Molecular basis of diverse ligand recognition and species-specific pharmacology at the complement anaphylatoxin receptor C3aR

Abstract

The complement anaphylatoxin receptor, C3aR, is a prototypical G protein-coupled receptor (GPCR) that plays multi-faceted roles in immune regulation, pro-inflammatory responses, neuroimmune modulation, and host defense mechanisms against pathogens. Recent structural coverage has provided important insights into C3aR activation and signaling, however, confounding observations in the literature related to ligand efficacy and functional responses present a major challenge for drug discovery efforts. Here, we systematically and comprehensively profile a broad set of natural and synthetic ligands at C3aR, and demonstrate functional specialization in terms of species-specific pharmacology, receptor activation, and signaling-bias. Taking a lead from this, we determine seven cryo-EM structures of C3aR, and employ structure-guided site-directed mutagenesis studies to decode a convergent molecular mechanism that orchestrates the recognition and efficacy of diverse agonists. This study elucidates key molecular insights into C3aR activation and signaling with direct implications for therapeutic design.

The complement anaphylatoxin receptor, C3aR is a potential drug target in various inflammatory disorders. Here, authors present insights into its activation, signaling, and pharmacology that is likely to facilitate therapeutic design going forward.

Introduction

The complement system is an integral part of the innate immune response mechanism that is critical for protecting the host against pathogenic infections, including bacteria, viruses and fungi^1,2. It involves the activation of a multi-step cascade resulting in the generation of a series of peptides in blood plasma with specialized functions coordinating to neutralize pathogens³. Complement proteins mediate and contribute towards the process of opsonization, formation of the membrane attack complex (MAC), and recruitment of immune cells to the site of infection, triggering inflammation⁴. The terminal products of the complement cascade include complement anaphylatoxins, namely C3a, C4a, and C5a, which are produced by the proteolytic cleavage of C3, C4, and C5 complement proteins^1,5. Of these, C3a exerts its cellular and physiological functions, such as chemotaxis, degranulation and cytokine production via complement anaphylatoxin receptor C3aR^6,7. C3aR is a prototypical G protein-coupled receptor (GPCR) that is expressed at the surface of a variety of immune cells, including mast cells, neutrophils, and macrophages, and its downstream signaling via heterotrimeric G-proteins and β-arrestins (βarrs) is crucial for the physiological responses exerted by the anaphylatoxin C3a^8–14. C3aR continues to be an important therapeutic target for acute anaphylaxis, sepsis and lung injury, ischemia/reperfusion injury, autoimmune disorders, such as vasculitis, rheumatoid arthritis, lupus, and multiple sclerosis, as well as chronic inflammatory conditions, such as inflammatory bowel disease and neurodegenerative disease^15–17.

Physiologically, C3a is rapidly proteolyzed by carboxypeptidases to remove the terminal arginine residue, leading to the generation of C3a^-d-Arg18,19. Such a mechanism is generally believed to help dampen the inflammatory response exerted by anaphylatoxins C3a and C5a, as C3a^-d-Arg/C5a^-d-Arg are typically believed to lose their ability to bind and activate their cognate receptors^18–23. More recently, a naturally occurring peptide derived from VGF, the large neuropeptide expressed in the peripheral and central nervous system, known as TLQP-21, has been identified as the second natural agonist of C3aR with a functional contribution in lipolysis, neurodegeneration, and metabolism^24–27. In particular, TLQP-21-induced, β-adrenergic receptor-mediated lipolysis in adipocytes, and increased resting energy expenditure in mouse model, have uncovered untapped therapeutic potential of C3aR in metabolic conditions including obesity^26–29. Moreover, TLQP-21-mediated activation of microglial BV2 cells promotes the clearance of extracellular amyloid-β fibrils. Additionally, the peptide modulates microglial function through C3aR signaling pathways, reducing neuropathology in mouse models^30,31. However, an incomplete understanding of species-specific functional responses encoded by the TLQP-21-C3aR axis poses a key challenge in therapeutic development^25,32,33.

Although peptides derived and modified from the carboxyl-terminus of C3a and C5a are known to behave as partial agonists at C3aR with potential therapeutic opportunities^34–37, a systematic study to decipher the intricate details of receptor subtype selectivity (e.g., C3aR vs. C5aR1), species-specificity (e.g., human vs. mouse), and signaling-bias (e.g., G-proteins vs. β-arrestins) is still lacking. For example, EP54, a C5a-derived peptide, significantly reduced mammary tumor development and progression in murine models, and it has been explored as an immunoadjuvant with multiple antigens^38–41. Although derived from C5a, EP54 is capable of activating C3aR; however, its species-specific pharmacology at C3aR has not been explored yet. Finally, a handful of small-molecule ligands have been described for C3aR, including the well-characterized C3aR antagonists SB290157 and JR14a^{15,16,42–44}. However, emerging evidence from multiple studies has hinted at their agonistic properties and species-specific pharmacology^{16,32,45–47}, which remains primarily unexplored at the molecular and structural levels, although JR14a-bound C3aR structures have been reported recently^47,48. Taken together, C3aR offers a rich tapestry of ligands and an opportunity to explore the molecular basis of diverse ligand recognition, species-specific pharmacology, and signaling-bias.

In this backdrop, here we present a comprehensive profiling of a diverse set of ligands at C3aR to uncover species-specific pharmacology of the natural and synthetic agonists as well as naturally encoded signaling-bias. Using cryo-EM structures of distinct agonist-receptor-transducer combinations, we rationalize the agonistic activity of diverse ligands, including those previously portrayed as antagonists and discover a key switch in the orthosteric binding pocket that is engaged to impart robust agonism. This study provides a important insights into C3aR pharmacology and activation with direct implications for novel drug discovery targeting these receptors.

Results

Species-specific pharmacology and signaling-bias at C3aR

We first set out to comprehensively profile a broad set of ligands at C3aR, including the natural agonists C3a and C3a^-d-Arg and TLQP-21, synthetic peptide agonist EP54, and small molecule ligands SB290157 and JR14a (Figs. 1A, 2A, 3A). We tested all possible combinations in terms of ligands of human and mouse origin on human and mouse C3aR in transducer-coupling assays, namely cAMP inhibition, G-protein dissociation and βarr1/2 recruitment using GloSensor and NanoBiT assays (Figs. 1B, C, 2B–E, 3B, C)^49–51. We observed that human and mouse C3a exhibited a modest but significant difference in potency at the human C3aR, with human C3a being more potent, although both were full agonists (Fig. 1B, C). On the other hand, both human and mouse C3a were equally potent and efficacious at the mouse C3aR (Fig. 1B, C).

Fig. 1. Pharmacological profiling of h/mC3a and h/mC3a^-d-Arg on h/mC3aR.

A A schematic depicting potential species-specific pharmacology of the human and mouse C3a/C3a^-d-Arg at h/mC3aR. Created in BioRender. Mishra, S. (2026) https://BioRender.com/9mj4a62. B C3a/C3a^-d-Arg-induced G-protein-coupling measured using NanoBiT-based G-protein dissociation and GloSensor-based cAMP assays upon stimulation of HEK-293 cells expressing the indicated receptors. Data was normalized with respect to the response at the lowest ligand concentration treated as 100% (mean ± SEM; n = 3-4 independent experiments, i.e., for G-protein dissociation with C3a at h/m C3aR: n = 4, for G-protein dissociation with C3a^-d-Arg at h/m C3aR: n = 3; for GloSensor assay with C3a at h/m C3aR: n = 4, for GloSensor assay with C3a^-d-Arg at hC3aR: n = 4, and at mC3aR: n = 3. C C3a/C3a^-d-Arg-induced βarr1/2 recruitment measured using a NanoBiT-based assay upon stimulation of HEK-293 cells expressing the indicated receptors. Data was normalized to the response at the lowest ligand concentration, which was set to 1 (mean ± SEM; n = 3 independent experiments). Comparison of Emax and EC₅₀ values of the corresponding data are presented in Supplementary Table 1. Source data are provided as a Source Data file.

Fig. 2. Pharmacological profiling of TLQP-21 and EP54 on h/mC3aR.

