Single Stabilizing Point Mutation Enables High‐Resolution Co‐Crystal Structures of the Adenosine A2A Receptor with Preladenant Conjugates

Abstract The G protein‐coupled adenosine A2A receptor (A2AAR) is an important new (potential) drug target in immuno‐oncology, and for neurodegenerative diseases. Preladenant and its derivatives belong to the most potent A2AAR antagonists displaying exceptional selectivity. While crystal structures of the human A2AAR have been solved, mostly using the A2A‐StaR2 protein that bears 9 point mutations, co‐crystallization with Preladenant derivatives has so far been elusive. We developed a new A2AAR construct harboring a single point mutation (S913.39K) which renders it extremely thermostable. This allowed the co‐crystallization of two novel Preladenant derivatives, the polyethylene glycol‐conjugated (PEGylated) PSB‐2113, and the fluorophore‐labeled PSB‐2115. The obtained crystal structures (2.25 Å and 2.6 Å resolution) provide explanations for the high potency and selectivity of Preladenant derivatives. They represent the first crystal structures of a GPCR in complex with PEG‐ and fluorophore‐conjugated ligands. The applied strategy is predicted to be applicable to further class A GPCRs.


Expression of A2AAR constructs in Sf9 insect cells
Sf9 insect cells (Expression Systems) were cultured in ESF921 cell culture medium (Expression Systems) at 27 °C and were frequently tested negative for mycoplasma contamination (PCR-test, in-house). A 2A AR constructs were expressed utilizing a baculoviral expression system (Bac-to-Bac, ThermoFisher). For this purpose, the DNA sequence encoding for the wt A 2A AR and the A 2A -ΔC construct (comprising A 2A AR residues 2-316) was cloned into a modified pFastBac1 vector (ThermoFisher) that has its original polyhedrin promoter substituted with a GP64 promoter (GP64-pFastBac1). [2] The overall expression cassette is identical to previously described A 2A AR crystallization constructs [3] including an N-terminal auto-cleavable influenza hemagglutinin (HA) signal sequence [4] followed by a FLAG-tag and a C-terminal deca-histidine tag. The C-terminal histidine tag is not present in the wt A 2A AR plasmid. For A 2A -ΔC-bRIL, the third intracellular loop (residues 209-218) of A 2A -ΔC was substituted with a thermostabilized apocytochrome bRIL. [5] The S91 3.39 K point mutation was introduced using site-directed mutagenesis to create A 2A -PSB1-bRIL and A 2A -PSB1. The DNA sequence encoding for the A 2A -StaR2-bRIL [6] comprising nine point mutations (A54 2.42 L, T88 3.36 A, R107 3.55 A, K122 4.43 A, N154 ECL2 A, L202 5.63 A, L235 6.37 A, V239 6.41 A, S277 7.42 A) was gene-synthesized by BioCat and subcloned into the same GP64-pFastBac1 vector. The final plasmids were transfected into Sf9 insect cells at a cell density of 1.0 mio cells per ml as previously described [7] to generate the initial P0 virus. 400 µl of the P0 viral solutions were used to infect 40 ml of Sf9 insect cells at a density of 2.0-3.0 mio cells per ml following incubation at 27 °C for 48 h at 140 revolutions per minute. Both, the P1 virus and cells expressing the different A 2A AR constructs were harvested by centrifugation. Proteins purified from these cells were used to assess protein stability. For upscaling, 6 ml of P1 viruses were used to infect 900 ml of Sf9 insect cells at a density of 2.0-3.0 mio cells per ml following incubation at 27 °C for 48 h at 140 revolutions per minute. Cells were harvested by centrifugation, washed with phosphate-buffered saline (PBS) and stored at -80 °C for further use.

