Structural models for the complex of chemotaxis inhibitory protein of Staphylococcus aureus with the C5a receptor

Abstract

The study presents structural models for the complex of the chemotaxis inhibitory protein of Staphylococcus aureus, CHIPS, and receptor for anaphylotoxin C5a, C5aR. The models are based on the recently found NMR structure of the complex between CHIPS fragment 31-121 and C5aR fragment 7-28, as well as on previous results of molecular modeling of C5aR. Simple and straightforward modeling procedure selected low-energy conformations of the C5aR fragment 8-41 that simultaneously fit the NMR structure of the C5aR 10-18 fragment and properly orient the NMR structure of CHIPS_31-121 relative to C5aR. Extensive repacking of the side chains of CHIPS_31-121 and C5aR_8-41 predicted specific residue-residue interactions on the interface between CHIPS and C5aR. Many of these interactions were rationalized with experimental data obtained by site-directed mutagenesis of CHIPS and C5aR. The models correctly showed that CHIPS binds only to the first binding site of C5a to C5aR not competing with C5a fragment 59-74, which binds the second binding site of C5aR. The models also predict two elements of CHIPS, fragments 48-58 and 97-111, may be used as structural templates for potential inhibitors of C5a.

Introduction

CHIPS, a 121-membered chemotaxis inhibitory protein of Staphylococcus aureus, is a potent inhibitor of neutrophil and monocyte chemotaxis involving C5a anaphylatoxin or formylated peptides [1]. CHIPS selectively binds both C5a receptor (C5aR) and formylated peptide receptor (FPR) with nanomolar affinities [2], which makes it a promising lead for development of new anti-inflammatory compounds. Mutational studies have previously demonstrated that different epitopes of CHIPS inhibit signaling through C5aR or FPR: the first six residues of CHIPS blocked FPR but not C5aR [3], whereas the truncated CHIPS fragments, 31-121 and 31-113, inhibited C5a binding to C5aR as effectively as the entire CHIPS, but lacked FPR antagonism [4, 5].

C5a binds and activates the C5aR utilizing two distinct binding sites, the first located in the N-terminus of C5aR and the second located in a helical crevice between the extracellular loops which accommodate the C-terminus of the C5a ligand (residues 65–74) [6–8]. Mutational studies of C5aR have mapped CHIPS binding to the N-terminal fragment of C5aR (residues 1–38) and demonstrated an essential role for the C5aR fragment 10-18 [9]. In addition, CHIPS does not affect activation of C5aR by a peptide mimic of the C5a 65-74 fragment [9]. These studies demonstrated also that CHIPS blocks C5a activation by inhibiting the binding of the intact ligand to the N-terminus of the C5a receptor and also implied that CHIPS does not interact appreciably with the extracellular loops of the C5aR.

Previously, NMR spectroscopy provided the three-dimensional solution structure of the isolated CHIPS 31-121 [4] and, very recently, in complex with the C5aR 7-28 fragment (Tyr¹¹ and Tyr¹⁴ of C5aR were sulfated) [10]. The NMR structure of the C5aR_7-28:CHIPS_31-121 complex confirmed the important role of C5aR fragment 10-18 in binding CHIPS. The C-terminal portion of the C5aR_7-28 peptide was highly flexible, which prevented elucidation of specific interactions between residues 22–28 of C5aR_7-28 with CHIPS in the NMR structure [10]. Also, mutational studies using tethered N-terminal peptides that either introduced point mutations into this region or deleted it altogether did not impact binding of CHIPS, thereby supporting the notion that this region does not interact directly with CHIPS [9]. However, in the context of the intact C5aR, residues 22 – 28 may adopt a very different set of conformations, especially when one considers that this region is directly connected to the transmembrane helix of C5aR (TM1, residues 38–63 [11]).

