2 (macroglobulin-2) domains of C3b. Thus, it seems that stabilization of the CUB domain with respect to the core of the C3b molecule is central for its CFA. Identification of CFA-critical regions in Kaposica guided experiments in which the equivalent regions of membrane cofactor protein were swapped into decay-accelerating factor. This strategy allowed CFA to be introduced into decay-accelerating factor, suggesting that viral and human regulators use a common mechanism for CFA.

The complement system is the central component of innate immunity that controls invading pathogens directly by lysis or inactivation and indirectly by recruiting and boosting the pathogen-specific adaptive immunity (1, 2). Although the system efficiently targets the invaders, it is equally deleterious to the host cells, and therefore, an effective regulation is needed to control complement activation on the cell surface. This regulation is primarily achieved by a family of proteins termed regulators of complement activation (RCA), which are located on the cell surface, such as decay acceleration factor (DAF; CD55), membrane cofactor protein (MCP; CD46), and complement receptor 1 (CR1; CD35), and in solution, like factor H (FH) and C4b-binding protein (3). The RCA proteins are formed by 4–59 complement control protein (CCP) modules that are separated by small linkers and regulate complement by inactivating C3 convertases by two distinct mechanisms dubbed as “cofactor activity” (CFA) and “decay-accelerating activity,” which work in concert to achieve the robust regulation. It is, therefore, not inexplicable that mutations and polymorphisms in RCA proteins are linked to various diseases, like age-related macular degeneration, atypical hemolytic uremic syndrome, and dense deposit disease (4–6).

Complement regulation owing to CFA refers to serving of RCA protein as a cofactor during serine protease factor I-mediated inactivation of C3b and C4b (the subunits of C3 convertases). To examine the structural requirements for the CFA of RCA proteins, initial studies focused on the identification of minimum CCP modules required for this activity and identified three CCPs as the smallest structural unit essential for this activity (3, 7, 8). Later, mutagenesis studies primarily performed on CR1 and MCP identified residues within the CCPs and concluded that functional sites are located in each of the three functionally critical CCPs (9, 10). A major advance in understanding the molecular mechanism of CFA came only recently because of the availability of structural and biochemical data (11–14). In particular, structure of FH(1–4) in complex with C3b identified the contact regions of FH(1–4) and their footprints on C3b (13). Also, the structure of factor I revealed allosteric regulation in factor I and identified the putative factor I interaction sites on the regulator and C3b (14). Although the aforementioned studies elucidated that interaction of three to four modules of a regulator with the target proteins C3b or C4b forms a docking platform for factor I, resulting in binding and reorientation of factor I for cleavages in the CUB (complement C1r-C1s, Uegf, Bmp1) domains of C3b/C4b, there are questions that still remain unanswered. (i) Among the contact regions for C3b/C4b found in each of the functional domains of a regulator (13), which region(s) is most critical for imparting CFA and why? (ii) Which residues of a regulator participate in the interaction with factor I? (iii) Do different regulators use a common molecular mechanism to inactivate C3b and C4b?

Among the human pathogens, RCA proteins are known to be encoded by pox (e.g., variola) and herpes viruses [e.g., Kaposi's sarcoma-associated herpesvirus (KSHV)]. Also, intriguingly, these mimics also regulate C3 convertases through their decay-accelerating activity and CFA (15). Thus, these viral proteins can serve as useful tools to understand the functioning of RCA proteins. Extensive mutagenesis studies in these viral regulators have led to the understanding that CCP modules 2 and 3 are critical for imparting CFA and that the factor I binding site resides in these domains (16–19). However, the underlying molecular mechanisms of complement regulation by viral RCA are still enigmatic.