Protein purification for crystallization experiments
Protein purification for crystallization experiments was performed according to previously described procedures. [3] Sf9 insect cells from 900 ml infected cell culture were lyzed by osmotic shock in low osmotic buffer [10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl and EDTA-free cOmplete protease inhibitor cocktail (Roche)] using a dounce homogenizer and washed repeatedly using a high osmotic buffer that is identical to the low osmotic buffer but with the addition of 1 M NaCl. Membranes were finally resuspended in 50 ml resuspension buffer [10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl, 30% (v/v) glycerol] and stored at -80 °C for further use. Purified membranes corresponding to a volume of 25 ml were incubated with 4 mM theophylline and 2 mg per ml iodoacetamide for 60 min. The purified receptor complex was concentrated to volume of 20-30 µl using 100 kDa cut-off Vivaspin concentrators (Sartorius). The protein was immediately used for crystallization experiments while protein purity, monodispersity and thermostability were assessed using analytical size-exclusion chromatography, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and a thermal shift assay, respectively.

Protein purification for stability screening
A 2A ARs from 40 ml of Sf9 insect cell culture were purified accordingly but with minor modifications. Cells were lyzed by osmotic shock in low osmotic buffer and washed once with high osmotic buffer. Purified membranes were resuspended in 3 ml resuspension buffer and stored at -80 °C until further use. Prior to solubilization, membranes were incubated with 2 mg per ml iodoacetamide for 30 min at 4 °C. No ligand was added during purification so that the A 2A ARs were obtained in their APO state. Solubilization was performed using an equal volume of solubilization buffer II [100 mM HEPES pH 7.5, 1.6 M NaCl, 2.0 % (w/v) DDM and 0.4% (w/v) CHS] over the time course of 3 h while shaking (end-over-end) at 4 °C. Solubilized proteins were obtained by centrifugation at 14,000 g. The supernatant was supplemented with 20 mM imidazole pH 7.5 and 12.5 µl cobalt-based IMAC medium (TALON Superflow, cytiva) following overnight incubation at 4 °C. The IMAC medium was washed with 60 CVs of ligand-free wash buffer I (without the addition of ATP and MgCl 2 ) and 40 CVs of ligand-free wash buffer II. Finally, the proteins were eluted using 10 CVs of ligand-free elution buffer without further protein concentration. The protein was used directly for analytical size-exclusion chromatography, SDS-PAGE and thermal shift assays.

Thermal shift assay
The thermostability of different A 2A AR protein preparations was determined as previously described [8] using the thiol-specific fluorescent dye N- [4-(7-diethylamino-4-methyl-3-coumarinyl)phenyl]maleimide (CPM). Briefly, CPM (final concentration 2 µg per ml) was incubated for 10 min with an appropriate amount of protein in a buffer consisting of 25 mM HEPES pH 7.5, 500 mM NaCl, 2% (v/v) glycerol, 0.05% (w/v), 0.01% (w/v) CHS and the tested A 2A AR ligand (ligand stock solutions were prepared at a concentration of 1 mM in DMSO) or control. The protein thermostability was assessed on a Rotor-Gene Q real-time PCR cycler (Qiagen) using the excitation wavelength of 365 ± 20 nm and the detection wavelength of 460 ± 20 nm. Data was collected over a temperature range of 30 °C to 90 °C with a ramp of 1 °C per min and a fluorescence gain of 1. The T M values were calculated from the turning point of the non-linear regression (Boltzmann sigmoidal fit) in GraphPad Prism 7.0 after subtraction of a respective buffer control value. Thermostability data was obtained from at least three independent measurements.

Data collection and structure determination
X-ray diffraction data were collected on the P11-high-throughput-MX beam line at PETRA III, Hamburg, Germany. Data were collected at 100 K using a microfocused beam (20 × 20 µm 2 ) of ~12.0 keV (1.0332 Å) with 1 % transmission at a rate of 100 ms per frame and an oscillation-range of 0.1°. Data were collected until 1.8 Å and 2.5 Å for the A 2A -PSB1-bRIL-PSB-2113 and A 2A -PSB1-bRIL-PSB-2115 crystals, respectively. XDS, XSCALE and XDSCONV [10] were used for data processing (see Table S2 for details). The PSB-2113 structure was determined by phenix.phaser [11] using the previously solved A 2A -ΔC-bRIL structure [3] as a model (PDB 4EIY) (translation function Z score (TFZ) -67.0 and log-likelihood gain (LLG) -9667.819). The PSB-2115 structure was determined similarly (TFZ -61.2 and LLG -6257.201), but instead using the A 2A -PSB1-bRIL-PSB-2113 coordinates as the search model. Each model went through phenix.autobuild [12] once, which included density modification, iterative-model building and refinement. Then, each model was built with respective ligands (PSB-2113 and PSB-2115) and other components like cholesterol hemisuccinate, monoolein and PEG using COOT. [13] The newly determined protein structures were consistent with the published PDB 4EIY structure, the RMSD being 0.28 Å. For the ligands, well-defined electron densities were observed within the orthosteric binding site, while no significant electron density was observed for the fluorophores and the connecting flexible PEG linker. The refined structures of the PSB-2113 and PSB-2115 complexes showed good agreement with the obtained data (R work / R free ratios of 0.190 / 0.235 and 0.187 / 0.245 (2.6 Å), respectively). Detailed refinement statistics are reported in Table S2.