Our previous studies developed 3D models of the C5aR:C5a complex involving various possibilities for flexible extracellular loops and the N-terminal fragment of C5aR [11]. The models were validated by comparison with the available data of site-directed mutagenesis of C5aR [11] and the results of a novel technique of disulfide trapping by random mutagenesis [12]. In the present study, we employed the results of the previous modeling of the C5aR and the NMR data on the C5aR_7-28:CHIPS_31-121 complex to generate new models for interaction between CHIPS and the C5aR. Our modeling procedure selected low-energy conformations of the C5aR fragment 8-41 that simultaneously fit the NMR structure of the C5aR 10-18 fragment and properly orient the NMR structure of CHIPS_31-121 relative to C5aR. Extensive repacking of the side chains of CHIPS_31-121 and C5aR_8-41 predicted specific residue-residue interactions on the interface between CHIPS and C5aR. Many of these interactions are rationalized with experimental data obtained by site-directed mutagenesis of CHIPS and C5aR.

Materials and methods

All energy calculations were performed using the ECEPP/2 force field with rigid valence geometry [13, 14] and trans-conformations of Pro residues; residues Arg, Lys, Glu and Asp were regarded as charged species. The 3D structures of the C5aR_8-41:CHIPS_31-121 complex were obtained by energy minimization of the system consisting of two components, namely the selected conformation(s) of C5aR_8-41 and the NMR-derived structure of CHIPS_31-121. The modeling procedure was essentially the same as that applied earlier for determining the 3D structures of the transmembrane regions of GPCRs (see, e.g., [11] for details). The backbone structures of each component were regarded as rigid bodies (with the fixed values of the backbone dihedral angles), while the side chains of both C5aR_8-41 and CHIPS_31-121 were extensively repacked prior to energy minimization. For CHIPS mutants, the corresponding side chains were replaced in the NMR structure. The algorithm of side chain repacking [15, 16] involved a stepwise grid search (with the grid step of 30°) in the space of the dihedral angles of the side chains, χ_i. To ensure thorough sampling of the χ_i space, two options of the search were applied independently. The first option sampled the grid varying sequentially all χ₁ angles of the side chains of CHIPS_31-121 and C5aR_8-41, then all χ₂ angles, then all χ₃ angles, etc. until all side chain angles of both components were rotated. The second option sampled all χ_i angles for each side chain in order of their number in protein sequence, i.e., from 31 to 121 for CHIPS_31-121 and from 8 to 41 for C5aR_8-41. In both cases, sampling was performed until convergence to energetically optimal packing of side chains was achieved. If necessary, specific side chains (as those on CHPS:C5aR interface) were repacked additionally. Finally, energy minimization in the space of possible rotations and translations of the two components as well as in the space of the dihedral angles of the side chains followed. Results of all options were pooled together to cover energetically plausible conformational possibilities for the side chains. To account to some extent for the absence of solvent and membrane environment, energy calculations were performed using two values of macroscopic dielectric constant, ε = 2 (the generic value for the ECEPP force field) and ε = 80, which is closer to the water dielectric constant. In most cases, distributions of the side chain rotamers obtained by calculations with both values were identical.

Results and discussion

Computational models of the CHIPS_31-121:C5aR_8-41 complex

Previous modeling of the isolated N-terminal fragment C5aR_8-41 identified 185 low-energy backbone structures [11]; it was reasonable to assume that possible conformations of C5aR_8-41 in complex with CHIPS should be selected from this pool of structures. 59 of 185 structures contained spatial positions of fragment C5aR 10-18 that fit to that observed in the C5aR_7-28:CHIPS_31-121 complex by NMR spectroscopy (according to the rms cut-off of 2 Å calculated for the Cα atoms of residues Asp¹⁰, Tyr¹¹, Gly¹², Tyr¹⁴, Asp¹⁵ and Asp¹⁸; NMR model #1 from the PDB entry 2K3U was used for fitting). In turn, only 17 of 59 structures showed no steric clashes with the NMR structure of CHIPS_31-121 (i.e., no intermolecular Cα-Cα distance was less than 3 Å). When these 17 structures were fit to the helical stem of TM1 (residues 38–41) of our model of C5aR, only four structures did not contain steric clashes with either the TM region of C5aR or with all possible conformations of the extracellular loops that have been found previously [11]. Visual inspection showed that, in two of the four structures, CHIPS_31-121 was significantly embedded into the membrane space, which is highly unlikely; those two structures were discarded from subsequent considerations.

