GABA receptor associated protein changes the electrostatic environment around the GABA type A receptor

Abstract We have performed fully atomistic molecular dynamics simulations of the intracellular domain of a model of the GABAA receptor with and without the GABA receptor associated protein (GABARAP) bound. We have also calculated the electrostatic potential due to the receptor, in the absence and presence of GABARAP. We find that GABARAP binding changes the electrostatic properties around the GABAA receptor and could lead to increased conductivity of chloride ions through the receptor. We also find that ion motions that would result in conducting currents are observed nearly twice as often when GABARAP binds. These results are consistent with data from electrophysiological experiments.


| INTRODUCTION
The family of GABA A receptors is responsible for the majority of fast neuronal inhibition in the mammalian central nervous system, and is a target of general anesthetics, benzodiazepines, barbiturates and neurosteroids. These pentameric proteins belong to the cys-loop family of ligand-gated ion channels that includes the nicotinic acetylcholine, glycine, and 5HT 3 receptors. The GABA A receptors are composed of five subunits arranged pseudosymmetrically around the central ion channel. 1 The subunits, of which 19 have thus far been identified, are separated into classes based on their sequence similarity: there are six α-subunits, three β, three γ, three ρ, and single representatives of δ, ϵ, θ, and π. 2 The precise subunit isoform composition of the pentamer defines the recognition and biophysical characteristics of the particular receptor subtype. The most ubiquitous subtype, which accounts for approximately 30% of GABA A receptors in the mammalian brain, 3 contains two α 1 -, two β 2 -, and a single γ 2 -subunit. 4 The GABA A receptors can be divided into three structural domains, the extracellular (EC) domain, the transmembrane (TM) domain, and the intracellular (IC) domain. When GABA binds to the GABA A receptor, the central ion channel opens to let chloride ions through. 5 This opening is quickly followed by a period of desensitization of the receptor. 6 GABA A receptors with a γ-subunit are often associated with the GABA A -receptor associated protein, GABARAP. This protein, first described by Wang et al., 7 consists of 117 amino acids and has a relative molecular mass of 13 900. Experimental work 7,8 shows that it binds to the intracellular domain of the γ 2 -subunit of the GABA A receptor. Its function is most probably twofold: anchoring the GABA A receptor to the cytoskeleton, and modulating the function of the receptor. Amino acids near the N-terminal of GABARAP could bind to tubulin, 9 whilst the amino acids nearer the C-terminal bind to the GABA A receptor. 8 Moreover, Chen et al. 10 showed that GABARAP caused GABA A receptor clustering, and clustered receptors exhibited lower affinity for GABA (EC 50 increased from 5.74 ± 1.4 μM to 20.27 ± 3.8 μM), and they Benedict W. J. Irwin and Clara C. Wanjura have contributed equally to this study. desensitized less quickly (the desensitisation time constant τ increased from 1 to 2 s). Luu et al. 11 show that GABARAP binding increases the conductance of the GABA A receptor from below 40 pS to above 50 pS, and the mean opening times from about 2 ms to about 6 ms. For the GABARAP structure, we use dock 54a of structure 15 of the NMR solution structure (PDB code: 1KOT 15 ) from previous study. 12 Figure 1 shows the interaction between GABARAP and the GABA A receptor intracellular pentahelix viewed from the extracellular space towards the cytoplasm. Figure 2 shows the interaction between GABARAP and the GABA A receptor intracellular pentahelix from the side, with two amino acids from GABARAP and two amino acids from the pentahelix labeled.

