Structure-Function Relationships of Human C5a and C5aR¹

Human C5a (C5a) is a 74-aa glycoprotein generated upon activation of the complement system via the classical, alternative, or lectin-binding pathways (1). In concert with C4a, C3a and the membrane attack complex (C5b-9), C5a plays a central role in host defenses. Excessive complement activation leading to elevated plasma levels of C5a is known to be associated with many clinical conditions, including sepsis (2), adult respiratory distress syndrome (3), rheumatoid arthritis (4), Alzheimer’s disease (5), and ischemic heart disease (6), to name just a few conditions. Biochemical and pharmacological studies indicate that both C3a and C5a bind and interact with receptors that belong to the G protein-coupled superfamily of rhodopsin-like receptors (7, 8, 9, 10). The C5aR (CD88) is expressed widely on inflammatory (9, 10, 11, 12) and noninflammatory cells (13, 14, 15, 16, 17, 18). For C5a-C5aR interaction, multiple discontinuous residues in two major structural elements within the C5a molecule have been reported to be required, as identified by nuclear magnetic resonance (NMR) structure analysis and site-directed mutagenesis experiments (19, 20). The first site of interaction is the highly ordered disulfide bridge-stabilized core of C5a (aa 1–63) with its four α helices juxtaposed in an antiparallel topology and connected by three peptide loops. The second area of C5a implicated in interaction with C5aR is the relatively disordered and flexible carboxyl-terminal tail of C5a (aa 64–74) (20). To understand the interplay between C5a and C5aR that trigger signal pathways, a “two-site binding” model of C5a-C5aR interaction has been proposed (21). Whereas the disulfide-linked core region of C5a may be recognized and have binding interactions with both the N-terminal segment as well as the third extracellular loop of C5aR (“recognition site”), the C-terminal region of C5a may fit into a binding pocket formed around the fifth transmembrane domain (“effector site”), leading via G protein activation to signal transduction (21). In recent studies, a genetic screen of C5aR mutants has identified in C5aR two distinct clusters of residues that possibly serve as a C5a-activated “molecular switch” (22). One cluster, located at the extracellular face of the receptor (binding area), may be required for interaction with the C5a molecule and the second cluster, at the core of the transmembranous helix bundle, may transmit the signal (via conformational changes) from the extracellular C5a ligand to the G protein (22, 23, 24).

For the human C3aR, direct agonist binding and calcium mobilization assays have shown that the large second extracellular loop (between the fourth and fifth transmembrane domain) is necessary not only for high-affinity C3a binding but also for the subsequent functional responses (25). In contrast, for the C5aR and IL-8_{A and B} receptors, the N terminus and the third extracellular loop have been suggested to contain structures critical for binding (26) but redundant for activation. With regard to the ligands, all binding energy is contributed by the C terminus of C3a, in contrast to C5a where the central portion contributes a considerable amount of receptor interaction energy (27), suggesting, that different regions of the C5a molecule have to act in concert to achieve full potency. Whereas many studies have focused on the C terminus of C5a (which interacts with the “effector site” of the C5aR), little is known about different regions within the middle of C5a (and their interactions with C5aR), especially the inter-α helical peptide loops. Therefore, in the present study the structure-function relationships of different selected peptide regions of C5a were investigated. Loop peptide homologues, which are chemically conserved within the family of C5a species and are structurally different from the corresponding residues in the family of C3a species (20), were investigated as candidates for additional interaction of C5a with its receptor. Based on present data, a “three-site” interaction of C5a with C5aR is proposed.

Human rC5a was purchased from Sigma-Aldrich (St. Louis, MO). All materials were obtained from Sigma-Aldrich unless otherwise indicated.

Based on structure-function studies (19, 20) of human complement C5a (C5a), different regions of the molecule were chosen and synthesized for evaluation of their ability to interact with the C5aR and affect the biological responses of neutrophils. The amino-terminal “A” region with the sequence MLQKKIEEIAAKYKHSVVKK (corresponding to aa residues 1–20), the middle “M” region CCYDGASVNNDETCEQRAAR (corresponding to aa residues 21–40), and the carboxyl-terminal “C” region CVVASQLRANISHKDMQLGR (corresponding to aa residues 55–74) were selected as peptides of C5a. A scrambled version of the C peptide was also synthesized (C scrambled: CRVGVLAQSMQDLKRHASIN). Additionally, based on the proposed three-dimensional (3D) structure (20), smaller peptide regions, representing the first, second, and third inter-α helical loop regions of C5a molecule were synthesized as follows: D₁ (KYKHSVVKK, aa residues 12–20), D₂ (VNNDET, aa residues 28–33), and D₃ (AARISLGPR, aa residues 38–46) (Fig. 1 A). In addition, scrambled versions of D₂ peptide were also synthesized (D₂-scrambled I: TENFNV; D₂-scrambled II: ENTNDV). These peptides were synthesized and purified by HPLC (Research Genetics, Huntsville, AL and Open Biosystems, Huntsville, AL).

