Contributions of Conserved Residues at the Gating Interface of ...

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 286, NO. 40, pp. 35129 –35136, October 7, 2011 © 2011 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

Contributions of Conserved Residues at the Gating Interface of Glycine Receptors*□ S

Received for publication, June 6, 2011, and in revised form, August 2, 2011 Published, JBC Papers in Press, August 11, 2011, DOI 10.1074/jbc.M111.269027

Stephan A. Pless1, Ada W. Y. Leung, Jason D. Galpin, and Christopher A. Ahern From the Departments of Anesthesiology, Pharmacology, and Therapeutics and the Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, British Columbia, V6T 1Z3, Canada Glycine receptors (GlyRs) are chloride channels that mediate fast inhibitory neurotransmission and are members of the pentameric ligand-gated ion channel (pLGIC) family. The interface between the ligand binding domain and the transmembrane domain of pLGICs has been proposed to be crucial for channel gating and is lined by a number of charged and aromatic side chains that are highly conserved among different pLGICs. However, little is known about specific interactions between these residues that are likely to be important for gating in ␣1 GlyRs. Here we use the introduction of cysteine pairs and the in vivo nonsense suppression method to incorporate unnatural amino acids to probe the electrostatic and hydrophobic contributions of five highly conserved side chains near the interface, Glu-53, Phe-145, Asp-148, Phe-187, and Arg-218. Our results suggest a salt bridge between Asp-148 in loop 7 and Arg-218 in the pre-M1 domain that is crucial for channel gating. We further propose that Phe-145 and Phe-187 play important roles in stabilizing this interaction by providing a hydrophobic environment. In contrast to the equivalent residues in loop 2 of other pLGICs, the negative charge at Glu-53 ␣1 GlyRs is not crucial for normal channel function. These findings help decipher the GlyR gating pathway and show that distinct residue interaction patterns exist in different pLGICs. Furthermore, a salt bridge between Asp-148 and Arg-218 would provide a possible mechanistic explanation for the pathophysiologically relevant hyperekplexia, or startle disease, mutant Arg-218 3 Gln.

The glycine receptor (GlyR)2 chloride channel is a member of the Cys-loop receptor family, a subfamily of the pentameric ligand-gated ion channel (pLGIC) superfamily (1). GlyRs mediate fast inhibitory neurotransmission in the nervous system, and recent studies have provided a wealth of insight into the structure and function of the GlyR and other pLGICs. One of

* This work was supported by Canadian Institutes of Health Research Grant 56858, the Heart and Stroke Foundation of Canada, the Michael Smith Foundation for Health Research (to C. A. A.), and a postdoctoral fellowship by the Heart and Stroke Foundation of Canada (to S. A. P.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Figs. 1 and 2. 1 To whom correspondence should be addressed: 2350 Health Science Mall, Vancouver, BC V6T 1Z3, Canada. Tel.: 604-827-4189; E-mail: spless@mail. ubc.ca. 2 The abbreviations used are: GlyR, glycine receptor; pLGIC, pentameric ligand-gated ion channel; Akp, 2-amino-4-ketopentanoic acid; GABAA and GABAC, ␥-aminobutyric acid, types A and C, respectively; GluCl, glutamategated chloride channel; LBD, ligand binding domain; M1-M4, transmembrane segments 1– 4; Nha, nitrohomoalanine; TMD, transmembrane domain; nAChR, nicotinic acetylcholine receptor.

OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40

the more studied regions, the N-terminal ligand binding domain (LBD), is composed of a 10-strand ␤-sheet sandwich interconnected by 9 loops and the ligand binding pocket situated at the interface between adjacent subunits (2– 8). The transmembrane domain (TMD) contains four ␣-helical segments (M1-M4), including the pore-lining M2 helices. How is the binding of an agonist molecule in the LBD communicated to the channel gate in the TMD, almost 60 Å away? This question has gained considerable attention in the past (9 –11), and previous studies have identified a number of charged residues likely to couple the LBD to the TMD via electrostatic interactions in different pLGICs (12–17). A number of critical residues have been identified at the interface of LBD and TMD of ␣1 GlyRs (18 –22), but evidence for direct electrostatic interactions is thus far missing. The interface of LBD and TMD in ␣1 GlyRs is of interest not only because of its proposed role in channel gating but also because inherited mutations of side chains near this interface cause hyperekplexia or startle disease (23): Ala-52 in loop 2 (24), Leu-184 in loop 9 (25), Arg-218 in the pre-M1 domain (26), and Arg-271 and Lys-276 in the M2-M3 linker (27, 28). Based on sequence alignment and structural data from other pLGICs, we have identified five highly conserved charged and aromatic amino acid side chains that line the interface of LBD and TMD: Glu-53 in loop 2, Phe-145 and Asp-148 in loop 7 (or Cys-loop), Phe-187 in loop 9, and Arg-218 in the pre-M1 domain (Fig. 1). These charged side chains are of particular interest because the equivalent side chain to Arg-218 has been proposed to interact with the equivalent residues to Glu-53 in GABAC receptors (15, 16), whereas it is thought to interact with the side chain equivalent to Asp-148 in the ␤2 subunit of GABAA receptors (12). We used natural and unnatural amino acid side-chain substitutions to probe the contributions of these side chains to channel gating and tested for possible electrostatic interactions between these residues.

