Crystal structure of adenosine A_2A receptor in complex with clinical candidate Etrumadenant reveals unprecedented antagonist interaction

Adenosine receptors (ARs), G protein-coupled receptors (GPCRs), play crucial roles in transcellular signaling. Four subtypes exist: A₁AR, A_2AAR, A_2BAR, and A₃AR, each with distinct G protein coupling preferences (primarily G_i for A₁AR and A₃AR, G_s for A_2AAR and A_2BAR). Adenosine's immunosuppressive and anti-inflammatory effects are mediated by A_2AAR and A_2BAR. A_2AAR blockade benefits various conditions with increased adenosine-A_2AAR signaling, exemplified by the approved A_2AAR antagonist Istradefylline for Parkinson's disease. Preclinical studies suggest potential benefits in Alzheimer's disease. Recently, A_2AAR and A_2BAR, expressed by immune and cancer cells, emerged as immunotherapy targets (purinergic immune checkpoints). Etrumadenant (AB928), a dual-acting A_2A/A_2B receptor antagonist with a unique poly-substituted 2-amino-4-phenyl-6-triazolylpyrimidine core, is in clinical development for cancer treatment. Despite its clinical advancement, its binding mode is largely unknown. This study aims to elucidate the high-resolution crystal structure of Etrumadenant bound to A_2AAR, providing insights into its binding mechanism and facilitating the development of more selective antagonists.

Extensive research has been conducted on adenosine receptors and their roles in various physiological processes and disease states. The development of selective antagonists has been a key focus, leading to the discovery of drugs like Istradefylline. The purinergic signaling pathways, involving adenosine receptors and other purinergic components, are becoming increasingly recognized as key players in cancer. Studies have shown the involvement of A_2AAR and A_2BAR in tumor immune evasion, thereby motivating the development of antagonists that block these receptors and enhance anti-tumor immunity. Existing literature has highlighted the complexity of interactions within this system, emphasizing the need for high-resolution structural data. Prior studies had limitations in providing a clear understanding of the precise binding interactions of antagonists with A_2AAR, particularly in the absence of key interactions caused by mutations in receptor constructs.

Two high-resolution crystal structures of the A_2AAR in complex with Etrumadenant were determined using two different thermostabilized A_2AAR constructs. The first construct, A_2A-PSB2-BRIL, contained only two point mutations (S91^3.39K and N154^ECL2A), minimizing interference with ligand binding and improving thermostability. The second construct, A_2A-StaR2-bRIL-A277S, was a widely used construct with a T88^3.36A mutation in the binding pocket. The A_2A-PSB2-BRIL construct was created by adding a glycosylation-site mutation to A_2A-PSB1-BRIL. SDS-PAGE analysis confirmed the removal of N-linked glycosylation. Both A_2AAR-Etrumadenant complexes were crystallized using the lipidic cubic phase method. X-ray diffraction data were collected at DESY (A_2A-PSB2-BRIL) and SLS (A_2A-StaR2-bRIL-A277S) synchrotrons. Data processing utilized XDS, autoPROC, CCP4, and other related software. Structure solution and refinement were performed using PHENIX and Coot. In vitro pharmacological characterization was conducted using radioligand binding assays with CHO and Sf9 cell membranes expressing the various AR subtypes. G protein dissociation assays using a BRET-based approach measured the inhibitory effects of Etrumadenant on NECA-induced G protein activation. Sequence alignments of AR subtypes were used to analyze the conservation of amino acids involved in Etrumadenant binding.

The crystal structure of A_2A-PSB2-BRIL-Etrumadenant (PDB ID 8C9W) showed a 2.1 Å resolution and revealed an unprecedented interaction: a hydrogen bond between the cyano group of Etrumadenant and T88^3.36 (N-O distance 2.8 Å). This interaction was not observed in the A_2A-StaR2-bRIL-A277S-Etrumadenant structure (PDB ID 8CIC), which harbors a T88^3.36A mutation. Etrumadenant displayed additional interactions: π-π interactions with H250^6.52 and W246^6.48; contacts with V84^3.32, L85^3.33, and F168^ECL2; interactions with a water network connecting helices II and III; π-π stacking with F168^ECL2 and hydrogen bonds with N253^6.55 and E169^ECL2; π-π interactions with F168^ECL2; and water-mediated hydrogen bonds. The triazolyl ring showed π-π stacking with F168^ECL2 and water-mediated hydrogen bonds. The 2-hydroxyisopropyl residue displayed multiple rotamers. The A_2A-StaR2-bRIL-A277S-Etrumadenant structure, while showing nearly identical binding poses despite the mutation, displayed a 47-fold lower affinity compared to the wild-type A_2AAR (K_i 39.8 nM vs 0.851 nM), while the A_2A-PSB2-BRIL construct maintained unaltered affinity (K_i 1.12 nM). Radioligand binding assays showed high nanomolar affinity for A₁, A_2A, and A_2BARs and potent blockade of G protein activation. The low selectivity (ninefold A_2A vs A₁AR) is explained by high conservation of residues interacting with Etrumadenant across A₁, A_2A, and A_2BAR subtypes. Differences in the A₃AR binding pocket account for reduced affinity. BRET-based G protein dissociation assays confirmed potent antagonism across all four AR subtypes.

The discovery of the hydrogen bond between Etrumadenant's cyano group and T88^3.36 is a significant finding, highlighting the importance of using wild-type receptor constructs for structure-activity relationship studies. The 47-fold decrease in affinity for the A_2A-StaR2-bRIL construct underscores the impact of the T88^3.36A mutation, which is present in the majority of previously reported A_2AAR crystal structures. This mutation prevents the observation of key interactions such as the hydrogen bond. The high conservation of amino acid residues involved in Etrumadenant binding across the A₁, A_2A, and A_2BAR subtypes explains the lack of subtype selectivity. This lack of selectivity might be a limiting factor in its therapeutic application in cancer therapy, where A₁AR blockade might counteract the beneficial effects of A_2AAR/A_2BAR antagonism. The study provides a structural basis for designing improved, more selective AR antagonists for cancer and neurodegenerative disease therapies.

This study provides the first high-resolution crystal structures of the adenosine receptor antagonist Etrumadenant bound to the A_2AAR. The key finding is the identification of an unprecedented interaction, a hydrogen bond between Etrumadenant and T88^3.36. This highlights the importance of using wild-type or minimally modified receptor constructs in structural studies of drug binding. The lack of selectivity observed for Etrumadenant, particularly against A₁AR, provides valuable information for future drug design. Future studies could focus on modifying the structure of Etrumadenant to improve its selectivity while maintaining its potent antagonist activity.

The study focuses primarily on the A_2AAR and did not comprehensively investigate the binding of Etrumadenant to A_2BAR, limiting our ability to establish a direct comparison between the interaction mechanisms at both receptor subtypes. Functional assays were conducted using overexpressed receptors, limiting the direct translation of results to physiological conditions. The limited number of AR subtypes investigated (A1, A2A, A2B, A3) does not provide a full picture of the overall pharmacological profile. Furthermore, the crystal structures obtained were of the inactive receptor conformations. Further studies are needed to determine the binding of Etrumadenant to the active receptor conformation to fully elucidate the functional effects.