Crystal structure of adenosine A2A receptor in complex with clinical candidate Etrumadenant reveals novel antagonist interaction

doi:10.21203/rs.3.rs-2602909/v1

Download PDF

Article

Crystal structure of adenosine A2A receptor in complex with clinical candidate Etrumadenant reveals novel antagonist interaction

https://doi.org/10.21203/rs.3.rs-2602909/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 01 Jun, 2023

Read the published version in Communications Chemistry →

Version 1

posted

You are reading this latest preprint version

The G_s-protein-coupled adenosine A_2A receptor (A_2AAR) represents an emerging drug target for cancer immunotherapy. The clinical candidate Etrumadenant was developed as an A_2AAR antagonist with ancillary blockade of the A_2BAR subtype. It constitutes a novel chemotype featuring a poly-substituted 2-amino-4-phenyl-6-triazolylpyrimidine core structure. Herein, we report two crystal structures of the A_2AAR in complex with Etrumadenant, obtained with differently thermostabilized A_2AAR constructs. This led to the discovery of an unprecedented interaction, a hydrogen bond of T88^3.36 with the cyano group of Etrumadenant. T88^3.36 is mutated in most A_2AAR constructs used for crystallization, which has prevented the discovery of its interactions. In-vitro characterization of Etrumadenant indicated low selectivity versus the A₁AR subtype, which can be rationalized by the structural data. These results will facilitate the future design of AR antagonists with desired selectivity. Moreover, they highlight the advantages of the new A_2AAR crystallization construct devoid of ligand binding site mutations.

Biological sciences/Structural biology/X-ray crystallography

Biological sciences/Drug discovery/Medicinal chemistry

G protein-coupled receptors (GPCRs) activated by the nucleoside adenosine are widely distributed and play important roles in transcellular signaling.^1,2 Four subtypes of adenosine receptors (ARs) exist, the preferentially G_i protein-coupled A₁- and A₃ARs, and the G_scoupled A_2A- and A_2BARs.³ Coupling to additional G proteins has been described, e.g., to G_q proteins.^4–6 Adenosine acts as a “stop signal” resulting in strong anti-inflammatory and immunosuppressive effects, mediated by the A_2A- and A_2BAR subtypes.⁷ Blockade of the A_2AAR is beneficial for several pathological conditions, in which adenosine-A_2AAR signaling is increased.^8,9 For example, the A_2AAR antagonist Istradefylline has been approved in Japan and the USA for the treatment of Parkinson’s Disease.¹⁰ Preclinical studies suggest major effects of A_2AAR antagonists against Alzheimer’s Disease.^11,12 The A_2AAR, and later on also the related A_2BAR, both of which are expressed by immune cells and may be upregulated in cancer cells, have recently emerged as novel targets for the immunotherapy of cancer, constituting purinergic immune checkpoints.¹³

Etrumadenant (3-[2-amino-6-[1-[[6-(2-hydroxypropan-2-yl)pyridin-2-yl]methyl]− 4-yl]pyrimidin-4-yl]-2-methylbenzonitrile, also known as AB-928) was developed as one of the first dual-acting adenosine A_2A/A_2B receptor antagonists.¹⁴ It constitutes a novel chemotype featuring a poly-substituted 2-amino-4-phenyl-6-triazolylpyrimidine core structure. The drug has entered clinical development and has been evaluated in several clinical phase I and phase II trials for the treatment of cancer.¹⁵ Despite its advanced stage in drug development, the characterization of Etrumadenant is limited, and the exact drug-receptor binding mode is unknown.

Here, we determined the high-resolution crystal structure of Etrumadenant in complex with a new thermostabilized A_2AAR construct comprising only two point mutations that do not interfere with ligand binding. The structure reveals unique binding pocket interactions of Etrumadenant including a novel interaction of its cyano group with T88^3.36. For comparison, we also determined the high-resolution crystal structure of Etrumadenant in complex with a widely used A_2AAR construct that contains a T88^3.36A mutation in the binding pocket (A_2A-StaR2-bRIL-A277S). The structural findings were complemented with an in-vitro pharmacological characterization of Etrumadenant at all AR subtypes. The compound was found to display high affinity in the low nanomolar range for A₁-, A_2A-, and A_2BARs, and it potently blocked G protein activation by these subtypes. Structural data provided an explanation for the compound’s lack of selectivity.

Exploring the A_2AAR binding pocket of Etrumadenant using optimized crystallization constructs

We previously developed an optimized A_2AAR crystallization construct designated A_2A-PSB1-bRIL, that contains a single point mutation (S91^3.39K) inside the allosteric sodium binding pocket to stabilize the inactive conformation which significantly enhanced protein thermostability.¹⁶ For the co-crystallization of Etrumadenant, we used the same modification but inserted an additional point mutation (N154^ECL2A) to remove a putative glycosylation site on extracellular loop (ECL) 2 of the receptor. This new construct is designated A_2A-PSB2-bRIL (PSB: Pharmaceutical Sciences Bonn, bRIL refers to thermostabilized apocytochrome b₅₆₂RIL¹⁷). Mutation of the asparagine in position 154 to either alanine or glutamine had previously been utilized to eliminate post-translational N-linked glycosylation of the A_2AAR, as protein glycosylation is expected to inhibit crystal growth due to microheterogeneity.^18,19 Evidence of N-linked glycosylation is missing, and N154^ECL2 is not surface-exposed in available A_2AAR crystal structures,²⁰ indicating that the non-glycosylated form of the A_2AAR crystallizes predominantly. Here, we additionally employed sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) to demonstrate that A_2A-PSB1-bRIL (bearing the wild type (wt) N154^ECL2) is still partially glycosylated, whereas A_2A-PSB2-bRIL (bearing an N154^ECL2A mutation) had lost N-linked glycosylation (Figure 1). Glycosylated proteins typically migrate more slowly in SDS-PAGE and generate higher molecular weight smearing.²¹ Despite the fact that only a single glycosylation site is present, smearing could be observed for A_2A-PSB1-bRIL, whereas a sharper band was detected for the N154^ECL2A mutant (A_2A-PSB2-bRIL) as well as for N154^ECL2Q and N154^ECL2D mutants, indicating the loss of glycosylation (Figure 1). Alternatively, the glycosyl residues of A_2A-PSB1-bRIL could be cleaved off by the enzyme peptide-N-glycosidase F (PNGase F)²² when added to the purified protein. Glycosylation could also be prevented during receptor expression by addition of the glycosylation inhibitor Tunicamycin.²³