| Molecular dynamics simulation of GABARAP and intracellular helices
The systems were prepared using the CHARMM-GUI freely available on the web. 16 The molecular dynamics package NAMD 2 17 was used in this study. We used the CHARMM potential for all our simulations. 18 Each system was minimized for 10 000 steps with all the protein atoms frozen. Molecular dynamics at 310 K was initialized for 10 000 time- Only the intracellular helices of the GABA A receptor is shown in this diagram. GABARAP is shown in cyan, the intracellular helix of the γ 2 -subunit in red, that of the α 1 -subunit in yellow and the helix of the β 2 -subunit is shown in green lengthened to 2 fs over 30 000 time-steps, during which period all main-chain nitrogen atoms of the three helices were tethered with a force constant of 2 kJ/mol/Å 2 . These helices are part of a large protein and the helical structures are stabilized by neighboring structures, some of unknown configuration. In this study, we included only the helices and so to stabilize them, we imposed the tethers. A 50-ns equilibration was carried out on the initialized system, followed by a data collection period of 100 ns. Equilibration was confirmed by a stable r.m.s. deviation from the starting structure and, in the case of the pentahelix/GABARAP complex, a stable intermolecular distance. Configurations were output every 20 ps. A convex hull was created using the following 10 amino acids as vertices (they are at the end points of the five intracellular helices) using a previously developed method 19  We tracked the trajectories of the ions to locate movements which are similar to Cl À ion movements when the GABA A receptor is conducting naturally. We define a "natural" ion movement where the Cl À ion moves into the convex hull from the membrane side across the plane where z $ 20 Å and where it exits from one of the five side portals at positions where z < 15 Å (see Figure 3); previous experiments show that these side portals are the exit routes for ions. 21

| Evaluation of electrostatic potential
We calculated the electrostatic potential around the GABA A receptor intracellular pentahelix, in the absence and presence of GABARAP.
From the 100-ns data production run of the molecular dynamics simulation, we took a configuration at every 10 ns to obtain 10 configurations. The water molecules and ions were removed from these configuration and, for each configuration, we calculated the electrostatic potential due to the CHARMM partial charges 18 on the protein atoms using simple Coulombic interactions; the dielectric constant was taken as one and nonperiodic boundary conditions were applied.
We then averaged the potential over the 10 configurations and compared them in the absence and presence of GABARAP.

| Molecular dynamics simulations
The GABA A receptor intracellular pentahelix atoms moved little during the course of the 100 ns data collection simulation, as they were tethered. GABARAP was not tethered, but it stayed in close proximity of the pentahelix. The volume enclosed by the amino acids at the end points of the intracellular helices were calculated using a previously developed method 22    We observed ions moving from the membrane side of the convex hull, through the hull of the pentahelix, then exiting from one of the five portals on the side, at positions where z < 15 Å. Figure 6 shows examples of such movements. Note that these ion passage trajectories usually last <1 ns, and they are short events on the timescale of the simulation. We observed 32 such events when GABARAP was absent but 60 such events when GABARAP was present during the 100 ns molecular dynamics simulations.

| Electrostatic potential
We visualize the electrostatic potentials due to the protein(s) by displaying the values on different planes using a color-coded scheme. In Figure 7, the electrostatic potential is displayed on a plane perpendicular to the central axis of the receptor. In the absence of GABARAP, the electrostatic potential is more positive in the region around the β 2 -subunits. In the presence of GABARAP, there is a finger-like "extension" of more positive electrostatic potential through the slit made by the β 2 -and γ 2 -subunits next to the bound GABARAP. The region over which the electrostatic potential increases is largely outside the pentahelix.
In Figures 8 and 9, the electrostatic potential is displayed on five planes through each of the five slits formed by the GABA A receptor intracellular helices. In Figure 8, the electrostatic potential due to the intracellular helices alone is displayed. In Figure 9, the electrostatic potential due to the intracellular helices and GABARAP is shown. It can be seen that the effect of GABARAP on the electrostatic potential in planes (B) and (C) is small. However, in planes (A), (D) and (E), the electrostatic potential is more positive in the region outside the intracellular helices (Figure 9). To make it easier to visualize these changes in electrostatic potential, we plot the difference potential in Figure 10; this is the difference in electrostatic potential between the case where GABARAP is absent and the case where GABARAP is present. A positive difference means that the electrostatic potential in the presence of GABARAP is more positive than in its absence. It can be seen from Figure 10 that most regions outside the pentahelix become electrostatically more positive due to the presence of GABARAP, but some regions towards the cytoplasmic end inside the pentahelix become more negative. We suggest that this increase in electrostatic potential outside the receptor with a concomitant decrease in potential inside the receptor leads to the increase in Cl À ion conductance.

