Role of Extracellular Loops and Membrane Lipids for Ligand Recognition in the Neuronal Adenosine Receptor Type 2A: An Enhanced Sampling Simulation Study

Human G-protein coupled receptors (GPCRs) are important targets for pharmaceutical intervention against neurological diseases. Here, we use molecular simulation to investigate the key step in ligand recognition governed by the extracellular domains in the neuronal adenosine receptor type 2A (hA2AR), a target for neuroprotective compounds. The ligand is the high-affinity antagonist (4-(2-(7-amino-2-(furan-2-yl)-[1,2,4]triazolo[1,5-a][1,3,5]triazin-5-ylamino)ethyl)phenol), embedded in a neuronal membrane mimic environment. Free energy calculations, based on well-tempered metadynamics, reproduce the experimentally measured binding affinity. The results are consistent with the available mutagenesis studies. The calculations identify a vestibular binding site, where lipids molecules can actively participate to stabilize ligand binding. Bioinformatic analyses suggest that such vestibular binding site and, in particular, the second extracellular loop, might drive the ligand toward the orthosteric binding pocket, possibly by allosteric modulation. Taken together, these findings point to a fundamental role of the interaction between extracellular loops and membrane lipids for ligands’ molecular recognition and ligand design in hA2AR.


SUPPORTING INFORMATION
Section S1: Well-tempered Metadynamics Well-tempered metadynamics (WTM) [1,2] works by introducing of a history-dependent potential V acting on a selected number of slow degrees of freedom, the collective variables (CVs). This forces the dynamics to explore conformations that were not previously visited and discourages the system from returning to these regions. Therefore, it allows the system escaping minima along low free energy paths and exploring other minima in the free energy landscape. This occurs independently of the minima one starts from. From the potential V, one can calculate the free energy.
Let us consider only one CV (s hereafter) for simplicity. The free energy surface along s, F(s), can be then calculated from the equation: In the well-tempered formulation [1] (the ones implemented here), the potential used to bias the dynamics in order to accelerate the sampling is: where ω has the dimension of an energy rate, ∆T is a temperature and N(s, t) comes from the biased simulation. As pointed out by the authors [1], an important property of this formulation is that, it insures the bias eventually to converge, yet "slow enough for the final result not to depend on the initial condition V (s, 0)." Section S2: H264 7.29 -E169 ECL2 salt bridge: intramolecular interactions. H264 7.29 is predicted to be doubly protonated by PROPKA [3] at neutral pH. Hence, it is expected to form a salt-bridge with E169 ECL2 . 18 out of 19 X-ray hA2AR/ligand structures at pH<7.6 feature this putative salt bridge (see Table S1). The only exception occurs with the agonist UK43907 (PDBid: 3QAK). Here, because of steric reasons, the bulky ring of UK43907 causes a displacement of H264 7.29 and E169 ECL2 , with an increase of 3~4 Å of the Cα-Cα distance). 1 The salt bridge might behave as a 'gate' closing the ligand in the inner binding site and separating it from the vestibular binding site [4][5][6].
The following analysis shows that the two groups are fairly conserved across all A2ARs, while this is not the case for all the other human adenosine receptors.
Results. H264 7.29 is replaced by R and V for the 33% and 17% of the cases, respectively ( Figure  S7). In the R264 variants, the residues in position 169 are always E. Hence, the putative salt bridge (either formed by E169 and H264, or by E169 and R264) is conserved for slightly more than 60% across all A2ARs.
The pairwise sequence identities of the other human adenosine receptors (hA1R, hA2BR, and hA3R) with A2AR ranges from 58% to 39% ( Figure S8). Position 169 is E hA1R and hA2BR while it is V in hA3R (see Figure S7). Position 264 is H in hA1R, E in hA3R, N in hA2BR. Here, N is embedded between two K's in a "KNK" motif.
The E169 ECL2 -H264 7.29 positions are not conserved across all A2BRs and A3Rs ( Figure S10-S11). It may be replaced by residues which could form H-bond interactions (as opposed to a salt bridge) or a salt bridge other than that formed here: indeed, in the case ofA3Rs, the E169/K264 pair is present in 30% of the sequences.