The remaining two models of the C5aR_8-41:CHIPS_31-121 complex are depicted in Figs 1A and 1B. The interfaces between C5aR and CHIPS were established by subjecting both models to extensive side chain repacking and energy calculations (see Methods). The system of possible residue-residue contacts on the interface in both models is described in Table 1 (a contact was defined as situation when at least one distance between atoms belonging to the corresponding side chains was less than 5.5 Å). The two models feature significantly different potential strong salt bridges and/or hydrogen bonding at the interfaces. The strongest interactions in the first model were between Y48/K51 of CHIPS and Asp¹⁰ of C5aR (the corresponding residues are shown in Table 1 in bold; different notations are applied to distinguish between amino acid residues of C5aR and CHIPS). In the second model, the strongest interactions were K51 - Thr⁸, K54 – Ser³⁰, K100 – Thr²⁹ and Y108 – Thr²⁹. Importantly, many of interactions on the interface listed in Table 1 should be regarded as possibilities, since some side chains possess more conformational freedom and therefore may also interact with residues not on the interface (see also below).

Fig. 1.

Right panels show cartoon sketches of complexes of CHIPS_31-121 (green) and C5a (magenta). Left panels show more detailed sketches of complexes CHIPS_31-121:C5aR_8-41 focusing on possible strong residue-residue interactions. Side chains are shown as sticks and labeled by one letter notation (CHIPS residues) or by three-letter notation (C5aR residues). Residues K40 – R46 are shown in cyan, interface residues Y48 – Y108 are in green, Asp¹⁰ – Ser³⁰ are in magenta. A and B relate to the first and second suggested models, respectively.

Table 1.

Contacts between side chains of CHIPS_31-121 and C5aR_8-41. Residues participating in strongest intermolecular interactions are shown in bold.

Residues of CHIPS	Residues of C5aR8-41
	First model of complex	Second model of complex
R44	Asp10
Y48	Pro9, Asp10, Tyr14	Thr8, Pro9, Asp10
K51	Thr8, Pro9, Asp10, Tyr14	Thr8, Tyr11
T53	Pro25, Thr29	Tyr14, Thr29
K54	Val26, Thr29, Ser30	Pro25, Val26, Ser30
N55	Thr29	Thr29
Q58		Thr29, Thr32
K95	Asp18
Y97	Asp18
F98		Tyr14
K100	Thr32	Pro25, Lys28, Thr29, Thr32, Leu33, Arg34
E103		Arg34
K105		Asp18, Arg34
S107	Lys28, Tyr14, Thr32	Lys28, Arg34
Y108	Tyr14, Pro25	Tyr14, Pro25, Thr29
V109	Tyr14	Tyr14
I110	Tyr14	Pro9, Asp10, Tyr14
N111		Asp10, Tyr14

Rationalizing of site-directed mutagenesis data by computational models of the CHIPS_31-121:C5aR_8-41 complex

