Structural insights into the agonist selectivity of the adenosine A3 receptor

Abstract

Adenosine receptors play pivotal roles in physiological processes. Adenosine A₃ receptor (A₃R), the most recently identified adenosine receptor, is expressed in various tissues, exhibiting important roles in neuron, heart, and immune cells, and is often overexpressed in tumors, highlighting the therapeutic potential of A₃R-selective agents. Recently, we identified RNA-derived N⁶-methyladenosine (m⁶A) as an endogenous agonist for A₃R, suggesting the relationship between RNA-derived modified adenosine and A₃R. Despite extensive studies on the other adenosine receptors, the selectivity mechanism of A₃R, especially for A₃R-selective agonists such as m⁶A and namodenoson, remained elusive. Here, we identify tRNA-derived N⁶-isopentenyl adenosine (i⁶A) as an A₃R-selective ligand via screening of modified nucleosides against the adenosine receptors. Like m⁶A, i⁶A is found in the human body and may be an endogenous A₃R ligand. Our cryo-EM analyses elucidate the A₃R-G_i complexes bound to adenosine, 5'-N-ethylcarboxamidoadenosine (NECA), m⁶A, i⁶A, and namodenoson at overall resolutions of 3.27 Å (adenosine), 2.86 Å (NECA), 3.19 Å (m⁶A), 3.28 Å (i⁶A), and 3.20 Å (namodenoson), suggesting the selectivity and activation mechanism of A₃R. We further conduct structure-guided engineering of m⁶A-insensitive A₃R, which may aid future research targeting m⁶A and A₃R, providing a molecular basis for future drug discovery.

Fig. 1. Screening of modified nucleosides against human A₃R.

a Heatmaps showing the activation of adenosine receptor subtypes by modified nucleosides, as measured by the TGF-α shedding assay. The color scale represents % receptor activation compared to TPA (12-O-tetradecanoylphorbol 13-acetate)-mediated receptor activation, which induces the maximum alkaline phosphatase(AP)-fused TGFα-shedding response independently of the receptor, and the adenosine receptor-dependence of signals were calculated by subtracting the response in mock-transfected cells from the response in adenosine receptor-expressing cells. The tested compounds’ concentrations were 100 nM for hA₁R and hA₃R, 1 µM for hA_2AR, and 5 µM for hA_2BR. Values are shown as an average of three independent experiments. Comparison of AP-TGFα-shedding response curves for hA₃R between modified nucleosides (b: m⁶A, c: i⁶A, d: m^6,6A, and e: t⁶A) and adenosine. Response curves for mock-transfected cells are shown in the same graph. Ligand-induced AP-TGF-α release ratio into conditioned media is quantified. Symbols and error bars represent mean ± SEM, respectively, of 3–6 independent experiments with each performed in triplicate. Source data are provided as a Source Data file.

Fig. 2. Overall structure of the A₃R-G_i complex.

a Structures of the agonists used for structural analyses. Atom numbers for the adenosine moiety are colored blue. Modifications at N⁶, C2, and 5′ positions are highlighted in green, red, and blue, respectively. b Cryo-EM maps of the A₃R-G_i complex bound to adenosine, NECA, m⁶A, i⁶A, and namodenoson. Densities of each agonist are shown in the top-right corner of each map. c Superimposition of each agonist-bound A₃R. d Overall structure of the A₃R-G_i complex bound to NECA. e Superimposition of A₁R (PDB 6D9H), A_2AR (PDB 6GDG), A_2BR (PDB 8HDP), and A₃R. f Close-up view of ECLs of adenosine receptors. ECL2 and ECL3 show remarkable differences among the adenosine receptors. The disulfide bond between ECL2 and TM3 of A₃R is labeled. Receptor-G-protein interactions around α5 (g) and ICL2 (h). The residues involved in the interactions are represented by stick models. Black dashed lines indicate hydrogen bonds.