EXPERIMENTAL PROCEDURES Molecular Biology—Human GlyR ␣1 and ␤ subunit cDNA (both in the pGEMHE vector) were used, and mutants were generated using conventional site-directed mutagenesis. Successful incorporation of mutations was confirmed by automated sequencing. Capped mRNA was generated with the T7 mMessage mMachine kit (Ambion, Austin, TX). Stage V-VI oocytes from female Xenopus laevis frogs were prepared as previously described (29) and injected with 50 nl of mRNA alone or mRNA plus tRNA (see below). After injection, oocytes were incubated at 18 °C for 12–72 h. JOURNAL OF BIOLOGICAL CHEMISTRY

35129

A Proposed Salt Bridge Crucial for Glycine Receptor Gating

FIGURE 1. Sequence alignment and molecular model. A, shown is sequence alignment of various pLGICs. Shown are loops 2, 7, and 9 as well as the pre-M1 domain, with highly conserved aromatic (Phe-145, Phe-187), basic (Arg-218), and acidic (Glu-53, Asp-148) side chains highlighted in gray, blue, and red, respectively. B, a model of GluCl ␣ (PDB code 3RHW) shows the LBD and the TMD. The inset shows the five side chains as highlighted in A. Note that the residues 53 and 145 have been mutated in silico to the corresponding ␣1 GlyR side chains with no further energy minimization (Glu-53 is a Val in GluCl ␣; Phe-145 is a Tyr in GluCl ␣).

Unnatural Amino Acids and in Vivo Nonsense Suppression— Fluorinated phenylalanine derivates were purchased from Asis Chem (Watertown, MA) or Sigma. The synthesis of the unnatural amino acids 2-amino-4-ketopentanoic acid (Akp) and nitrohomoalanine (Nha) was performed as previously described (30). Unnatural amino acids were protected with a nitroveratryloxycarbonyl group, then activated as the cyanomethyl ester and finally coupled to the dinucleotide pdCpA (Dharmacon, Lafayette, CO) (31). This aminoacyl dinucleotide was enzymatically ligated to a 74-mer THG73 tRNACUA from Tetrahymena thermophila (32). UV irradiation was used for deprotection of the aminoacylated tRNA-amino acids (33) immediately before co-injection with the mRNA. In a typical experiment, 25–100 ng of tRNA-amino acids and 25 ng of mRNA were injected. In control experiments, mRNA containing the TAG stop codons was injected with tRNA-pdCpA (with no amino acid attached). Under these conditions we never observed currents bigger than 10 nA for Phe-145 3 TAG and Phe-187 3 TAG; for Glu-53 3 TAG and Asp-148 3 TAG, high (3 mM) concentrations of glycine elicited 60 ⫾ 40 and 170 ⫾ 90 nA of current, respectively (n ⫽ 5– 6 each). This result suggests that endogenous amino acids from the oocyte are not incorporated at a significant rate and that the tRNAs are not reacylated with naturally occurring amino acids by endogenous synthetases. Characterization of Mutant Receptors—Voltage-clamped chloride currents were recorded using an OC-725C voltage clamp (Warner Instrument, Hamden, CT). Microelectrodes were backfilled with 3 M KCl and had resistances between 0.2 and 2 megaohms. All recordings were performed in a standard Ringer’s solution (116 mM NaCl, 2 mM KCl, 1 mM MgCl2, 0.5 mM CaCl2, 5 mM HEPES, pH 7.4) at room temperature with a

35130 JOURNAL OF BIOLOGICAL CHEMISTRY

holding potential of ⫺60 mV. Glycine, ␤-alanine, taurine, strychnine, dithiothreitol (DTT), and mercuric chloride (HgCl2) (all Sigma) application was achieved with an automated perfusion system (AutoMate Scientific, San Francisco, CA). The Hill equation (fitted with a nonlinear least squares algorithm) was used to calculate values for ligand-induced halfmaximal concentrations for activation (EC50) and Hill coefficients (nH) (SigmaPlot 9.0, Systat Software, Point Richmond, CA). All data are the mean ⫾ S.E. Computations—Gaussian09/GaussView 5.0 (Wallingford, CT) was used in ab initio gas phase calculations (Hartree-Fock 6 –31G level) to generate compound structures and electrostatic surface potential distributions. The molecular model was generated with PyMOL 0.99 (DeLano Scientific LLC, San Francisco, CA).