In this study, we, therefore, have used Kaposica, the complement regulator of Kaposi’s sarcoma-associated herpesvirus, as a model protein to further probe the molecular mechanism underlying CFA. By performing extensive mutagenesis and biochemical characterization of Kaposica mutants and combining these data with the earlier structural data (13, 14), we propose that, apart from providing a docking surface for factor I, bridging of the MG2 (macroglobulin-2) and CUB domains of C3b/C4b by Kaposica is critical for imparting CFA. Furthermore, based on the DAF-MCP chimera data, we propose that such bridging is also vital for CFA of the human RCA proteins.

Results

CCP Modules 2 and 3 Are Essential for the CFA of Kaposica.

Structurally, Kaposica is composed of four CCP modules followed by a serine/threonine (S/T) -rich region and a transmembrane domain for membrane attachment, wherein regulatory activities are provided by the CCP modules. Thus, as a first step toward identification of its functional site(s) essential for the CFA, we sought to map its CCPs critical for this activity. Because domain swapping allows for discerning the importance of the individual modules in the context of the full-length molecule, we used this approach here to identify which modules are critical for the CFA of Kaposica. There is, however, a caveat—swapping strategy, although superior than domain deletions, may lead to structural alteration because of disruption of the interaction between modules. Having said the above, we would like to point out that this strategy did prove successful in identifying the functional modules in another viral RCA: vaccinia virus complement control protein (VCP) (17). We, thus, swapped each of the Kaposica modules with the homologous modules of human DAF, which is also composed of four CCP modules but lacks CFA. Because modules 1–3 of Kaposica are homologous to modules 2–4 of DAF, we swapped the respective modules of Kaposica with those of DAF and named the mutants based on the modules that we swapped from DAF (Fig. 1). These chimeric mutants along with the WT protein were then expressed and purified to homogeneity (SI Appendix, SI Materials and Methods). Next, the purified chimeras were analyzed for the CFA by using a fluid-phase assay. The results showed that swapping of module 1 of Kaposica with that of the corresponding DAF module had no effect on its C3b and C4b CFA but that swapping modules 2 or 3 severely decreased its CFA (Fig. 1 and SI Appendix, Fig. S1A and Table S1), suggesting that these modules are critical for the CFA of Kaposica.

Fig. 1.

CFA resides in domains 2 and 3 of Kaposica (Kapo). (A) Diagrammatic representation of Kapo, DAF, and the domain swap mutants. Domains of Kapo as well as DAF are numbered. The names of mutants (D2–D4) were assigned based on the incorporated DAF domain. The domain boundaries have been marked by vertical lines, and the numbers denote the respective boundary residues. (B) SDS/PAGE analysis of purified Kapo, DAF, and the domain swap mutants. (C and D) CFA of Kapo, DAF, and the domain swap mutants for (C) C3b and (D) C4b. Data shown in the graphs are one of three experiments summarized in SI Appendix, Table S1. MW, molecular mass.

Intercysteine Regions Adjacent to Module 2–3 Linker Are Critical for the CFA of Kaposica.

Because the above data clearly pointed toward the crucial requirement of modules 2 and 3 of Kaposica for its CFA, we next pursued the determination of the region(s) within domains 2 and 3 that are decisive for this activity. Each CCP module contains four invariant cysteines; therefore, to localize the critical regions within the CCPs, we swapped the intercysteine regions of modules 2 and 3 of Kaposica with the homologous regions of DAF. Homologous mutagenesis has also been performed earlier in CR1 between CCP1 and CCP2 and homologous CCP8 and CC9 to identify sites critical for C3b/C4b binding and CFA (20). In this exercise, we constructed six region swap chimeras, which were named according to the swapped region of DAF (Fig. 2). These chimeras were then expressed, purified, and assessed for CFA. Although such mutations could affect the structure, all of the mutants maintained proper conformation (SI Appendix, SI Materials and Methods). It became apparent from the results that two chimeras, namely D3R3 and D4R1, were almost completely devoid of C3b CFA, whereas the other chimeras retained the activity. Intriguingly, the same two chimeras also showed considerable loss in C4b CFA, although the effect was less pronounced in D3R3 chimera (Fig. 2 and SI Appendix, Fig. S1B and Table S1). Collectively, these data suggest that the intercysteine regions adjacent to the linker between modules 2 and 3 (residues M113–F128 and H135–E160) are vital for the CFA and that there are subtle differences in the structural requirements for C3b and C4b CFA of Kaposica.