A A schematic depicting the generation of TLQP-21 from VGF, and design of EP54 based on the carboxyl-terminus of C5a. Created in BioRender. Mishra, S. (2026) https://BioRender.com/klmhtgn. B,C TLQP-21/EP54-induced G-protein-coupling measured using NanoBiT-based G-protein dissociation and GloSensor-based cAMP assays upon stimulation of HEK-293 cells expressing the indicated receptors. Data was normalized with respect to the response at the lowest ligand concentration treated as 100% (mean ± SEM; n = 3-4 independent experiments, i.e., for G-protein dissociation with TLQP-21 at hC3aR: n = 3, and at mC3aR: n = 4, G-protein dissociation with EP54 at h/mC3aR: n = 3; for GloSensor assay with TLQP-21 at hC3aR: n = 4, and at mC3aR: n = 3, for GloSensor assay with EP54 at h/mC3aR: n = 4. D,E TLQP-21/EP54-induced βarr1/2 recruitment measured using a NanoBiT-based assay upon stimulation of HEK-293 cells expressing the indicated receptors. Data was normalized to the response at the lowest ligand concentration, which was set to 1 (mean ± SEM; n = 3-4 independent experiments, i.e., for βarr1 recruitment with TLQP-21 at h/mC3aR: n = 3, for βarr1 recruitment with EP54 at hC3aR: n = 4, and at mC3aR: n = 3; for βarr2 recruitment with TLQP-21 at h/mC3aR: n = 4, for βarr2 recruitment with EP54 at hC3aR: n = 4, and at mC3aR: n = 3). Comparison of Emax and EC₅₀ values of the corresponding data are presented in Supplementary Table 2. Source data are provided as a Source Data file.

Fig. 3. Pharmacological profiling of SB290157 and JR14a at h/mC3aR.

A Chemical structures of small molecule ligands of C3aR, SB290157 and JR14a, drawn using ChemDraw software. B SB290157/JR14a-induced G-protein-coupling measured using NanoBiT-based G-protein dissociation and GloSensor-based cAMP assays upon stimulation of HEK-293 cells expressing the indicated receptors. Data was normalized with respect to the response at the lowest ligand concentration treated as 100% (mean ± SEM; n = 3 independent experiments). C SB290157/JR14a-induced βarr1/2 recruitment measured using a NanoBiT-based assay upon stimulation of HEK-293 cells expressing the indicated receptors. Data was normalized to the response at the lowest ligand concentration, which was set to 1 (mean ± SEM; n = 3 independent experiments). Comparison of Emax and EC₅₀ values of the corresponding data are presented in Supplementary Table 3. Source data are provided as a Source Data file.

C3a undergoes proteolytic cleavage under physiological conditions, resulting in the removal of terminal arginine and generation of C3a^-d-Arg, and this is generally believed to be a mechanism to dampen the inflammatory response to limit prolonged inflammation^18,19. Accordingly, we observed near-complete loss of transducer-coupling for human C3a^-d-Arg on the human C3aR; however, mouse C3a^-d-Arg exhibited a weak but measurable activity on the mouse C3aR (Fig. 1B, C). We note here that mC3a^-d-Arg may exhibit full efficacy at higher concentrations, although it remains technically challenging to measure.

In addition to C3a as the natural agonist, C3aR also recognizes another naturally encoded peptide known as TLQP-21, which is derived from VGF, a member of the extended granin family with primary expression in the central and peripheral nervous system^24,25,52. Interestingly, TLQP-21 is reported to promote energy expenditure and modulate inflammatory pain and gastric contractility in mouse models⁵³, and also inhibit feeding in hamsters^54,55, and therefore, the TLQP-21-C3aR axis has been gaining traction as a potential therapeutic target in obesity and metabolic disorders⁵⁶. Earlier studies have established the selectivity of TLQP-21 for C3aR over C5aR1^29,52, and our functional profiling data presented here revealed that both human and mouse TLQP-21 are more potent on the mouse C3aR compared to the human C3aR, which is also in agreement with previous studies⁵² (Fig. 2B, D). More interestingly, both human and mouse TLQP-21 display a very low level of βarr recruitment on the human C3aR, while they are nearly as efficacious as C3a on the mouse C3aR, albeit with a lower potency (Fig. 2D). These findings underscore the species-specific pharmacology exhibited by TLQP-21 and naturally encoded signaling-bias at C3aR, making it an interesting system for structural analysis.

In addition to the naturally derived peptide agonists of C3aR, several synthetic peptides derived and modified based on the carboxyl-terminus of C3a and C5a have also been explored in terms of their agonism and signaling-bias and receptor selectivity^34,57–62. Of these, EP54, a decapeptide designed based on the carboxyl-terminal sequence of human C5a, has also been characterized as an effective vaccine adjuvant^63,64. However, there are confounding reports in the literature at the level of a specific complement anaphylatoxin receptor that it engages to elicit the functional responses⁶⁵. Therefore, we first profiled EP54 on the human and mouse C3aR in the transducer-coupling assays. We observed that EP54 exhibited weaker potency but nearly full efficacy at C3aR in G-protein and βarr assays, although it was slightly less efficacious at mouse C3aR (Fig. 2C, E).

Finally, a series of small-molecule ligands portrayed as antagonists has been developed and characterized at C3aR^42,43. However, the most well studied compounds, namely SB290157 and JR14a, have also been reported to exhibit agonistic activity in various cellular assays^45–47. Therefore, we profiled both molecules in our transducer-coupling assays on the human and mouse C3aR and observed that both elicited agonistic activity at C3aR (Fig. 3B, C). Compared to C3a, SB290157 displayed a partial agonist profile at the human C3aR in terms of G-protein activation but a weak response in βarr recruitment (Fig. 3B, C). However, it did not activate mouse C3aR in either G protein activation or β-arrestin recruitment assays, suggesting strong selectivity for human C3aR. On the other hand, JR14a was nearly identical to C3a at both human and mouse C3aR in terms of G-protein activation but behaved as a partial agonist in the βarr recruitment assay. These data suggest that, similar to naturally-occurring peptide agonists, i.e., TLQP-21, the small molecule compound SB290157, but not JR14a, also encodes species-specific pharmacology at C3aR. Receptor expression as measured by cell surface ELISA in the functional assays showed at least two-fold signal over the mock-transfected cells (Supplementary Fig. 1). The statistical analyses of efficacy and potency of various ligands measured on human C3aR and mouse C3aR are presented in Supplementary Table 1–3.

We note that the ligand pharmacology presented here is largely consistent with previous studies; however, there are some minor differences, likely arising from differences in assay systems and experimental conditions. For example, in several of the previous studies, a C3a fragment containing the carboxyl-terminal residues (63-77) has been used as a reference ligand, while we have used full-length C3a in our study^32,66. Moreover, some of the previous studies have also used GRK2 overexpression in the βarr recruitment assays^32,66. while our assay system utilizes the endogenously expressed GRKs. In addition, the precise details of the receptor species (i.e., human or mouse) in some of the previous studies have not been explicitly mentioned³², and the apparent difference may simply reflect a better-defined receptor-ligand pairing employed in the current study.

Taken together, comprehensive profiling of ligands on C3aR uncovered modest but significant species-specific differences in anaphylatoxin pharmacology, and near-complete loss of function of C3a^-d-Arg. We also revealed a striking pattern of selectivity in human and mouse TLQP-21 at C3aR, efficient activation of human C3aR by C5a-derived peptide EP54, and robust agonism by SB290157 and JR14a at C3aR with striking selectivity of SB290157 for the human C3aR vs. mouse C3aR.

Structural coverage of ligand receptor complexes

In order to understand the molecular basis of the intriguing functional responses exhibited by these diverse set of ligands, we next determined seven structures of the human and mouse C3aR in complex with various agonists and heterotrimeric G-proteins using cryo-EM (Fig. 4). The experimental details of complex reconstitution, cryo-EM data collection and processing, and overall regions of different components resolved in the structures are summarized in Supplementary Figs. 2–4, Supplementary Tables 4 and 5 and Source Data 1–3. The cryo-EM density enabled us to model the overall architecture of the complexes and most of the side chains in the regions of sufficient local resolution; however, the side chains in the regions of weak cryo-EM density and poor resolution were left unmodeled. In the mC3a-mC3aR structure, the density corresponding to the core domain of the ligand was poorly resolved in the final map despite focused local refinement using a ligand-centered mask. Still, however, the distal carboxyl-terminus of mouse C3a was well resolved to occupy the orthosteric ligand binding pocket in the receptor (Fig. 4 and Supplementary Fig. 5A). This likely reflects a relatively lesser involvement of the core domain of the ligand in the interaction with the receptor. In addition, the density corresponding to the proximal stretch of the N-terminus of the mouse C3aR, and the majority of the long 2nd extracellular loop (ECL2) i.e., from Ser¹⁷⁸ to His³²¹ were also not observed. For the human and mouse TLQP-21 in complex with the mouse C3aR, the proximal stretch of the ligand was not resolved; however, the distal segment, ranging from R13 to R21, occupying the orthosteric binding pocket was resolved well (Fig. 4 and Supplementary Fig. 5B, C). The interaction of EP54 on mouse C3aR, was clearly visible with discernible interaction with the receptor (Fig. 4 and Supplementary Fig. 5D). The density for the small-molecule ligand, i.e., SB290157, in human C3aR and JR14a in the orthosteric binding pocket of both human C3aR and mouse C3aR was clearly observed, allowing the modeling of the ligands and clear identification of the differences between the two ligands in terms of their interaction with the receptor (Fig. 4 and Supplementary Fig. 5E–G). As anticipated, the overall structures of the complexes of the human and mouse C3aR with different ligands are similar in terms of receptor conformation with an RMSD of <1 Å. We also note that C3a^-d-Arg and TLQP21^-d-Arg exhibit very weak activity at the human and mouse C3aR^25,30,67, if any, and therefore, reconstituting a complex for cryo-EM-based structure determination is likely to be challenging. Taken together with previously reported structures of the human and mouse C3aR^68,69, these structures offer a rich tapestry for extracting structure-guided insights into receptor pharmacology and signaling-bias.