RESULTS Contributions of Asp-148—Asp-148 is a highly conserved acidic charge in loop 7 (or Cys-loop) of pLGICs (Fig. 1), and previous studies on the GlyR have shown that mutations at this position have severe effects on channel function (18, 20, 21), suggesting that the negative charge at this position is crucial for channel gating (20). However, Asp-148 replacement via conventional mutagenesis presents an interpretive challenge as no naturally occurring hydrophilic amino acids exist that are neutral at physiological pH without a strong propensity to donate hydrogens. For instance, the relatively conservative Asp to Asn substitution neutralizes (while retaining a hydrophilic head group) but also replaces a strong hydrogen bond acceptor (carboxyl group) with a potent hydrogen bond donor (amide group). This seemingly simple substitution can be detrimental due to introduced hydrogen bond clashes, especially in protein VOLUME 286 • NUMBER 40 • OCTOBER 7, 2011

A Proposed Salt Bridge Crucial for Glycine Receptor Gating domains that are crucial for gating (30, 34). To investigate the role of charge in a more subtle way thus requires the use of unnatural amino acid side chains. To this end, we replaced the negative charge in position 148 with a neutral analog of Asp, Akp (Fig. 2) (30), which is identical in size, carries no charge,

FIGURE 2. Natural and unnatural side chains used in this study. Shown are the chemical structures of Asp and Akp (A) and Glu and Nha (B) as well as of Phe and the fluorinated Phe derivatives, 4-F-Phe, 3,5-F2-Phe, and 3,4,5-F3Phe (C).

and has minimal propensity for hydrogen bond donation. Incorporation of Akp in position 148 led to robust glycinegated currents that displayed less desensitization than WT channels (Fig. 3A). Importantly, even this very subtle neutral substitution of Asp-148 led to a dramatic decrease in glycine sensitivity compared with WT channels (Fig. 3B, Table 1). Furthermore, the relative efficacies of the partial agonists ␤-alanine and taurine were significantly reduced, suggesting that the removal of charge at position 148 affects channel gating (Fig. 3C, Table 1). A Proposed Salt Bridge between Asp-148 and Arg-218—If the negative charge at position 148 plays a crucial role in channel gating, is it possible that Asp-148 interacts electrostatically with a nearby positive charge? A previous study ruled out electrostatic interactions with Lys-279 in the M2-M3 linker (18), but mutations of a highly conserved basic residue in the pre-M1 domain, Arg-218, have been shown to severely impact channel function by reducing channel expression, decreasing glycine sensitivity and partial agonist efficacy, ultimately resulting in

FIGURE 3. A negative charge in position 148 is crucial for channel function. A, glycine-induced currents were recorded for either Asp (WT) or Akp in position 148 (upper and lower panel, respectively). B, concentration-response curves for Asp (WT) or Akp in position 148 are shown. Insets show ball-and-stick representations and electrostatic surface potentials of the side chains (red ⫽ ⫺100 kcal/mol; green ⫽ 0 kcal/mol). C and D, shown is relative efficacy of Asp (WT) versus Akp in position 148 for ␤-alanine (C) and taurine (D); asterisks indicate significant difference (Student’s t test; *, p ⬍ 0.05).

TABLE 1 Concentration-response data and relative efficacies of partial agonists for constructs tested in this study Data are shown as the mean ⫾ S.E.; the asterisks indicate significant difference (Student´s t test). NC indicates constructs that yielded no measurable currents even at high (100 –300 mM) glycine concentrations. Construct

Glycine EC50

nH

␮M

WT WT (⬍1 ␮A) Asp-148 3 Akp Asp-148 3 Cys Asp-148 3 Cys,Arg-218 3 Cys Arg-218 3 Cys Arg-218 3 Asp Asp-148 3 Arg,Arg-218 3 Asp Glu-53 3 Nha Glu-53 3 Akp Glu-53 3 Cys Glu-53 3 Arg Glu-53 3 Arg,Arg-218 3 Gluc Glu-53 3 Cys,Arg-218 3 Cysc Phe-187 3 Ala Phe-187 3 Phe (WT) Phe-187 3 F-Phe Phe-187 3 F2-Phe Phe-187 3 F3-Phe Phe-145 3 Phe (WT) Phe-145 3 F-Phe Phe-145 3 F2-Phe Phe-145 3 F3-Phe a b c

64 ⫾ 3 377 ⫾ 43a 392 ⫾ 52a NC 133 ⫾ 14b NC NC NC 75 ⫾ 13 104 ⫾ 8a 713 ⫾ 29a NC NC NC NC 77 ⫾ 4 155 ⫾ 8a 88 ⫾ 9 99 ⫾ 16 63 ⫾ 7 31 ⫾ 3a 65 ⫾ 2 NC

I␤-ala/Igly

ITau/Igly

%

%

1.5 ⫾ 0.1 1.7 ⫾ 0.3 1.1 ⫾ 0.2

89 ⫾ 7 80 ⫾ 3 58 ⫾ 5a

80 ⫾ 4 30 ⫾ 1a 21 ⫾ 1a

1.5 ⫾ 0.2

45 ⫾ 5b

51 ⫾ 5b

0.7 ⫾ 0.1a 1.9 ⫾ 0.2 1.5 ⫾ 0.1

73 ⫾ 13 90 ⫾ 5 10 ⫾ 1a

57 ⫾ 8 45 ⫾ 4a 7 ⫾ 2a

1.9 ⫾ 0.2 1.6 ⫾ 0.1 2.1 ⫾ 0.4 1.6 ⫾ 0.4 1.5 ⫾ 0.2 2.0 ⫾ 0.4 1.7 ⫾ 0.1

89 ⫾ 3

79 ⫾ 7

93 ⫾ 4

92 ⫾ 4

n 5 4 4 5 8 5 5 5 3 4 3 5 4 4 7 3 4 4 3 4 4 3 6

p ⬍ 0.05 compared to WT. p ⬍ 0.05 compared to WT (⬍1 ␮A). Constructs also co-injected with a 10-fold access of GlyR ␤ subunit.

OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40

JOURNAL OF BIOLOGICAL CHEMISTRY

35131

A Proposed Salt Bridge Crucial for Glycine Receptor Gating hyperekplexia (26). Intriguingly, recent structural studies have suggested that the residues homologous to Asp-148 and Arg218 are in close proximity (2–3 Å) to each other in pLGICs (4 – 6, 8) in both open and closed channel conformations. Furthermore, the residues equivalent to ␣1 GlyR Asp-148 and Arg218 have been proposed to interact in the ␤2 subunit of GABAA receptors (12). To test the possibility that Asp-148 and Arg-218 are in close enough proximity to form a salt bridge in vivo in ␣1 GlyRs, we attempted to cross-link them via a disulfide bridge with introduced Cys side chains. Single cysteines at either position 148 or position 218 produced no glycine-gated currents, even at very high (300 mM) glycine concentrations (Fig. 4, A and B; Table 1). However, the double cysteine mutant Asp-148 3 Cys,Arg-218 3 Cys resulted in significant glycine-gated inward currents (Fig. 4C, upper panel) that were sensitive to the GlyRspecific competitive antagonist strychnine and could be modulated by the reducing agent DTT but not by the oxidizing agent HgCl2 (supplemental Fig. 1). As the currents from Asp-148 3 Cys,Arg-218 3 Cys receptors were significantly smaller than those from WT (0.4 ⫾ 0.1 versus 21.5 ⫾ 4.9 ␮A, n ⫽ 5– 8) and low expression levels of ␣1 GlyRs in X. laevis oocytes have been shown to reduce both glycine sensitivity and the relative efficacy of partial agonists (35, 36), we reduced WT expression to levels comparable with those observed for the Asp-148 3 Cys,Arg-218 3 Cys double mutant (⬍1 ␮A) (Fig. 4C, lower panel). Under these conditions the Asp-148 3 Cys,Arg-218 3 Cys showed a 3-fold higher glycine sensitivity than the lowexpressing WT channels (Fig. 4D, Table 1). To test the effect of the Asp-148 3 Cys,Arg-218 3 Cys double mutant on channel gating in more detail, we established the relative efficacies of the partial agonists, ␤-alanine and taurine. Although significantly reduced compared with WT, we found ␤-alanine to be a potent agonist of the Asp-148 3 Cys,Arg-218 3 Cys double mutant receptor (Fig. 4F). Interestingly, taurine displayed a significantly higher efficacy than at the low expressing WT channels (Fig. 4G), indicating that the gating pathways of different agonists may not be identical and, hence, are not affected to the same extent by the Asp-148 3 Cys,Arg-218 3 Cys double mutant. In contrast, the double charge reverse mutant Asp-148 3 Arg, Arg-218 3 Asp (as well as the single charge reverse mutants, Asp-148 3 Arg and Arg-218 3 Asp) did not result in functional channels (Table 1). Another Potential Salt Bridge; Glu-53 and Arg-218—The residue homologous to ␣1 GlyR Arg-218 has been proposed to be involved in salt bridge formation with the homologous residue to ␣1 GlyR Glu-53 in nAChRs (14) and GABAC receptors (15, 16) (but not in the related 5-hydroxytryptamine receptors (16)). In support of this idea, recent structural evidence from pLGICs (4 – 6, 8) has placed the corresponding residues in close proximity, although it should be noted that a negative charge at the side chains equivalent to ␣1 GlyR Glu-53 is not present in a pLGIC from Erwinia chrysanthemi (ELIC) and GluCl ␣, the latter sharing a high sequence identity (34%) with the ␣1 GlyR (Fig. 1). Using Glu-53 3 Cys and cysteine-reactive charged adducts, a recent study had suggested that a negative charge in position 53 is likely to be important for channel function in ␣1 GlyRs (22). However, given the dramatically changed stereo-


FIGURE 4. Cysteine cross-linking suggests a salt bridge between Asp-148 and Arg-218. A and B, individual cysteine mutations in positions 148 (A) and 218 (B) resulted in non-functional receptors. C, glycine-induced currents recorded from the Asp-148 3 Cys,Arg-218 3 Cys (Cys-Cys) double mutant (upper panel) and from WT channels yielding less than 1 ␮A of glycine-induced currents (lower panel) are shown. D, shown are concentration-response curves for WT (⬍1 ␮A) in black circles, for Cys-Cys double mutant in empty circles, and for WT channels expressed at normal levels in gray circles (reproduced from Fig. 3). E and F, relative efficacy for ␤-alanine (E) and taurine (F) for WT (⬍1 ␮A), the Cys-Cys double mutant, and WT channels expressed at normal levels, color coded as in D; asterisks indicate significant difference (Student’s t test; *, p ⬍ 0.05). Note that the double charge reversal mutant (Asp-148 3 Arg,Arg-218 3 Asp) did not generate functional receptors (see Table 1).

