Identification of Two Distinct Sites for Antagonist and Biased Agonist Binding to the Human Chemokine Receptor CXCR3

Abstract

The chemokine receptor CXCR3 is a G protein-coupled receptor that conveys extracellular signals into cells by changing its conformation upon ligand binding. We previously hypothesized that small-molecule allosteric CXCR3-agonists do not bind to the same allosteric binding pocket as 8-azaquinazolinone-based negative allosteric modulators. We have now performed molecular-dynamics (MD) simulations with metadynamics enhanced sampling on the CXCR3 system to refine structures and binding modes and to predict the CXCR3-binding affinities of the biased allosteric agonist FAUC1036 and the negative allosteric modulator RAMX3. We have identified two distinct binding sites; a "shallow" and a second "deeper" pocket to which the biased allosteric agonist FAUC1036 and negative allosteric modulator RAMX3 bind, respectively.

An Efficient Metadynamics-Based Protocol To Model the Binding Affinity and the Transition State Ensemble of G-Protein-Coupled Receptor Ligands

“…Recent simulations on a variety of GPCRs have shown that the extracellular vestibule of the receptor can pre-orient the ligand [34][35][36][37][38][39] and thus provide a well-defined extracellular end-point for docking pathways, and thus simplifying the path-sampling task, often by a form of electrostatic focusing 40 but also by a simple mechanical effect in which part of the ligand is anchored, Additionally, to speed up the sampling, this approach can be easily combined with multiplereplica metadynamics approaches, such as parallel tempering 29 or the multiple-walker technique, 39 which uses several "walkers" to converge one free-energy profile. In our case, we found the trivially parallelizable (and asynchronous) multiple-walker approach more suitable as it is very flexible (the number of replicas can be dynamically changed to use the available resources) and ensures maximum computational efficiency even on non-homogenous clusters.…”

“…40 The ligand is first unbound using a faster protocol (see the Methods section for details). The position on the reaction coordinate at which the ligand is completely hydrated and unbound is determined and representative structures This convergence is due to multiple re-crossing events (Fig.…”

Differences between G‐Protein‐Stabilized Agonist–GPCR Complexes and their Nanobody‐Stabilized Equivalents

“…It has been pointed out [23] that targeting specific GPCR conformations represents the future of GPCR drug discovery.T he increasing recognition [24] of functional bias and the need to modulate GPCRs specifically are emerging as important aspects of GPCR pharmacology and drug discovery research. Given that the relevant highaffinity binding (active) state for GPCR agonists must be stabilized by complexation with the G-protein [2,[5][6][7][8][9][10][11][12][13][14][15][16][17] or arrestin, [25][26][27] refinement of X-ray structures by simulations on the biological system, or even extensive metadynamics predictions of free energies of binding, [11] will be necessary for effective structure-based drug design. Thes imulations can play another important role when designing antagonists, which often occupy ab inding site other than the orthosteric one preferred by agonists.…”

“…It has been pointed out [23] that targeting specific GPCR conformations represents the future of GPCR drug discovery.T he increasing recognition [24] of functional bias and the need to modulate GPCRs specifically are emerging as important aspects of GPCR pharmacology and drug discovery research. [11][12][13][14] Thus,d esign principles are slowly emerging for functionally selective GPCR ligands.F or agonists,t he orthosteric binding site exhibits considerable plasticity,s ot hat exact knowledge of the binding site for activating the individual pathways is necessary.F or antagonists,multiple binding sites are likely involved, and the relevant agonist binding sites may even be different for subtypes of the same receptor. Thes imulations can play another important role when designing antagonists, which often occupy ab inding site other than the orthosteric one preferred by agonists.…”

“…Both experimental studies [9,10] and atomistic molecular dynamics (MD) simulations have suggested multiple-step binding for different GPCRs. [11][12][13][14][15][16][17] Thee arlier binding sites along the binding path (those closer to the extracellular medium) are sometimes separated from the deeper orthosteric activation binding site by significant energy barriers.W e have previously combined unbiased MD simulations with metadynamics enhanced sampling to show that accurate freeenergy profiles for ligand binding can be determined computationally, [11][12][13][14] and we have described ac omputational method for these simulations with an analysis of the agreement of the calculated free energies of binding with experiment. [11] Root mean square deviations (RMSDs) between calculated free energies of binding and those derived from experimental binding constants are of the order of one kcal mol À1 for agonists,a ntagonists and reverse agonists,a nd binary and ternary complexes.Simulations that start from Xray structures for the receptor and those based on homology models equilibrated for several microseconds perform similarly.…”