Fig. 4. cryo-EM structures of ligand-bound human and mouse C3aR-Gα_o-scFv16 complexes.

cryo-EM density maps and models corresponding to the indicated complexes are presented and the density for the corresponding ligands in the structures is also shown separately. Color scheme: light sea green, human C3aR; salmon orange, mouse C3aR; prime gold, Gα_o; pink, Gβ; skobeloff, Gγ; dark gray, scFv16.

A convergent mechanism of diverse ligand recognition

The binding of complement anaphylatoxins C3a and C5a to C3aR and C5aR1, respectively, has been conceived to utilize a two-site mechanism involving the N-terminus of the receptor referred to as site 1, and the extracellular side of the transmembrane bundle referred to as site 2 or orthosteric binding pocket⁶¹. While C5a binding to C5aR1 has been directly visualized to involve an interface involving the N-terminus, direct involvement of the N-terminus of C3aR in C3a binding has not been observed in the cryo-EM structures^68–70. Structural comparison of C3aR structures determined here reveals that the distal carboxyl-terminus of the human and mouse C3a penetrate in the orthosteric binding pocket, and the carboxyl-terminal residues make multiple contacts in the orthosteric binding pocket including Arg^5.42, Tyr^6.51, and Asp^7.35 in C3aR (Figs. 5A, 6). In particular, the terminal arginine (R77/R78 in hC3a/mC3a) is involved in strong ionic or cation-π interaction with at least two of these five residues (Fig. 6). Therefore, the absence of this terminal arginine residue in C3a^-d-Arg is likely to disrupt these conserved ionic and cation-π interactions, leading to a significant loss of binding and receptor activation as observed in pharmacological assays. Moreover, the carboxyl-terminal stretch of TLQP-21 and EP54 also occupies a similar and nearly overlapping binding pocket on the corresponding receptors, and their terminal arginine residue also engages a similar set of residues (Figs. 5B–G, 6). Therefore, similar to C3a^-d-Arg, a deletion or mutation of Arg21 in TLQP-21 is also expected to disrupt the ionic and hydrogen-bonding contacts with critical residues in the C3aR orthosteric site, and thereby, destabilize ligand engagement. This aligns with previous studies demonstrating that a deletion of the carboxyl-terminus of TLQP-21, or the mutation of terminal arginine to alanine, nearly abolishes the activity at C3aR^25,30,67.

Fig. 5. Overall positioning and binding pose of various ligands in the orthosteric pocket of h/mC3aR.

A–K Structural comparison of the indicated structures, highlighting the key differences in the overall positioning of various ligands in the orthosteric pocket of the human and mouse C3aR. PDB: 8I9L for hC3a-hC3aR, 9KZK for mC3a-mC3aR, 9KV8 for hTLQP-21-mC3aR, 9KV6 for mTLQP-21-mC3aR, 8I95 for EP54-hC3aR, 9UMJ for EP54-mC3aR, 9KUT for JR14a-hC3aR, 9L0H for JR14a-mC3aR and 9KVP for SB290157-hC3aR complex.

Fig. 6. Key interactions of various ligands in the orthosteric pocket of h/mC3aR.

Selected interactions of the indicated ligands in the orthosteric pocket of the human and mouse C3aR, highlighting the conserved interactions with those of C3a. PDB: 8I9L for hC3a-hC3aR, 9KZK for mC3a-mC3aR, 9KV8 for hTLQP-21-mC3aR, 9KV6 for mTLQP-21-mC3aR, 8I95 for EP54-hC3aR, 9UMJ for EP54-mC3aR, 9KUT for JR14a-hC3aR, 9L0H for JR14a-mC3aR and 9KVP for SB290157-hC3aR complex.

Interestingly, a closer inspection of SB290157-C3aR and JR14a-C3aR structures reveals that these ligands also occupy nearly the same orthosteric pocket on C3aR and exhibit similar interactions as other agonists (Figs. 5H–K, 6). For example, the arginine moiety of SB290157 and JR14a occupy a similar position as that of R77 in C3a with the guanidino group engaging Asp^7.35 and Tyr^6.51 through hydrogen bond and cation-π interaction (Figs. 5H–K, 6). In addition, the terminal oxygen of these ligands interacts with Tyr174^ECL2 through a non-bonded contact (Fig. 6), which is also corroborated by the recently reported JR14a-human C3aR structures^47,48.

Structural basis of species-specific pharmacology at C3aR

Functional profiling suggests that human and mouse C3a differ modestly but significantly in potency at the human vs. mouse C3aR. Structural analysis of the human and mouse C3a with the corresponding receptors shows an overall similar hook-like conformation in both, which overlaps well in the orthosteric binding pocket (Fig. 5A). Interestingly however, we observed that the positioning of human and mouse C3a diverge significantly on their respective receptors at an angle of ~81° when compared with respect to L72 in human C3a as a pivot point (Fig. 5A). As a result, there is a discernible difference in the interaction pattern of the two ligands on their corresponding receptors and analysis of the C3a-interacting residues on human C3aR with orthologous residues on mouse C3aR reveals few notable differences, such as Arg/IIe171^ECL2, Val/IIe^3.33, Phe/Val^6.54, and Lys/Glu^7.29 (human C3aR vs mouse C3aR) in positions that could modulate ligand interaction and functional outcome (Fig. 5A and Supplementary Figs. 6, 7), despite the conserved engagement of the five key residues as mentioned above.

Human and mouse TLQP-21 display a much weaker response on the human C3aR, while robustly activating mouse C3aR. Therefore, we used mouse C3aR activated by human and mouse TLQP-21 to investigate their agonism and compare them with the mC3a-mC3aR structure. The binding of TLQP-21 primarily involves the orthosteric pocket of the receptor, occupying a buried surface area of ~1700 Ų, and the N-terminal segment adopts a helical conformation similar to C3a (Fig. 5B–D). Strikingly, human and mouse TLQP-21 are also positioned on mouse C3aR in an orientation that shows a cross-angle of ~80° compared to human C3a on the human C3aR (Fig. 5B–D). Carboxyl-terminal residues adopt a hook-like structure, analogous to C3a, and occupy a similar position in the orthosteric binding pocket in line with a conserved mechanism of agonism (Fig. 5B–D). As predicted, the overall structural features of C3aR in complex with TLQP-21 are similar to those of the C3a-bound receptor, representing a prototypical active conformation with outward movement of TM6 and 7, and the rearrangement of key structural motifs. Structure-guided mutagenesis of C3aR combined with transducer-coupling assays further validates the key contribution of the interacting residues as visualized in the structures (Fig. 7A, B and Supplementary Fig. 8 and Supplementary Tables 6 and 7).

Fig. 7. Experimental validation of TLQP-21 interacting residues.

A Structural snapshots highlighting key residue interactions between TLQP-21 and mC3aR. B C3aR TLQP-21-induced cAMP response and βarr1 recruitment measured using the GloSensor assay and NanoBiT-based assay, respectively, upon stimulation of HEK-293 cells expressing the indicated receptor mutants. Data was normalized with the response for the wild-type treated as 100% (mean; n = 3 independent experiments). Comparison of ligand pharmacology of the above data are presented in Supplementary Table 6 and 7. Source data are provided as a Source Data file.

Although we do not have the structure of TLQP-21 in complex with human C3aR, it is plausible that the divergent positioning of TLQP-21 on human vs. mouse C3aR, and similar to that observed for C3a, drives this species-specific pharmacology (Fig. 5B–D). Specifically, structural comparison reveals several non-conserved residues at the ligand-receptor interface, including Val/Ile^3.33, Arg/Ile171^ECL2, Phe/Val^6.54, Ser/Leu^6.58, and Cys/Ser^7.38 (human C3aR vs. mouse C3aR). These substitutions might alter the physicochemical properties of the orthosteric pocket, potentially creating a less favorable environment in human C3aR and thereby impairing TLQP-21 binding. Additionally, differences in the ECL2 region sequence between human C3aR and mouse C3aR might further contribute to species-specific effects (Supplementary Figs. 6, 9, 10).