chemistry inherent to this experimental approach, a secondary effect could not be ruled out. Furthermore, although mutations at Glu-53 have been shown to affect ␣1 GlyR channel function (18, 22, 37), it is not clear if an electrostatic interaction between Glu-53 and Arg-218 is present in the ␣1 GlyR. We, therefore, aimed to first probe the contribution of charge at position 53 VOLUME 286 • NUMBER 40 • OCTOBER 7, 2011


FIGURE 5. A negative charge in position 53 not required channel gating. A, glycine-induced currents recorded with Nha (upper panel) or Akp (lower panel) in position 53 are shown. B, concentration-response curves for Glu (WT) (black circles), Cys (gray circles), Nha (green triangles), or Akp (blue triangles) in position 53 are shown. Insets show ball-and-stick representations and electrostatic surface potentials for Glu and Nha (red ⫽ ⫺100 kcal/mol; green ⫽ 0 kcal/mol). C and D, relative efficacy of Glu (WT), Cys, Nha, or Akp in position 53 for ␤-alanine (C) and taurine (D) is color-coded as in B; the asterisk indicates significant difference (Student’s t test; *, p ⬍ 0.05).

using the unnatural amino acid Nha, a side chain that is isosteric to Glu but contains a neutral nitro group instead of a negatively charged carboxyl group (Fig. 2B) (30). Incorporation of Nha in position 53 resulted in robust glycine-gated currents (Fig. 5A), and in contrast to conventional neutral side chain replacements, such as Cys, which lead to large changes in the glycine sensitivity, Nha-53 displayed a WT-like glycine EC50, although with a significantly reduced Hill coefficient (Fig. 5B, Table 1). Similarly, another subtle neutral replacement of Glu53, Akp-53, had only a minor effect on the glycine EC50 and the Hill coefficient (Fig. 5B, Table 1). Furthermore, conventional mutations, such as Glu-53 3 Cys, dramatically reduced the relative efficacies of both ␤-alanine and taurine. The more subtle side-chain neutralization using Nha, on the other hand, restored their efficacies to values that were not significantly lower than those observed in WT channels (Fig. 5, C and D, Table 1). Even the less conservative neutral side chain of Akp fully restored a WT-like ␤-alanine efficacy, although taurine efficacy remained reduced (Fig. 5, C and D). Together, these results suggest that a negative charge is not necessary for normal channel function, arguing against an electrostatic contribution of Glu-53 in general and a salt bridge interaction between Glu-53 and Arg-218 in particular, reminiscent of results obtained for the equivalent residues in 5-hydroxytryptamine receptors (16) and a recent study on nAChRs (38). In support of this notion, the charge reversal mutant Glu-53 3 Arg,Arg-218 3 Glu as well as the double cysteine mutant Glu-53 3 Cys,Arg-218 3 Cys failed to produce glycinegated currents even at very high (300 mM) glycine concentrations (n ⫽ 5–8, Table 1). As both the Glu-53 3 Arg,Arg-218 3 Glu mutant and the Glu-53 3 Cys,Arg-218 3 Cys may have displayed strongly altered channel properties (due to the drastically altered stereochemistry of a possible interaction between residues 53 and 218), we sought co-express these constructs with the GlyR ␤ subunit (which itself fails to produce homomeric receptors), as a recent study has shown that the GlyR ␤ subunit can functionally rescue mutations in the ␣1 subunit that fail to produce currents as homomers (25). However, coinjection of Glu-53 3 Arg,Arg-218 3 Glu mRNA or Glu-53 3 Cys,Arg-218 3 Cys mRNA with a 10-fold excess of GlyR ␤ subunit mRNA did not result in glycine-gated currents, even in response to very high (100 mM) glycine concentrations (n ⫽ 4 each, see also Table 1). OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40