Fig. 2.

CFA is mapped to the central intercysteine regions of Kaposica (Kapo). (A) Diagrammatic representation of Kapo, DAF, and the intercysteine region swap mutants. The mutant names (D3R1, D3R2, D3R3, D4R1, D4R2, and D4R3) were assigned based on the incorporated DAF region. The colored lines and the numbers mark the boundaries of the R1, R2, and R3 regions of D3 and D4 domains of DAF. (B) SDS/PAGE analysis of purified Kapo, DAF, and the region swap mutants. (C and D) CFA of Kapo, DAF, and the region swap mutants for (C) C3b and (D) C4b. Data shown in the graphs are one of three experiments summarized in SI Appendix, Table S1. MW, molecular mass.

Ala Scanning Mutagenesis of Intercysteine Regions Reveals Functional Sites for CFA of Kaposica.

Having localized the intercysteine regions critical for the CFA of Kaposica (Fig. 3A), we next followed the identification of functional residues within these regions by Ala substitution mutagenesis of the residues predicted to be surface-exposed. Because 3D structure of Kaposica is not available, we built a homology model of this molecule using the structure of the murine complement regulator Crry as the template (Fig. 3B) and used it to identify the putative surface-exposed residues. This exercise led to the identification of 25 residues in the functionally critical intercysteine regions of CCP2 (K2R3; residues M113–F128) and CCP3 (K3R1; residues H135–E160) (Fig. 3A), which were then selected for mutagenesis. In addition, we also selected three other residues (R35, F102, and Y108) outside of these regions based on the previous mutagenesis data of FH and smallpox inhibitor of complement enzymes (SPICE) (18, 21) (Fig. 3 A and B) and one residue (E130) predicted earlier that was based on docking of factor I onto the C3b–FH(1–4) structure (14) (Fig. 3 A and B). The Ala substitution mutants were then generated as single, double, or triple site-directed substitution mutants, wherein residues for mutations were clubbed together primarily based on their properties, except in the K2 mutant, wherein residues were selected as per the predicted factor I binding site (14). Consequently, we generated 16 mutants, of which 11 were multiresidue mutants and 5 were single-residue mutants (Fig. 3 C and D).

Fig. 3.

Model of Kaposica and gel analysis of its Ala substitution mutants. (A) Kaposica sequence identifying CCP domains (black) and linkers (blue); invariant cysteine residues are highlighted in pink. The intercysteine regions highlighted in yellow (CCP2 region, K2R3; CCP3 region, K3R1) have been identified as critical regions for the CFA of Kaposica. The earlier predicted factor I-interacting residues are highlighted in green. The superscript numbers denote the first residue number of each CCP domain. (B) Surface representation of Kaposica model depicting positions of the Ala substitution mutations in color. (C) Table showing the list of mutants. (D) SDS/PAGE analysis of Kaposica (Kapo) and its Ala substitution mutants. MW, molecular mass; Sr. no., serial number.

Evaluation of the Ala substitution mutants for CFA showed significant loss of C3b CFA in eight mutants (K2, K3, K6, K9, K10, K12, K13a, and K15) and C4b CFA in four mutants (K2, K6, K12, and K15) (Fig. 4 and SI Appendix, Fig. S1 C and D and Table S1). It is, therefore, evident from these data that, although some residues of Kaposica participate in both C3b and C4b CFAs, there are residues that participate only in the C3b CFA. Interestingly, among all of the mutants that were found to be important for CFAs, only two mutants (K2 and K3) belonged to the K2R3 region. Also, among these mutants, only one mutant (K2) showed significantly low C3b and C4b CFA. Among the residues selected for mutagenesis that were outside the critical intercysteine regions, E130 was found to be significant for both C3b and C4b CFA, whereas F102 was found to be important only for C3b CFA.