Utilizing the optimized construct A_2A-PSB2-bRIL, we obtained the crystal structure of the A_2AAR in complex with Etrumadenant at 2.1 Å resolution (see Table 1 for detailed data collection and refinement statistics). Etrumadenant was well resolved within the orthosteric ligand binding pocket (Figure 2a and b). Its novel scaffold shows unique interactions within the A_2AAR’s orthosteric binding pocket. Importantly, the cyano group forms a direct hydrogen bond to T88^3.36 (N-O distance 2.8 Å) (Figure 2a and b) representing a novel interaction that has so far not been observed in A_2AAR co-crystal structures with various antagonists. T88^3.36 is conserved within the AR family and was shown to be directly involved in A_2AAR agonist binding (illustrated for 5’‑N‑ethylcarboxamideadenosine (NECA) in Figure 2c).²⁴ It undergoes significant conformational changes during receptor activation.²⁴ The interaction of Etrumadenant with T88^3.36 by direct hydrogen bonding represents a novel interaction to stabilize the A_2AAR in its inactive state. Notably, the A_2A-StaR2-bRIL construct that is extensively used to determine inactive state A_2AAR crystal structures¹⁶ harbors a T88^3.36A mutation (see Figure 3a), that can be expected to affect the affinity of Etrumadenant and possibly other antagonists. In fact, the affinity of Etrumadenant is ~47-fold lower for A_2A-StaR2-bRIL as compared to the wt A_2AAR (K_i values of 39.8 nm compared to 0.85 nm, see Table 2). In contrast, the affinity of Etrumadenant for our optimized crystallization construct A_2A-PSB2-bRIL remained unaltered (K_i 1.12 nm).

Besides the hydrogen bond to T88^3.36, Etrumadenant shows multiple additional receptor-ligand interactions. The phenyl ring of Etrumadenant is stabilized by π‑π interactions to H250^6.52 (T-shaped) and W246^6.48 (stacked) (Figure 2b). Its 2-methyl group comes in contact to V84^3.32, L85^3.33 and F168^ECL2. It is additionally exposed to a water network connecting the ligand to helices II and III (Figure 2a). The 2‑aminopyrimidine core is stabilized by π-π stacking interactions to F168^ECL2 (Figure 2a) and forms key anchoring interactions by hydrogen bonding of the N3 and the exocyclic NH₂-group to N253^6.55 (Figure 2a and b). Hydrogen bonding interactions of the side-chain of N253 are also observed for other ligands including agonists (Figure 2c) and antagonists (see blue rectangles in Figure 3). The hydrogen bond network is extended by a direct interaction of the exocyclic NH₂-group of Etrumadenant with E169^ECL2 (Figure 2b). E169^ECL2 forms a salt bridge to H264^ECL3 that is frequently observed in A_2AAR crystal structures, but was found to be dependent on the structure of the antagonist and the pH value during crystallization.¹⁶

Table 1. Data collection and refinement statistics. Statistics for the highest resolution shell are shown in parentheses.

	A_2A-PSB2-bRIL-Etrumadenant (PDB ID 8C9W)	A_2A-StaR2-bRIL-A277S-Etrumadenant (PDB ID 8CIC)
Data collection
Number of crystals used	1	1
Wavelength (Å)	0.97625	0.99997
Space group	C222₁	C222₁
Cell parameters a, b, c (Å)	39.16, 178.23, 139.60	39.36, 179.09, 140.57
Number of reflections processed	284104	391306
Number of unique reflections	21413	29632
Resolution (Å)	38.24 - 2.114 (2.316 - 2.114)	46.86 - 2.10 (2.16 - 2.10)
Max. Resolution aniso. (Å)	2.16 (a), 2.11 (b), 2.50 (c*)	not applied
R_merge	0.130 (1.697)	0.172 (2.177)
CC_1/2	0.9992 (0.5288)	0.999 (0.499)
Mean I/σ (I)	13.054 (1.299)	11.1 (1.2)
Completeness spherical	0.748 (0.159)	1.000 (0.996)
Completeness ellipsoidal	0.8980 (0.4125)	not applicable
Redundancy	13.27 (11.08)	13.2 (13.4)
Refinement
Resolution (Å)	38.24 - 2.11 (2.23 - 2.11)	44.78 - 2.10 (2.14 - 2.10)
Number of reflections (test set)	21405 (994)	55857 (2840)
R_work	0.1965 (0.3298)	0.1904 (0.3081)
R_free	0.2665 (0.5429)	0.2144 (0.3764)
Number of non-hydrogen atoms
A_2AAR	2369	2349
bRIL	689	705
Ligand	32	32
Lipids, polyethylene glycol (PEG) and waters	284	604
Overall B values (Å²)
A_2AAR	40.5	41.8
bRIL	70.8	76.7
Ligand	29.6	32.2
Lipids, PEG and waters	48.3	64.4
Root-mean-square deviation (RMSD)
Bond lengths (Å)	0.012	0.003
Bond angles (°)	1.305	0.55
Ramachandran plot statistics
Favored regions (%)	97.65	99.48
Allowed regions (%)	2.09	0.52
Disallowed regions (%)	0.26	0

The triazolyl ring of Etrumadenant (Figure 2d), connected to the 6-position of the core aminopyrimidine, and bearing a substituted pyridylmethylene residue, forms π-π stacking interactions with F168^ECL2 and water-mediated hydrogen bonding to H278^7.43 and to the backbones of A59^2.57, I80^3.28 and A81^3.29 (Figure 2a). The pyridine ring is located in close proximity to the entrance of the orthosteric ligand binding pocket at the extracellular ends of helices I and II with direct contacts to S67^2.65 and Y271^7.36. The 2-hydroxyisopropyl residue that is attached to the pyridine of Etrumadenant shows three ambiguous rotamers. We chose to model the rotamer conformation with the hydroxy group in close proximity to a nearby water molecule thereby forming an intramolecular water-mediated hydrogen bond to the pyrimidine N1-nitrogen (Figure 2a and b).

