Forced unbinding of GPR17 ligands from wild type and R255I mutant receptor models through a computational approach

Abstract

Background: GPR17 is a hybrid G-protein-coupled receptor (GPCR) activated by two unrelated ligand families, extracellular nucleotides and cysteinyl-leukotrienes (cysteinyl-LTs), and involved in brain damage and repair. Its exploitment as a target for novel neuro-reparative strategies depends on the elucidation of the molecular determinants driving binding of purinergic and leukotrienic ligands. Here, we applied docking and molecular dynamics simulations (MD) to analyse the binding and the forced unbinding of two GPR17 ligands (the endogenous purinergic agonist UDP and the leukotriene receptor antagonist pranlukast from both the wild-type (WT) receptor and a mutant model, where a basic residue hypothesized to be crucial for nucleotide binding had been mutated (R255I) to Ile.

Results: MD suggested that GPR17 nucleotide binding pocket is enclosed between the helical bundle and extracellular loop (EL) 2. The driving interaction involves R255 and the UDP phosphate moiety. To support this hypothesis, steered MD experiments showed that the energy required to unbind UDP is higher for the WT receptor than for R255I. Three potential binding sites for pranlukast where instead found and analysed. In one of its preferential docking conformations, pranlukast tetrazole group is close to R255 and phenyl rings are placed into a subpocket highly conserved among GPCRs. Pulling forces developed to break polar and aromatic interactions of pranlukast were comparable. No differences between the WT receptor and the R255I receptor were found for the unbinding of pranlukast.

Conclusions: These data thus suggest that, in contrast to which has been hypothesized for nucleotides, the lack of the R255 residue doesn't affect the binding of pranlukast a crucial role for R255 in binding of nucleotides to GPR17. Aromatic interactions are instead likely to play a predominant role in the recognition of pranlukast, suggesting that two different binding subsites are present on GPR17.

Figure 1

Superimposition of α-helical domains of the β₂ AR-Fab, β₂ AR-T4, β₂AR(E122W)-T4 β₁ AR and A_2A R structures to GPR17 model. Ribbon representation of the β₂AR-Fab (2R4R), β₂AR-T4 (2RH1), β₂AR(E122W)-T4 (3D4S), β₁AR (2VT4) and A_2AR structures after alignment of the α-helical domains to GPR17 model (in gray) are reported in cyan, orange, green, magenta and yellow respectively.

Figure 2

Macroscopic view of three best configurations of pranlukast docked to GPR17. The picture shows the three potential binding poses (CI, CII and CIII) obtained for the antagonist pranlukast (stick representation) on GPR17 (cartoon representation), by means of docking studies and 6 ns of molecular dynamic simulations. The chance of pranlukast to assume different and energetically comparable configurations, as for CI, CII and CIII, it is probably due to the high flexibility of the molecule yielding its high conformational freedom.

Figure 3

Comparison of the energetic profile of the MD simulations of the three docking configurations pranlukast. The plot shows the total energy profile as a function of time of the entire system membrane-receptor-ligand system during 6 ns of MD simulations, for the three different pranlukast configurations, here indicated as CI (in black), CII (in red) and CIII (in green).

Figure 4

Comparison of the RMSD of C-α and ligand atoms of the MD simulations of the three docking configurations of pranlukast. The plot a shows the RMSD of C-α atoms of the protein as a function of time obtained for the MD runs of CI (in black), CII (in red) and CIII (in green). In panel b, the same colour are used to indicate the RMSD versus time of ligand atoms obtained for CI, CII and CIII simulations.

Figure 5

Free energy estimate of the binding of pranlukast. The free energy estimate of the binding of pranlukast for the 6 ns of MD simulations performed for the three different docking configurations is reported in black for CI, in red for CII and in green for CIII.

Figure 6

Model of the pranlukast conformation CII. Model of the complex formed by pranlukast and GPR17 after 6 ns of conventional MD simulation. Pranlukast is displayed in orange within the detailed binding pocket.

Figure 7

Superimposition of UDP and pranlukast within the putative GPR17 binding pockets. The picture shows the superimposition of UDP (in red) with the three docking poses of pranlukast (CI, in blue; CII, in green; CIII, in gray). For each simulation, the correspondent target residue R255, here highlighted with arrows, is reported in the same colors of either UDP or pranlukast. The hypothetical interactions between R255 and ligands are represented by white dotted lines.

Figure 8

Forced unbinding profile of pranlukast. Panel a and b compare the unbinding simulations of pranlukast from the WT (in red) and the R255I (in black) receptor models: panel a shows the work developed to unbind pranlukast; panel b shows the displacement of the COM of the ligand from its starting position. Panel c and d show the distances between groups of atoms of the ligand that form polar or hydrophobic interactions with atoms of the WT or the R255I models, respectively.

Figure 9

EL2 dynamical behaviour. Representative frames of the open and closed form of EL2, extracted from the SMD simulation, are reported in green and red, respectively. The picture shows a detailed view of the hinge movement of the loop that exhibits an extension up to 6.6 Å.