Fig. 3. Binding modes of adenosine and NECA.

a Superimposition of adenosine-bound A₁R (PDB 6D9H), A_2AR (PDB 2YDO), A_2BR (PDB 8HDP), and A₃R. b Superimposition of NECA-bound A_2AR (PDB 6GDG), A_2BR (PDB 7XY7), A₃R. Binding modes of adenosine (c) and NECA (d). Black dashed lines indicate hydrogen bonds. e Comparison of the interactions around the 5′ tail of NECA. f Comparison of the binding modes of adenosine to each adenosine receptor. Representative residues involved in the ligand-receptor interaction are represented by stick models. Residues of A₃R are labeled. Key interactions are conserved among adenosine receptors. g Alignment of residues comprising the orthosteric pockets of adenosine receptors.

Fig. 4. Binding mode of m⁶A.

a Superimposition of adenosine-bound and m⁶A-bound A₃R. b Binding mode of m⁶A. Black dashed lines indicate hydrogen bonds. c Close-up view of the hydrophobic pocket of the m⁶A-bound model. d CPK models of the hydrophobic pocket of A₃R and those of A₁R, A_2AR, A_2BR, and adenosine-bound A₃R (PDB 6D9H, 2YDO, 8HDP, and this study, respectively). The methyl group of m⁶A is especially tightly packed. e–h Relative RAi values of adenosine, m⁶A, and m^6,6A for human A₃R mutants, as determined by the TGFα-shedding assay. RAi values are expressed as fold change of the values for the WT. The LogRAi cutoff value was set to −2. Mutations that reduce (e) and enlarge (f) the size of the hydrophobic pocket. g Mutations that replace the original residue to a hydrophilic residue. h Designed mutants for selectively reducing the potency of m⁶A. Data are presented as mean values ± SEM from at least three independent experiments performed in triplicate (the n values are represented by circles). Source data are provided as a Source Data file.

Fig. 5. Binding mode of i⁶A.

a Superimposition of adenosine- and i⁶A-bound A₃R. b Binding mode of i⁶A. Black dashed lines indicate hydrogen bonds. c Close-up view of the hydrophobic pocket of the i⁶A-bound structure. d–f Relative RAi values of adenosine and i⁶A for human A₃R mutants, as determined by the TGFα-shedding assay. The LogRAi cutoff value was set to −2. Mutations that reduce (d) and enlarge (e) the size of the hydrophobic pocket. f Mutations that replace the original residue to a hydrophilic residue. Data are presented as mean values ± SEM from at least three independent experiments performed in triplicate (the n values are represented by circles). The values for adenosine used references in (d–f) are the same as those for adenosine in Fig. 4e–g. Source data are provided as a Source Data file.

Fig. 6. Binding mode of namodenoson.

a Superimposition of adenosine- and namodenoson-bound A₃R. b Binding mode of namodenoson. Black dashed lines indicate hydrogen bonds. c Close-up view of the hydrophobic pocket of the namodenoson-bound structure. The 3-iodobenzyl group of namodenoson wedges into the hydrophobic pocket. d Close-up view of the interactions around the chloro group of namodenoson. e Close-up view of the interactions around the N-methylcarboxamide group of namodenoson. f Cross section of the orthosteric pocket of A₃R. Namodenoson fills the orthosteric pocket comprehensively.

Fig. 7. Structural insight into the selective recognition of adenosine receptors.

a–f Binding modes of the adenosine receptors and their selective drugs. a m⁶A-bound A₃R. b Namodenoson-bound A₃R. c DU172-bound A₁R (PDB 5UEN). d ZM241358-bound A_2AR (PDB 3PWH). e BAY60-6583-bound A_2BR (PDB 7XY6). f UK-432097-bound A_2AR (PDB 3QAK). While occupying the orthosteric pocket, each drug exhibits interactions with the extracellular portion of the receptor. These extracellular interactions would be essential for the selectivity among the adenosine receptors.