Radioligand binding assays
Radioligand binding assays were performed on Sf9 insect cell and CHO-S cell membranes. Sf9 membrane preparations were obtained from 40 ml of baculovirus infected Sf9 insect cells using a dounce homogenizer with one step low osmotic buffer and one step high osmotic buffer as described above, but resuspended in the following resuspension buffer: 10 mM HEPES pH 7.5, 10 mM MgCl 2 , 20 mM KCl. CHO-S cells were disrupted in a buffer consisting of 50 mM tris(hydroxymethyl)aminomethane (Tris) pH 7.4 and 2 mM EDTA using an Ultra-Turrax homogenizer and cell membranes were resuspended in a 50 mM Tris pH 7.4 buffer. Total protein concentrations were determined using a Bradford assay [14] with bovine serum albumin (BSA) as a reference, and 15 µg of protein per well was used in the radioligand binding experiments. Competition binding experiments to determine the affinity (K i ) of A 2A AR ligands were performed using the A 2A AR-selective antagonist radioligand [ 3 H]MSX-2 (specific activity 85 Ci per mmol, final concentration 1 nM) [15] in a buffer consisting of 50 mM Tris buffer pH 7.4 (supplemented with 2 U per ml adenosine desaminase). Nonspecific binding was determined in the presence of 10 µM CGS15943, and total binding was determined in the presence of DMSO. The assays were incubated for 30 min at RT followed by filtration through GF/B glass fiber filters (Whatman) using a 48-well harvester (Brandel). Filters were pre-incubated for 30 to 60 min in a solution of 0.3 % (w/v) polyethylenimine to reduce nonspecific binding. After harvesting, filters were washed with ice-cold Tris buffer (50 mM Tris, pH 7.4), transferred into scintillation vials, and incubated with 2.5 ml of scintillation cocktail (ProSafe FC plus) for 6 h. Subsequently, the radioactivity was determined on a liquid scintillation counter (Tricarb 2810TR, Perkin Elmer, efficiency of 53%). At least three independent experiments were performed. Homologous competition with unlabeled MSX-2 was performed to determine the equilibrium dissociation constant (K d ) of each A 2A AR construct. K i values of A 2A AR antagonists and agonists were calculated using the Cheng-Prusoff equation (GraphPad Prism 7.0).