The sidechain of Y271^7.36 was observed to be highly flexible when comparing different A_2AAR co-crystal structures.^19,25 It adapts the hydrophobic pocket to the size of the ligand (as depicted for a selection of ligands in Figure 3). The relatively large Etrumadenant molecule requires a significant sidechain movement of Y271^7.36. This sidechain is located much closer to the orthosteric binding pocket in the ZM241385-bound A_2AAR crystal structure, where it is hydrogen-bonded to the water network around the ligand (Figure 3).²⁰ In that structure, an additional oleic acid molecule occupies the space which Y271^7.36 adopts in the current Etrumadenant structure, where the hydrophilic head group of the oleate is displaced by the rotation of Y271^7.36 (also compare structures of Imaradenant²⁵ and Vipadenant²⁶) (Figure 3).

Next, we additionally obtained the crystal structure of Etrumadenant in complex with a modified A_2A-StaR2-bRIL receptor construct in which the S277^7.42A mutation had been reverted to wt (designated A_2A-StaR2-bRIL-A277S), but which still harbored the T88^3.36A mutation in the binding pocket. A co-crystal structure could be obtained at the same high resolution of 2.1 Å (see Table 2 for detailed refinement statistics). Surprisingly, even though a major interaction partner of the ligand was mutated, the binding pockets of A_2A-PSB2-bRIL-Etrumadenant and A_2A-StaR2-bRIL-A277S-Etrumadenant are largely similar with only subtle differences (Figure 4). Notably, the cyano group of Etrumadenant in the A_2A-PSB2-bRIL structure is slightly tilted, relative to the plane of the phenyl ring, towards the hydroxy group of T88^3.36 and deviates from the ideal planar orientation by ~8° (Figure 4b). The same cyano moiety is planar in the T88^3.36A mutated structure, but is unable to form the same hydrogen bond interaction due to the mutation. Another difference between both structures can be identified in the rotamers of the 2‑hydroxyisopropyl residue and the adjacent sidechain of Y271^7.36 (Figure 4a) which confirms the initially observed flexibility of these moieties.

Pharmacological characterization of Etrumadenant

In the original patent describing Etrumadenant, affinity ranges were reported, but no specific K_i or half-maximal inhibitory concentration (IC₅₀)values were provided.¹⁴ In order to complement the pharmacological characterization of Etrumadenant, we determined its affinities for all human AR subtypes as well as for the crystallization constructs by radioligand binding assays (Table 2). To this end, we employed membrane preparations of Chinese hamster ovary (CHO) cells or Spodoptera frugiperda (Sf9) insect cells recombinantly expressing the respective AR subtype, or crystallization construct, respectively. Additionally, we investigated the inhibitory effects of Etrumadenant in G protein dissociation assays (Figure 5).

In addition to its high affinity for the A_2A- and A_2BAR subtypes (K_i values: A_2A, 0.851 nm; A_2B, 3.16 nm), Etrumadenant was found to exhibit high affinity for the A₁AR (K_i value: 7.59 nm versus the antagonist radioligand [³H]DPCPX, and 7.08 nm versus the agonist radioligand [³H]CCPA). Thus, the compound showed only about 9-fold selectivity comparing A_2A- with A₁AR affinity, and only 2-fold selectivity for the A_2B- versus A₁AR subtype. In contrast, Etrumadenant exhibited high selectivity versus the A₃AR (>100-fold), as determined in radioligand binding studies. For comparison, we determined the affinities of standard AR antagonists using the same assays (Table 2). While ZM241385 showed a moderate preference for the A_2AAR (K_i values: A_2A, 2.04 nm; A_2B, 29.5 nm; 12-fold difference), the A_2AAR antagonist Preladenant displayed similarly high A_2A affinity as Etrumadenant (K_i: A_2A, 0.884 nm ²⁷ ) but showed high A_2A-selectivity. The A_2BAR antagonist PSB-603 was somewhat more potent than Etrumadenant (K_i: A_2B, 0.553 nm ²⁸ ) showing high selectivity for the A_2BAR subtype.

Subsequently, functional assays were performed to determine concentration-dependent antagonistic effects of Etrumadenant on receptor activation. To this end, we performed bioluminescence resonance energy transfer (BRET) based G protein dissociation assays employing Renilla Luciferase 8 (RLuc8) fused to Gα subunits and green fluorescent protein (GFP) attached to the Gγ subunit.^29–31 AR activation was induced with the non-selective agonist NECA at a concentration where it shows 80% of its maximal effect (EC₈₀). The preferentially G_i protein-coupled A₁- and A₃AR subtypes were co-expressed with Gα_i1-RLuc8, Gβ₃, and Gγ₉-GFP proteins, whereas the Gs protein-coupled A_2A- and A_2BAR subtypes were co-expressed with Gα_s-RLuc8, Gβ₃, and Gγ₉-GFP proteins. Etrumadenant was able to block the activation of all four AR subtypes in a concentration-dependent manner. The antagonist was found to be most potent at the A_2BAR followed by the A_2AAR, but also showed significant antagonistic activity at the other AR subtypes, A₁ and A₃ (see Figure 5). Blockade of the G_i protein-coupled A₁- and A₃ARs will lead to an increase in intracellular cyclic adenosine monophosphate (cAMP) levels thereby counteracting the effects of the G_s protein-coupled A_2A- and A_2BARs.³² For this reason, A₁- and A₃ARs can be regarded as anti-targets in the development of AR antagonists for cancer therapy, and the lack of selectivity may contribute to side-effects. ³

Table 2. Binding affinities of Etrumadenant and selected antagonists for the human adenosine receptors and for crystallization constructs.^a