Fig. 4.

Identification of residues important for CFA of Kaposica (Kapo). CFA of Kapo and its site-directed mutants for (Upper) C3b and (Lower) C4b. Data shown in the graphs are one of the three experiments summarized in SI Appendix, Table S1.

Because CFA is a summation of interaction between the regulator and the target protein (C3b or C4b) as well as the regulator and factor I, we measured direct binding of the mutants to C3b and C4b using surface plasmon resonance to determine whether reduction in CFA of the substitution mutants is owing to decrease in their ability to bind to C3b or C4b. We observed that four [K9 (F144A/Y153A), K10 (K145A/K148A), K12 (E152A/D155A), and K15 (Y150A)] of eight mutants that displayed reduced C3b CFA also showed substantial reduction in C3b binding, whereas only one [K12 (E152A/D155A)] of four mutants that displayed reduced C4b CFA showed attenuated binding to C4b (Fig. 5 and SI Appendix, Fig. S2). It is, therefore, apparent that residues mutated in the above-mentioned four mutants are involved in binding to C3b/C4b. Furthermore, it may also be inferred that the residues that are vital for CFA but do not participate in binding to C3b/C4b are likely to be involved in factor I interaction.

Fig. 5.

Binding analysis of Ala substitution mutants of Kaposica (Kapo). Sensogram overlays for the interactions of Kapo and its Ala substitution mutants (K1–K15) with (Upper) C3b and (Lower) C4b. Binding was measured by injecting 4 or 1 µM of each mutant over the flow cell immobilized with C3b or C4b, respectively. Binding is represented in the form of response units (RUs). Mutants labeled in blue showed significant reduction in binding to C3b/C4b. Data shown are one of the three experiments shown in SI Appendix, Fig. S2.

Mapping of the Functionally Important Residues onto the Modeled Structure of C3b–Kaposica–Factor I Complex.

The above functional and binding analyses identified the Kaposica residues critical for CFA and binding to C3b/C4b. However, to obtain a detailed view of the interactions of the vital residues with C3b and factor I, we mapped the functionally important residues of Kaposica onto the modeled structure of C3b–Kaposica–factor I complex (Fig. 6A). To build the trimolecular complex model, we first generated the C3b–Kaposica model from the available C3b–FH(1–4) structure (Protein Data Bank ID code 2WII) by replacing FH(1–4) with the Kaposica homology model. When we examined the locations of the residues important for C3b CFA in this bimolecular complex, we observed, as expected, that the C3b-interacting residues of Kaposica (F144, K145, K148, Y150, E152, Y153, and D155) (Fig. 5) were present at the C3b–Kaposica interface (Fig. 6B and SI Appendix, Figs. S3A and S4) and that the C3b noninteracting residues (F102, S117, S123, S125, F128, E130, and H135) were clearly exposed to the solvent. Furthermore, mapping of the footprints of C3b-interacting residues of Kaposica onto C3b suggested that these residues interact with the CUB and MG2 domains of C3b (Fig. 6C), implying that bridging of these two domains of C3b by Kaposica is critical for its CFA. Next, we built the trimolecular complex of C3b, Kaposica, and factor I to determine whether C3b noninteracting residues interact with factor I. The model suggested that these residues, indeed, make contact with factor I, except H135 (Fig. 6D and SI Appendix, Fig. S3B).

Fig. 6.

Mapping of the functional sites in C3b–Kaposica (Kapo)–factor I trimolecular complex. (A) Model of C3b–Kapo–factor I trimolecular complex. Kapo model was superimposed with the coordinates of FH in C3b–FH(1–4) structure (2WII), and then, factor I was docked onto the C3b–Kapo structure to generate the ternary complex. C3b (cyan) and Kapo (gray) are represented by solid surface, whereas factor I is represented by cartoon (olive). Residues of Kapo that affect its activity (Fig. 4) are labeled in orange and pink. (B) Zoomed view of C3b interaction sites (orange) of Kapo. C3b domains are represented by cartoon (cyan). (C) Footprint of Kapo interaction sites on C3b. The footprints are seen in MG2 and CUB domains in C3b. (D) Zoomed view of factor I contact sites (pink) of Kapo.

