Molecular basis of anaphylatoxin binding, activation, and signaling bias at complement receptors

Abstract

The complement system is a critical part of our innate immune response, and the terminal products of this cascade, anaphylatoxins C3a and C5a, exert their physiological and pathophysiological responses primarily via two GPCRs, C3aR and C5aR1. However, the molecular mechanism of ligand recognition, activation, and signaling bias of these receptors remains mostly elusive. Here, we present nine cryo-EM structures of C3aR and C5aR1 activated by their natural and synthetic agonists, which reveal distinct binding pocket topologies of complement anaphylatoxins and provide key insights into receptor activation and transducer coupling. We also uncover the structural basis of a naturally occurring mechanism to dampen the inflammatory response of C5a via proteolytic cleavage of the terminal arginine and the G-protein signaling bias elicited by a peptide agonist of C3aR identified here. In summary, our study elucidates the innerworkings of the complement anaphylatoxin receptors and should facilitate structure-guided drug discovery to target these receptors in a spectrum of disorders.

Keywords: GPCRs; anaphylatoxins; arrestin-coupled receptors; beta-arrestins; cellular signaling; complement cascade; complement receptors; cryo-EM; drug discovery; signaling bias.

Graphical abstract

Figure 1. Activation of complement receptors and downstream functional outcomes

(A) Triggering of the various complement pathways leads to the generation of complement peptides and subsequent activation of cognate complement receptors. An overview of the activation of C3aR, C5aR1, and C5aR2 and their signaling has been illustrated. (B) To study Gαi activation, forskolin-elevated decrease in cAMP level is measured using GloSensor assay downstream of C3aR in response to indicated ligands (mean ± SEM; n = 4; normalized with starting value for each ligand as 100%). (C and D) βarr1/2 recruitment to C3aR in response to indicated ligands as measured by NanoBiT assay (receptor-SmBiT + LgBiT-βarr1/2), respectively (mean ± SEM; n = 4; normalized with the luminescence signal at maximal ligand dose of C3a as 100%). (E and F) βarr1/2 trafficking to the endosomes downstream of C3aR in response to indicated ligands as measured by NanoBiT assay (receptor + SmBiT-βarr1/2 + LgBiT-FYVE) (mean ± SEM; n = 4; normalized with the luminescence signal at minimal ligand dose of each condition as 1). (G and H) C5a (top) and C5a^pep (bottom) driven Gαi-mediated second messenger response as measured by agonist-dependent decrease in forskolin-induced cytosolic cAMP levels downstream to C5aR1. Respective logEC50 values are mentioned in the inset. Data (mean ± SEM) represent four independent experiments, normalized with respect to the highest signal (measured as 100%) for each receptor. (I–P) C5a/C5a^pep induced βarr1/2 recruitment and trafficking as measured by NanoBiT assay. Respective logEC50 values are mentioned in the inset. Data (mean ± SEM) represent four independent experiments, fold normalized with respect to luminescence observed at the lowest dose (measured as 1) for each receptor. (Q and R) Comparison of C5a/C5a^pep-mediated cAMP response downstream of human (top) and mouse (bottom) C5aR1 reveals reduced potency of C5a^pep as compared with C5a. Respective logEC50 values are mentioned in the inset. Data (mean ± SEM) represent four independent experiments, normalized with respect to the highest signal (measured as 100%) in response to each ligand. (S–Z) Measurement of βarr1/2 recruitment and trafficking to human (top) and mouse (bottom) C5aR1 upon stimulation with C5a and C5a^pep. Respective logEC50 values are mentioned in the inset. Data (mean ± SEM) represent four independent experiments, fold normalized with respect to luminescence observed at the lowest dose (measured as 1) for each ligand. Bias factor (β value) determined by taking C5a as reference elucidates the G-protein-biased nature of C5a^pep that has been provided in insets. Plots explaining the functional bias (Q–Z) of C5a^pep are from the data shown in (G)–(P) and presented separately to highlight the effects compared with C5a in both human and mouse C5aR1. See also Figure S1.

Figure 2. Structures of complement receptor signaling complexes

(A–G) Cryo-EM density maps (top) and corresponding models (bottom) of Apo-C3aR-Go (Glacios), C3a-C3aR-Go, EP54-C3aR-Go, EP54-C3aR-Gq, C5a-hC5aR1-Go, C5a-mC5aR1-Go, and C5a^pep-mC5aR1-Go. Cryo-EM density maps of respective ligands have been shown in gray dotted circles (top corner). (Rosy brown, C3aR; green, C3a; pale blue, EP54; light gray, hC5aR1; slate gray and blue, mC5aR1; cyan, C5a; yellow, C5a^pep; teal, Gαo; deep blue, Gαq; gold, Gβ1; purple, Gγ2; dark gray, ScFv16.) See also Table S1 and Methods S1.