Interfaces between C5aR and CHIPS suggested by both models agree well with the two independent sets of data on NMR titration of C5aR_1-37 in presence of 15N-CHIPS_0-121 [17] and of the sulfated C5aR_7-28 in presence of 15N-CHIPS_31-121 [10]. Both studies demonstrated that two segments of CHIPS (ca. 45 – 61 and 98 – 111) directly interact with the C5aR fragments; these segments figure prominently in the binding interfaces of both computational models. Mutations of K95 (K95A-CHIPS_1-121 [4], K95S-CHIPS_31-113 [5]) or Y97 (Y97A-CHIPS_31-121 [10], Y97K-CHIPS_31-113 [5]) significantly lowered the ability of mutants to block activation of C5aR by C5a. K95 and Y97 likely are directly involved in interactions with the C5aR residues on C5aR:CHIPS interface, since CD spectroscopy showed that the above mutations do not disturb structural integrity of CHIPS (see Fig S3 in [10]). At the same time, interactions K95 - Asp¹⁸ and Y97 – Asp¹⁸ are available in the first computational model but not in the second one that features interaction K105 - Asp¹⁸ (see Table 1 and Fig. 1). Other experimental data on CHIPS mutants also support the first model. Specifically, mutations at Y48A-CHIPS_31-121 [10] and K51A-CHIPS_1-121 [4] led to significant loss of CHIPS ability to block C5aR activation by C5a by likely affecting the strong Y48 - Asp¹⁰ and K51 - Asp¹⁰ interactions present in the first model. Alternatively, the K51A mutation may interrupt the strong K51 – Thr⁸ interaction characteristic for the second model. Also, S107A-CHIPS_31-121 [10] and S107N-CHIPS_31-113 [5] only partially inhibit C5aR interaction with C5a, which may be due to loss of the possible interaction S107 - Lys²⁸ present in both models (Lys²⁸ may alternatively interact either with S107 or with Asp¹⁵, which is shown in Fig. 1A). In both models, S106 closely contacts the side chain of Y97 that may change orientation in the S106A-CHIPS_31-121 mutant. This mutation significantly impairs CHIPS ability to block C5aR activation [10], which might reflect loss of interaction Y97 - Asp¹⁸ present in the first but not in the second model. At the same time, mutation K105A in CHIPS_1-121 [4] also disrupts CHIPS interactions with C5aR, which supports the proposed contact K105 - Asp¹⁸ in the second model.

The mutational data obtained from mutational studies of the C5aR demonstrated essential roles for residues Asp¹⁰, Asp¹⁵ and Asp¹⁸ for interaction with CHIPS [9]. These observations also are more consistent with the first model, where Asp¹⁰ and Asp¹⁸ may be involved into direct interactions with Y48 and K51 as well as with K95 and Y97, respectively, and Asp¹⁵ may interact with Lys²⁸ influencing possible interaction of the latter with S107. In the second model, on the contrary, only Asp¹⁸ may interact directly with the K105 residue of CHIPS.

The mutations in R44A-CHIPS_1-121 and R46A-CHIPS_1-121 also impair the ability of CHIPS to block C5aR activation by C5a [4], while not altering the secondary structure of CHIPS_31-121 [10]. In both computational models, R44 and R46 are not directly involved in interactions with the C5aR residues. To gain some insights into how these mutations might impair C5aR interactions, we separately modeled the complexes between these two CHIPS mutants and C5aR_8-41. Extensive side chain repacking for the mutant complexes C5aR_8-41:R44A-CHIPS_31-121 and C5aR_8-41:R46A-CHIPS_31-121 revealed more subtle influences of R44A or R46A mutations on possible interactions between C5aR and CHIPS. The side chains of both residues are involved in plethora of potential salt bridges/hydrogen bonds with the side chains of other residues of CHIPS (see Fig. 1). The negatively charged D42 and E43 residues may interact with the positively charged residues K40, R44 and R46 located in the same helical stretch of CHIPS (segment 38–51), forming the salt bridges K40 – E43, D42 – R46 and E43 – R46. They also may interact with K85 located in the loop connecting two β-sheets of CHIPS, 70–76 and 94–100. Also, the side chain of R44 may form strong hydrogen bond with the side chain of Y48, which participates in the direct interaction with Asp¹⁰ in the first model. Accordingly, removal of the R44 side chain in R44A-CHIPS_1-121 may increase conformational flexibility of the Y48 side chain thus affecting direct interactions with C5aR. In R46A-CHIPS_1-121, the salt bridges D42 – R46 and E43 – R46 are absent, and the salt bridges K40 – E43, D42 – K85 and E43 – K85 become dominant. Addition of the latter two strong residue-residue interactions between structural elements of CHIPS may influence overall conformational stability of the molecule, as was noted in the earlier experimental study [4]. On the other hand, conformational possibilities of the R44 side chain and, as a consequence, possibilities of the Y48 side chain still may be affected in R46A-CHIPS_1-121.