Compounds	Human A₁AR	Human A_2AAR	Human A_2BAR	Human A₃AR
	[³H]DPCPX (or [³H]CCPA)	[³H]MSX-2	[³H]PSB-603	[³H]PSB-11 (or [¹²⁵I]-AB-MECA^b)
	pK_i ± SEM	pK_i ± SEM	pK_i ± SEM	pK_i ± SEM
Etrumadenant	8.12 ± 0.08 (8.15 ± 0.02)	9.07 ± 0.14	8.50 ± 0.06	6.50 ± 0.14
ZM241385	6.65^b	8.69 ± 0.20	7.53 ± 0.20	(<5.00^b)
PSB-603	(<5.00^c)	<5.00^c	9.26^c	<5.00^c
Preladenant	(6.53^d)	9.05^d	<6.00^d	<6.00^d
	A_2A-PSB2-bRIL	A_2A-StaR2-bRIL
	[³H]MSX-2	[³H]MSX-2
	pK_i ± SEM	pK_i ± SEM
Etrumadenant	8.95 ± 0.12	7.40 ± 0.05

^apK_i values were determined as means from at least three independent experiments ± standard error of the mean (SEM) performed on CHO cell membranes expressing the respective human wt AR, or on Sf9 insect cell membranes for the two crystallization constructs A_2A-PSB2-bRIL and A_2A-StaR2-bRIL. [³H]CCPA and [¹²⁵I]-AB-MECA represent agonist radioligands whereas all other radioligands are antagonists at ARs. ^bOngini et al.³³ ^cBorrmann et al.²⁸ ^dBurbiel et al.²⁷

For comparison, we additionally investigated the prototypical non-selective A_2A/A_2BAR antagonist ZM241385, the A_2A-selective antagonist Preladenant, and the A_2B-selective antagonist PSB-603. Preladenant inhibited the A_2AAR with similar potency as Etrumadenant in this assay (IC₅₀ values 85.1 nm, 53.7 nm), whereas the potency of ZM241385 (IC₅₀: 178 nm) was lower than that of Etrumadenant (IC₅₀: 4.57 nm) at the A_2BAR, but similar at the A_2AAR (IC₅₀ values: 100 nm; 53.7 nm). PSB-603 showed similarly high potency at the A_2BAR as Etrumadenant (IC₅₀ values: 3.02 nm; 4.57 nm). It should be kept in mind that the employed functional G protein activation assays require overexpression of receptors and G proteins and .³¹ Nevertheless, these data confirm that Etrumadenant is a potent antagonist of A_2A- and A_2BARs, but its selectivity versus the G_i protein-coupled ARs is low. To explain this observation, we performed a sequence alignment of all AR subtypes and analyzed the conservation of amino acids that interact with Etrumadenant as observed in the A_2AAR co-crystal structures (Figure 4). In fact, these amino acid residues are largely conserved in the A₁-, A_2A-, and A_2BAR subtypes, which is consistent with the high affinity of Etrumadenant for all three subtypes. One notable difference is S67^2.65 of the A_2AAR that is exchanged for N70^2.65 in the A₁AR. S67^2.65 forms direct contacts with the pyridine core of Etrumadenant at the extracellular ends of the ligand binding pocket. Its exchange to asparagine might therefore affect the binding of Etrumadenant and explain the slightly reduced affinity of Etrumadenant for the A₁AR.

The A₃AR, on the other hand, shows significant differences being the least conserved AR subtype regarding Etrumadenant’s binding pocket residues. Three hydrophobic amino acid residues that form direct Etrumadenant contacts in the A_2AAR (I66^2.64, V84^3.32, and A273^7.38) are exchanged in the A₃AR for different hydrophobic amino acids or glycine (V72^2.64, L90^3.32, and G267^7.38). Moreover, we observed direct interactions of Etrumadenant to the side chains of E169^ECL2 and H250^6.52 as well as a water-bridged hydrogen bond to T256^6.58. These residues are conserved among the A₁-, A_2A-, and A_2BARs, whereas the A₃AR contains V169^ELC2, S247^6.52, and I253^6.58 in the analogous positions (Figure 4, red boxes). Variation of these interacting residues in the A₃AR provides an explanation for the decreased affinity of Etrumadenant for the A₃AR.

The novel adenosine receptor antagonist Etrumadenant represents a promising clinical candidate for the treatment of cancer, with high affinity for A_2A and A_2BARs. Both AR subtypes represent purinergic immune checkpoints inhibiting the immune system and showing ancillary direct cancer proliferation-enhancing and proangiogenic effects.^7,32 Blockade of A_2A- and A_2BARs consequently is expected to exert anti-cancer activity. We show that the affinity of Etrumadenant for the A₁AR, which has been discussed as an anti-target in cancer therapy, is similarly high as for the A_2A- and A_2BAR, with a K_i value in the single-digit nanomolar range. The first A_2AAR co-crystal structures of Etrumadenant in complex with the A_2AAR provide novel insights into the A_2AAR ligand binding pocket. They revealed that Etrumadenant stabilizes the inactive state of the A_2AAR by hydrogen bond interaction to T88^3.36 through its cyano group. A direct hydrogen bond to T88^3.36 has thus far not been observed in antagonist-bound A_2AAR crystal structures. Importantly, the A_2A-StaR2 construct that has been used for the vast majority of A_2AAR co-crystal structures¹⁶ contains two mutations inside the ligand binding pocket (T88^3.36A and S277^7.42A) preventing the discovery of this hydrogen bond. Nevertheless, the structure of a modified A_2A-StaR2-bRIL construct (that has the S277^7.42A mutation reverted to the wt residue) in complex with Etrumadenant revealed nearly identical binding poses of Etrumadenant, despite the T88^3.36A mutation. However, we show that the affinity of Etrumadenant for the A_2AStaR2bRIL crystallization construct is reduced by 47-fold compared to the wt A_2AAR (K_i 39.8 nm vs. 0.851 nm) whereas the affinity to the employed optimized crystallization construct A_2A-PSB2-bRIL is unaltered (K_i 1.12 nm). The newly discovered interaction with T88^3.36 will be relevant for the design and optimization of new A_2AAR antagonists as well as for dual A_2A/A_2BAR antagonists and pan-AR antagonists, in particular, since T^3.36 is conserved among the AR subtypes, not only in humans, but also in rats.³⁴ Its conservation likely also contributes to Etrumadenant’s high affinity for the A₁AR. The A₃AR contains several non-conserved residues that are involved in A_2AAR binding, which could explain the selectivity of Etrumadenant versus the A₃AR subtype as observed in radioligand binding experiments (more than two orders of magnitude). The present Xray structure will serve as a basis for the future design of tailored AR antagonists, which have great potential for the treatment of cancer as well as neurodegenerative diseases.