Section S4: Sodium allosteric binding site.
This site has a fundamental importance in allosteric modulation of GPCRs [9,10]. The residues that mostly contribute to sodium binding along the GPCRs family, i.e S 3.39 , N 7.45 and D 2.50 are described in literature as highly conserved. Position 2.50, in particular is an aspartic acid in 90% of the eukaryotic GPCRs, accordingly to our alignments taken from a curated multiple sequence alignment from GPCRdb. This residue can highly modulate the function of GPCRs. Indeed, the role of sodium modulation is well known for several GPCRs [10]. Mutagenesis studies on residues involved in the sodium ion coordination, and in particular D 2.50 , highlighted the different effects that allosteric sodium may have in various class A GPCRs signaling [9]. Indeed, D 2.50 replacement with uncharged amino acids can drastically reduce the agonist-induced G protein activation [11,12] [13] [14,15] or modulate the allosteric effect of the G-protein on ligand binding [16]. The presence of sodium ions in the allosteric cavity can also exert different effects on the constitutive signaling of GPCRs [17,18]. In many cases, the presence of bound sodium seems to stabilize the inactive conformation of the receptor reducing the constitutive G-protein [13,15], whereas in other receptors the substitution of sodium coordinating D 2.50 abolishes the constitutive G-protein coupling and activation without affecting the agonist-stimulated activity [19]. Exhaustive studies have also revealed that the sodium pocket collapses due to the activation-related movements of the transmembrane helixes [18]. Recently, we have shown that the disruption of the sodium binding site of GPR3 strongly biased the receptor to the inactive state [20]. Thus, most of the studies agree with the fact that the constitutive activity can be dramatically affected by mutations in this cavity. Indeed, a constitutive active mutant (CAM) on human mu-opioid receptors, has been observed to disrupt the allosteric sodium binding cavity, favoring the exploration of active-like conformations even in the apo state [21]. In particular, for A2A receptors, very recently, White and collaborators [22] obtained the crystal structures of agonist complexes for two variants sodium binding site, D 52 2.50 N and S91 3.39 A. In both cases the structures are active-like but, the variants induce important changes in the activation motif NPxxY. The authors, combining several experimental techniques provide a basis for understanding the role of the sodium-coordinating residues on stability and G-protein signaling.

Tables
Table S1. H264 7.29 protonation state across hA2AR X-ray structures. The table reports: the resolution of the X-ray structures, the type of ligand (antagonists and agonists are colored in red or blue respectively), the pH of crystallization, the protonation state of H264 7.29 at corresponding crystallization pH value (H264 7.29 protonation), as predicted using PROPKA [3], the presence of H264 7.29 and E169 ECL2 salt-bridge/HB interactions and H264 7.29 CA -E169 ECL2 CA distances (Å ).  Table S4. Amino acid coevolution profile computed using the Coeviz tool within the web server polyview-2d [41]. The chi-squared covariance, weighted by phylogeny derived from alignments of hA2AR sequence (as defined in PDBid 3PWH [29]) against NCBI NR database with 90% identity [41].
For each amino acid, the genetic number and binding site location are annotated.  Table S5. Presence of residue coevolution between orthosteric binding site (OBS) and extracellular loops (ECLs) of human receptors in class A, B, C and F. X-ray structures of 27 human GPCRs with OBS-bound ligand were used for evolutionary correlation analysis with Coeviz tool in polyview-2d

VBS
webserver [41]. The threshold of PCS score was chosen as >0.3 to identify evolutionarily correlated residue pairs. Among them, 22 GPCRs have residue-based ECL-OBS coevolution relation. '0" ECL is missing in X-ray structure; "Yes" presence of residue-based evolutionary correlation; "No" absence of residue-based evolutionary correlation.       Figure 1 are shown in A and B panels, respectively, as 3D representation. The protein backbone is rendered as cartoon, ZMA is shown as a green licorice, residues interacting with ZMA are shown as grey lines. The E169 ECL2 and H264 7.29 residues are shown in cyan licorice. Hydrogen, oxygen and nitrogen atoms are specifically colored in white, red and light blue, respectively. C-D) 2D scheme of binding poses in A) and B), respectively.

Figure S4
Superimposition of hA2AR representative structure in the minima B (yellow tube) and C (cyan tube). ZMA is shown in yellow and cyan licorice representation for B and C minima, respectively. Figure S5. Conservation of solvent-exposed motif of ECL2 in human Adenosine receptor subfamily.
This multiple sequence alignment was generated using the web server of CLUSTAL O (1.2.2) [7].
Amino acid residues are colored according to this scheme: small and hydrophobic residues including aromatic residues are colored in red, acidic residues are colored in blue, basic residues are colored in magenta, and hydroxyl, sulfhydryl, amine residues and glycine are colored in green. Same alignment method and coloring schemes are applied in the following Figures S6-11. Figure S6. Conservation of solvent-exposed motif of ECL2 in Adenosine receptor A2R across different species. Color-code and alignment method as in Figure S5. Figure S7. Conservation of H264 7.29 and E169 ECL2 in Adenosine receptor A2AR across different species. Color-code and alignment method as in Figure S5. Figure S8. Conservation of H264 7.29 and E169 ECL2 in human Adenosine receptor subtypes hA1R, hA2AR, hA2BR, hA3R. Color-code and alignment method as in Figure S5. Figure S9. Conservation of H264 7.29 and E169 ECL2 in Adenosine receptor A1R across different species.
Color-code and alignment method as in Figure S5. Figure S10. Conservation of H264 7.29 and E169 ECL2 in Adenosine receptor A2BR across different species.
Color-code and alignment method as in Figure S5. Figure S11. Conservation of H264 7.29 and E169 ECL2 in Adenosine receptor A3R across different species. Color-code and alignment method as in Figure S5.