| DISCUSSION
Cys-loop ligand-gated ion channels often interact with cytoplasmic proteins, and this interaction serves many purposes, amongst them the clustering of ion channels and the modulation of channel function. interacts with the cytoplasmic protein rapsyn. Rapsyn has a molecular weight of about 43 000, 23 and electron microscopy showed that the nAChR are interconnected by rapsyn dimers. Up to three rapsyn dimers can contact each nAChR in specific regions in the nAChR intracellular domain. 24 This tight network probably anchors the nAChR in the plane of the cell membrane and allows nAChR to be concentrated at the neuromuscular junction motor end-plate. 24 Another example is gephyrin. This protein was first identified as a bridge between the glycine receptor and tubulin. 25 Sola et al. 26  Gephyrin also interacts with the GABA A receptor through the receptor α 2 -subunit 27 and α 3 -subunit. 28 It is unclear if gephyrin binds the α 1 -subunit of the GABA A receptor; some experiments failed to show any interaction, 29 but others showed a weak interaction. 30 Maric et al. 31 co-crystallized segments of the α 3 -subunit of the GABA A receptor with segments of gephyrin, and identified the undecapeptide T 367 FNIVGTTYPIN 381 from the GABA A receptor as important for interaction with gephyrin. They showed that there were similarities between the binding of the GABA A receptor and of the glycine receptor to gephyrin: in particular, T 367 FNIVGTT 374 from the GABA A receptor, and F 398 SIVGSL 404 from the glycine receptor β-subunit adopted similar conformations.
In addition to gephyrin, the GABA A receptor also interacts with collybistin; there are two types of collybistin, which consist of 413 and 493 amino acids, respectively. 32 Saiepour et al. 29 showed that collybistin interacted with the intracellular domain of the α 2 -subunit of the GABA A receptor, and its binding site for the α 2 -subunit overlapped that for gephyrin. Collybistin was later shown to be important for clustering gephyrin and the GABA A receptor. 33 The GABA A receptor also interacts with GABARAP. GABARAP binds specifically to the γ 2 -subunit of the GABA A receptor. Binding of that GABARAP is not involved in altering GABA A receptor F I G U R E 8 Electrostatic potential around GABA A receptor intracellular helices. The α 1 -subunit is shown in yellow, the β 2 -subunit is shown in green and the γ 2 -subunit is shown in red. The top left panel shows five planes, each cutting through one of the five slits formed by the helices. The electrostatic potential due to the protein alone is calculated and displayed in a color-coded scheme. Panels (A-E) show the electrostatic potential on the five planes F I G U R E 9 Electrostatic potential around GABA A receptor intracellular helices and GABARAP. The α 1 -subunit is shown in yellow, the β 2 -subunit is shown in green, the γ 2 -subunit is shown in red and GABARAP shown in cyan. The top left panel shows five planes, each cutting through one of the five slits formed by the helices. The electrostatic potential due to the proteins is calculated and displayed in a colorcoded scheme. Panels (A-E) show the electrostatic potential on the same five planes as in Figure 8 F I G U R E 1 0 Difference electrostatic potential (the difference in electrostatic potential between the case where GABARAP is absent and the case where GABARAP is present) around GABA A receptor intracellular helices. The α 1 -subunit is shown in yellow, the β 2 -subunit is shown in green, the γ 2 -subunit is shown in red and GABARAP shown in cyan. The top left panel shows five planes, each cutting through one of the five slits formed by the helices. The difference electrostatic potential due to the protein alone is calculated and displayed in a color-coded scheme; a positive difference means that the electrostatic potential in the presence of GABARAP is more positive than in its absence. Panels (A-E) show the difference electrostatic potential on the same five planes as in Figure 8 conductance. Tierney 36 suggests that adjacent GABA A receptors interact via their solitary γ 2 -subunit MA helices; the ionic conductance is thus increased by this interaction. However, in the suggested mechanism, the γ 2 -subunit of one GABA A receptor swings out to interact with the γ 2 -subunit of another receptor, which involves a large structural change. These results seem to contradict previous experimental findings. 11,35 To define the interaction between GABARAP and the GABA

CONFLICT OF INTEREST
The authors declare no conflict of interest.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/prot.26241.

DATA AVAILABILITY STATEMENT
"The simulation data are available on request from the corresponding author (pc104@pasteur.fr) but in due course they will be put on a server where they can be downloaded by readers." ORCID P.-L. Chau https://orcid.org/0000-0003-3614-1561