Expression, purification and crystallization of the A_2A-PSB2-bRIL-Etrumadenant complex

The crystallization construct A_2A-PSB2-bRIL was cloned using site-directed mutagenesis in order to add the glycosylation mutation N154^ECL2A to the previously reported crystallization construct A_2A-PSB1-bRIL.¹⁶ The N154^ECL2Q and N154^ECL2D mutations were cloned analogously. A_2A-PSB2-bRIL was expressed and purified as previously described for A_2A-PSB1-bRIL.¹⁶ Briefly, A_2A-PSB2-bRIL was expressed in Sf9 insect cells using GP64-pFastBac1 as baculoviral expression vector. Cells were disrupted by osmotic shock and membranes were repeatedly washed using a washing buffer that contained a high amount of NaCl. The resuspended cell membranes were subsequently incubated with 50 µm Etrumadenant (obtained from MedChemExpress, cat. #HY-129393) and 2 mg∙mL^-1 iodoacetamide for 1 h prior to solubilization. A_2A-PSB2-bRIL was solubilized and purified from Sf9 membranes as previously described for A_2A-PSB1-bRIL.¹⁶ Etrumadenant was added to wash buffer I and to wash buffer II at 50 µm and 25 µm concentration, respectively. The protein was eluted with four column volumes of elution buffer containing 25 µm Etrumadenant, 25 mm HEPES pH 7.5, 800 mm NaCl, 10% (v/v) glycerol, 220 mm imidazole, 0.025 % (w/v) dodecyl-β-d-maltoside (DDM), and 0.005% (w/v) cholesteryl hemisuccinate (CHS).

The A_2A-PSB2-bRIL-Etrumadenant complex was concentrated to a volume of 20-30 µL using 100 kDa cut-off Vivaspin concentrators (Sartorius), and immediately used for crystallization experiments. The complex was reconstituted into lipidic cubic phase using the two-syringe method³⁵ by mixing the protein with a molten lipid mixture [90% (w/w) 1-oleoyl-rac-glycerol (Sigma), 10 % (w/w) cholesterol (Sigma)] in a 2 to 3 ratio. Crystallization experiments were performed using an automatic crystallization robot (Formulatrix NT8) by overlaying 50 nL of mesophase with 800 nL of precipitant solution. The A_2A-PSB2-bRIL-Etrumadenant complex crystallized in 30 % (w/v) PEG400, 7 % (w/v) Tacsimate pH 7.0 (Hampton Research, cat. #HR2-755), ³⁶ 100 mm HEPES-Na pH 7.4, 1.8 % (w/v) 2,5-hexandiol (Molecular Dimensions, cat. #MD2-100-226). Crystals were harvested with micromounts (MiTeGen) and flash-frozen in liquid nitrogen without further cryo-protection.

X-ray diffraction data collection and structure elucidation of the A_2A-PSB2-bRIL-Etrumadenant structure

X-ray data collection was carried out at 100 K on EMBL beamline P14 of the DESY synchrotron (Hamburg, Germany). An Eiger2 16M detector was placed at a distance of 340 mm behind the crystal, which was rotated for 360° while diffraction images were recorded at 0.15° steps with exposure for 0.01 s. All datasets were indexed, integrated, scaled, and converted to structure factor amplitudes using ISPyB software: autoPROC,³⁷ XDS,³⁸ CCP4,³⁹ POINTLESS,⁴⁰ AIMLESS,⁴¹ STARANISO. ⁴² Crystallographic statistics are presented in Table 1. PDB ID 5IU4⁴³ was used as the starting model for refinement with phenix.⁴⁴ Coot ⁴⁵ was used for model building. The stereochemical restraints for the ligand were generated with the GRADE web server. ⁴⁶

Expression, purification and crystallization of the A_2A-StaR2-bRIL-A277S-Etrumadenant complex