Whole blood from healthy donors was drawn into syringes containing 0.1 ml/ml blood of anticoagulant ACD (Baxter, Health Care, Deerfield, IL). Neutrophils were isolated using Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) gradient centrifugation (1500 rpm, room temperature, 30 min). The cell pellet was resuspended in dextran (final concentration, 1%) followed by a 45-min sedimentation step of erythrocytes. Supernatant fluids containing neutrophils were centrifuged (1500 rpm, room temperature, 5 min) and residual RBCs were removed by a hypotonic lysis step. After washing, neutrophils were resuspended in assay buffer and evaluated in binding studies, chemotaxis assays, and oxidative burst measurements.

For binding studies, commercially available recombinant human C5a (C5a) was labeled with ¹²⁵I using the chloramine-T-based protocol (28), which allows gentle oxidation and preserves the chemotactic activity of C5a for neutrophils (data not shown). To avoid receptor internalization, all of the following steps were performed at 4°C. To block nonspecific surface binding sites, isolated neutrophils were incubated for 1 h in binding buffer (HBSS, without Ca²⁺, containing 1% BSA). After a gentle washing step, neutrophils (2 × 10⁶ cells) were incubated in binding buffer (HBSS, with Ca²⁺, containing 0.1% BSA) in 1.5-ml microcentrifuge tubes (in a final volume of 200 μl) with 100 pM ¹²⁵I-C5a (sp. act., 33.24 μCi/μg) in the absence or presence of increasing amounts of unlabeled C5a (ranging from 10⁻¹¹ to 10⁻⁶ M). After an incubation interval of 20 min, cell suspensions were layered over 20% sucrose and sedimented by centrifugation at 11,000 × g (Beckman Microfuge B; Beckman Coulter, Palo Alto, CA) for 2 min. The tubes were then frozen at −80°C and the tips, containing the cell pellet, were cut off to determine the cell-bound ¹²⁵I-C5a using a gamma counter (1261 Multigamma; Wallac, Gaithersburg, MD). Binding affinities (K_d values) of C5a were calculated in the conventional manner (9). The saturation binding curve was generated using increasing amounts of ¹²⁵I-labeled C5a (1–20 nM) in the absence or presence of 100-fold excess of unlabeled C5a and analyzed by nonlinear regression analysis using the GraphPad Prism program (GraphPad Software, San Diego, CA). Interference of the different synthetic peptides (1–1000 nM) with C5a binding was determined in the absence or presence of 10 nM ¹²⁵I-labeled C5a using 2 × 10⁶ neutrophils in 200 μl of binding buffer.

Following neutrophil isolation, cells were fluorescein-labeled with 2′,7′-bis-(2-carboxyethyl)-5-(and 6)-carboxyfluorescein acetoxymethyl ester) (Molecular Probes, Eugene, OR) for 30 min at 37°C. After a washing step, labeled neutrophils (5 × 10⁶ cells/ml) were loaded into the upper chamber of a 96-well mini chamber (NeuroProbe, Gaithersburg, MD), separated by a polycarbonate filter with a porosity of 3 μm (NeuroProbe). The lower chambers were loaded with different concentrations of C5a or fMLP in the absence or presence of different concentrations of synthetic peptides of C5a (ranging from 10 to 1000 nM). Cells were then incubated for 30 min at 37°C. The number of cells that migrated through the polycarbonate membrane to the lower surface was determined by cytofluorometry (Cytofluor II; PerSeptive Biosystems, Framingham, MA). For each measurement, samples were done at least in quadruplicate.