Structural basis of small molecule agonism at C3aR

The analysis of SB290157-bound human C3aR and JR14a-bound human and mouse C3aR structures provides the molecular basis for small molecule agonism at C3aR. These ligands occupy a largely similar orthosteric binding pocket on the receptor, thereby engaging the critical interactions as identified earlier for other agonists (Figs. 5H–K, 6, 8A). However, unlike peptide agonists, JR14a does not engage Arg^5.42 in either human or mouse C3aR, whereas SB290157 is able to interact with this residue (Fig. 6). In addition, the overall structural features of C3aR in complex with these ligands are nearly identical to that of C3a-bound receptor representing a prototypical active conformation with outward movement of TM6 and 7, and the rearrangement of structural motifs (Fig. 8B and Supplementary Fig. 11). Structure-guided mutagenesis of C3aR combined with transducer-coupling assays further validates the key interaction network of these two ligands (Fig. 8C, D and Supplementary Fig. 8, and Supplementary Table 8 and 9). As noted in Fig. 3, SB290157 displayed striking selectivity for the human C3aR, while JR14a is able to recognize both human and mouse receptors. Although the overall structures of SB290157- and JR14a-bound human C3aR are very similar, the spatial positioning of the carboxylate group in SB290157 is marginally deeper in the orthosteric pocket compared to that of JR14a (Fig. 8A). Moreover, while the overall interaction network of JR14a with the human and the mouse receptor is largely conserved within the orthosteric binding pocket, some notable differences are also observed. Specifically, His^7.36 engages in a π-π interaction with one of the 4-chlorotoluene rings in the JR14a-bound mouse C3aR structure, which is missing in the human counterpart (Supplementary Fig. 11). Thus, it is possible that the lack of 4-chlorotoluene rings in SB290157 drives its preference for the human C3aR over mouse C3aR. However, the lack of the SB290157-bound mouse C3aR structure limits a direct comparison, and therefore, future studies are essential to precisely define the structural basis of the species-specific pharmacology exhibited by SB290157 (Supplementary Fig. 12A).

Fig. 8. Interaction of SB290157 and JR14a with C3aR, and receptor activation.

A Structural comparison of SB290157 and JR14a binding pose to human and mouse C3aR. B Structural comparison of SB290157- and JR14a-bound C3aR structure with the inactive C5aR1 reveals activation-dependent structural changes, including the outwards movement of TM6, TM7 and Helix8. C,D SB290157/JR14a-induced cAMP response and βarr1 recruitment measured using the GloSensor assay and NanoBiT-based assay, respectively, upon stimulation of HEK-293 cells expressing the indicated receptor mutants. Data were normalized with the response for the wild-type treated as 100% (mean; n = 3 independent experiments). Comparison of ligand pharmacology of the above data are presented in Supplementary Tables 8, 9. Source data are provided as a Source Data file.

Molecular insights into C3aR activation by EP54

As we have determined the structure of C3aR in complex with EP54, it allows us to systematically explore the molecular details of receptor activation and the efficacy of this peptide agonist. As mentioned earlier, EP54 is a decapeptide derived and modified from the distal carboxyl-terminus of human C5a; however, there are confounding reports in the literature about its cross-reactivity and selectivity towards C3aR and C5aR1⁷¹, and a previous study reported a distinct species-dependent binding profile of EP54 on C5aR1⁵⁹. Our systematic analysis of cellular responses demonstrates that EP54 is capable of activating both human and mouse C3aR with nearly similar potency and efficacy in terms of G-protein activation, although it appears to be relatively weaker at the mouse receptor in the βarr2 recruitment assay. The structural snapshots determined here reveal that the distal carboxyl terminus of EP54 adopts a hook-like conformation, occupying a similar orthosteric pocket of the receptor, and has a significantly conserved set of interactions as that of C3a including critical interactions of the terminal arginine with Arg^5.42, Tyr^6.51, Asp^7.35 through hydrogen bond, cation-π interaction, and salt-bridge interaction, respectively (Fig. 5E–G, 6). Interestingly, while the carboxyl-terminal segment of EP54 aligns well on human and mouse C3aR, their N-terminal segment diverges around a proline kink (Pro7), making an angle of ~65° and ~22°, respectively, between the two receptors (Fig. 5E, F), in addition to minor differences in the interacting interface of EP54 across the two receptors. This may explain the differences in the efficacy and potency of EP54 observed for the human vs. mouse C3aR, although future studies are required to test this hypothesis (Supplementary Fig. 12B, C).

Taken together, these observations suggest a convergent mechanism of diverse ligand recognition driving agonist efficacy at C3aR, where nearly every single agonist visualized so far engages a set of five common residues in C3aR. These residues namely Ser^2.60, Ile^3.32, Arg^5.42, Tyr^6.51, and Asp^7.35 are positioned in a manner to link together TM2-7 in the orthosteric binding pocket and we referred to them as the Five-Point-Switch (Fig. 9 and Supplementary Table 10). Therefore, we mutated each of these residues to alanine in human and mouse C3aR and measured ligand-induced transducer coupling in terms of G protein dissociation and βarr recruitment. In line with structural observations, we observed that site-directed mutagenesis of Arg^5.42, Tyr^6.51, and Asp^7.35 in the human and mouse C3aR resulted in a clear loss of both G-protein dissociation and βarr recruitment, while the effect of Ser^2.60 and Ile^3.32 was rather modest (Fig. 10, Supplementary Fig. 13 and Supplementary Tables 11, 12). These results complement previous findings on the indispensable roles of Arg^5.42, Tyr^6.51, and Asp^7.35 in C3aR ligand binding and activation⁷². Mechanistically, Asp^7.35 anchors the ligand’s terminal arginine via an ionic bond, while Arg^5.42 forms a hydrogen bond with its carboxylate. Tyr^6.51 interaction triggers the toggle-switch movement required for the TM6 outward kink and subsequent G-protein coupling^48,69. While we observed modest effects for Ser^2.60 and Ile^3.32 mutations, recent studies highlight their subtle roles in stabilizing the inactive state (Ile^3.32) and contributing to basal signaling (Ser^2.60)⁴⁸. Taken together, the site-directed mutagenesis data validate a predominant role of these residues in ligand recognition and receptor activation and provide experimental evidence for the “Five-Point-Switch” mechanism. Expectedly, there are additional ligand-specific contacts visualized in the structure, which are likely to drive further fine-tuning of the functional responses, including species-specific functional responses, as discussed in the following sections.

Fig. 9. Identification of a Five-Point-Switch for ligand-receptor interaction in C3aR.

A A heatmap representation of ligand-interacting residues in C3aR derived from the structural snapshots. The color coding reflects the frequency of residue-engagement across the compared structures and highlights five key residues (Ser².⁶⁰, IIe^3.32, Arg^5.42, Tyr^6.51, and Asp^7.35), which are engaged with nearly every ligand analyzed here. In addition to the structures reported here, previously published C3aR structures have also been included in this analysis (PDB: 8I9L for hC3a-hC3aR, 8I95 for EP54-hC3aR and 8J6D for EP141-hC3aR). B A schematic depiction demonstrating the broadly conserved engagement of the five key residues in C3aR across multiple ligands. C A structural representation of C3aR with the five key residues mapped on the ribbon diagram (left) and their spatial arrangement is indicated to highlight the overall positioning of the Five-Point-Switch (right).

Fig. 10. Structure-guided experimental validation of the Five-Point-Switch in C3aR.

Ligand-induced G-protein dissociation and βarr1 recruitment were measured using the NanoBiT-based assays upon stimulation of HEK-293 cells expressing the indicated receptor mutants. Data was normalized with the response for the wild-type treated as 100% (mean; n = 3-4 independent experiments i.e., for G-protein dissociation with all five mutants at hC3aR: n = 3, for G-protein dissociation with Ser^2.60Ala, Ile^3.32Ala and Arg^5.42Ala at mC3aR: n = 4, and for Tyr^6.51Ala and Asp^7.35Ala at mC3aR: n = 3; for βarr1 recruitment with Ser^2.60Ala, Ile^3.32Ala and Arg^5.42Ala at h/mC3aR: n = 4, and for Tyr^6.51Ala and Asp^7.35Ala at h/mC3aR: n = 3). The raw data with SEM are presented in Supplementary Table 11 and 12. Source data are provided as a Source Data file.