Characterizing Phe-187—Although highly conserved and likely to be situated at the interface of LBD and TMD, little is known about the role of Phe-187 in ␣1 GlyR gating. The putative proximity of this side chain to Arg-218 (Fig. 1B) has led us to speculate that these side chains interact via an electrostatic, cation-␲ interaction (39). The general importance of this site can be seen with the Phe-187 3 Ala mutation, which resulted in non-functional channels, as we could not observe glycineinduced currents even in response to very high (300 mM) glycine concentrations (n ⫽ 7). To more directly explore the role of Phe-187 in channel function, we introduced side chains with more subtle alterations of Phe in this position that retained the size and hydrophobicity while only changing the electrostatic surface potential of the side chain by serially fluorinating Phe187. The approach of fluorinated aromatic amino acids to test for cation-␲ interactions has been used extensively in the past to test for receptor-ligand-type interactions (3, 7, 31, 40 – 47). However, functional evidence for intramolecular cation-␲ interactions has proven far more elusive to establish, with only a single example presently known; that is, a strong cation-␲ interaction between Lys-374 in the charge-carrying S4 segment and an introduced Trp in position 290 of Shaker potassium channels (34). To test if the electrostatic surface potential of Phe-187 is crucial for channel gating, we introduced fluorinated Phe derivatives (Fig. 2C) at position 187 and observed robust glycine-gated currents with all fluorinated Phe analogs tested (Fig. 6A). If the electrostatic surface potential of Phe-187 was engaged in a cation-␲ interaction with Arg-218 (or another basic residue), we would expect a stepwise, linear shift in the EC50 with every added fluorine, as each of these electronegative substituents results in an equally significant reduction of the electronegative surface potential on the face of the aromatic. However, as shown in Fig. 6B, only F-Phe produced a small increase in the EC50, whereas F2-Phe and F3-Phe displayed WTlike glycine sensitivities (Table 1). Consistent with this observation, the relative efficacies of ␤-alanine and taurine of F3-Phe were identical to those of WT channels. Characterizing Phe-145—Phe-145 is located in the highly conserved loop 7 (or Cys-loop), which has been identified as an important component of the gating machinery of pLGICs (13, 18, 21, 48, 49). A previous study of the Phe-145 3 Ala mutation in the ␣1 GlyR found this non-conservative mutation to have a severe impact on channel function, with an increase in the EC50 JOURNAL OF BIOLOGICAL CHEMISTRY

35133


FIGURE 6. The electrostatic surface potentials of Phe-187 and Phe-145 do not contribute to channel gating. A and E, glycine-induced currents recorded for F3-Phe in position 187 (A) and F2-Phe in position 145 (E), respectively, are shown. B and F, concentration-response curves for Phe (WT) and fluorinated Phe derivatives in position 187 (B) or 145 (F) are shown. Panels C and D and panels G and H, shown is relative efficacy of Phe (WT) and F3-Phe in position 187 or Phe (WT) and F2-Phe in position 145 for ␤-alanine and taurine, respectively.

of almost 30-fold, substantially lowered maximal currents, and decreased efficacy of the partial agonist taurine (21). These results pointed to a crucial role for Phe-145 in channel gating. Additionally, recent crystallographic evidence from other pLGICs suggested that the aromatic side chains homologous to Phe-145 in ␣1 GlyRs reside in close proximity to the positively charged Arg-218 (Fig. 1B) (4 – 6, 8), suggesting that Phe-145 could stabilize the cationic charge at Arg-218 via an electrostatic cation-␲ interaction. To test this possibility directly, we introduced fluorinated phenylalanine derivatives at position 145, with exemplified currents recorded from Phe-145 3 F2-Phe shown in Fig. 6E. Although the addition of a single fluorine significantly decreased the EC50, adding a second fluorine restored WT-like glycine sensitivity (Fig. 6F and Table 1). Additionally, replacing Phe with F2-Phe did not affect the relative efficacy of ␤-alanine or taurine.

DISCUSSION The mechanistic nature of the linkage between the LBD and the TMD in pLGICs has garnered much attention, and significant progress has been made using structural (4 – 6, 8) and functional approaches (12–17, 38, 50). However, in ␣1 GlyRs, little is known about possible electrostatic interactions that link the LBD to the TMD in other pLGICs. Here, we show that although individual cysteine mutations at positions 148 and 218 result in non-functional receptors, the double cysteine mutant Asp-148 3 Cys,Arg-218 3 Cys produces robust glycine-gated currents and displays functional properties that resemble those of WT receptors. This is a surprising finding given the severity of the side chain replacements in positions 148 and 218 (Asp to


Cys and Arg to Cys, respectively), and the mutual rescue implies a strong physical coupling between the two cysteines. We thus propose that the two cysteine side chains are likely to form a disulfide bridge. As disulfide bridges can only form over short (2–3 Å) distances, our results suggest that the side chains in positions 148 and 218 may be in close (and likely static) proximity during channel gating. Moreover, as such short distances are ideal for salt bridge formation, we propose that Asp-148 and Arg-218 form in the native receptor. This notion is further supported by the requisite nature of charge at positions 148 and 218, negative and positive, respectively, for normal channel function in the ␣1 GlyR. The fact that the double charge reverse mutant Asp-148 3 Arg,Arg-218 3 Asp (as well as the single charge reverse mutants, Asp-148 3 Arg and Arg-218 3 Asp) did not result in functional channels could be due to the severe disruption of the highly conserved charge pattern of the interface between LBD and TMD (17), which shows a strong preference for negative charges in the LBD, whereas the TMD primarily contains positive charges. The idea of an electrostatic interaction between Asp-148 and Arg-218 is further supported by another study that suggested an interaction between the equivalent residues in the ␤2 subunit of GABAA receptors (12). The proposed interaction between Asp-148 and Arg-218 in the ␣1 GlyR is of particular interest as mutations at side chains equivalent to ␣1 GlyR Arg-218 have been shown to be crucial for channel function in different pLGICs (12, 15, 38, 51–53) and especially because the ␣1 GlyR Arg-218 3 Gln mutation can give rise to the inherited hyperekplexia or startle disease (26). The idea that Arg-218 contributes to a salt bridge crucial for VOLUME 286 • NUMBER 40 • OCTOBER 7, 2011