A_2A construct (A_2A-StaR2-bRIL-A277S) containing the same thermostabilizing and deglycosylation site mutations as PDB ID 5IU4²⁶ (except for the S277^7.42A mutation, which is reverted to wt) was codon-optimized and cloned between the BamHI and HindIII sites of pFastBac1 (Trenzyme). The bacmid was generated by transforming this plasmid into E. coli strain DH10EMBacY (MultiBac, Geneva Biotech). Isolated bacmid DNA was transfected into Sf9 insect cell using Cellfectin (Invitrogen) to generate baculovirus. For large-scale expression, High Five insect cells growing at 27 °C in Sf900 II medium (Invitrogen) at 2.5∙10⁶∙mL^-1 were infected with P2 baculovirus and harvested at 48 h post infection. Cells were harvested by centrifugation and the pellet was stored at ‑80 °C. Cells were thawed at room temperature and resuspended in 40 mm Tris‑HCl pH 7.6, 1 mm ethylenediaminetetraacetic acid (EDTA), and cOmplete EDTA-free protease inhibitors tablets (Roche). Membranes were fractionated by passing the cell once using microfluidizer (Microfluidics) operated at 8000 psi. Membranes were centrifuged at 42000 rpm using a Ti45 rotor (Beckman) and washed once with 40 mm Tris-HCl pH 7.6, 1 M NaCl, and cOmplete EDTA-free protease inhibitors tablets. Membranes were centrifuged again and resuspended in 40 mm Tris−HCl pH 7.6, cOmplete EDTA-free protease inhibitor cocktail tablets and frozen at -80 °C. Unless otherwise stated, all purification procedures were carried out at 4 °C. Membranes were solubilized with 1.5 % (w/v) decylmaltoside (DM) and 0.1 % (w/v) CHS for 1 h. Insoluble fractions were pelleted by centrifugation at 42000 rpm using a Ti45 rotor (Beckman) for 30 min. A_2A-StaR2-bRIL-A277S was purified by loading the supernatant (supplemented with 8 mm imidazole) into a 5 mL cartridge containing Ni-NTA super flow resin (Qiagen) at 2 mL∙min^-1 using an ÄKTA pure system (Cytiva). The resin was first washed with 25 mL of 40 mm Tris-HCl pH 7.5, 400 mm NaCl, 0.15 % (w/v) DM, 0.002 % (w/v) CHS, and 8 mm imidazole and then washed once more with 25 mL of 40 mm Tris-HCl pH 7.5, 400 mm NaCl, 0. % (w/v) DM, 0.002 % (w/v) CHS, and 72mm imidazole. The protein was eluted in the same buffer containing 272 mm imidazole, concentrated using Vivaspin turbo 15 mL with a molecular weight cut-off of 50 kDa (Sartorius) and loaded onto a Superdex 200 10/300 GL increase column equilibrated in 40 mm Tris-HCl pH 7.4, 200 mm NaCl, 0.15 % (w/v) DM, and 0.002 % (w/v) CHS. Fractions containing A_2A-StaR2-bRIL-A277S were pooled, concentrated to 22.5 mg∙ml^‑1 aliquots and frozen at -80 °C. Protein purity and homogeneity were controlled by SDS-PAGE and fluorescence size exclusion chromatography (FSEC).

For crystallization, frozen A_2A-StaR2-bRIL-A277S aliquots were thawed on ice and centrifuged at 18213 g for 10 min. Lipidic cubic phase (LCP) crystallization was performed by mixing the protein into monoolein (containing 10 % (w/w) cholesterol) at 2:3 (w/w) ratio. The resulting LCP was dispensed using the mosquito LCP (SPT Labtech) using a bolus/precipitant solution ratio of 40:800 nL. Crystals were obtained using precipitant solution containing 0.1 M sodium citrate pH 5.0, 50 mm sodium thiocyanate, 3 % (v/v) 2-methyl-2,4-pentanediol (MPD), 21-32 % (w/v) PEG400, and 2 mm theophylline. Crystals appeared overnight and grew to full size (50-60 µm in the longest dimension) over a week. To prepare the A_2A-StaR2-bRIL-A277S-Etrumadenant complex, crystals were soaked overnight in the same precipitant solution by replacing theophylline with 1 mm Etrumadenant.²⁶ After soaking overnight, crystals were harvested with MicroLoops LD (mitogen) and frozen in liquid nitrogen.

X-ray diffraction data collection and structure elucidation of the A_2A-StaR2-bRIL-A277S-Etrumadenant structure

Diffraction data were collected at the Swiss Light Source (SLS) beamline PXII. Crystal was exposed to a 25x13 µm X-ray beam at 25 % transmission. A total of 180° of rotational data were collected using 0.1° oscillation and 0.05 s exposure per image. Dataset was processed and scaled to 2.1 Å using XDS³⁸ (built 20161205), then merged and converted to mtz file format with AIMLESS⁴¹ (version 0.7.3 from CCP4³⁹ distribution 7.0.066). The structure was solved by molecular replacement with PHASER⁴⁷ (version 2.8.2 from CCP4 distribution 7.0.066), using PDB ID 5IU4²⁶ as the search model. The model was rebuilt and refined using COOT⁴⁵and PHENIX⁴⁸ (version 1.14) using TLS and optimizing 8CIC and ADP weight. After structure refinement, the model was validated using MolProbity⁴⁹ (from PHENIX version 1.14).

Radioligand binding studies

Radioligand binding assays were performed on CHO cell membranes or Sf9 insect cell membranes expressing the respective human wt adenosine receptor or A_2AAR crystallization constructs (A_2A-PSB2-bRIL or A_2A-StaR2-bRIL) as previously described.^16,34 The agonist [³H]CCPA or the antagonist [³H]DPCPX were employed as radioligands for the A₁AR, the antagonist [³H]MSX-2 was used for the A_2AAR, the antagonist [³H]PSB-603 for the A_2BAR and the antagonist [³H]PSB-11 for the A₃AR. All assays were performed in 50 mm Tris buffer (pH 7.4 at room temperature) at a final volume of 400 µL (A₁-, A_2A-, and A₃ARs) or 1000 µL (A_2BAR). Test compounds were dissolved in dimethyl sulfoxide and incubated at room temperature with the respective membranes and radioligand for 90 min (A₁AR, [³H]CCPA), 60 min (A₁AR, [³H]DPCPX), 30 min (A_2AAR), 75 min (A_2BAR) and 45 min (A₃AR). The final dimethyl sulfoxide concentration was 1 %. Then, the mixture was filtered through GF/B glass fiber filters using a cell harvester (Brandel). Filters were then washed three times with 2 mL ice-cold Tris-buffer (50 mm, pH 7.4 at room temperature). Filters containing the A_2BAR were washed with the same ice-cold Tris-buffer but with the addition of 0.1 % BSA. The remaining radioactivity was quantified after incubation for 9 h with scintillation cocktail (Beckmann Coulter) using a scintillation counter (Tricarb 2700TR).

G protein dissociation assays

TRUPATH BRET² assays³¹ were performed as previously described⁵⁰ (TRUPATH was a gift from Bryan Roth (Addgene kit #1000000163)). In case of Gα_i/o-coupled adenosine receptors (A₁ and A₃ARs), a biosensor composed of Gα_i1‑RLuc8, Gβ₃, and Gγ₉-GFP2 was used. In case of the Gα_s-coupled A_2Aand A_2BARs, the biosensor consisted of Gα_s-RLuc8, Gβ₃, and Gγ₉-GFP2. Antagonist solution was incubated with the cells for 20 min before the addition of luciferase substrate solution (50 µm Deep Blue C, Biomol). Agonist solution (NECA) was added 5 min after the addition of the substrate solution at its EC₈₀ concentrations (A₁AR 20 nm, A_2AAR 1 µm, A_2BAR 5 µm, A₃AR 20 nm) and incubated for 5 min until measurement. GFP2 fluorescence (515 nm emission filter) was divided by RLuc8 luminescence (395 nm emission filter) to obtain BRET ratios. Data was normalized to controls (100 % activation = NECA without antagonist, 0 % activation = no agonist), and IC₅₀ values were obtained by a four-parameter sigmoidal curve fit in GraphPad PRISM v8.0 (GraphPad, San Diego, CA).