Discussion

The complement anaphylatoxin receptors continue to be sought-after therapeutic targets due to their critical role in mediating the physiological responses of the terminal peptides generated following complement cascade activation^3,5. Moreover, from a receptor pharmacology point of view, they exemplify an intriguing system encoding diverse ligand recognition and signaling-bias at the level of ligands and the receptors⁶. The spectrum of structural and associated pharmacological data presented here uncovers several intriguing mysteries encoded by these receptors. In particular, the identification of a Five-Point-Switch in the orthosteric binding pocket of C3aR not only provides a convergent mechanism of diverse ligand recognition but also raises a tantalizing possibility that additional naturally occurring peptides, or those provided by the human microbiota, may also serve as endogenous modulators of these receptors. Although such a scenario remains to be systematically explored and experimentally validated, the emerging paradigm of intracellular complement activation with the conceptual framework of the complosome may already provide preliminary support for such a hypothesis⁷³.

The striking pharmacology and selectivity observed here for some of the agonists may have significant implications for novel drug discovery targeting these receptors in future studies. For example, TLQP-21 has emerged as an interesting lead molecule to engage C3aR from the therapeutic perspective of obesity considering its ability to mediate lipolysis and resting state energy expenditure in mouse models^26,33. However, due to its limited ability to activate the human C3aR, native hTLQP-21 is unlikely to be a promising therapeutic candidate^25,32. We also note that due to the relatively weaker affinity of the human and mouse TLQP-21 on the human C3aR, reconstitution and structural analysis of TLQP-21 with the human receptor were not feasible. However, our structural data now provide an opportunity to further explore the sequence space encoded by TLQP-21 to gain reactivity towards human C3aR, and may support the ongoing efforts to develop TLQP-21-based lead molecules with therapeutic potential⁶⁶. Moreover, direct structural visualization of TLQP-21 interaction with the C3aR also helps rationalize the pharmacological data reported previously. For example, as mentioned earlier, similar to C3a, the carboxyl-terminus TLQP-21 occupies the orthosteric pocket with several critical interactions involving the terminal arginine. This aligns with TLQP-21 structure-function studies reported earlier, demonstrating that carboxyl-terminal truncation and mutation of Arg21 in TLQP-21 leads to a loss of activity at C3aR^25,30,67. There have been confounding reports in the literature regarding EP54-mediated beneficial outcomes, where the specific contribution of C3aR vs. C5aR1 has been unclear^74,75. Our study now reveals that EP54 can activate both human and mouse C3aR, and provides a template to further optimize the efficacy and selectivity of this ligand with potential for an improved therapeutic window⁷⁶.

Our study also unequivocally establishes the agonistic nature of SB290157 and JR14a, which were originally and are still widely reported as antagonists and provides a structural explanation for their ability to activate the receptor. Not only do these ligands make critical interactions in the orthosteric binding pocket that overlap with those of the natural agonist, C3a, but they also induce a receptor conformation that is similar to that stabilized by C3a. These direct structural observations further reiterate the agonistic activity of these ligands and also align with additional structural studies reported recently^47,48. Although we cannot entirely rule out a context-dependent pharmacological behavior of these small-molecule ligands, it is likely that their inherent agonistic activity, leading to receptor desensitization upon pre-incubation, may have resulted in incorrect classification as antagonists.

At the molecular level, our study also provides further support that the N-terminus of C3aR is likely not critical for C3a interaction, and this helps rationalize pharmacology data regarding the relatively better potency and efficacy of peptide agonists at C3aR. In contrast, structural studies of C5aR1 have demonstrated a more substantial contribution of the receptor N-terminus to ligand engagement. In summary, we elucidate multiple layers of functional specialization encoded in the complement anaphylatoxin receptor C3aR using an integrative experimental approach, and we anticipate that these molecular insights will help overcome the existing hurdles to guide therapeutic design targeting these receptors.

Methods

General reagents and chemicals

Most of the general reagents were purchased from Sigma-Aldrich unless otherwise specified. Dulbecco’s Modified Eagle Medium (DMEM), Trypsin-EDTA, Fetal Bovine Serum (FBS), Phosphate-Buffered Saline (PBS), Hank’s Balanced Salt Solution (HBSS), and Penicillin-Streptomycin solution were obtained from Thermo Fisher Scientific. HEK-293 cells (ATCC, Cat. No: CRL-3216) were maintained in DMEM (Gibco, Cat. No: 12800-017) supplemented with 10% (v/v) FBS (Gibco, Cat. No: 10270-106), 100 U mL^-1 penicillin, and 100 μg mL^-1 streptomycin (Gibco, Cat. No: 15140122) at 37 °C under 5% CO₂. Sf9 cells (Expression System, Cat No: 94-001 F) were cultured in protein-free media (Gibco, Cat. No: 10902-088) at 27 °C with shaking at 125 rpm. Constructs for C3aR expression in HEK-293 and Sf9 were designed as described previously⁶⁹. The constructs used in the NanoBiT-based βarr1/2 recruitment assay and the bystander mode of the βarr1 recruitment assay were designed as described previously^69,77,78. The G-protein subunit constructs used in the dissociation assays were generously provided by Asuka Inoue. Mutants of human C3aR and mouse C3aR were generated using the Q5 Site-Directed Mutagenesis Kit (NEB, Cat. No: E0554S). The primers used to generate receptor mutants were synthesized by Sigma-Aldrich; their sequences are listed in Supplementary Tables 13, 14. All DNA constructs were verified by sequencing at Macrogen. Peptides TLQP-21 and EP54 were either purchased from GenScript or synthesized as previously described (Purity >95%)⁵². Small-molecule compounds SB290157 and JR14a were obtained from MedChem Express (Cat. No: HY-101502A for SB290157; Cat. No: HY-138161 for JR14a). M2-HRP-coupled anti-FLAG antibodies were purchased from Sigma-Aldrich (Cat No: A8592).

GloSensor assay to measure cAMP response

Change in intracellular cAMP levels was measured using the GloSensor assay to assess the effect of ligand-induced Gi-mediated signaling⁵⁰. Briefly, HEK-293 cells were transfected with 3.5-5 μg of either N-terminal FLAG-tagged human C3aR or mouse C3aR (WT and mutants), along with 3.5 μg of the F22 plasmid (Promega, Cat. No: E2301). After 14–16 h of transfection, cells were trypsinized and resuspended in assay buffer (20 mM HEPES, pH 7.4, 1X HBSS, and 0.5 mg mL^-1 D-luciferin (GoldBio, Cat. No: LUCNA-1G). Later, 100 µL of transfected cells were seeded into 96-well plates at a density of 0.2 million cells per well. The cells were incubated at 37 °C for 1.5 h, followed by an additional 30 min incubation at room temperature. Basal luminescence was measured for three cycles using the FLUOstar Omega plate reader. After the basal readings, 5 μM forskolin was added to each well, and luminescence was recorded for eight cycles until the readings stabilized. Subsequently, ligand was added at the indicated concentration, and luminescence was recorded for 15 cycles. The signal obtained was normalized to forskolin values, with the lowest ligand concentration set to 100%. Normalized data were subjected to non-linear regression analysis, and dose-dependent response was plotted using GraphPad Prism 10 software.

NanoBiT-based G-protein dissociation assay

Ligand-induced G-protein activation was measured using a NanoBiT-based G-protein dissociation assay⁷⁹. G-protein subunits consisting of the LgBiT-tagged Gα subunit and SmBiT-tagged Gγ2 (C68S mutation) subunit, along with the untagged Gβ1 subunit (pcDNA3.1), were co-expressed with the N-terminal FLAG-tagged hC3aR/mC3aR receptors. The ligand-induced change in luminescence signal was then measured. Briefly, HEK-293 cells were transfected with a plasmid mixture containing 1 µg LgBiT-G_oA, 4 µg Gβ1, 4 µg SmBiT-Gγ2 (C68S), and 1 µg of receptor plasmid (human C3aR or mouse C3aR). After 14-16 h of transfection, cells were harvested by using Trypsin-EDTA and DMEM supplemented with 10% FBS and resuspended in NanoBiT buffer (5 mM HEPES, pH 7.4, 1X HBSS, 0.01% BSA, and 10 µM coelenterazine (Gold Bio, Cat. No: CZ2.5). 100 μL of resuspended cells were then seeded into a 96-well plate at a density of 0.1 million cells per well. The plate was then incubated at 37 °C for 1.5 h, followed by an additional 30 min incubation at room temperature. Basal luminescence was recorded for three cycles using a FLUOstar Omega plate reader. Following basal luminescence measurements, cells were stimulated with ligand at dose-dependent concentrations, resulting in a reduction in the luminescence signal. Ligand-induced response at 10 min was taken for analysis, wherein the basal corrected response was percentage normalized relative to the lowest ligand concentration. The resulting data were plotted using GraphPad Prism 10 software.