To examine whether residues involved in the interactions proposed above are conserved in other complement regulators, we next aligned the sequence of Kaposica with other human and viral RCA proteins (SI Appendix, Fig. S4). It was observed that four of seven putative factor I-interacting residues of Kaposica are well-conserved in position in other complement regulators. For example, the hydrophobic and negatively charged residues at positions comparable with F128 and E130 are conserved in all of the regulators, except DAF. Similarly, S/T and F/Y are conserved at positions comparable with S123 and F102, respectively. Although S/T has not been mutated earlier at this position, mutation of F/Y at the collinear position resulted in a loss of CFA in SPICE (18) and MCP (10). Residues at the position corresponding to S125 have also been identified as critical in CR1, MCP, VCP, SPICE, and complement control protein homolog of herpesvirus saimiri (CCPH) (10, 18, 19, 22). Conservation is also observed among the putative C3b-interacting residues of Kaposica. The most conserved residue is Y/F at the position corresponding to Y153. In addition, polar residues are also conserved at sites occupied by K145, E152, and D155 and hydrophobic at F144. Although previous mutation data are not available for these positions [except for CR1 (20)], residues in this region of FH are in contact with C3b in FH(1–4)–C3b structure (13).

Viral and Human RCA Proteins Use a Common Mechanism to Inactivate C3b and C4b.

Viral RCA proteins are structural and functional mimics of the human RCA proteins. We, therefore, hypothesized that human and viral RCA proteins may inactivate C3b and C4b by using a common mechanism. To test this premise, we determined whether CFA can be incorporated into human DAF by swapping DAF intercysteine regions with MCP intercysteine regions homologous to those found to be crucial for the activity in Kaposica. We, thus, created two DAF-MCP chimeras: one where the D3R2–D4R1 region (residues N158–A219) of DAF was swapped with the M2R2–M3R1 region (residues N94–S156) of MCP and one where the D3R3–D4R1 region (residues L172–A219) of DAF was swapped with the M2R3–M3R1 region (residues E108–S156) of MCP (Fig. 7 A and B). Examination of CFA of these chimeras showed that both the chimeras possess the ability to inactivate C3b and C4b (Fig. 7 C and D), suggesting that the intercysteine regions found critical for imparting CFA in Kaposica are also critical for imparting CFA in MCP. The rate of C4b inactivation, however, was faster than C3b inactivation.

Fig. 7.

Substitution of the putative MG2-CUB–bridging region and factor I-interacting region of MCP in DAF imparts CFA to the DAF molecule. (A) Diagrammatic representation of DAF, MCP, and the DAF-MCP intercysteine region swap mutants DM1 and DM2. The vertical lines and the numbers mark the boundaries of the R1, R2, and R3 regions of the M2 and M3 domains of MCP. (B) SDS/PAGE analysis of purified DM1 and DM2 mutants. (C) CFA of DM1 and DM2. (D) Graphical representation of the percentage of α′-chain of C3b and C4b vs. time. Data shown in the graph are means of two experiments. MW, molecular mass.

Discussion

Physiological activation of the complement system can occur in the fluid phase as well as on the cell surface, and therefore, efficient regulatory mechanisms are necessary to prevent inappropriate activation at both of these sites to avoid fluid-phase complement consumption and host cell damage (23, 24). RCA proteins effectively regulate complement at both of these locates, and therefore, delineation of mechanisms that govern their regulatory processes is central to understanding the biology of complement and complement-mediated pathologies. Notably, here, we have identified the core functional regions and residues of Kaposica that are essential to impart the CFA to this protein—a mechanism that allows proteolytic inactivation of the key complement components C3b and C4b with the help of serine protease factor I. Furthermore, we show that similar determinants also participate in imparting CFA in human RCA proteins.