SDS-PAGE analysis

Proteins for SDS-PAGE were expressed and purified from 40 mL of Sf9 insect cell culture, as previously described for thermostability assessment.¹⁶ Proteins were analyzed on homemade 10 % SDS-PAGE gels casted using bis-2-amino-2-(hydroxymethyl)propane-1,3-diol (bis-Tris) buffer. Protein samples were prepared using NuPAGE loading dye (ThermoFisher, cat. #NP0007) supplemented with a final concentration of 50 mm dithiothreitol (DTT). Samples were incubated for 30 min at 37 °C prior to SDS-PAGE analysis using 3-(N-morpholino)propanesulfonic acid (MOPS) running buffer without addition of sodium hydrogen sulfite. SDS-PAGE gels were stained with Coomassie brilliant blue R-250 and destained using hot water. In order to investigate the effect of Tunicamycin on A_2AAR glycosylation, the respective insect cell culture was treated with 1 µg∙mL^-1 of Tunicamycin (CaymanChemical, cat. #11445) during infection. PNGase F (New England Biolabs, cat. #P0704S) was used to cleave the glycosylation in the purified protein prior to SDS-PAGE analysis using 375 units in a total reaction volume of 22.5 µL followed by incubation at 16 °C for 16 h.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. The coordinates and structure factors for the obtained crystal structures have been deposited to the Protein Data Bank (https://www.rcsb.org/) under accession IDs 8C9W (A_2A-PSB2-bRIL-Etrumadenant) and 8CIC (A_2A-StaR2-bRIL-A277A).

Acknowledgments

The authors thank Christin Vielmuth for her assistance with radioligand binding experiments. N.S. thanks the Deutsche Forschungsgemeinschaft for financial support (CRC 1423, project A6). The authors thank the team at Merck KGaA, Germany, for donation of AB928 X-ray structural data for A_2A-StaR2-bRIL-A277S-Etrumadenant used in the present work. This X-ray structure was originally generated by LeadXPro, contracted on behalf of Merck KGaA, Germany.

Author contributions

T.C., J.G.S., J.H.V., V.J.V.: Investigation, Methodology, Validation, Formal analysis; R.K.Y, S.M.M, D.B.: Investigation, Methodology, Review & Editing.; T.C.: Conceptualization, Writing – Original draft, Visualization; R.H.W.: Investigation, Formal analysis, N.S.: Investigation, Supervision; C.E.M: Project administration, Conceptualization, Supervision, Funding acquisition, Writing – Review & Editing

Competing interests

The authors declare no competing interests.