Computational models of the CHIPS_31-121:C5aR_8-41 complex compared to NMR structure and to model of C5a:C5aR complex

Both computational models are not mutually exclusive. Both models are similar to that suggested by the NMR spectroscopy [10]. It is not surprising, since one important element of our models, namely spatial positions of Cα atoms in the C5aR 10-18 fragment, was required to be compatible to the NMR model. Nevertheless, many details of the C5aR:CHIPS interface suggested by our models and by the NMR model are different. The NMR study used the sulfated Tyr¹¹ and Tyr¹⁴ in C5aR_7-28 to ensure forming of strong complex with CHIPS_31-121 (ca. 100 times stronger than that using the non-sulfated C5aR_7-28 [10]), while our study considered both tyrosines as non-sulfated. (The role of the sulfated/non-sulfated Tyr¹¹ and Tyr¹⁴ in C5aR in binding CHIPS is disputable, since mutations Y11F in C5aR_1-37 and Y14F in the entire C5aR did not affect CHIPS binding [9]; this point recently was discussed further [10].) Accordingly, in the NMR model, the sulfate group of sTyr¹¹ was coordinated by interactions with the side chains of R44, Y48 and K51 (the former, though, was not well-resolved in the NMR structure [10]), and the sulfate group of sTyr¹⁴ interacted with the backbone amide hydrogens of K54 and N55 as well as with the side chain of Q58. These interactions, obviously, are absent in our models; according to Table 1, we might rather expect interactions of K51 either with sTyr¹⁴ in the first model, or with sTyr¹¹ in the second one.

Our models predict the specific orientations of CHIPS relative to C5aR (see Fig. 1) that exclude interactions between CHIPS and the transmembrane region of C5aR. This agrees with the experimental observation that CHIPS does not interfere with binding of C5a 59-74 to C5aR [9], which binds to a cavity in the transmembrane region of C5aR between TM3, TM4, TM5 and TM6 [11]. Interestingly, our first model features a backbone conformation of C5aR_8-41 that was selected previously as capable to interact with C5a (conformation a, see Table 1 in [11]). However, the interface between C5aR_8-41 and CHIPS_31-121 in our first model includes Thr⁸, Pro⁹, Asp¹⁰, Tyr¹⁴, Asp¹⁸, Pro²⁵, Val²⁶, Lys²⁸, Thr²⁹, Ser³⁰ and Thr³², while the interface between C5aR_8-41 and C5a in conformation a includes different residues, namely Leu²², Asp²⁷, Asn³¹, Val³⁵, Pro³⁶ and Asp³⁷.

Concluding remarks

Our modeling study yielded two structural models of the C5aR:CHIPS_31-121 complex based on conformational possibilities of the N-terminal fragment C5aR_8-41, established earlier and on the NMR-derived structure of the C5aR_7-28:CHIPS_31-121 complex. The models elucidated the general orientation of CHIPS_31-121 relative to C5aR and interface between C5aR_8-41 and CHIPS_31-121. The models provide plausible rationalization of the available data of site-directed mutagenesis of CHIPS and the N-terminal fragment of C5aR. In agreement with experimental data, the models showed that CHIPS binds only to the first binding site of C5a to C5aR, and, therefore, CHIPS does not compete with C5a 59-74, which binds the second binding site of C5aR. The models also predict that two elements of CHIPS, namely fragments 48-58 and 97-111 may be used as structural templates for potential inhibitors of C5a. The atomic coordinates of the C5aR_7-28:CHIPS_31-121 complexes are available from the authors by request.

Abbreviations

GPCR

G-protein-coupled receptor

C5a

anaphylotoxin C5a

C5aR

C5a receptor

CHIPS

chemotaxis inhibitory protein of Staphylococcus aureus