Figure 3. Complement peptide binding to complement receptors

(A) Structure of C3a (top) and C5a (middle), showing four-helix bundle with a short C-terminal tail. Free C3a (PDB: 4HW5) and free C5a (PDB: 1KJS) have been superimposed with C3a and C5a, respectively. Structural alignment of C3a and C5a in receptor-bound forms (bottom). (B) Side view of C3aR (top), C5aR1 (middle), and both (bottom) bound to their endogenous ligands. (C) The C termini of the ligands change their conformation upon binding to their corresponding receptors compared with the basal states. (D) The C-terminal tails of C3a (top) and C5a (middle) adopt a hook-like conformation upon entering deep into the orthosteric pocket of respective receptors. C3aR and C5aR1 have been shown in surface slice, and C3a/C5a as cartoon representation. (E) The distal C-terminal portions of C3a/C5a make extensive contacts with C3aR/C5aR1. Residue at the interface between the H4 and C termini of C3a/C5a and C3aR/C5aR1 is depicted. (F) C3a and C5a are shown in topology diagrams. Residues making contact with receptors are highlighted in yellow circles. See also Figures S1 and S2 and Table S2.

Figure 4. Structural determinants of reduced functional efficacy of C3a^des-Arg and C5a^des-Arg

(A) Proteolytic cleavage of the C-terminal Arg⁷⁷ from C3a and Arg⁷⁴ from C5a results in C3a^des-Arg and C5a^des-Arg, respectively (top). C3a^des-Arg is reported to elicit almost no functional outcomes upon binding to C3aR, whereas C5a^des-Arg retains minimal potency in driving downstream signaling via C5aR1 (bottom). (B) Interactions of Arg⁷⁷ in C3a (top) and Arg⁷⁴ in C5a with the residues of C3aR and C5aR1 have been illustrated. (C) Cryo-EM density map (left) and corresponding model (right) of C5a^des-Arg-hC5aR1-Go. Cryo-EM density map of C5a^des-Arg has been shown in gray dotted circles. (D) Structural superimposition of C5a- and C5a^des-Arg-bound C5aR1 highlighting a similar binding pose. The hook-like C termini of both the ligands are shown within the ligand-binding pocket. (E) The sub-pocket in C5aR1 occupied by the guanidinium side chain of Arg⁷⁴ is empty in case of C5a^des-Arg. (F) Conformational switching of Leu⁷² and Gly⁷³ located toward the distal end of C5a^des-Arg is shown. (G) Residue contacts by Arg⁷⁴ of C5a in C5aR1 have been shown in circles. Interactions of Gln⁷¹, Leu⁷², and Gly⁷³ compensate for the missing terminal Arg in C5a^des-Arg are shown as green circles. Residues highlighted in red do not make any interactions in C5a^des-Arg-bound C5aR1 structure. See also Figures S1 and S7, Table S2, and Methods S1.

Figure 5. Binding poses of C3a- and C5a-derived peptides on C3aR and C5aR1 and species-specific insights into ligand binding of C5aR1

(A) Sequence of EP54 derived from the C terminus of C5a (top right). Side view of EP54 (transparent surface) binding to C3aR (ribbon) (left). EP54 docks into C3aR and forms a hook-like structure (bottom right). (B) Sequence of C5a^pep derived from the C terminus of C5a (top right). Side view of C5a^pep (transparent surface) binding to C5aR1 (ribbon) (left). C5a^pep docks into C5aR1 and forms a hook-like structure (bottom right). (C and D) Comparative analysis of common and specific interactions of C3a/EP54 with C3aR and C5a/C5a^pep with C5aR1. (E) Superimposition of agonist-C5aR1 structures. C5a and C5a^pep in the ligand-binding pocket are shown. (hC5aR1 in surface slice and ligands in cartoon representation.) (F) Helical shifts in the core domain of C5a upon binding to mouse and human C5aR1 are shown (left). Conformational changes in the N termini of mouse and human C5aR1 upon interaction with C5a (right). (G) Schematic representation of residue contacts between C5a and C5a^pep with mC5aR1. Residues present at the orthosteric pocket of ligand binding are depicted. Residues from C5a-hC5aR1 structure shown in red boxes, C5a-mC5aR1 structure in dark gray, C5a^pep-mC5aR1 in blue, and common residue contacts at the interface have been shown in yellow boxes. Different regions from the receptors have been highlighted in green boxes. (H) Measuring βarr1/2 recruitment in response to C5a^pep downstream to a series of mC5aR1 mutants mimicking the corresponding hC5aR1 residues show dramatic increase in both potency and efficacy of βarr1/2 recruitment compared with the wild-type mouse receptor. Data (mean ± SEM) represent six independent experiments, fold normalized with respect to luminescence observed at the lowest dose (measured as 1) for each receptor. See also Figures S1 and S3 and Table S2.

