An Activation Switch in the Ligand Binding Pocket of the C5a Receptor*

Although agonists are thought to occupy binding pockets within the seven-helix core of serpentine receptors, the topography of these binding pockets and the conformational changes responsible for receptor activation are poorly understood. To identify the ligand binding pocket in the receptor for complement factor 5a (C5aR), we assessed binding affinities of hexapeptide ligands, each mutated at a single position, for seven mutant C5aRs, each mutated at a single position in the putative ligand binding site. In ChaW (an antagonist) and W5Cha (an agonist), the side chains at position 5 are tryptophan and cyclohexylalanine, respectively. Comparisons of binding affinities indicated that the hexapeptide residue at this position interacts with two C5aR residues, I116 (helix III) and V286 (helix VII); in a C5aR model these two side chains point toward one another. Both the I116A and the V286A mutations markedly increased binding affinity of W5Cha, but not that of ChaW. Moreover, ChaW, the antagonist hexapeptide, acted as a full agonist on the I116A mutant. These results argue that C5aR residues I116 and V286 interact with the side chain at position 5 of the hexapeptide ligand to form an activation switch. Based on this and previous work, we present a docking model for the hexapeptide within the C5aR binding pocket. We propose that agonists induce a small change in the relative orientations of helices III and VII, and that these helices work together to allow movement of helix VI away from the receptor core, thereby triggering G protein activation.

Serpentine receptors transmit a diverse array of extracellular stimuli to heterotrimeric G proteins located on the cytoplasmic face of the plasma membrane.
These receptors promote exchange of GTP for GDP bound to the α subunit of the heterotrimer, allowing the α and βγ subunits to disengage from one another and activate intracellular effectors (1). Of several serpentine receptor families (2), the rhodopsin-like family is the largest (3). Low-resolution models of the 3D structure of the seven-helix bundle in the serpentine receptor core were based on patterns of conserved primary structure, biochemical observations with many receptors, and a low-resolution electron cryomicroscopy structure of rhodopsin (4). Three such models, constructed independently (3,5,6), predict 3D structures of the transmembrane helices that are remarkably similar to one another and to a recent 3D crystal structure of rhodopsin at atomic resolution (7). These similarities make the crystal structure a promising platform for designing and interpreting experiments aimed at elucidating structure and molecular mechanisms of other members of the rhodopsin-like family of serpentine receptors.
The mechanism of receptor activation, which is probably highly conserved also, resides in the transmembrane helices (8), where most of the evolutionarily conserved residues are located. Indeed, extra-or intracellular loops and termini can be exchanged between different receptors, leaving intact the receptors´ capacity to be activated, while swapping ligand-or G protein specificity (9). How is this conserved activation switch activated by an enormously diverse set of agonist ligands, widely differing in size and chemical character, which must occupy similarly diverse binding pockets in the receptors? Small agonists, including biogenic amines and chromophores, are thought to bind exclusively to a transmembrane receptor pocket; larger ligands, such as oligopeptides and proteins, interact in addition with extracellular domains of their receptors (2,10,11). Progress beyond this generalization has proved difficult; binding interactions have been studied in biochemical detail only for relatively small ligands, Ligand Binding to the C5a Receptor 4 such as adrenergic amines and retinal, the chromophore of the visual pigment, rhodopsin (reviewed in (10)).
The receptor for complement factor 5a (C5a), a 74 amino acid protein, furnishes an instructive experimental model for studying how larger peptides bind to and activate serpentine receptors (reviewed in (12)). Deletion mutants of the C5a receptor (C5aR) indicate that C5a interacts both with the receptor´s N-terminus and with the transmembrane bundle; the latter interaction is required for activation of the C5aR (13,14). Molecular details of the C5a-C5aR interaction have been elucidated by pharmacological characterization of C5a mutants and peptides derived from the Cterminal amino acid sequence of C5a (13,(15)(16)(17)(18)(19)(20)(21)(22)(23). In addition, 3D structures have been determined for C5a (24) and a hexapeptide antagonist (22) Me-F-K-P-dCha-W-dR (hereafter termed ChaW; dCha = D-cyclohexylalanine). Taken together, these studies show that receptor activation is mediated by interaction of C-terminal residues of C5a with the receptor´s transmembrane helix bundle; hexapeptide ligands analogous to the C-terminal eight residues of C5a interact exclusively with the transmembrane pocket (13). One residue in transmembrane helix V of the receptor, R206, is essential for receptor activation by a hexapeptide agonist; the guanidinium group of R206 interacts with the terminal carboxylate of this agonist, as shown by characterization of simultaneously altered ligand and receptor mutants (21). Unlike its C-terminal residues, remaining regions of C5a sequence interact with the N-terminus of the C5aR (14); this interaction enhances ligand binding affinity, but is not required for receptor activation.