TRUPATH assay
Human embryonic kidney 293 (HEK293) cells were cultured at 37° C and 5% CO 2 in Dulbecco's modified eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS), 100 U penicillin per ml, and 100 µg streptomycin per ml. Repeated in-house PCR tests confirmed the cells to be free of mycoplasma contamination. The TRUPATH bioluminescence resonance energy transfer (BRET) assay was conducted with minor variations to the originally published protocol. [16] TRUPATH biosensors were a gift from Bryan Roth (Addgene kit #1000000163). On the first day, HEK293 cells were detached from the flask with trypsin and seeded into 6-well plates at a density of 10 6 cells per well in 2 ml DMEM. After 2 h, the cells were transiently transfected with the biosensors and the GPCR of interest. For this purpose, the A 2A AR constructs with or without bRIL fusion protein were PCR-amplified without N-terminal HA-and C-terminal His-tags and subcloned into pcDNA3.1(+) using BamHI and EcoRI. For each well, 3 µl lipofectamine 2000 (ThermoFisher) per µg DNA were mixed with OptiMEM medium (ThermoFisher) to a volume of 250 µl per well and then incubated for 10 min at RT. Then, the lipofectamine solution was mixed with an equal volume of OptiMEM-DNA mixture (100 ng of each pcDNA5/FRT/To-Gα s-short -RLuc8, pcDNA3.1-Gβ 3 , pcDNA3.1-Gγ 9 -GFP2, and pcDNA3.1-FLAG-GPCR) and incubated for 30 min at RT. The mixture was added to the cells following overnight incubation. The next day, the cells were detached from the 6-well plate by pipetting and were seeded into a white-bottom 96-well plate (Greiner BioOne) at a density of 30,000 cells per well. After 24 h, the cells were washed with assay buffer (Hank's balanced salt solution (ThermoFisher) plus 20 mM HEPES pH 7.4). Then, 60 µl of assay buffer supplemented with 1000 U per ml of adenosine desaminase (Roche) and 10 µl of luciferase substrate (50 µM coelenterazine 400a (Biomol) in assay buffer) was added to the cells and incubated for another 5 min. Agonist solution (30 µl, 1% final DMSO concentration) was added and the BRET measurement was started in a Mithras LB940 plate reader after 5 min (RLuc8 emission at 395 nm, GFP2 emission at 510 nm). The BRET2 ratio was calculated by division of the GFP2 signal by the RLuc8 signal. Data analysis was performed by means of GraphPad PRISM 8.0. netBRET values were calculated by subtraction of buffer control BRET2 ratio from the BRET2 ratio of each data point. Concentration-response curves were then fitted by a sigmoidal doseresponse curve with variable slope (four parameters) to yield EC 50 and E max values. Data was obtained in at least three independent experiments performed in duplicate.

Synthesis of carboxy-functionalized Preladenant analog 2
The synthesis of the precursor 2 is depicted in Scheme S1 below.

Synthesis of the side-chain (a) Benzyl 4-(4-hydroxyphenyl)piperazine-1-carboxylate (13)
N CbzN OH 4-(Piperazin-1-yl)phenol (12, 500 mg, 2.8 mmol, 1.0 eq.) is dissolved in 2.8 ml of methanol, and benzyl chloroformate (402 µl, 2.8 mmol, 1.0 eq.) is added dropwise. The reaction mixture is stirred for 12h at RT. Then, the solvent is removed in vacuo, and the residue is dissolved in 50 ml of a saturated aq. NaHCO 3 solution. After extraction with ethyl acetate (3 x 50 ml) the united organic phases are dried over MgSO 4 , and the solvent is subsequently removed in vacuo. The product is purified by column chromatography on silica gel with dichloromethane/methanol (95:5) yielding 661 mg (2.1 mmol) of 13 (yield: 76 %).   (1 ml). Under an atmosphere of argon (using an argon-filled ballon), 10% Pd/C (10 mg) is added. Then, the argon atmosphere is replaced by an atmosphere of hydrogen gas (H 2 ) again using a balloon (a T-connector is used for switching between the two gas-containing balloons), and the solution is stirred for 2h at RT. The solution is filtered over Celite ® , and the solvent is removed in vacuo. The product is obtained in 99% yield (67 mg, 0.23 mmol).

Synthesis of amino-substituted PEG derivatives 3a-3f
The following azido-functionalized PEG esters used as starting materials were synthesized as previously described [19,20] and reduced using Method A or Method B, respectively, to yield the desired amino-functionalized PEG derivatives.

Method A
The appropriate ester (1.0 eq.) is dissolved in dichloromethane (1 ml per 100 mg of ester) under an atmosphere of argon (using a ballon), and 10% Pd/C (10 mg per 100 mg of ester) is added. The argon atmosphere is exchanged for hydrogen gas (H 2 ) using a balloon, and the mixture is stirred for 8h at RT. Then, the solution is filtered through Celite ® , and the solvent is subsequently removed in vacuo. The product is purified by column chromatography on silica gel using 7N ammonia solution in methanol/dichloromethane as eluent; for details see below.