H₂O₂ production was determined by an established assay (29). Briefly, isolated neutrophils (2 × 10⁶ cells/ml) were pretreated for 1 h at 37°C with either C5a (10 nM) or the helical peptide regions (described above) A, M, C (each at 1–1000 nM), or the loop peptides D₁, D₂, and D₃ (each at 10–1000 nM), or HBSS (control) alone in the presence of 1 mM sodium azide (to prevent endogenous catalases from destroying H₂O₂). To stimulate neutrophils, PMA (25 ng/ml) was added at the end of the preincubation and the cell suspensions were then incubated at 37°C for 15 min. The reaction was then terminated by addition of 0.1 ml of trichloroacetic acid (50% w/v). The samples were centrifuged for 10 min at 500 × g and ferrous ammonium sulfate and potassium thiocyanate were added to the supernatant fluid at final concentrations of 1.5 mM and 0.25 M, respectively. The absorbance of the ferrithiocyanate complex was measured at 480 nm and compared with a standard curve generated from dilutions of reference solutions of H₂O₂.

To analyze the molecular structures of C5a and C5aR and to localize possible C5a-C5aR interaction sites, the published sequences of C5a and C5aR were molecular graphics in conjunction with examined using the PROSITE database (30) and PHD coil prediction (31), and based on our current in vitro data a three-site binding model was designed.

All values were expressed as mean ± SEM. Data sets of binding, chemotaxis, and H₂O₂ assays were analyzed with one-way ANOVA; differences in the mean values among experimental groups were then compared using the Tukey multiple comparison test. Results were considered statistically significant where p < 0.05.

Based on the sequence and the NMR structure of C5a, different regions of C5a were chosen for synthesis of peptides. For reference purposes, Fig. 1,A depicts the NMR structure indicating positions of the designated inter-α-helical loop peptides (D₁, D₂, D₃). This figure was generated using MOLSCRIPT (32) and POV-Ray. For reference purposes, the structure of C5a and its proposed interactions with C5aR (shown by the red shaded regions) are shown in Fig. 1 B.

These studies were undertaken to assess the binding of ¹²⁵I-C5a to blood neutrophils in the presence or absence of the different selected peptide regions of C5a. Using 100 pM ¹²⁵I-C5a, competition experiments with unlabeled C5a revealed a K_d value of 1 nM (data not shown), which is consistent with published data (10). The saturation binding curve indicating a plateau of binding beyond 10 nM C5a is shown in the inset of Fig. 2,A. Six different peptides termed A, M, and C (aa residues 1–20, 21–40, and 55–74, respectively), and loop peptides D₁, D₂,and D₃ (described in Fig. 1) were evaluated for their ability to inhibit the binding of 10 nM ¹²⁵I-C5a to human neutrophils. Except for loop peptide D₃, all peptides (each at 1 μM) significantly reduced binding of C5a by >50% (Fig. 2). Lower concentrations of peptides (1–100 nM) revealed a dose dependency, with no significant alteration of C5a binding when 1–10 nM of peptides were used and >25% reduction in the presence of 100 nM for all peptides, exclusive of peptide D₃, which again failed to alter binding (data not shown). The pattern of the reduced binding response suggests additional important and, as yet unknown, binding sites present in the middle region of C5a represented by M peptide and in the interhelical loop peptides represented by D₁ and D₂ in addition to the already designated binding sites in the N- and C-terminal regions of C5a (19, 20, 21). As would be expected, intact C5a had the most competitive activity in terms of interfering with binding of ¹²⁵I-C5a to human neutrophils.

