The amino hydroxy methyl isoxazolepropionic acid
The α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic Fructose synthesis (AMPA) receptors are related ionotropic glutamate receptors (iGluRs) that mediate fast excitatory neurotransmission, and are also typically heteromeric ion channels, composed of GluA2 subunits in conjugation with GluA1, GluA3, or GluA4 subunits (Isaac et al., 2007, Herguedas et al., 2016). Like NMDA receptors, they have functionally modular domains separated into the ATDs, LBDs, TMD, and CTDs, but differ in that activation of the channel requires binding of only a single agonist, glutamate, to the separate LBDs. Recent computational and electrophysiology studies have elucidated the binding mechanisms of glutamate to the AMPA receptor and found binding intermediates that assist the process of ligand binding (Yu and Lau, 2017, Yu et al., 2018).
Molecular simulation has proven useful in describing the dynamics and energetics of NMDA receptor LBD cleft closure and the potentiating role of glycans (Yao et al., 2013, Dai and Zhou, 2013, Dai and Zhou, 2015, Dai and Zhou, 2016, Dutta et al., 2015, Omotuyi and Ueda, 2015, Sinitskiy and Pande, 2017, Sinitskiy et al., 2017). Here, using an aggregate of ∼75 μs of unbiased all-atom molecular dynamics (MD) simulations, we investigate the molecular mechanisms of binding for both glutamate and glycine to the LBDs of the NMDA receptor subunits, GluN2A and GluN1. We directly simulate glutamate binding to the GluN2A LBD, as well as glycine binding to the GluN1 LBD, and subsequent cleft closure. Analysis of the cross-lobe interactions formed during cleft closure suggests these interactions may contribute to previously observed dynamical motions (Yao et al., 2013). Free energy landscapes indicate that the dynamic mechanisms responsible for glutamate binding are distinct from that of glycine binding in NMDA receptors.
Discussion In the present study, we simulated glutamate binding to the GluN2A LBD and glycine binding to the GluN1 LBD. In our simulations, glycine binds in the same orientation seen in crystal structures, whereas glutamate binds in an inverted pose. This observation does not contradict glutamate binding in a non-inverted pose; it simply suggests an inverted pose is possible. Pre-binding intermediates for glutamate were able to interconvert between a crystal structure orientation, in which the α-carboxylate contacts R518 (in lobe 1 of GluN2A), and an inverted orientation, in which the γ-carboxylate contacts R518. It is unknown whether, or to what extent, NMDA receptors can be activated by glutamate bound in the inverted conformation. In addition to the amount of cleft closure, the orientation of the agonist within the binding pocket may also be a factor in activation. This study complements prior studies that examined the free energy of conformational change between ligand-bound and apo GluN1 and GluN2A LBDs (Yao et al., 2013, Dai and Zhou, 2015). The binding mechanisms in NMDA receptors for glutamate and its co-agonist, glycine, are distinct. Glutamate-binding PMFs for GluN2A and GluA2 feature continuous densities linking the periphery of the LBD to the binding pocket. The glycine-binding PMF for GluN1, on the other hand, features disconnected density between the periphery and the pocket. These distinct binding mechanisms serve as examples from two opposing paradigms for protein-ligand binding, one in which peripheral residue-ligand interactions may contribute substantially to the binding process, and the other in which protein-ligand interactions at the binding site are the only important factors in ligand binding. Glutamate binding to GluN2A is representative of the former; binding occurs along several preferred pathways, whereby positively charged residues mediate protein-ligand interactions at the periphery of the LBD and assist binding via a guided-diffusion mechanism. These binding intermediates are similar to those found in AMPA receptors, and electrophysiological recordings of AMPA receptor mutants lacking these metastable interactions exhibit slowed activation and deactivation kinetics (Yu et al., 2018). In contrast, glycine binding to GluN1 proceeds via an unguided-diffusion mechanism, where binding is largely a random diffusive process, and the most important protein-ligand interactions are those made between the lobe 1 binding site residues and the ligand. It is possible that GluN2A LBDs behave similarly to AMPA receptors when metastable interactions between the protein and ligand are abolished, whereas GluN1 LBDs are more resistant to such alterations.