Figure 6. Mechanism of activation and G-protein coupling at C3aR and C5aR1

(A and B) Dynamic changes in TMs of activated C3aR/C5aR1 compared with the inactive state of C5aR1. The TM6, TM7, and H8 from different receptor complexes shown are from receptors mentioned in boxes. Solid lines (active receptors) and dotted lines (inactive C5aR1) indicate direction of movement. The respective degrees of movements in corresponding TMs have also been mentioned. (C and D) Close-up views of the conserved DRY, NPxxY, CWxP, and PIF motifs show conformational changes upon C3aR and C5aR1 activation. The names of the hallmark microswitches are noted inside respective boxes. Polar contacts are depicted as black dashed lines. (E and F) α5 helix of Gαo docks into the cytoplasmic core of C3aR and C5aR1. Only receptor and Gαo are shown in ribbon representations to highlight the binding pose of G proteins with receptor core. Surface slice presentations (top) and cryo-EM maps are shown in inset boxes to highlight the direct docking of Gα to receptors. (G and H) Magnified view of the interactions between TM2, TM3, TM6, TM7, ICL2, and ICL3 of C3aR and hC5aR1 with Gαo. Ionic bonds are depicted as black dashed lines. See also Figures S4, S5, S6, and S7 and Table S2.

Figure 7. Identification of a biased agonist and schematic of complement recognition by C3aR and C5aR1

(A) G-protein activation and βarr1/2 recruitment were studied using GloSensor assay and NanoBiT-based assay (receptor-SmBiT + LgBiT-βarr1/2), respectively, first panel: forskolin-induced cAMP level downstream of C3aR in response to indicated ligands (mean ± SEM; n = 4; normalized with the lowest ligand concentration for each ligand as 100%); second panel: βarr1 recruitment to C3aR (mean ± SEM; n = 4); and third panel: βarr2 recruitment to C3aR (mean ± SEM; n = 3); normalized with the highest ligand concentration of C3a as 100% (top). G-protein activation and βarr1/2 recruitment downstream of C5aR1 in response to indicated ligands, first panel: forskolin-induced cAMP level decrease downstream of C5aR1 in response to indicated ligands (mean ± SEM; n = 5; normalized with the lowest concentration of each ligand as 100%), second panel: βarr1 recruitment to C5aR1 (mean ± SEM; n = 5) and third panel: βarr2 recruitment to C5aR1 (mean ± SEM; n = 4); normalized with the highest ligand concentration of C5a as 100% (bottom). Bias factors (β value) determined taking C3a as reference elucidate the G-protein-biased nature of EP141 have been shown in insets. Bias factor was calculated using https://biasedcalculator.shinyapps.io/calc/. (B) Intracellular calcium response, ERK1/2 activation, cytokine release upon stimulation with C3a, and EP141 were studied in human macrophages. Top left: calcium response (mean ± SEM; n = 9, data were normalized to the bottom [0%] and top [100%] values of the C3a dose-response curve). Bottom left: ERK1/2 activation (mean ± SEM; n = 9, normalized to the peak C3a-induced response for that donor). For cytokine release, top right: TNF-α release (mean ± SEM; n = 4), and bottom right: IL-6 release (mean ± SEM; n = 5), data were normalized to the medium only (0%) and LPS (100%) triggered response, analyzed using two-way ANOVA, Tukey’s multiple comparison test. The exact p values are as follows: for TNFα release, LPS vs. LPS+C3a: (p = 0.0066), LPS only vs. LPS+EP141: (p = 0.3947). For IL6 release, LPS vs. LPS+C3a: (p = 0.0002), LPS vs. LPS+EP141 (p = 0.1048) (**p < 0.01, ***p < 0.001, ns, non-significant). (C) Cryo-EM density map and corresponding model of EP141-C3aR-Go. Cryo-EM density map of EP141 has been shown in gray dotted circles (left). Sequence of EP141 derived from the C terminus of C3a. Side view of EP141 (surface) bound to C3aR (ribbon). (D) The ligand binding pocket has been shown as surface slice representation to highlight the positioning of C3a, EP54, and EP141 on C3aR (right). (E) Unique residue contacts between EP141 and C3aR at the ligand binding pocket might help explain the phenotypic behavior exhibited by EP141. Arg⁶⁹ of EP141 makes extensive interactions with residues from TM6 and ECL3 (left), which are absent in C3a-C3aR (middle) and EP54-C3aR complexes (right). Black dotted arrows represent probable movements of Arg⁶⁹ in C3a and Lys in EP54 with respect to Arg⁶⁹ in EP141. (F) Interactions by Arg⁶⁹ of EP141 with C3aR result in outward movements in TM6, TM7, and ECL3 in C3aR. (G) Schematic diagram showing ligand binding of the complement receptors C3aR and C5aR1. The C-terminal tail of C3a/C5a changes its conformation to form a hook-like loop upon interacting with C3aR/C5aR1. In addition to the ECLs and the extracellular side of TMs of both the receptors, C5a engages with an extra interface on N terminus of C5aR1. Absence of this interaction network on N terminus might explain the differential positioning of C3a compared with C5a. A group of charged residues on C3aR/C5aR1 interact and stabilize the critical terminal Arg with polar contacts from the C terminus of complement peptides. C5a also binds to C5aR2, but the mechanism behind this interaction has yet to be explored in high detail. See also Figures S1 and S7, Table S2, and Methods S1.