Fig. 3 shows the ability of the different synthetic C5a peptides (each at 1 μM) to reduce chemotactic responses of neutrophils to C5a. Human neutrophils were isolated from whole blood and chemotactic responses were assessed using 10 nM C5a (this concentration being associated with peak chemotactic response (Fig. 3,A, left inset, arrow)). C5a was added to the lower compartments of chemotactic chambers in the presence or absence of the six different peptides, A, M, C (Fig. 3,A) and D₁, D₂, and D₃ (Fig. 3,B). Using concentrations ranging from 10 to 1000 nM, all peptides failed to induce migrational responses of neutrophils in the absence of C5a (Fig. 3, all open bars representing peptides at 1 μM). The dose responses for A, M, and C peptide interference with chemotactic responses to C5a are shown in the right inset for Fig. 3,A. In the presence of 10 nM C5a, 1 μM peptide M or C (representing residues 21–40 and 55–74, respectively) resulted in a significant reduction (>60%) in the chemotactic responses. In separate and independent experiments, the C peptide (1 μM) reduced the neutrophil chemotactic response to C5a by 87%, whereas the scrambled C peptide (1 μm) reduced the response by only 8.5% (Table I). The N-terminal helix peptide A failed to significantly alter C5a-induced migration of neutrophils, albeit interference with C5a binding to neutrophils was found with the same peptide (Fig. 2,A), indicating some differences in binding ability and subsequent effects on functional responses to C5a. Whereas loop peptide D₃ did not significantly alter chemotactic responsiveness of neutrophils to C5a, significant reductions (>60%) by synthetic loop peptide D₂ occurred, while nearly total inhibition by peptide D₁ was found (Fig. 3,B). D₃ peptide failed to show any inhibitory effect on the chemotactic response to C5a, consistent with the relatively poor reduction in binding of ¹²⁵I-C5a (Fig. 2,B). In contrast to the inhibitory effects of D₂ peptide on chemotactic responses to C5a (85% reduction, Fig. 3,B and Table I), the scrambled D₂ peptides (D₂-scrambled I and D₂-scrambled II) had no effects on the chemotactic responsiveness of neutrophils to C5a (<7% reductions, Table I). Complete dose responses for the inhibitory activity of each D peptide, normalized to the 10 nM C5a response, are shown in the right-hand inset of Fig. 3 B. These data suggest that peptides from the middle (M) and carboxyl (C)-terminal regions as well as from the loop regions (D₁ and D₂, but not D₃) have the ability to compete functionally with intact C5a, suggesting multiple sites for the binding of C5a to C5aR.

As shown in Figs. 4,A and 5,A, the dose-dependent chemotactic response of neutrophils to 10 nM C5a was examined in the absence and presence of A, M, and C peptides as well as D₁, D₂, and D₃ peptides (all at 1 μM). As expected, both the helical A peptide and the loop D₃ peptide failed to significantly alter the dose-dependent C5a-induced chemotactic response of neutrophils over a wide dose range of C5a (0.1 pM to 100 nM of C5a). In contrast, the presence of M and C peptides as well as the D₁ and D₂ peptides significantly suppressed the C5a-induced dose-dependent chemotactic responses (Figs. 4,A and 5,A). Companion experiments were conducted to see whether any of the six peptides (all at 1 μM) would affect the chemotactic responses of human neutrophils to a variety of concentrations of fMLP, another potent chemotactic agonist for human neutrophils. As shown in Figs. 4,B and 5 B, none of the peptides affected the chemotactic responses of human neutrophils to fMLP. These results demonstrate the specificity of these peptides to C5a-induced responses of human neutrophils.

We have recently shown that in vitro exposure of human neutrophils to human C5a in a dose- and time-dependent manner reduces the ability of human neutrophils to generate H₂O₂ following stimulation with PMA (2, 33). Human neutrophils in suspension were incubated for 1 h at 37°C in the presence or absence of peptides A, M, and C or D₁, D₂, and D₃ (in doses ranging from 1 to 1000 nM). As indicated, cells were then exposed to PMA (25 ng/ml) and the H₂O₂ response was assessed as described elsewhere (29). The PMA concentration chosen was according to our recently published study, in which the 25-ng/ml dose of PMA exhibited a half-maximal oxidative burst response of neutrophils (33). As shown in Fig. 6, in the absence of PMA, C5a and the six different peptides failed to induce significant H₂O₂ responses (gray bars). The scrambled C and D₂ peptides by themselves also failed to induce H₂O₂ responses (data not shown). When neutrophils were incubated for 1 h at 37°C in buffer and then exposed to PMA in the absence of peptides (none), robust H₂O₂ responses were found (Fig. 6). In both sets of experiments, pre-exposure of neutrophils to 10 nM C5a for 1 h at 37°C almost totally abolished the H₂O₂ response (Fig. 6), which has recently been shown to be caused by a C5a-induced inhibition of the NADPH oxidase assembly (33). Pre-exposure of neutrophils to 1 μM of peptides A, M, or C significantly reduced PMA-induced generation of H₂O₂ (Fig. 6,A). In separate and independent experiments, pre-exposure of neutrophils to 1 μM native C peptide reduced the H₂O₂ response by 85%, whereas the scrambled C peptide (1 μM) reduced the response by only 35% (Table I). Dose responses for all peptides are shown in the insets in Fig. 6. Pre-exposure of neutrophils with 1 μM of loop peptides D₁ or D₂ resulted in nearly total abolition of the H₂O₂ response to PMA, whereas exposure to the D₃ peptide did not significantly alter the H₂O₂ response to PMA (Fig. 6,B). In separate and independent experiments, 1 μM D₂ peptide reduced the H₂O₂ response by 78.5%, while 1 μM scrambled D_2I and D_2II peptides reduced the H₂O₂ responses by only 1.5% and 15.7%, respectively, (Table I). These data are consistent with the rank order of effects of peptides described in Figs. 2–4, as demonstrated by their ability to interfere with C5a binding to human neutrophils and to inhibit C5a-induced chemotactic responses.

