Microsecond simulations indicate that ethanol binds between subunits and could stabilize an open-state model of a glycine receptor

Abstract

Cys-loop receptors constitute a superfamily of ion channels gated by ligands such as acetylcholine, serotonin, glycine, and γ-aminobutyric acid. All of these receptors are thought to share structural characteristics, but due to high sequence variation and limited structure availability, our knowledge about allosteric binding sites is still limited. These sites are frequent targets of anesthetic and alcohol molecules, and are of high pharmacological importance. We used molecular simulations to study ethanol binding and equilibrium exchange for the homomeric α1 glycine receptor (GlyRα1), modeled on the structure of the Gloeobacter violaceus pentameric ligand-gated channel. Ethanol has a well-known potentiating effect and can be used in high concentrations. By performing two microsecond-scale simulations of GlyR with/without ethanol, we were able to observe spontaneous binding in cavities and equilibrium ligand exchange. Of interest, it appears that there are ethanol-binding sites both between and within the GlyR transmembrane subunits, with the intersubunit site having the highest occupancy and slowest exchange (∼200 ns). This model site involves several residues that were previously identified via mutations as being crucial for potentiation. Finally, ethanol appears to stabilize the GlyR model built on a presumably open form of the ligand-gated channel. This stabilization could help explain the effects of allosteric ligand binding in Cys-loop receptors.

Figure 1

GlyRα1 embedded in a DOPC lipid bilayer after 200 ns of simulation. GlyRα1 is represented in cartoon. Water and DOPC molecules are displayed as van der Waals spheres, Na⁺ and Cl⁻ ions are shown as single spheres, and the larger ethanol molecules are shown in space-filling representation.

Figure 2

RMSD of GlyRα1 from the initial homology model structure. The black and gray lines represent the GlyRα1 with and without ethanol, respectively. The solid lines indicate the C_α RMSD of the whole protein, and the dashed lines are the C_α RMSD of the TMD, including residues 220–410 (the TM3-TM4 loop was replaced with three glycines).

Figure 3

Effects of ethanol on the (A) quaternary arrangement and (B) TM2 tilt, computed as averages and standard error estimates from four shorter simulations, each with (black) and without (gray) ethanol. Horizontal lines indicate estimates for open (solid) and closed (dashed) structures based on the prokaryotic GLIC and ELIC templates. For both properties, the nonethanol systems exhibit more drift, occasionally even being closer to the ELIC, whereas the ethanol ones stay closer to the initial GLIC-based structure.

Figure 4

Correlation between the average number of ethanols present per site versus the instantaneous amount of water and volume of the allosteric binding site. Crosses indicate the number of waters (left axis) in a cavity as a function of the number of ethanols in the same site, from 1-ns averages. The solid black line indicates the average for each amount of ethanol, with standard deviations. A linear fit (dotted black) shows that each ethanol replaces ∼1.5 waters. The average van der Waals volume of the cavity (gray dashed curve, right axis) also exhibits a clear increase with the average number of ethanols present in the cavity.

Figure 5

Binding site. Conformations of a single ethanol molecule over 229 ns are superimposed on the structure (from 176 to 405 ns) as semitransparent sticks. Dashed thin lines show hydrogen bonds between this ethanol and S267.

Figure 6

Ethanol binding and exchange. (A) Representation of GlyRα1 TMD showing the potential allosteric intersubunit binding site (site 1), an intrasubunit site (site 2), a pore-accessible site (site 3), and a surface/membrane-facing site (site 4). (B) Average occupancy of ethanol molecules per single binding site over the simulation, colored according to the site. The black line shows the sum of all deep inter- and intrasubunit sites (occupancy clearly is highest in the intersubunit site). (C) Autocorrelation occupancy/nonoccupancy of ethanol molecules that exhibit occupancy at some point in the simulation, colored according to the site. An exponential fit indicates exchange times of ∼200 ns for the deep binding sites. The average exchange time for the surface sites is <10 ns, but there is also a small component (<20%) with an ∼55 ns exchange time for these sites.