Towards high throughput GPCR crystallography: In Meso soaking of Adenosine A2A Receptor crystals

Here we report an efficient method to generate multiple co-structures of the A2A G protein-coupled receptor (GPCR) with small-molecules from a single preparation of a thermostabilised receptor crystallised in Lipidic Cubic Phase (LCP). Receptor crystallisation is achieved following purification using a low affinity “carrier” ligand (theophylline) and crystals are then soaked in solutions containing the desired (higher affinity) compounds. Complete datasets to high resolution can then be collected from single crystals and seven structures are reported here of which three are novel. The method significantly improves structural throughput for ligand screening using stabilised GPCRs, thereby actively driving Structure-Based Drug Discovery (SBDD).

readily crystallises in meso yielding thick ~60 µm long plates (Fig. 1C), typically diffracting to 2.0 Å and containing a theophylline molecule in the A 2A R orthosteric binding site 20 (Fig. 2A). Crystals with theophylline have also been used previously to generate a structure with another xanthine, PSB36 20 .
The utility of the in meso soaking system for diverse ligands from chemical series other than xanthines was then investigated. A 2A -StaR2-b RIL 562-Theophylline crystals were soaked in mother liquor supplemented with A 2A antagonists Tozadenant 21 (pK D = 8.4), LUAA47070 22 (pK D = 6.5) or Vipadenant 23 (pK D = 9.0), and their diffraction characterised. Crystals from these experiments diffracted in spacegroup C222 1 to 2.0-3.1 Å resolution (Table 1). Tozadenant, LUAA47070 and Vipadenant are all well defined in the electron density maps (Fig. 2B-D). For these ligands the basal region of the orthosteric site is delimited by Trp246 6.48 , which engages in Van der Waals contacts to the Tozadenant benzothiazole ring, the LUAA47070 thiazole ring or the Vipadenant furan ring (Fig. 2B-D). These ligands explore different regions at the apical end of the orthosteric site. The 4-hydroxy,4-methylpiperidine moiety of Tozadenant sits upright on the benzothiazole ring, and hydrogen bonds to Thr256 6.58 . The N,2,2-trimethylpropanamide group of LUAA47070 extends obliquely towards transmembrane helix 1 (TM1), and engages in water-mediated contacts with ECL2 Glu169 and Tyr9 1.35 . Further this structure shows how the experimentally defined water mediated interactions of the amide group of LUAA47070 to both Asn253 6.55 and His278 7.42 contribute to this ligand binding pose. Finally, the Vipadenant 2-methylaniline moiety points laterally towards TM1, and is hydrogen-bonded to Tyr9 1. 35 . We find that, despite adopting a range of orientations in the orthosteric binding site, ligands from different chemical series can be effectively soaked into A 2A -StaR2-b RIL 562-Theophylline crystals, and used in crystallographic structural studies to identify their binding modes. Contrary to poorly diffracting, bespoke A 2A -StaR2-b RIL 562-Tozadenant crystals, likely resulting from the disruption of the salt bridge between extracellular loop 2 (ECL2) Glu169 and ECL3 His264, interfering with crystal packing, co-crystals from soaking experiments yielded good quality structural data, highlighting the versatility of the in meso soaking system.
The validity of structural results obtained by the in meso soaking method was checked using ZM241385 24 (pK D = 8.6), a well-characterised A 2A R antagonist that increases A 2A -StaR2-b RIL 562 stability by ~12 °C (Fig. 1A). The crystal structure of the receptor in complex with ZM241385 resulting from in meso soaking, was compared with similar complexes obtained from bespoke crystallisation setups using either A 2A -StaR2-b RIL 562 25 or A 2A -b RIL 562 26 (Fig. 2E). Overlaying these structures shows a remarkably similar structural conformation of residues in the orthosteric located within 5 Å of the ligand with an all atom r.m.s.d. of only 0.074 Å (soaked v/s  (Fig. 2E). Such a high degree of structural conservation across different crystallisation methods (and A 2A R constructs) benchmarks and underlines the robustness of the in meso soaking system described here.