A Proposed Salt Bridge Crucial for Glycine Receptor Gating channel gating in the native receptor provides an explanation for molecular basis for a pathophysiologically relevant mutant; the neutral Arg-218 3 Gln mutation would prevent the formation of a strong salt bridge and, hence, give rise to the drastic effect on gating observed with this mutant. It further explains why Arg-218 3 Gln results in functional receptors, albeit with severe functional impairments, whereas other replacements such as Cys and Asp do not; in the proposed salt bridge interaction with Asp-148, Arg-218 would (in addition to contributing a positive charge) act as a strong hydrogen bond donor. Although uncharged, Gln is a potent hydrogen bond donor, in contrast to Cys or Asp, and could interact with the hydrogen bond accepting Asp-148, although to a weaker extent than Arg. The emerging picture for the role of Arg-218 thus points to two crucial roles; first, Arg-218 contributes to an interaction with Asp-148 that is crucial for channel gating. This functionally important interaction can be restored by the proposed disulfide bond in the Asp-148 3 Cys,Arg-218 3 Cys double mutant. Second, and in agreement with previous studies on the equivalent residue of different nAChRs (38, 51) and ␣1 GlyR Arg-218 (26), ␣1 GlyR Arg-218 is important for channel expression. Our data show that in contrast to the functionally relevant salt bridge with Asp-148, which can be mimicked by the proposed disulfide bond, the positive charge of Arg-218 is required for efficient expression, as the Asp-148 3 Cys,Arg-218 3 Cys double mutant, although functionally similar to WT, shows dramatically reduced expression levels (Table 1). The available structural data suggest that the proposed salt bridge between Asp-148 and Arg-218 is flanked by Phe-145 and Phe-187. Interestingly, our data suggest that large hydrophobic side chains are required in positions 145 and 187, although aromaticity is not, thus ruling out contributions of the electrostatic surface potential of these side chains to channel gating. This finding not only excludes the possibility of a cation-␲ interaction between Arg-218 and either of the two aromatic side chains, but it also raises an intriguing possibility about the role of Phe-145 and Phe-187; given that salt bridges are significantly stronger in hydrophobic, compared with aqueous, environments (34, 54), it is possible that it is the primary role of Phe-145 and Phe-187 to provide/promote a hydrophobic environment for the formation of an energetically significant salt bridge between Asp-148 and Arg-218. Additionally, Phe-145 and Phe-187 could participate in hydrophobic interactions with Ile-51 (22) and Leu-274 (55), respectively. Furthermore, a recent study on the ␣1 GlyR Phe-145 equivalent in nAChRs proposed an interaction with the equivalent of ␣1 GlyR Pro-146 (Pro-136 in nAChR) (48). However, it should be noted that the same study also found a strong preference for aromaticity at the position equivalent to Phe-145 in nAChRs, a finding not in agreement with our data, suggesting that, although conserved, the same residues may support different functional roles in GlyRs and nAChRs. Although our data do not directly speak to a possible interaction with the adjacent Pro-146, we propose that the principal role of Phe-145 (and Phe-187) is to provide a hydrophobic framework for a strong electrostatic interaction between Asp-148 and Arg-218, a notion that is further supported by recent structural data from GluCl ␣ (supplemental Fig. 2). OCTOBER 7, 2011 • VOLUME 286 • NUMBER 40