The complement system is a main buttress of the host defense system. The complement activation product, C5a, is known to be a potent mediator of the monocyte-phagocytic cell system and binds to the seven-transmembrane spanning receptor, C5aR (10). The parts of the C5a molecule mainly involved in binding to C5aR and in subsequent biological responses are still matters of debate (34, 35, 36). This is in striking contrast to C3a in which the C-terminal peptide (residues 69–77) contains virtually all of the biological function of intact C3a (37), while C-terminal peptides of C5a contain some biological activity (38, 39, 40), albeit this is extremely limited. Only at very high concentrations do these peptides show weak agonist activity when compared with intact C5a. This is consistent with the concept that interaction of C5a with C5aR involves several discontinuous regions of C5a and C5aR (22, 23, 24, 32, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50). Site-directed mutagenesis studies of C5a have suggested that regions in the middle of the C5a molecule as well as in the C and N terminus ends facilitate C5a binding to C5aR on neutrophil membranes (19). Our findings of substantial inhibition of ¹²⁵I-C5a binding to intact human neutrophils in the presence of synthetic peptides representing the amino-terminal, middle, and carboxyl-terminal regions of C5a implicate an involvement of all three regions in binding of C5a to C5aR (Fig. 2). Although many studies have focused on the amino- or C-terminal regions, little is known about the binding properties of the middle peptide region of C5a or the loop regions of C5a. In one report, a disulfide-linked core peptide including residues 20–37 obtained by tryptic digestion of C5a_desarg was shown to exhibit weak competitive binding of C5a to C5aR (47), which would support our findings.

However, not only linear protein segments but also posttranslational changes (51) and assembled topographic sites (i.e., arrays of amino acids juxtaposed by the 3D folding of the helical region of C5a) are possible targets for binding interactions between C5a and C5aR (51, 52). Therefore, based on 3D NMR structure studies, which showed closely conserved internal residues of the C3a and C5a molecules but completely different surface residues (43, 44), we chose different inter-α helical loop peptide regions containing residues of interest (K-14, H-15, V-17, K-19, E-32, T-33, A-39, R-40, and L-43) as candidates for additional interactions of C5a with C5aR (20) (Fig. 1). In competitive binding assays, peptides representing inter-α helical loop peptides (D₁ and D₂) significantly reduced C5a binding, but blocking activity was not found with the loop D₃ peptide (Figs. 1 and 2). The efficacy of the D₁ peptide is supported by other data indicating that a double replacement of K-19 and K-20 in this C5a region reduced binding of C5a to C5aR (45). Further evidence for a possible receptor-interaction site in the C5a peptide loop D₁ has been suggested by a greatly increased affinity of a conformationally biased C-terminal decapeptide when linked to C5a_12–20 (39). Loop D₂, which represents a small peptide region within the M peptide of C5a, revealed equal effects on blocking the binding of C5a to C5aR in contrast to the M peptide, suggesting that the hexapeptide D₂ represents a receptor interaction domain of the middle region. In support of this conclusion, C5a_20–37 residues that contain the D₂ region have been shown to represent an important receptor recognition domain (47). Inconsistent results have been reported regarding the interaction of the C5a peptide loop D₃ (including residues R-40 and R-46) with C5aR. Although substitution of R-40 by G changed C5a binding behavior to neutrophil membranes (19), no participation of this loop inclusive of R-40 and P-45 in receptor binding was found (38, 45) and no functional activity (39, 44) was demonstrated, in confirmation of our findings.

In the current studies, neutrophils were exposed to C5a, resulting in a defective H₂O₂ response of stimulated neutrophils. When human neutrophils were first exposed to the various peptides representing regions of C5a, all designated peptides except of D₃ were capable of profoundly compromising H₂O₂ responses in a dose-dependent manner (Fig. 5). The D₁, M, and D₂ peptides, but not the scrambled D₂ peptides, were potent molecules for inhibiting H₂O₂ generation, suggesting that important biological interaction sites may exist in the middle region of C5a. Mutations in the D₁ region ((A-19, A 20)-rhC5a), but not in the D₃ region, have been shown to influence myeloperoxidase release from human neutrophils (45). In addition, the amino-terminal peptides 1–17 of porcine C5a_{des arg} have been reported to cause neutrophil degranulation in guinea pigs and to contribute to the biological potency of the intact molecule (43). Other studies, which have investigated enzyme release from human neutrophils in the presence of cytochalasin B, indicated an effector site in the C-terminal decapeptide C5a_65–74, with residues 65–69 contributing to biological activity (41, 53). This is consistent with our findings suggesting the biological efficacy of the C peptide region of C5a, in addition to the central (middle) region and D₁ loop peptide region.

Almost the same pattern of sites in the C5a molecule interactive with C5aR could be identified in chemotaxis assays using different peptides in the presence of C5a. Neutrophil exposure to these peptides did not affect the chemotactic responses of neutrophils to fMLP (Figs. 3 and 4), suggesting specificity of the peptides. Similarly, the scrambled peptides had no effects on chemotactic and H₂O₂ responses of neutrophils when compared with the nonscrambled C and D₂ peptides. Mutation of the N terminus of C5a did not alter chemotactic responses to C5a, whereas chemotactic responses of neutrophils to R-74, G-73, L-72, K-68, or R-40-substituted mutants of C5a were significantly reduced (19). Additionally, substitution in C5a at A-26 appeared from NMR studies to reduce chemotactic potency of neutrophils by carrying a long-distance conformational change in D₁ around H-15, again suggesting discontinuous central regions and the C terminus to be required to achieve full biological response (19, 41).

In an attempt to localize sites within C5aR (9, 10) that could interact with the C5a molecule, the secondary structure and potential interactions of C5a and C5aR were evaluated using the PROSITE database (30) and PHD coil analysis (31). Based on our in vitro data, we propose a three-site binding model (Fig. 1,B). There is much evidence that a first binding site is located between the negatively charged amino residues in region 21–30 of the C5aR (21, 34, 54) and the four positively charged K-12/14/19/20 residues within the D₁ region. The negatively charged N-terminal residues 10–18 of C5aR were reported not to be directly involved in binding of C5a but it was suggested that they interact with the positively charged and hydrophobic stretch of residues within extracellular C5aR loop II, which is close to the fifth transmembrane helix, presumably forming a stable C5a binding pocket (34). On the other end of the extracellular C5aR loop II (close to the fourth transmembrane helix), there is an additional positively charged region that could interact with the negatively charged D₂ region of C5a which therefore might be considered as a second, newly recognized binding site. The disulfide bond between C-109 in the extracellular loop I and C-189 in the extracellular loop II (34) could stabilize this proposed second binding site. There is also fairly good evidence for the existence of a third C5a-C5aR interaction site (21, 34, 35, 54, 55) in which R-206 in the fifth transmembrane domain has been suggested to play an important role in binding the C terminus by acting as a counterion for R-74 of C5a (35). A recent report has suggested in neutrophils that further C-terminal interactions also may occur between K-68 of C5a and E-199 of the C5aR, close to the fifth transmembrane helix (56). Another study has shown that substitution of H-67 by F resulted in a significant increase in potency of C5a, suggesting an advantage in receptor binding by adding more hydrophobic residues to the ligand (41), facilitating interactions between the C5a C terminus and transmembrane helices of C5aR (21). Furthermore, in comparison to C5aR, the recently published orphan receptor C5L2 (57) (which binds C5a with high affinity but couples poorly to G protein-mediated signaling pathways) exhibit almost identical electrical charge patterns of the three above suggested interaction sites (Fig. 1 B), even if there is only a 35% amino acid identity with C5aR.

These data imply that within the C5a molecule at least three different discontinuous regions exist and interact with C5aR and suggest possible targets for the development of effective C5aR antagonists for intervention in the inflammatory response.

We thank F. A. Bjork and Dr. D. McConnell for ¹²⁵I labeling of C5a. We also thank Beverly Schumann and Peggy Otto for their excellent secretarial assistance in preparation of this manuscript.