To determine the feasibility of using the in meso soaking method system to support optimisation of novel A 2A R antagonists for drug discovery, Compound 4e, a 1,2,4-triazine derivative 19 , was investigated. Compound 4e is a low nanomolar affinity ligand (pK D = 9.6) for A 2A R and increases A 2A -StaR2-b RIL 562 stability by ~19 °C when compared to apo protein (Fig. 1A) and co-crystals were generated using either a bespoke protein preparation or by soaking A 2A -StaR2-b RIL 562-Theophylline crystals in mother liquor supplemented with Compound 4e for 1 or 24 hours (Fig. 1D,E). Crystal morphology remained unchanged regardless of soaking times (Fig. 1D,E) and crystals from these three experiments diffracted to 1.9-2.1 Å in spacegroup C222 1 . Structures generated from  (Table 1). Compound 4e sits lower in the orthosteric site than theophylline, with the triazine ring π stacking against Phe168 from ECL2, while also engaging in polar contacts with an extensive water network. The amine moiety on the triazine ring is further hydrogen-bonded to ECL2 Glu169 and Asn253 6.55 , whereas the hydroxyl group on the chlorophenol ring makes a hydrogen bond with His278 7.43 . In the basal region of the orthosteric site, the ligand benzyl ring makes Van der Waals interactions with Trp246 6.48 .
A pairwise comparison of residues located within 5 Å of all the different liganded structures presented here demonstrates all atom r.m.s.d. values ranging from 0.48 Å (between the A 2A -StaR2-b RIL 562-Compound 4e and -LUAA47070 structures) to 1.05 Å (between the A 2A -StaR2-b RIL 562-ZM241385 and -Tozadenant structures). Altogether, most of the mobility stems from Tyr271 7.35 , involved in water-mediated interactions with ZM241385, and from Glu169 in ECL2 and His264 which adopt different rotamer orientations in the A 2A -StaR2-b RIL 562-Tozadenant structure compared to the other ligand complexes.
In drug development, high-throughput X-ray crystallography expedites the elaboration of novel hits into lead compounds and drug candidates by providing multiple high resolution views of ligand-receptor complexes, which are key for understanding critical intermolecular interactions alongside interpretation of ligand-induced receptor conformational changes 27 . The accelerated availability of multiple receptor-ligand complexes provides a data-rich starting point for SBDD and medicinal chemistry 28 which, when correlated with in vitro biological activity, allows rapid incorporation of molecular modifications towards increasing ligand affinity for the binding site or improvement of their absorption, distribution, metabolism, excretion and toxicity (ADMET) properties.
We have demonstrated that an in meso ligand soaking methology can rapidly and efficiently yield multiple high-resolution co-crystal structures from a diverse set of ligands in complex with a given GPCR. Such soaking techniques have also been employed in-house for other discovery projects. The method described here has general applicability to further discovery campaigns with stabilised membrane proteins using LCP crystallisation setups, provided high quality crystals exist for the target in complex with low affinity stabilising carrier ligands with fast off-rates.

Methods
StaR generation. The thermostabilisation of the human A 2A receptor (resulting in A 2A -StaR2) using a mutagenesis approach 8 , has been previously described 29 . For each protein preparation, cells from 2 L cultures were resuspended in 40 mM TRIS buffer at pH 7.6 supplemented by 1 mM EDTA and Complete EDTA-free protease inhibitor cocktail tablets (Roche). Cells were disrupted at ~15 000 psi using a microfluidizer (Processor M-110L Pneumatic, Microfluidics). Membranes pelleted by ultra-centrifugation at 200 000 g for 50 minutes, were subjected to a high salt wash in a buffer containing 40 mM Tris pH 7.6, 1 M NaCl and Complete EDTA-free protease inhibitor cocktail tablets, before they were centrifuged at 200,000 g for 50 minutes. Washed membranes were resuspended in 50 mL 40 mM Tris pH 7.6 supplemented with Complete EDTA-free protease inhibitor cocktail tablets and stored at −80 °C until further use.
Protein preparations intended for soaking experiments were carried out in the presence of theophylline whereas the bespoke preparation of A 2A -StaR2-b RIL 562 in complex with Compound 4e was done in the presence of 5 µM ligand.
Membranes were thawed, resuspended in a total volume of 150 ml with 40 mM Tris-HCl pH 7.6, Complete EDTA-free protease inhibitor cocktail tablets (Roche), 3 mM theophylline (Sigma Aldrich) (or 5 µM Compound 4e), and incubated for 2 hours at room temperature. Membranes were then solubilized by addition of 1.5% n-Decyl-β-D-maltopyranoside (DM, Anatrace), and incubation for 2 hours at 4 °C, followed by centrifugation at 145 000 g for 60 min to harvest solubilised material.
