Molecular mechanism of naturally-encoded signaling-bias at the complement anaphylatoxin receptors

The conceptual framework of biased signaling has revolutionized our understanding of GPCR signaling and regulatory paradigms, and greatly impacted the efforts focused on the discovery of GPCR-targeted therapeutics. However, the mechanistic basis of biased signaling remains primarily defined based on synthetic ligands and receptor mutants with relatively limited progress in understanding naturally-encoded signaling-bias. Here, we present fundamental molecular and structural insights into naturally-encoded signaling-bias at the complement anaphylatoxin C5a receptors namely, C5aR1 and C5aR2. We first discover that C5a ^-d-Arg , the naturally-occurring version of C5a lacking the terminal arginine, exhibits robust G-protein signaling-bias at C5aR1, characterised by attenuated βarr recruitment. This signaling-bias manifests in both cytokine release from primary human immune cells, and in vivo, during neutrophil mobilization. We combine the cryo-EM structures of C5a/C5a ^-d-Arg -C5aR1 complexes with MD simulation, site-directed mutagenesis, and cellular experiments to elucidate that the G-protein-bias exhibited by C5a ^-d-Arg results from a distinct orientation of TM7 and helix 8 in C5aR1 leading to inefficient GRK recruitment and receptor phosphorylation. Next, we determine the first cryo-EM structures of C5aR2, a naturally-encoded β-arrestin-biased receptor, in an apo state, complexed with the natural agonists C5a and C5a ^-d-Arg , and three peptide agonists including a first-in-class, newly discovered C5aR2-selective agonist, R8Y. These structural snapshots reveal key differences between the binding of C5a and C5a ^{- d-Arg} to C5aR1 and C5aR2, and provide a molecular basis of functional specialization at these two receptors. Moreover, the structural insights also allow us to decipher the molecular basis of naturally-encoded signaling-bias at C5aR2 originating from a shallower cytoplasmic interface with hydrophobic interior pocket that is not permissive to efficient G-protein-coupling and activation. Finally, we also engineer and characterize loss-of-function and gain-of-function variants of C5aR1 and C5aR2, which in turn corroborate and validate the structural observations presented here. Collectively, our findings offer crucial insights into previously lacking molecular mechanisms of the naturally-encoded signaling-bias at GPCRs, which have broad implications not only for the general framework of biased-signaling, but also for novel therapeutic design.