Characterizing CCR2 homodimers using homology-derived and AI-predicted interfaces

Deciphering the functional roles of G protein-coupled receptors (GPCRs) oligomers requires defining their structural organization, which remains a challenge due to the inherent flexibility of their 7 transmembrane (TM) domains. Here, we focused on determining homodimer interfaces of the chemokine receptor CCR2, a major component of the inflammatory response, for which no antagonist has yet been approved. We generated models of CCR2 homodimers combining homology modeling, a conventional in silico approach, or an AI-based AlphaFold-Multimer predictions, with molecular dynamics simulations. We determined the biological relevance of our predicted models by in cellulo cross-linking experiments, introducing a cysteine in each predicted interface. This in cellulo assay supported the homology modeling predictions, but not those predicted by AlphaFold-Multimer, and demonstrated that CCR2 forms homodimers by at least two different interfaces involving the transmembrane domains TM5 (I₅) or TM5 and TM6 (I₅₆). This structural arrangement of the CCR2 homodimer represents a valuable model for further studies on the functional consequences of dimerization and for the rational design of CCR2-selective drugs.