The solubilised material was applied to a 5 ml Ni-NTA (nickel-nitrilotriacetic acid) Superflow cartridge Collected fractions were analyzed by SDS PAGE and fractions containing A 2a -StaR2-b RIL 562 were pooled and concentrated using an Amicon Ultra Ultracell 50 K ultrafiltration membrane to a final volume of ~800 µl. The protein sample was ultra-centrifuged at 436 000 g for 10 minutes before being applied to a Superdex200 size exclusion column (GE Healthcare) pre-equilibrated with 40 mM Tris pH 7.4, 200 mM NaCl, 0.15% DM, 1 mM theophylline (or 5 µM Compound 4e). Eluted fractions containing the protein were analyzed by SDS PAGE, pooled and concentrated to ~35 mg/ml using an Amicon Ultra Ultracell 50 K ultrafiltration membrane and subjected to an ultra-centrifugation at 436 000 g prior to crystallisation. Protein concentrations were measured using the DC assay (Bio-Rad), and confirmed using quantitative amino acid analysis.
Thermal unfolding experiments. A 2A -StaR2-b RIL 562 purified in DM in the presence of 500 µM theophylline was used for thermal unfolding experiments. The protein was diluted in 40 mM Tris pH 7.4, 200 mM NaCl, 0.15% DM to a final concentration of 0.2 mg/ml. Following heavy dilution (~70-fold) of the protein in a buffer without ligand, the sample was considered to be in an apo-like state. Samples were supplemented with the respective ligands to a final concentration of 50 µM, with a final DMSO concentration of 5% (v/v). The control sample was supplemented with DMSO to a final concentration of 5% (v/v). Samples were incubated 30 minutes on ice before being loaded into UV capillaries (NanoTemper Technologies) and experiments were carried out using the Prometheus NT.48. The temperature gradient was set to an +1 °C/min from 20 °C to 90 °C. Protein unfolding was measured by detecting the temperature-dependent change in tryptophan fluorescence at emission wavelengths of 330 and 350 nm. The experiment was repeated four times and data analysed with the one-way analysis of variance (ANOVA) with Dunnett's post-test. Tm values obtained for the three ligands are statistically different from the control sample with p < 0.001. Diffraction data collection and processing. X-ray diffraction data were measured on a Pilatus 6 M detector at beamline I24 (Diamond Light Source) using a 6 × 9 μm beam size of for crystals of A 2A -StaR2-b RIL 562 in complex with Compound 4e, Tozadenant or LUAA47070. Complete datasets were acquired from a single crystal for each of these complexes at wavelengths 0.96857 Å (Compound 4e and LUAA47070) or 0.96862 Å (Tozadenant), using an unattenuated beam and 0.2° oscillation per frame, with an exposure of 0.1 second per degree of oscillation. Diffraction data for the A 2A -StaR2-b RIL 562-Vipadenant complex were acquired from 3 different crystals on an Eiger 16 M detector at beamline X06SA (Swiss Light Source) at a wavelength of 1 Å, using 10% beam transmission and 0.1° oscillation per frame, with an exposure of 1 second per degree of oscillation. The A 2A -StaR2-b RIL 562-ZM241385 data was collected from a single crystal on an Eiger 16 M detector at beamline × 06SA at a wavelength of 1 Å, using 20% beam transmission and 0.25° oscillation per frame, with an exposure of 0.24 second per degree of oscillation. Data from individual crystals were integrated using XDS 31 , merged and scaled using AIMLESS 32 from the CCP4 suite 33 . Data collection statistics are reported in Table 1.

Crystallisation. The
Structure solution and refinement. The structures of the different A 2A -StaR2-b RIL 562-ligand complexes were solved by molecular replacement (MR) with Phaser 34 using the A 2A -StaR2-b RIL 562-theophylline complex structure 20 as the search model (PDB code: 5MZJ). Iterative rounds of model refinement performed using phenix. refine 35 , were interspersed with manual model building in COOT 36 . Both xray and B-factor restraint weights were optimised in phenix.refine, and 2 TLS groups corresponding to the receptor and to the b RIL 562 respectively were defined during refinement. Refinement was with positional and individual isotropic B-factor refinement. The final models were validated using MolProbity 37 . The final refinement statistics are presented in Table 1 Data Availability Statement. The data that support the findings of this study are available from the corresponding author upon reasonable request. Co-ordinates and structure factors have been deposited in the Protein Data Bank under the accession codes 5OM1, 5OM4, 5OLZ, 5OLV, 5OLO, 5OLH and 5OLG.