IJzerman, A. P., Jacobson, K. A., Müller, C. E., Cronstein, B. N. & Cunha, R. A. International union of basic and clinical pharmacology. CXII: adenosine receptors: a further update. Pharmacol. Rev. 74, 340–372 (2022).
Burnstock, G. Purinergic signalling and disorders of the central nervous system. Nat. Rev. Drug Discov. 7, 575–590 (2008).
Borea, P. A., Varani, K., Gessi, S., Merighi, S. & Vincenzi, F. The Adenosine Receptors (Springer International Publishing, Cham, 2018).
Voss, J. H., Mahardhika, A. B., Inoue, A. & Müller, C. E. Agonist-dependent coupling of the promiscuous adenosine A_2B receptor to Gα protein subunits. ACS Pharmacol. Transl. Sci. 5, 373–386 (2022).
Linden, J., Thai, T., Figler, H., Jin, X. & Robeva, A. S. Characterization of human A_2B adenosine receptors: radioligand binding, western blotting, and coupling to G_q in human embryonic kidney 293 cells and HMC-1 mast cells. Mol. Pharmacol. 56, 705–713 (1999).
Gao, Z.-G., Inoue, A. & Jacobson, K. A. On the G protein-coupling selectivity of the native A_2B adenosine receptor. Biochem. Pharmacol. 151, 201–213 (2018).
Borea, P. A., Gessi, S., Merighi, S., Vincenzi, F. & Varani, K. Pharmacology of adenosine receptors: the state of the art. Physiol. Rev. 98, 1591–1625 (2018).
Ohta, A. et al. A_2A adenosine receptor protects tumors from antitumor T cells. Proc. Natl. Acad. Sci. U. S. A. 103, 13132–13137 (2006).
Hatfield, S. M. & Sitkovsky, M. A_2A adenosine receptor antagonists to weaken the hypoxia-HIF-1α driven immunosuppression and improve immunotherapies of cancer. Curr. Opin. Pharmacol. 29, 90–96 (2016).
Chen, J.-F. & Cunha, R. A. The belated US FDA approval of the adenosine A_2A receptor antagonist istradefylline for treatment of Parkinson's disease. Purinergic Signal. 16, 167–174 (2020).
Faivre, E. et al. Beneficial Effect of a Selective Adenosine A_2A Receptor Antagonist in the APPswe/PS1dE9 Mouse Model of Alzheimer's Disease. Front. Mol. Neurosci. 11, 235 (2018).
Merighi, S. et al. A_2A adenosine receptor antagonists in neurodegenerative diseases. Curr. Med. Chem. (2021).
Ohta, A. A metabolic immune checkpoint: adenosine in tumor microenvironment. Front. Immunol. 7, 109 (2016).
Beatty, J. et al. Azolopyrimidine for the treatment of cancer-related disorders. International Patent, WO/2018/136700 (2018).
Thompson, E. A. & Powell, J. D. Inhibition of the adenosine pathway to potentiate cancer immunotherapy: potential for combinatorial approaches. Annu. Rev. Med. 72, 331–348 (2021).
Claff, T. et al. Single stabilizing point mutation enables high-resolution co-crystal Structures of the adenosine A_2A receptor with preladenant conjugates. Angew. Chem. Int. Ed., e202115545 (2022).
Chu, R. et al. Redesign of a four-helix bundle protein by phage display coupled with proteolysis and structural characterization by NMR and X-ray crystallography. J. Mol. Biol. 323, 253–262 (2002).
Hino, T. et al. G-protein-coupled receptor inactivation by an allosteric inverse-agonist antibody. Nature 482, 237–240 (2012).
Doré, A. S. et al. Structure of the adenosine A_2A receptor in complex with ZM241385 and the xanthines XAC and caffeine. Structure 19, 1283–1293 (2011).
Liu, W. et al. Structural basis for allosteric regulation of GPCRs by sodium ions. Science 337, 232–236 (2012).
Wang, G., Jong, R. N. de, van den Bremer, E. T. J., Parren, P. W. H. I. & Heck, A. J. R. Enhancing accuracy in molecular weight determination of highly heterogeneously glycosylated proteins by native tandem mass spectrometry. Anal. Chem. 89, 4793–4797 (2017).
Tarentino, A. L. & Plummer, T. H. in Guide to techniques in glycobiology, edited by W. J. Lennarz & G. W. Hart (Elsevier1994), pp. 44–57.
Duksin, D. & Mahoney, W. C. Relationship of the structure and biological activity of the natural homologues of tunicamycin. J. Biol. Chem. 257, 3105–3109 (1982).
Carpenter, B. & Lebon, G. Human adenosine A_2A receptor: Molecular mechanism of ligand binding and activation. Front. Pharmacol. 8, 898 (2017).
Borodovsky, A. et al. Small molecule AZD4635 inhibitor of A_2AR signaling rescues immune cell function including CD103⁺ dendritic cells enhancing anti-tumor immunity. J. Immunother. Cancer 8 (2020).
Rucktooa, P. et al. Towards high throughput GPCR crystallography: In meso soaking of adenosine A_2A receptor crystals. Sci. Rep. 8, 41 (2018).
Burbiel, J. C. et al. 2-Amino[1,2,4]triazolo[1,5-c]quinazolines and derived novel heterocycles: Syntheses and structure-activity relationships of potent adenosine receptor antagonists. ChemMedChem 11, 2272–2286 (2016).
Borrmann, T. et al. 1-Alkyl-8-(piperazine-1-sulfonyl)phenylxanthines: development and characterization of adenosine A_2B receptor antagonists and a new radioligand with subnanomolar affinity and subtype specificity. J. Med. Chem. 52, 3994–4006 (2009).
Galés, C. et al. Real-time monitoring of receptor and G-protein interactions in living cells. Nat. Methods 2, 177–184 (2005).
Galés, C. et al. Probing the activation-promoted structural rearrangements in preassembled receptor-G protein complexes. Nat. Struct. Mol. Biol. 13, 778–786 (2006).
Olsen, R. H. J. et al. TRUPATH, an open-source biosensor platform for interrogating the GPCR transducerome. Nat. Chem. Biol. 16, 841–849 (2020).
Fredholm, B. B. Adenosine receptors as drug targets. Exp. Cell Res. 316, 1284–1288 (2010).
Ongini, E., Dionisotti, S., Gessi, S., Irenius, E. & Fredholm, B. B. Comparison of CGS 15943, ZM 241385 and SCH 58261 as antagonists at human adenosine receptors. Naunyn Schmiedeberg's Arch. Pharmacol. 359, 7–10 (1999).
Alnouri, M. W. et al. Selectivity is species-dependent: Characterization of standard agonists and antagonists at human, rat, and mouse adenosine receptors. Purinergic Signal. 11, 389–407 (2015).
Caffrey, M. & Cherezov, V. Crystallizing membrane proteins using lipidic mesophases. Nat. Protoc. 4, 706–731 (2009).
McPherson, A. & Cudney, B. Searching for silver bullets: an alternative strategy for crystallizing macromolecules. J. Struct. Biol. 156, 387–406 (2006).
Vonrhein, C. et al. Data processing and analysis with the autoPROC toolbox. Acta Crystallogr. D Biol. Crystallogr. 67, 293–302 (2011).
Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010).
Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr. 67, 235–242 (2011).
Evans, P. Scaling and assessment of data quality. Acta Crystallogr. D Biol. Crystallogr. 62, 72–82 (2006).
Evans, P. R. & Murshudov, G. N. How good are my data and what is the resolution? Acta Crystallogr. D Biol. Crystallogr. 69, 1204–1214 (2013).
Tickle, I. J. et al. STARANISO. Global Phasing Ltd., Cambridge, United Kingdom (2018-2021).
Segala, E. et al. Controlling the dissociation of ligands from the adenosine A_2A receptor through modulation of salt bridge strength. J. Med. Chem. 59, 6470–6479 (2016).
Afonine, P. V. et al. Towards automated crystallographic structure refinement with phenix.refine. Acta Crystallogr. D Biol. Crystallogr. 68, 352–367 (2012).
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010).
Smart, O. et al. Grade, version 1.2.20. Global Phasing Ltd., Cambridge, United Kingdom (2011).
McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674 (2007).
Liebschner, D. et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr. D Struct. Biol. 75, 861–877 (2019).
Chen, V. B. et al. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D Biol. Crystallogr. 66, 12–21 (2010).
Temirak, A. et al. Irreversible antagonists for the adenosine A_2B receptor. Molecules 27 (2022).

There is NO Competing Interest.

Download PDF

Journal Publication

published 01 Jun, 2023

Read the published version in Communications Chemistry →

Version 1

posted

You are reading this latest preprint version

Crystal structure of adenosine A2A receptor in complex with clinical candidate Etrumadenant reveals novel antagonist interaction

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results And Discussion

Conclusions

Methods

Declarations

References

Additional Declarations

Status:

Journal Publication

Version 1