Our current understanding of the molecular events that occur during the CFA-mediated inactivation of C3b and C4b suggests that two steps direct this process: (i) recognition of C3b (or its homolog C4b) by the RCA protein and (ii) docking of factor I onto the C3b–RCA complex (or C4b–RCA complex) followed by the proteolytic cleavage of C3b (or C4b), leading to its inactivation. Earlier, several groups, including our group, focused on the identification of RCA residues involved in specific interactions with C3b/C4b, which provided a wealth of information (9–11, 19, 21, 25), but the clear picture about C3b–RCA interaction emerged only after the availability of the structure of FH(1–4) in complex with C3b (13). It identified four contact regions that spanned the entire length of the FH(1–4) molecule. What, however, remained unresolved was which of these contact region(s) is principal for imparting the CFA? Our results provided here clearly point out that CCP2 and CCP3 of Kaposica are most vital for its CFA, because swapping of these with the homologous domains of DAF results in complete loss of the CFA. These data also gain support from our earlier studies with viral RCAs Kaposica and VCP (16, 17). Furthermore, by swapping the intercysteine segments and by site-directed mutagenesis, we also identify the intercysteine regions and residues in CCP2 and CCP3 of Kaposica critical for its CFA. Mapping of these critical residues of CCP2 and CCP3 onto the modeled structure of the C3b–Kaposica complex depicts that CCP2 residues do not participate in C3b contacts, whereas CCP3 residues bridge the MG2 and CUB domains of C3b. It should be pointed out here that there exists a significant flexibility between the core of C3b (MG ring) and CUB/thioester-containing domain (TED) (26–28). We, therefore, propose that the bridge formed by CCP3 of Kaposica between MG2 and CUB holds the latter in a proper orientation relative to the core of C3b, which then allows its efficient cleavage by factor I. Although these data show the significance of MG2-CUB bridging for the CFA of Kaposica, generalization of this phenomenon for the CFA of human RCA proteins required the additional experimentation discussed below.

Human RCA proteins DAF and MCP along with a non-RCA protein CD59 are largely responsible for protecting the host cells from autologous complement, as these proteins are widely expressed in the normal cells (29). Both DAF and MCP are composed of four CCP modules and function in concert to prevent the host cell damage, because each one possesses only one of the two complement regulatory activities (i.e., DAF possesses only decay-accelerating activity, whereas MCP possesses only CFA) (30). Although domain requirements and critical residues of MCP necessary for the CFA have been identified earlier (10, 31), the minimum functional determinants that are necessary and sufficient for the CFA of MCP are still unknown. Thus, to identify these and determine whether MG2-CUB bridging is also essential for MCP function, we next asked whether CFA can be incorporated into DAF by substitution of the putative MG2-CUB–bridging region along with the factor I-interacting region of MCP. Our results clearly show that such substitution results in implantation of the CFA into DAF, indicating that MG2-CUB bridging is a common mechanism used by viral as well as human RCA proteins for imparting the CFA.

The structure of C3b–FH(1–4) revealed that one of four C3b contact regions of FH located in CCP4 forms a bridge between MG1 and TED. It was proposed that this would allow stabilization of the CUB-TED arrangement with the rest of the molecule and thus, would aid proper cleavage of CUB by factor I during C3b inactivation (13). The fact that CCP1–CCP3 of many RCA proteins show lower CFA compared with CCP1–CCP4 (3, 7, 8) clearly supports the above view, but it also suggests that the interactions between CCP4 and C3b play only a supportive role.

