In order to understand the roles of individual residues in receptors, it is critical to combine both structural and functional data. Here we have used such data to understand the interactions and importance of aromatic residues in the GlyR TMD. Aromatic residues are among many hydrophobic residues located, as would be expected, in the transmembrane α-helices, and many of these hydrophobic residues play some role in the assembly and/or correct function of the TMD. Aromatic residues, however, are especially important, as they have the potential for more interactions with adjacent residues because of their π rings. Indeed we observed that these residues frequently contribute to aromatic networks that exist between the different α-helices that constitute the TMD, and many are necessary for the efficient functioning of the protein; these interactions and networks are discussed in more detail below.
Y222 and Y223
Y222 and Y223 are located at the N-terminus of the M1 α-helix of the GlyR. Our data indicate both these residues are important for receptor function, with Ala substitution ablating function, although this is retained, albeit with an increase in EC50, with a Phe substitution. Our data do differ from a previous study [17] who found that Ala substitution did not ablate function. We cannot currently explain this, although in that study currents were reduced (e.g. Y223A currents were 14% of WT) and no EC50 data were given, so it is difficult to directly compare them with the current work. We did, however, observe that both Y222A and Y223A-containing GlyR were expressed in HEK293 cells, suggesting aromatic residues at these locations are not essential for receptor expression.
The structural data indicate that the only interaction that Y222 could make with adjacent residues is a π–π interaction with Y223. Such an interaction could help stabilise Y223, which has the potential for multiple interactions as shown in Fig. 4. Firstly it could form a bond between the aromatic ring of Y223 and the polarized CH in P146 (a CH–π bond) in the Cys-loop. A Pro in the Cys-loop is essential for pLGIC function, and it has been proposed this family of proteins should be renamed Pro-loop receptors because of the conservation of this Pro in all pLGICs [23]. Previous studies suggest this Pro may form a critical cis peptide bond [24, 25]; the intrinsically higher cis bias of Pro peptide bonds compared to other residues is consistent with this proposal. It is possible that a CH-π bond here could assist in stabilizing a cis peptide bond, and/or this interaction could hold the Cys-loop in a position or orientation that allows gating: the distance between Y223 and P146 differs in the open and closed states, and thus an interaction here might be important to stabilize the GlyR in the open state. GLIC and GABAAR structures show similar distances between equivalent residues, indicating that similar interactions may occur in these and perhaps all pLGICs.
Y223 could also form a hydrogen bond with Q186, located on the β9-β10 loop, another important region for channel gating (Fig. 4). However, the distance between these residues suggest a hydrogen bond here would be weak, and it is not conserved in GLIC and GABAAR. In addition the EC50 difference when the OH in Tyr is removed (i.e. in Phe) is small (< 4 fold). Thus we conclude there is no hydrogen bond here.
Y223 is also an appropriate distance and orientation from F145 in the Cys-loop to form a π–π interaction (Fig. 4). F145 in the GlyR has been shown to be important in providing a hydrophobic framework for a strong electrostatic interaction between D148 in the Cys-loop and R218 in M1 [26], and an interaction with Y223 could assist in the correct positioning of F145. Conservation of similar residues between GlyR, GLIC and GABAAR provide some support for this hypothesis, although the distances observed suggest that this interaction is less likely than that with P146, although both could contribute to Cys-loop stability.
Y228, W286, Y406, W407 and Y410
Y228 and W286 are located at the N-terminal end of the M1 α-helix and M3 α-helix respectively, and substitution with Ala ablates function, consistent with previous studies that show no surface expression [17], while substitution with an alternative aromatic (Phe or Tyr respectively) results in receptors that function well (Table 1). These data suggest that π rings are important in residues at these locations for receptor assembly, and the structure reveals Y228 and W286 are ideally placed to form a T-type π–π interaction (Fig. 5A). Replacement of Y228 with Phe would have little effect on such an interaction, consistent with a WT-like EC50 of Y228F mutant GlyR, while a W286 to Tyr substitution would have a greater impact, as the orientations of the 2 π rings would likely be less than ideal; our data showing an ~ 10 fold increase in EC50 in W286Y-containing GlyR are again consistent with this interpretation. These residues are not, however, the only aromatics in this location: Y406, W407 and Y410 at the C-terminal end of the M4 α-helix are close to Y228 and W286, and these five residues form an aromatic cluster between the M1, M3 and M4 α-helices at the extracellular side of the TMD (Fig. 5B). The loss of Gly-elicited responses in Y406A-containing GlyR, consistent with previous studies that show no surface expression [17], and increase in EC50 in W407A-and Y410A-containing GlyR, supports contributions of these aromatic residues to the assembly and function of the receptor as previously proposed [17]. These authors also demonstrated that these latter residues contribute to assembly as shown by their lower expression levels (and resulting lower maximal currents) with Ala substitution, and our data is similar, supporting this proposal (Fig. 3). They also suggested that Y228 interacts with F293, F402 and F405, but we now know that Y228 is too far from these residues for any interaction. Similarly Y406, W407 and Y410 were aberrantly suggested to face and interact with the lipid bilayer, stabilizing the M4 α-helix. The structure reveals they actually face the interface of M1, M3 and M4, and thus do likely have a role in stabilizing M4, but by interacting with M1 and M3 α-helices and not the lipid.