Although the general importance of loop 2 for channel gating is uncontested in ␣1 GlyRs (19, 22, 56, 57) and other pLGICs (14 –16, 58 – 62), we propose that in ␣1 GlyRs there is no crucial contribution of the negative charge of Glu-53 to channel gating. Instead, it appears more likely that the side chain in position 53 of ␣1 GlyRs does not require a negative charge per se but a hydrophilic head group with no propensity to donate hydrogens, as hydrogens could result in steric and/or electrostatic clashes with other residues at the interface of LBD and TMD. It should be noted that Glu and Asp are the only naturally occurring side chains that fulfill these criteria and that both these side chains (18) as well as Nha and Akp (Fig. 5) are the only side chains that result in WT-like gating behavior at position 53 of the ␣1 GlyR. This is in agreement with a recent study that proposed an alcohol binding pocket that is flanked by loop 2, the pre-M1 domain, M2, and the M2-M3 linker (63); odd-numbered residues in loop 2, such as Glu-53, are thought to point into this pocket (22), possibly explaining the strict requirement for a hydrophilic side chain in position 53 as this would help create and maintain a likely water-filled allosteric ligand binding pocket. We thus propose that although an electrostatic interaction between the equivalent residues of ␣1 GlyR Glu-53 and Arg-218 has been proposed in GABAC receptors (15, 16) and nAChRs (14), such an interaction is not present in ␣1 GlyRs. However, the altered Hill coefficient for Nha at position 53 could indicate that the negative charge mildly contributes to cooperativity between adjacent subunits. In conclusion, we propose a salt bridge between Asp-148 in loop 7 and Arg-218 in the pre-M1 domain that is crucial for channel gating in ␣1 GlyRs and has implications for the clinically relevant hyperekplexia mutant Arg-218 3 Gln. The data further support a role for two highly conserved aromatic residues in loop 7 and loop 9 in shielding the proposed salt bridge between Asp-148 and Arg-218 from a polar environment, which would otherwise weaken the interaction. The study thus highlights important aspects of the ␣1 GlyR gating pathway and provides insight into the molecular mechanisms for a mutation that gives rise to a channelopathy. Acknowledgment—We thank Dr. Ana Niciforovic for excellent technical assistance. REFERENCES 1. Lynch, J. W. (2004) Physiol. Rev. 84, 1051–1095 2. Grudzinska, J., Schemm, R., Haeger, S., Nicke, A., Schmalzing, G., Betz, H., and Laube, B. (2005) Neuron 45, 727–739 3. Pless, S. A., Millen, K. S., Hanek, A. P., Lynch, J. W., Lester, H. A., Lummis, S. C., and Dougherty, D. A. (2008) J. Neurosci. 28, 10937–10942 4. Bocquet, N., Nury, H., Baaden, M., Le Poupon, C., Changeux, J. P., Delarue, M., and Corringer, P. J. (2009) Nature 457, 111–114 5. Hilf, R. J., and Dutzler, R. (2008) Nature 452, 375–379 6. Hilf, R. J., and Dutzler, R. (2009) Nature 457, 115–118 7. Pless, S. A., Hanek, A. P., Price, K. L., Lynch, J. W., Lester, H. A., Dougherty, D. A., and Lummis, S. C. (2011) Mol. Pharmacol. 79, 742–748 8. Hibbs, R. E., and Gouaux, E. (2011) Nature 474, 54 – 60 9. Dougherty, D. A. (2008) Chem. Rev. 108, 1642–1653 10. Lester, H. A., Dibas, M. I., Dahan, D. S., Leite, J. F., and Dougherty, D. A. (2004) Trends Neurosci. 27, 329 –336 11. Miller, P. S., and Smart, T. G. (2010) Trends Pharmacol. Sci. 31, 161–174 12. Kash, T. L., Dizon, M. J., Trudell, J. R., and Harrison, N. L. (2004) J. Biol.

JOURNAL OF BIOLOGICAL CHEMISTRY

35135

A Proposed Salt Bridge Crucial for Glycine Receptor Gating Chem. 279, 4887– 4893 13. Kash, T. L., Jenkins, A., Kelley, J. C., Trudell, J. R., and Harrison, N. L. (2003) Nature 421, 272–275 14. Lee, W. Y., and Sine, S. M. (2005) Nature 438, 243–247 15. Wang, J., Lester, H. A., and Dougherty, D. A. (2007) J. Biol. Chem. 282, 26210 –26216 16. Price, K. L., Millen, K. S., and Lummis, S. C. (2007) J. Biol. Chem. 282, 25623–25630 17. Xiu, X., Hanek, A. P., Wang, J., Lester, H. A., and Dougherty, D. A. (2005) J. Biol. Chem. 280, 41655– 41666 18. Absalom, N. L., Lewis, T. M., Kaplan, W., Pierce, K. D., and Schofield, P. R. (2003) J. Biol. Chem. 278, 50151–50157 19. Cederholm, J. M., Absalom, N. L., Sugiharto, S., Griffith, R., Schofield, P. R., and Lewis, T. M. (2010) J. Neurochem 115, 1245–1255 20. Schofield, C. M., Jenkins, A., and Harrison, N. L. (2003) J. Biol. Chem. 278, 34079 –34083 21. Schofield, C. M., Trudell, J. R., and Harrison, N. L. (2004) Biochemistry 43, 10058 –10063 22. Crawford, D. K., Perkins, D. I., Trudell, J. R., Bertaccini, E. J., Davies, D. L., and Alkana, R. L. (2008) J. Biol. Chem. 283, 27698 –27706 23. Ryan, S. G., Sherman, S. L., Terry, J. C., Sparkes, R. S., Torres, M. C., and Mackey, R. W. (1992) Ann. Neurol. 31, 663– 668 24. Ryan, S. G., Buckwalter, M. S., Lynch, J. W., Handford, C. A., Segura, L., Shiang, R., Wasmuth, J. J., Camper, S. A., Schofield, P., and O’Connell, P. (1994) Nat. Genet. 7, 131–135 25. Chung, S. K., Vanbellinghen, J. F., Mullins, J. G., Robinson, A., Hantke, J., Hammond, C. L., Gilbert, D. F., Freilinger, M., Ryan, M., Kruer, M. C., Masri, A., Gurses, C., Ferrie, C., Harvey, K., Shiang, R., Christodoulou, J., Andermann, F., Andermann, E., Thomas, R. H., Harvey, R. J., Lynch, J. W., and Rees, M. I. (2010) J. Neurosci. 30, 9612–9620 26. Castaldo, P., Stefanoni, P., