The best version of A2AR protein currently available on the market is truncated at the C-terminus (96 amino-acids removed) and heavily mutated (8 mutations). The challenge that CALIXAR has been able to tackle is to produce the full-length and wild-type version of this membrane protein. Conditions to express A2AR were obtained in insect cells (Sf9).
Using a combination of DDM/CHS and a calixarene based detergent, good solubilization yield (~90%) was obtained. Most of A2AR could bind to the Talon-His column and eluted specifically with a good purity (>90%). Higher molecular weight gel migration of A2AR at ~80 kDa was observed. This corresponds most probably to SDS-resistant dimers since protein samples were not heated to avoid aggregation.
Then Radioligand binding of purified A2AR was assessed using ZM241385 (antagonist) and CGS-21680 (agonist). The Kd indicated here are very similar to those of Sf9 membrane bound forms.
These data show that purified A2AR has maintained its ligand binding properties during the expression, solubilization and purification processes. In addition to that, purified A2AR was able to bind to agonist and antagonist compounds. This was not the case for the well-known StaR construct (8 mutations and C-terminal truncation) that was heavily reported in the literature. Indeed, such thermostabilized mutant was locked in antagonist conformation and was not able to bind to agonist compounds.
To assess the behavior of purified A2AR in solution, we loaded his-tag affinity purified A2AR on a gel filtration column. Size exclusion chromatography reveals that our purified native A2AR was not aggregated. This was also the case after at least 7 days at room temperature, illustrating therefore that native A2AR was stabilized thanks to the use of calixarene (CALX) based detergent.
To confirm A2AR stability we performed a Western Blot-based thermal shift assay. This assay relies on the assumption that unstable heated proteins will aggregate after ultracentrifugation; the band intensity (Western Blot) corresponding to the protein will decay proportionally to its instability. The result shown here indicates that using CALX-R10/ DDM condition (in the presence of antagonist ZM241385), A2AR exhibits a Tm (melting temperature) of ~55°C. The same A2AR is less stable in DDM with a Tm of ~43°C as previously reported. As a comparative study, we have expressed A2ARStaR2, solubilized it using DDM/CHS and submitted it to the same thermal shift assay.
Here we show a 4°C higher stability of A2AR StaR2 in comparison to A2AR wild-type and full length solubilized using CALX-R10/ DDM. This is relatively minor considering that StaR2 contains 8 points mutations and a 96 amino acids truncation in the C-terminus. More importantly, such stabilized construct was not able to bind to agonistic compounds such as CGS21860 or NECA as reported above. Thus, we could stabilize native, glycosylated, non-aggregated and homogenous A2AR that maintained its ligand binding capacity.
STD experiment, which is a well-established NMR method for fragment screening against soluble therapeutic targets, was not yet been used against purified GPCRs. Thus, we first wanted to demonstrate the feasibility of the approach through the binding investigation of known antagonists and agonists to A2AR. Here, we showed that the STD binding signal of caffeine bound to A2AR. By comparison, in the absence of the protein, the STD signal was considerably weaker, showing that the unspecific binding of caffeine to the micelles was insignificant.
One expected advantage of the preparation of native A2AR is the possibility to observe the binding of agonists since the conformational flexibility of the receptor is not constrained in such a preparation. We had therefore investigated the binding of adenosine to A2AR. Here we showed the STD spectrum of adenosine bound to A2AR. As for the caffeine, the STD signalswere significantly weaker in the control experiment performed in the absence of the receptor.
The same experiment was performed with CGS-21680 showing binding to pure native A2AR.
A competition experiment was achieved by adding the agonist compound ZM-241385. The STD intensities of adenosine decreased in the presence of ZM-241385, showing the competition between adenosine and ZM-241385 that both bound in the same binding pocket. The observation of the binding of agonists and antagonists to native A2AR as well as the competition experiments demonstrated that the fragment screening can be achieved using STD-NMR on the A2AR preparation.
We then performed fragment screening against A2AR using a hundred fragments. The molecules were screened into mixtures of 5 to 10, as typically done with soluble proteins. Fragments were then classified into three groups displaying strong binding, weak binding or no binding, depending on the intensity of the STD signals observed. Nineteen fragments (19%) were shown to exhibit significant STD intensities upon A2AR binding. To further analyse the fragment screening results, cAMP cell-based assay was performed on ten fragments displaying either strong (fragments 4, 10, 12, 13, 14, 15) or weak binding (fragments 6, 7, 8 and 11)
We then tested the capacity of the fragment binders to induce cAMP production in HEK293 cells stably expressing A2AR. CGS21680 titration curves showed that cAMP production was A2AR expression dependent (A). Non-transfected HEK293 cells were used as negative control and a small increase in cAMP production was observed associated to high concentrations of CGS21680, probably due to the known presence of endogenous A2AR.
We then investigated the potential agonistic effect of the fragments. Stimulation of the A2AR cell line for 30 minutes at room temperature had no effect on cAMP production even at the highest concentration of 10 mM (B). CGS21680 and adenosine served as positive controls and showed robust agonistic effects as expected.
Accordingly, the well-established A2AR antagonist ZM 241385 did not show any effect in this agonistic assay, while its efficiency in inhibiting CGS21680-induced increase in cAMP production was confirmed (C). We then investigated the antagonistic effect of the fragments in the cAMP signaling assay. This test was performed by pre-incubating the A2AR cell line with the fragments (15 min, room temperature), followed by addition of the CGS21680 agonist (30 min, room temperature).
Fragments, 4, 10, 11, 12, 14 and 15 behaved as full antagonists at 10 mM. whereas fragment 6, 7, 8 and 13 were without effect. This result confirmed the value of combining NMR-STD experiments and cell-based assays to discover functionally relevant fragments. Thus, using stabilized native A2AR, we could identify fragments with antagonistic effects on A2AR.
The STD intensities of adenosine bound to A2AR not only were weaker upon addition of the agonist CGS-21680, but the profile of the STD intensities was also modified. Notably, the STD signal of the proton of the adenosine ribose moiety at 5.9 ppm was considerably smaller when CGS-21680 binds A2AR. This indicated that the adenosine ribose moiety was buried in A2AR in the absence of CGS-21680, while it is solvent-exposed when it bound A2AR in the presence of CGS-21680.
This observation corroborated with previous investigation of the binding mechanism of GPCR ligands using molecular dynamics simulation, showing the presence of transient binding sites also called metastable binding sites or ligand-entry sites as potential allosteric sites. A metastable binding site was proposed for adenosine bound to A2AR. It was shown that adenosine could bind at the entrance of the orthosteric binding site, with the ribose oriented towards the entrance, solvent-exposed, in agreement with the NMR observation. These results showed that the benefit of the STD-NMR experiment was to provide structural information for ligands bound to the receptor in the presence or absence of other compounds.
Side by side mice immunization trials were performed using the native A2AR as well as the StaR version (8 mutations and 96 aminoacids truncation). A significant signal was observed by ELISA when the serum of mice immunized by native A2AR. No signal at all was observed for the staR version. This indicates that the native A2AR was better immunogen than the mutant. A successful GPCR antibody discovery program did benefit from that since functional antibody against extracellular loops were generated (undisclosed).