NanoBiT-based β-arrestin recruitment assay

Ligand-induced recruitment of β-arrestin (βarr1/2) was performed using NanoBiT-based β-arrestin recruitment assay^80,81. Briefly, HEK-293 cells were transfected with 3.5 µg of C-terminal SmBiT-tagged human C3aR or mouse C3aR, along with N-terminal LgBiT-tagged βarr1/2. For human C3aR and mouse C3aR mutant constructs, the bystander mode of βarr1 recruitment was employed. In this case, HEK-293 cells were transiently transfected with 3–5 µg of receptor (WT and mutants), 5 µg of an N-terminal LgBiT-fused CAAX motif, and 2 µg of N-terminal SmBiT-tagged βarr1. After 16–18 h of transfection, the cells were trypsinized and harvested. The cell pellets were resuspended in assay buffer containing 5 mM HEPES, pH 7.4, 1X HBSS, 0.01% BSA, and 10 µM coelenterazine. The cells were seeded into a 96-well plate at a density of 0.1 million cells per well in a 100 µL volume. Seeded plates were incubated at 37 °C for 1.5 h, followed by an additional 30 min incubation at room temperature. After incubation, basal luminescence was recorded for three cycles using the FLUOstar Omega plate reader. Cells were then stimulated with the indicated ligand concentrations, and ligand-induced increase in luminescence were recorded for 15 cycles. The responses from cycles 5–10 were averaged, and fold normalization was performed, considering the lowest ligand concentration as 1. The fold-normalized values were then plotted in GraphPad Prism 10.

Receptor surface expression assay

To assess cell surface expression of the corresponding receptors, whole cell surface ELISA was performed⁸². Briefly, transfected cells were seeded at a density of 0.2 million cells per well 24 h post-transfection in 24-well plates (pre-coated with 0.01% poly-D-Lysine) and incubated for 24 h at 37 °C in a CO₂ incubator. Once cells were well adhered, the media was removed by aspiration, and the cells were washed once with 1X TBS. This was followed by treatment with 300 μL of 4% PFA for 20 min to fix the cells on ice. Excess PFA was removed by washing the cells thrice with 400 μL of 1X TBS. The wells were blocked by incubating with 200 μL 1% BSA for 1.5 h. 200 μL of anti-FLAG M2-HRP antibody (Sigma, Cat. no: A8592) at a dilution of 1:10,000 was added to the wells and incubated for an additional 1.5 h. To develop the signal, 200 μL of TMB-ELISA substrate (Thermo Scientific, Cat. no: 34028) was added to the wells and incubated till the development of adequate color. Reaction was quenched by transferring 100 μL of the solution to a 96-well plate containing 100 μL of 1 M H₂SO₄, and absorbance was measured at 450 nm using PerkinElmer VICTOR^TM X4 plate reader. To estimate the number of cells fixed in each well, excess TMB was first removed by washing once with 400 μL of 1X TBS, and the cells were incubated for 20 min in 0.2% Janus green B stain (Sigma, Cat. no. 201677). After removing excess stain by repeated washing with Milli Q water, a signal was developed by adding 800 μL of 0.5 N HCl to each well and read at 595 nm. Signal was normalized by dividing the reading obtained at 450 nm with the reading obtained at 595 nm. Receptor surface expression was fold-normalized to empty pcDNA3.1 vector-transfected cells and plotted in the GraphPad Prism 10 software.

Expression and purification of C3a and C3a^-d-Arg

Genes encoding full-length C3a and C3a^-d-Arg (human and mouse) were cloned into the pET-32a(+) vector with a Trx-6X-His tag at the N-terminal end and purified following the previously described protocol with some modifications⁸³. Briefly, freshly transformed E. coli SHuffle cells were inoculated in 50 mL of LB media containing 100 µg mL^-1 ampicillin for the starter culture at 30 °C. The overnight-grown primary culture was then inoculated into 1.5 L of LB media with 100 µg mL^-1 ampicillin, and the culture was allowed to grow at 30 °C. When the optical density (O.D.) reached approximately 0.8–1.0, the culture was induced with 1 mM IPTG and then shifted to 16 °C for overnight induction. The cells were harvested and incubated with 1 mg mL^-1 lysozyme in a buffer containing 50 mM HEPES, pH 7.4, 300 mM NaCl, 20 mM imidazole, 1 mM PMSF, and 2 mM benzamidine for 2 h at 4 °C. Following this, the cells were disrupted using ultrasonication, and the cell debris was removed by high-speed centrifugation. C3a/C3a^-d-Arg were enriched on Ni-NTA resins using gravity flow. Non-specific proteins were removed through extensive washing with a buffer containing 50 mM HEPES, pH 7.4, 1 M NaCl, and 20 mM imidazole. The fusion protein(s) were then eluted using a buffer composed of 50 mM HEPES, pH 7.4, 150 mM NaCl, and 300 mM imidazole. The Trx-His tag was cleaved by treating the protein with TEV protease (1:20 w/w, TEV: protein) for 8–10 h at room temperature. The purified proteins were further cleaned using cation-exchange chromatography and stored at −80 °C with a final glycerol concentration of 10%.

Expression and purification of C3aR

The human and mouse C3AR genes were codon-optimized and cloned into the pVL1393 vector with an N-terminal HA signal sequence, FLAG tag, and a 22-amino-acid stretch of the M4 receptor N-terminal region (amino acids 2-23, ANFTPVNGSSGNQSVRLVTSSS) for purification. The receptor was expressed and purified from Spodoptera frugiperda (Sf9) cells using a baculovirus-mediated expression system. Briefly, the insect cells were harvested 72 h post-infection and lysed through sequential douncing in the following buffers: a hypotonic buffer (20 mM HEPES, pH 7.4, 10 mM MgCl₂, 20 mM KCl, 1 mM PMSF, and 2 mM benzamidine), a hypertonic buffer (20 mM HEPES, pH 7.4, 1 M NaCl, 10 mM MgCl₂, 20 mM KCl, 1 mM PMSF, and 2 mM benzamidine), and a solubilization buffer (20 mM HEPES, pH 7.4, 450 mM NaCl, 2 mM CaCl₂, 1 mM PMSF, 2 mM benzamidine, and 2 mM iodoacetamide). After douncing, the receptor was solubilized for 2 h at 4 °C with continuous stirring in a solution containing 0.5% L-MNG (Anatrace, Cat. no: NG310) and 0.01% cholesteryl hemisuccinate (Sigma, Cat. no: C6512). Following solubilization, the salt concentration was reduced to 150 mM, and the receptor was purified using an M1-FLAG column. To remove non-specific proteins from the FLAG beads, three washes with a low-salt buffer (20 mM HEPES, pH 7.4, 150 mM NaCl, 2 mM CaCl₂, 0.01% CHS, 0.01% L-MNG) were alternated with two washes using a high-salt buffer (20 mM HEPES, pH 7.4, 350 mM NaCl, 2 mM CaCl₂, 0.01% L-MNG). The bound receptor was eluted using a FLAG elution buffer containing 20 mM HEPES, pH 7.4, 150 mM NaCl, 0.01% L-MNG, 2 mM EDTA, and 250 µg mL^-1 FLAG peptide. The purified receptor was incubated on ice with 2 mM iodoacetamide for 30 min. Subsequently, 2 mM L-cysteine was added, and the mixture was incubated for an additional 30 min. The purified receptor was concentrated using a 30 kDa MWCO concentrator and stored at −80 °C in 10% glycerol till further use. 100 nM of mC3a or 1 µM of hTLQP-21, mTLQP-21, EP54, SB290157 or JR14a were kept in all steps of receptor purification.

Expression and purification of G-proteins

The miniGα_o construct was engineered by incorporating an scFv16 binding sequence at the N-terminus of the previously described miniGα_o^84,85. The miniGα_o gene was cloned into a pET-15b(+) vector with an N-terminal 6X-His tag for expression. The engineered construct was transformed into E. coli BL21(DE3) cells for protein expression. A starter culture was grown in LB media and incubated at 37 °C with shaking at 160 rpm for 6-8 h. This starter culture was used to inoculate an overnight culture in LB media supplemented with 0.2% glucose and grown at 30 °C. 15 mL of the overnight culture was transferred to 1.5 L of Terrific Broth (TB) media. Protein expression was induced with 50 μM IPTG when the culture reached an optical density (O.D₆₀₀) of 0.8, followed by incubation at 25 °C for 18–20 h. Cells were harvested and lysed in buffer containing 40 mM HEPES, pH 7.4, 100 mM NaCl, 10 mM imidazole, 10% glycerol, 5 mM MgCl₂, 1 mM PMSF, and 2 mM benzamidine, supplemented with 1 mg mL^-1 lysozyme, 50 μM GDP, and 100 μM DTT. Cell debris was removed by centrifugation at 39,846 x g for 30 min at 4 °C. The protein was purified using Ni-NTA affinity chromatography. The bound protein was washed extensively with buffer containing 20 mM HEPES, pH 7.4, 500 mM NaCl, 40 mM imidazole, 10% glycerol, 50 μM GDP, and 1 mM MgCl₂. The protein was then eluted using 20 mM HEPES, pH 7.4, 100 mM NaCl, 10% glycerol, and 500 mM imidazole. The purified protein fractions were pooled and stored at −80 °C in 10% glycerol until further use.

The genes encoding the Gβ1 and Gγ2 subunits were cloned into a bicistronic pVL1392-based vector with an N-terminal histidine tag on Gβ1. These constructs were expressed in Sf9 cells using the baculovirus expression system. The cells were harvested 72 h post-infection and resuspended in a lysis buffer containing 20 mM Tris-Cl, pH 8.0, 150 mM NaCl, 10% glycerol, 1 mM PMSF, 2 mM benzamidine, and 1 mM MgCl₂. The cells were disrupted using a dounce homogenizer and then centrifuged at 4 °C for 40 min at 39,846 x g. The resulting pellet was resuspended and dounced again in a solubilization buffer (20 mM Tris-Cl, pH 8.0, 150 mM NaCl, 10% glycerol, 1% DDM, 5 mM β-ME (β-mercaptoethanol), 10 mM imidazole, 1 mM PMSF, and 2 mM benzamidine) and solubilized at 4 °C under constant stirring for 2 h. Cell debris was pelleted down by centrifuging at 49,192 x g for 60 min at 4 °C. The protein was enriched on Ni-NTA resin. After extensive washing with a buffer containing 20 mM Tris-Cl, pH 8.0, 150 mM NaCl, 30 mM imidazole, and 0.02% DDM, the protein was eluted with 300 mM imidazole in a buffer composed of 20 mM Tris-Cl, pH 8.0, 150 mM NaCl, and 0.01% L-MNG. Eluted protein was concentrated with a 10 kDa MWCO concentrator (Cytiva Cat no: GE28-9322-96) and stored at -80 °C with 10% glycerol.

Expression and purification of scFv16

The scFv16 encoding DNA was cloned into a pET-42a(+) vector with an in-frame N-terminal 10X-His-MBP tag followed by a TEV-protease cleavage site. The construct was transformed into E. coli Rosetta (DE3) cells for protein expression⁸⁶. An overnight culture was inoculated in 1 L of 2XYT media supplemented with 0.5% glucose and 5 mM MgSO₄. When the culture reached an optical density (O.D₆₀₀) ~ 1.0, protein expression was induced with 250 µM IPTG, and cells were grown at 18 °C for 16-18 h. The harvested cells were resuspended in lysis buffer containing 20 mM HEPES, pH 7.4, 200 mM NaCl, 2 mM benzamidine, and 1 mM PMSF, followed by incubation at 4 °C for 1 h with continuous stirring. The cells were lysed by ultrasonication, and cellular debris was removed by centrifugation at 39,846 x g for 40 min at 4 °C. Protein was enriched on Ni-NTA resins, and non-specifically bound proteins were removed by extensive washing with buffer containing 20 mM HEPES, pH 7.4, 200 mM NaCl, and 50 mM imidazole. The protein was then eluted using the same buffer containing 300 mM imidazole. Further purification was performed using amylose resin (NEB, Cat. no: E8021L), where the Ni-NTA eluate was applied and washed with buffer containing 20 mM HEPES, pH 7.4, and 200 mM NaCl to remove any remaining non-specific proteins. Protein was eluted with 10 mM maltose (prepared in 20 mM HEPES, pH 7.4, 200 mM NaCl), and the His-MBP tag was removed by overnight treatment with TEV protease (20:1, Protein to TEV). Following TEV protease cleavage, the tag-free scFv16 was separated from the cleaved tag by passing the mixture through Ni-NTA resin. The purified protein was then concentrated and subjected to final polishing by size exclusion chromatography using a Hi Load Superdex 200 PG 16/600 column (Cytiva Life Sciences, Cat. no: 17517501). Purified protein was flash-frozen and stored at −80 °C with 10% glycerol.

Reconstitution of the receptor G-protein complexes

Purified receptor was incubated with 1.2 molar excess of Gα_o, Gβ1γ2, and scFv16 at room temperature for 2 h in the presence of 25 mU mL^-1 apyrase (NEB, Cat. no: M0398S) and either 1 µM of mC3a, or 10 µM hTLQP-21, mTLQP-21, EP54, SB290157 or JR14a, for complex formation. The G-protein complex was separated from unbound components by loading on a Superose 6 increase 10/300 GL SEC column and analyzed by SDS-PAGE. Complex fractions were pooled and concentrated to ~10–15 mg mL^-1 using a 100 kDa MWCO concentrator (Cytiva, Cat. no: GE28-9323-19) and stored at −80 °C until further use.

Cryo-EM sample preparation and data collection

The purified ligand-receptor-Gα_o complexes (mTLQP-21-mC3aR-Gα_o, hTLQP-21-mC3aR-Gα_o, mC3a-mC3aR-Gα_o, EP54-mC3aR-Gα_o) were incubated with excess ligand at 4 °C for 1 h before grid freezing. Quantifoil R1.2/1.3 Au 200-mesh grids were glow discharged at 20 mA current for 60 s glow, and 10 s hold using a PELCO easiGlow glow discharge system (Ted Pella). For vitrification, a 2.5 µL sample of the above-mentioned ligand-receptor-Gα_o was applied onto glow-discharged grids and blotted for 5-6 s with blot-force 0 at 4 °C and 100% humidity and plunged into liquid ethane using a Mark IV Vitrobot (Thermo Fisher Scientific). The cryo-EM data for mTLQP-21-mC3aR-Gα_o and hTLQP-21-mC3aR-Gα_o were collected using a TFS Titan Krios microscope operating at 300 kV, equipped with a BioQuantum GIF energy filter (Gatan) with 20 eV slit width, with a K3 direct electron detector (Gatan) operating in super-resolution counting mode at 130,000x magnifications with a pixel size of 0.647 Å (2-fold hardware binning) using EPU data acquisition software (Thermo Fisher Scientific). Movies were recorded with 1.8 s exposure time, a defocus range of -0.8 to -3.4 µm, and a total dose of 40–60e^- Å^-2. The cryo-EM data for the EP54-mC3aR-Gα_o sample were collected on a TFS Glacios microscope operating at 200 kV with a Falcon 4 direct electron detector in counting mode. The movies were recorded at 150,000x magnification (0.92 Å pixel size) with an exposure time of 7.74 s, a defocus range of −0.8 to −3.0 µm, and a total dose of 58.6 e^-Å^-2 using EPU data acquisition software. The cryo-EM data for mC3a-mC3aR-Gα_o were collected on a TFS Glacios (200 kV) microscope operating in Fringe-Free Imaging mode, enabling multiple shots per hole. The single-particle movies for mC3a-mC3aR-Gα_o were collected at 150,000x magnification (0.92 Å pixel size) with an exposure time of 4.66 s, a defocus range of -0.8 to -3.2 µm, and a total dose of 50 e^-Å^-2.

Grids for JR14a-hC3aR-Gα_o, and JR14a-mC3aR-Gα_o, were prepared by dispensing 3 μL of the complex samples (~10-14 mg mL^-1) onto glow-discharged holey grids (Quantifoil Au or Cu 1.2/1.3 300 mesh). The excess sample was blotted off for 4 s with Whatman no. 1 filter paper and plunge-frozen in liquid ethane (−181 °C) using a FEI Vitrobot Mark IV at 100% relative humidity and 4 °C. The grids were clipped and stored in liquid nitrogen for data collection later. Data collection was carried out on a TFS Titan Krios microscope operating at 300 kV using a GIF Quantum energy filter and K3 detector (Gatan) in counting mode. Movies were recorded across a defocus range of -0.8 to -1.8 μm, at a nominal magnification of 105,000x, pixel size of 0.86 Å/pix using EPU software. Movies were fractionated into 50 frames with a total dose of ~55 e^-Å^-2.For the SB290157-hC3aR-Gα_o complex sample, 3 μL of the sample were dispensed onto glow discharged Quantifoil holey carbon grids (Au R1.2/1.3) using a PIB-10 glow-discharge (Vacuum Device Co., Ltd., https://www.shinkuu.co.jp/) and blotted using Ø55/20 mm, Grade 595 (Ted Pella, Inc.) filter paper. The grids were subsequently flash frozen in liquid ethane (-181 °C) using a Vitrobot Mark IV maintained at 100% humidity and 4 °C. Data collection was performed on a 300 kV TFS Titan Krios microscope (G3i) equipped with a K3 direct electron detector (Gatan) and BioQuantum K3 imaging filter. Movies were recorded in counting mode across a defocus range of −0.8 to −1.6 μm at a pixel size of 0.83 Å/px using EPU software (Thermo Fisher Scientific). Movies were fractionated into 48 frames with electron exposures of 47.4 e^- Å^-2.

Data processing

Data processing of all the datasets was performed using cryoSPARC v4.0⁸⁷ unless otherwise stated. Movies were imported and aligned using Patch motion correction (multi), followed by estimation of contrast transfer function using Patch CTF (multi). All the datasets of C3aR complexes were processed using a similar strategy as outlined in Supplementary Figs. 3, 4. Briefly, particles were picked with the blob picker sub-program within the cryoSPARC v4.0 suite, followed by particle extraction with the appropriate box size. The extracted particle stacks were subjected to multiple rounds of 2D classification to discard junk particles. A clean particle stack against the 2D class averages with clear secondary features was used to generate ab-initio maps. Multiple rounds of heterogeneous refinement were performed to clean the particle stacks further using the ab-initio models as reference. The 3D class with features of the GPCR-G protein complex was subjected to non-uniform refinement or local refinement with a mask on the micelle. The resolution of all maps was calculated internally in cryoSPARC v4.0 against the half-sets using the gold standard Fourier shell correlation at 0.143. For the SB290157-hC3aR-Gα_o dataset, receptor-focused local refinement was performed to improve the local resolution of the transmembrane domain and ligand-binding pocket of hC3aR. The corresponding local refinement workflow and FSC curves have now been added to the Supplementary Methods and are presented in the Supplementary Fig. 6C. For the mTLQP-21-mC3aR-Gα_o dataset, 8,573 micrographs with CTF resolution better than 4 Å were selected for further processing. Blob picking produced 2D classes with limited orientations. To remove orientation bias, TOPAZ was used to train on a particle set with particles picked from denoised micrographs using the ResNet8 convolutional neural network model, using a down-sampling factor of 8, and the expected number of particles per micrograph set to 500. The particles picked with TOPAZ were extracted and cleaned through iterative rounds of 2D classification, followed by multiclass ab-initio/hetero classification. Particles corresponding to the best class were subjected to non-uniform refinement, followed by local refinement with a mask on the complex to eliminate noise around the reconstruction. Data collection, processing and model refinement statistics have been included as Supplementary Table 5. Estimation of local resolution was performed with the Blocres sub-program, and map sharpening was performed with either an inbuilt program within the cryoSPARC v4.0 suite or deepEMhancer included in the Cosmic cryo-EM webserver⁸⁸ to improve features for model building.

Model building and refinement

For the mC3aR-Gα_o complex, the starting model of mouse C3aR was generated with Swiss-model⁸⁹ using the active human C3aR structure as the template. For the human C3aR-Gα_o complexes, the coordinates of PDB ID: 8I9L were used as the starting models for model building. Initial coordinates of Gα_o, Gβ1 and Gγ2 were obtained from the cryo-EM structure of EP54-C3aR-Gα_o (PDB: 8I95). These initial models were used to rigidly dock into the sharpened map using Chimera^90,91, followed by flexible fitting of the coordinates using the “all-atom-refine” sub-module in COOT 0.9.6^92–94. This was followed by iterative manual adjustments in COOT 0.9.6 and Phenix 1.20 within the Phenix suite^95,96. The final models showed good statistics, with most of the residues lying in the most favored region of the Ramachandran plot (Supplementary Table 5). All structural figures used in the manuscript have been prepared with UCSF Chimera-1.17 and UCSF ChimeraX-1.9 softwares^90,91.

Data quantification and statistical analysis

All experiments were performed in at least triplicate. Data are presented as mean ± SEM and were analyzed using GraphPad Prism (version 10.0). For dose-response experiments, data were normalized and analyzed using non-linear curve fitting to the log (agonist) versus response (three-parameter) curves. Values were normalized to the appropriate experimental controls, as specified in the respective figure legends. Comparisons between two groups were measured by using two-tailed Weltch’s t-test, differences among three or more groups were examined by one-way or two-way analysis of variance (ANOVA), followed by Dunnett’s multiple comparison test or Šídák’s multiple comparisons test.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

S.M. and A.D. carried out the functional assays in HEK-293 cells with help from D.T. and P.S.; M.K.Y. prepared the complexes for cryo-EM with help from D.T., N.R., N.B., and C.A.D.; R.Y. collected and processed the cryo-EM data on mC3a-mC3aR-Gαo, mTLQP-21-mC3aR-Gαo, hTLQP-21-mC3aR-Gαo, and EP54-mC3aR-Gαo, complexes with help from H.A.K. and H.Z.; K.S. collected the data and solved the structure for SB290157-hC3aR-Gαo complex with F.K.S.; J.N.F., J.M., and J.D.L. helped characterize TLQP-21 response in primary cells with additional experimental support from C.S.C.; R.C. synthesized TLQP-21 peptides with experimental validation by X.X.L.; M.G. and R.B. processed the cryo-EM data on JR14a-hC3aR-Gαo and JR14a-mC3aR-Gαo complexes collected at IIT Kanpur, refined the structures, and carried out the analyses; M.K.Y., A.D., S.M., N.R., M.G., and R.B. prepared the figures with help from D.T. and G.R.; T.M.W., O.N., C.G., and A.K.S. supervised the overall project. All authors contributed to writing and editing the manuscript.

Peer review

Peer review information

Nature Communications thanks Can Cao and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

Research on complement anaphylatoxin receptors in A.K.S.’s laboratory is currently supported by the Indian Council of Medical Research (EMDR/SG/14/2024-01-02127 and EMDR/CAR/13/2024-01-00013) and Department of Science and Technology (DST/TTI/TC/AMR/COE/2023/5). Part of the cryo-EM data was collected at the National cryo-EM Facility at IIT Kanpur established with the support from ANRF/SERB (IPA/2020/000405). A.K.S. is the Sonu Agrawal Memorial Chair Professor. This research was also supported by the US National Institutes of Health grant R01AI182224 to C.G., the National Health and Medical Research Council grant 2009957 to T.M.W, and and the Australian Research Council Discovery project DP220103549 to R.J.C. and T.M.W. This research was also supported by the JSPS KAKENHI grant numbers JP25KJ0866 (K.S.), JP21H05037 (O.N.) and JP23KJ0491 (F.K.S.), Japan Agency for Medical Research and Development (AMED) grant numbers JP26gm7110004 (F.K.S.) and JP223fa627001 (F.K.S.), the Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research [BINDS]) (AMED) grant numbers JP22ama121012 and JP22ama121002 (O.N.)

Data availability

The data supporting this study are presented in the manuscript and supplementary information files. The cryo-EM density maps and structures generated in this study have been deposited in the PDB and EMDB with accession numbers, PDB ID-9KV8 [https://doi.org/10.2210/pdb9KV8/pdb] and EMD-62588 [https://www.ebi.ac.uk/emdb/EMD-62588] for hTLQP-21-mC3aR-Gαo, PDB ID-9KV6 [https://doi.org/10.2210/pdb9KV6/pdb] and EMD-62586 [https://www.ebi.ac.uk/emdb/EMD-62586] for mTLQP-21-mC3aR-Gαo, PDB ID-9KZK [https://doi.org/10.2210/pdb9KZK/pdb] and EMD-62665 [https://www.ebi.ac.uk/emdb/EMD-62665] for mC3a-mC3aR-Gαo,‌ ‌PDB ID-9KUT [https://doi.org/10.2210/pdb9KUT/pdb] and EMD-62580 [https://www.ebi.ac.uk/emdb/EMD-62580] for JR14a-hC3aR-Gαo, PDB ID-9L0H [https://doi.org/10.2210/pdb9L0H/pdb] and EMD-62712 [https://www.ebi.ac.uk/emdb/EMD-62712] for JR14a-mC3aR-Gαo, PDB ID-9KVP [https://doi.org/10.2210/pdb9KVP/pdb] and EMD-62598 [https://www.ebi.ac.uk/emdb/EMD-62598] for SB290157-hC3aR-Gαo-Full map, PDB ID-9KVS [https://doi.org/10.2210/pdb9KVS/pdb] and EMD-62600 [https://www.ebi.ac.uk/emdb/EMD-62600] for SB290157-hC3aR-Gαo ligand-receptor-focused map, and PDB-9UMJ [https://doi.org/10.2210/pdb9UMJ/pdb] and EMD-64276 [https://www.ebi.ac.uk/emdb/EMD-64276] for EP54-mC3aR-Gαo^97–112. Source data are provided with this paper.

Competing interests

The authors declare no competing interests relevant to this work.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76518-z.