It is long known that proteolytic inactivation of C3b (or C4b) by factor I requires the presence of RCA protein (32). Earlier, it was thought that this is primarily because of considerable enhancement in the affinity of factor I for the target protein (33). However, the recently solved crystal structure of factor I and modeling of the C3b–FH–factor I ternary complex showed a complex mechanism. According to this study, factor I stays in its zymogen state until it is bound to the C3b–RCA complex. After bound, factor I undergoes a conformational change that allows release of its allosteric inhibition and formation of the active site around the substrate loop (14). The questions that arise, therefore, are which residues of RCA protein allow anchoring of factor I onto the C3b–RCA complex and where are these located. The transient nature of factor I interaction made it difficult to identify such factor I-interacting residues in the RCA proteins. Nonetheless, a few RCA residues have been identified in the past (11, 18). Our efforts here led to a comprehensive identification of the possible factor I-interacting residues in Kaposica and indicated that they reside in CCP2 as well as in CCP3. Additionally, mapping of these residues onto the modeled C3b–Kaposica–factor I complex indicated that most residues that we identified are buried at the Kaposica–factor I contact region and interact with the serine protease domains of factor I (SI Appendix, Fig. S3B). The fact that DAF gains CFA after substitution of MCP regions corresponding to factor I-interacting and MG2-CUB–bridging regions of Kaposica suggests that functionally critical contact sites are spatially conserved in various RCA proteins.

Because mutations and polymorphisms in the genes encoding RCA proteins are linked to diseases, we investigated if disease-associated mutations and polymorphisms are located in the region that we have identified as critical for the CFA and mapped them onto the modeled ternary complex to understand the consequences. We found that a few of the atypical hemolytic uremic syndrome-associated mutations in FH (A161S and M162V) and MCP (P165S, E179Q, D185N, and Y189D) are located in the putative MG2-CUB–bridging regions of these proteins (34). Interestingly, among these mutations, the FH mutations and two of the MCP mutations (E179Q and D185N) are buried at the C3b–FH interface and the predicted C3b–MCP interface, respectively (SI Appendix, Fig. S3C), suggesting a likely effect on the CFA. Mutational analysis of one of four MCP residues (D185) show that it is critical for the CFA (35).

In summary, our study provides a structural definition of the contact regions critical for the CFA of Kaposica and suggests that similar regions are also crucial for the CFA of human RCA proteins. Based on our data and earlier studies on the viral as well as human RCAs, we propose that, while serving as a cofactor, RCA proteins serve two key functions and that both are vital for imparting the CFA: (i) they provide an anchoring surface for factor I onto the C3b/C4b–RCA complex, leading to binding and reorientation of factor I, and (ii) they bridge the MG2 and CUB domains of the target protein, which stabilizes C3b in a proper confirmation and allows factor I to cleave the scissile bonds in the CUB domain.

Materials and Methods

Construction, Expression, and Purification of Domain/Region Swap and Ala Substitution Mutants.

The soluble forms of Kaposica (CCP1–CCP4) and human DAF (CCP1–CCP4) cloned in pGEM-T Easy Vector were used as templates for generation of the mutants. The construction, expression, and purification of Kaposica mutants and DAF-MCP chimeras are detailed in SI Appendix, SI Materials and Methods. The domain swap mutants of Kaposica were expressed in Pichia pastoris, whereas the region swap and the Ala substitution mutants of Kaposica and DAF-MCP chimeras were expressed in Escherichia coli BL21 cells.

CFA Assay and Surface Plasmon Resonance Measurements.

Details are in SI Appendix, SI Materials and Methods.

Molecular Modeling Analysis.

The X-ray crystal structure of Crry protein (Protein Data Bank ID code 2XRB) (36) was selected as the template for building the homology models of Kaposica. The final model of Kaposica was superimposed against the coordinates of FH in the C3b–FH(1–4) complex (2WII) to comprehend its interactions with complement C3b. Furthermore, to identify the putative interaction sites of Kaposica with factor I, a ternary complex of C3b–Kaposica–factor I was generated as described earlier (14). The details of model validation and molecular modeling methods are mentioned in SI Appendix, SI Materials and Methods.