The structural data also predict a hydrogen bond between Y228 and Y406, but the WT-like response of Y228F containing GlyR indicates that if there is a hydrogen bond here it does not contribute to receptor function. Y406 could alternatively hydrogen bond with the backbone of A282 (Fig. 5C) and, given the increase in EC50 with Y406F containing GlyR, we propose this interaction does occur, and has a role in linking M1 and M4 that is perhaps important for assembly.
These suggested interactions are supported by studies in the GABAAR, where Y474 (equivalent to Y406) is predicted to hydrogen bond with Y289 (Y228), and Y474 has a likely π–π interaction with Y346 (W286) on M3 [27]. In GLIC, these aromatic residues are not conserved. However, I202 (Y228), I259 (W286), L310 (Y406) and A311 (W407) could perhaps form a hydrophobic patch that plays a similar role.
W239, F293 and F399
These residues are located in the M1/M3/M4 interface just below the aromatic cluster described above. Mutation of W239 and F399 to Ala ablated function, consistent with previous studies that show no surface expression [17], while function was retained with an alternative aromatic (Phe or Tyr respectively; Table 1), indicating a residue with a π ring is necessary for assembly. An aromatic at F293 is less important: F293A containing GlyR had an ~ 4 fold increase in EC50 which was reduced to ~ 2 fold with a Tyr substitution, suggesting that a small aromatic residue is preferred but not essential. We propose these residues interact with each other via π–π interactions (Fig. 6), which plays a role in stabilizing the transmembrane domain of the glycine receptor. Haeger et al. [17] also suggested there are interactions here, but details were inaccurate due to the lack of structural data. Such interactions are further supported by GLIC and GABAA receptor structural data: the Phe equivalent to F399 in M4 could interact with the Trp equivalent of W239 in M1 and the Phe equivalent of F293 in M3.
W239, F242 and F395
These residues on M1 (W239 and F242) and M4 (F395) could form π–π interactions (Fig. 7A, B) with W239 linking this group of aromatic residues with the one described above. W239 is the most sensitive of these residues to Ala substitution, where function is ablated as described above, whereas there are EC50s increases of 3–4 fold and decreases in maximal currents (as previously reported [17]) for F242A and F395A-containing GlyR (Table 1, Fig. 3) indicating roles in both assembly and function. WT-like responses with W239F containing GlyR suggest an aromatic residue is important here, and the altered distance between this residue and F395 in the open and closed states of the receptor suggest a π–π interaction that plays a role in stabilizing the open state. Data from other pLGIC indicate the equivalent residues in GLIC F216(M1) and F299(M4), ELIC F222(M1) and F303(M4) and GABAρR F303 (M1) and F463 (M4) may have similar interactions (e.g. Fig. 7C, D).
Y301, F306 and W243
These residues are located towards the intracellular side of the TMD. Y301 and F306 are on M3 and have not been previously studied. Our data indicate they play a role in function but not expression, as Ala substitutions resulted in changes to EC50s (Table 1) but not Imax (Fig. 3), while W243 on M1, where Ala substitution results in changes to both EC50 and Imax could have roles in both expression and function.
Ala substitution of Y301 results in an 8 fold increase in EC50. This residue has the potential to hydrogen bond with the backbone of M246 on M2 and/or form a cation–π interaction with R252, also on M2 (Fig. 8A–E). The distance between Y301 and M246 varies by 1Å between the open and closed states, whereas the distance between Y301 and R252 is similar. These data suggest a possible role of the hydrogen bond in stabilising the open state, while the cation–π interaction links M2 or M3, which could be important for information transfer between the transmembrane α-helices. Equivalent residues in the GABAA receptor could also form both the hydrogen bond and the cation–π interaction (Fig. 8F, G), providing some support for our suggestion, although no equivalent aromatic residue is present in GLIC.
Ala substitution of F306 or W243 both result in an ~ 4 fold increase in EC50. F306 is sufficiently close to R392 (M4) to form a cation–π interaction in the open but not the closed state, while W243 could form a cation–π interaction in the closed but not the open state (Fig. 9); such interactions could play a role in receptor opening and/or closing. Phe and Trp residues are conserved at these locations in the GABAA receptor and GLIC, supporting this hypothesis. W243 is also less than 4 Å from the backbone of W239, and thus could help stabilize this residue, which, as discussed above, is important for receptor expression.
F295 and F405
Ala substitution of F295 and F405 produce receptors that are similar to WT, and the structural data reveal both these residues face away from the core of the protein into the lipid bilayer. This indicates, as expected, that any hydrophobic residue could likely be located at these positions, as the presence of a π ring is not necessary.