A genetic screen of C5aR mutants expressed in S. cerevisiae (25), designed to assess the functional importance of amino acids in receptor helices III, V, VI and VII, identified a cluster of residues situated at extracellular ends of the transmembrane helices that were required for C5aR signaling but that are not evolutionarily conserved.
These residues are located at positions cognate to positions of residues that are thought

Construction of Receptor Mutants and
Yeast Assays˙Mutated C5aRs were created as described (25). Analysis of receptor signaling in yeast was performed by replica plating onto different concentrations of aminotriazole (AT), as previously described (25). For COS-7 cell assays, the various C5aR sequences were subcloned into plasmid pDM8 (Invitrogen, Carlsbad, CA).
Transfections of WT and mutant receptors were performed using a DEAEdextran/adenovirus method, as described (26).
Membrane Preparations˙Membranes of COS-7 cells transfected with WT or mutant C5aR were prepared by a modification of a previously described method (26).  these are colored orange and yellow in Figure 1. Of these, we chose seven (orange in the Figure) for further analysis, based on two criteria: (a) a strict requirement for receptor function (the corresponding single site mutants show no detectable activity in our yeast assay; see Table I and (25)); (b) localization on a helical face pointing into the transmembrane pocket between helices III, V, VI, and VII.
These criteria are compatible with the idea that these seven residues form part of the C5a binding pocket. If so, amino acid substitutions at these positions, which block C5a-triggered signals, should not affect signaling by a constitutively active receptor whose activity is independent of the agonist. In an intramolecular epistasis experiment,   (Table II). We imagine that the strong contribution of the receptor´s N-terminus to its binding affinity for C5a (13)  Mutant cycle analysis, which has successfully identified amino acids involved in intermolecular protein-protein interactions (33)(34)(35)(36), is based on the following intuitive principle: If residue a of the protein ligand (ChaW or its derivatives) interacts with residue b of the C5aR, the effect of mutating a on the ligand´s binding affinity should depend on whether receptor residue b is mutated as well. In practice, we simply compared the relative changes of binding affinities by dividing the respective inhibition constants (K I ; see the legend of Table II). We assessed binding affinities of ChaW and four singly substituted derivatives to the WT C5aR and C5aR mutants with substitutions at each of the seven residues in the putative ligand binding pocket.
Results of these experiments are summarized in Table II and Figure 2B, and illustrated in Figure  In contrast to the results in yeast (Table I) identification of I116 and V286 as sites for interaction with a specific residue of hexapeptide ligands furnishes stringent constraints for orienting ChaW relative to the C5a receptor binding site, provided that the ligand structure is known. We therefore solved the NMR structure of ChaW in DMSO (see Materials and Methods). Despite its small size, ChaW shows an ordered structure in solution, comprising an inverse γ-turn formed by the Lys-Pro-D-Cha residues and a distorted type II β-turn that includes the residues of the γ-turn and the tryptophan at position 5. This structure (Figure 4) appears to agree with an earlier NMR study that identified a well-defined backbone conformation composed of the same structural motifs (22). Because the authors of this analysis chose not to make their ChaW coordinates available to us, we cannot directly compare our structure with theirs.
Docking and modeling of a ChaW/C5aR ligand-receptor complex˙With the structure of ChaW in hand, we used intermolecular contacts as constraints in building a model of the receptor-antagonist complex. Two of these, detected by mutant cycle analysis (see above), are interactions of the W5 residue of ChaW with I116 in helix III and V286 in helix VII of the receptor. We also used constraints identified in a previous study (21) of the receptor´s interaction with the C-terminal dArg6 residue of ChaW. In this study, DeMartino and coworkers used an R206A receptor mutant and a series of ChaW derivatives to show that the side chain of R206 acts as a ˆgate-keeper,˜ allowing hexapeptide ligands to bind to the transmembrane receptor core only when the ligands present a C-terminal carboxylate, which appears to interact with the guanidinium group of R206. In addition to the carboxylate, the side chain of dArg6 was also required for receptor interaction, leading these investigators to suggest that it interacts with the transmembrane receptor pocket.
These constraints, together with the known backbone structure of ChaW, orient the hexapeptide in relation to the C5aR model. Figure 4 shows ChaW docked in the The orientation of ChaW with respect to the ligand binding pocket correlates nicely with results of detailed pharmacological analyses of ChaW and its derivatives (23,24).
Thus K2 and P3 of the hexapeptide, sites at which substitutions do not affect binding affinity or antagonistic potential, point away from the putative ligand binding pocket of the C5aR model ( Figure 4). In contrast, substitutions for F1 greatly decreased binding affinity of ChaW (23). The location of the F1 side chain close to the extreme extracellular end of helix V may also favor contacts with the N-terminal end of the receptor´s second extracellular loop, a region that appears to mediate high affinity binding of C5a: replacement of this region in a C5a/formyl peptide receptor chimera abolished C5a binding (37). Although the dCha4 side chain is close to that of L112 (helix III) in the receptor binding pocket (Figure 4), mutant cycle analysis did not indicate an energetically important interaction between the two side chains.
In our docking model the long arginine side chain at position 6 of ChaW penetrates deeply into the receptor´s helix bundle, between helices III, V, and VI ( Figure 4C). Its guanidinium group, located one helical turn lower than any of the genetically conserved residues in the putative ligand binding pocket, is cradled in a pocket of aromatic, mostly conserved residues in helices V and VI (green in Figure 4C).
Interaction of a positively charged side chain (arginine or lysine) with aromatic residues is not unlikely in protein structures; indeed, a recent analysis (38) of interactions between cationic groups and delocalized π-electron systems in proteins found such interactions to be much more frequent and energy-rich than previously assumed. This predicted location for the dR6 side chain of the ligand is especially intriguing in light of previous evidence (21) that this side chain is necessary for receptor activation. Thus, we speculate that the receptor is activated by insertion of a hydrophobic group (e.g., cyclohexylalanine) between helices III and VII, but that robust activation is not compatible with the greater separation of these helices that would be induced by  (40)). Moreover, constitutive activity of rhodopsin results from mutations that prevent formation of this helix III/VII salt bridge (that is, substitutions for either E113 or K296).
We are thus left with strong hints that receptors share a common activation trigger located at about the same level of the helix III/VII interface, though none of the available evidence tells us how the trigger works.
Perspective˙Our model of the ChaW-C5aR complex (Fig. 4)  We infer from our observations that C5a agonists activate the receptor by interacting with a trigger zone for activation, located between neighboring residues in helices III and VII. As summarized above, quite different experimental approaches in   from three independent experiments, each performed with duplicate or triplicate determinations. The cooperation value (Ω) for an intermolecular interaction between two residues (one on the receptor and one on the ligand, respectively) is defined as Ω = (K i wtwt x K i mtmt ) / (K i wtmt x K i mtwt ), where K i wtwt is the inhibition constant of the WT C5aR with ChaW , K i mtwt the inhibition constant of an individual C5aR mutant with ChaW, K i wtmt the inhibition constant of the WT C5aR with an individual ChaW mutant, and K i mtmt the inhibition constant of a mutant receptor with a ChaW mutant. For ease of comparison, the cooperation value of each pairing has been normalized by setting as = 1.0 the relative affinity change of ChaW vs. a substituted ligand for the WT C5aR. In addition, where the Ω for an interaction is less than 1.0, the value is reported as its reciprocal.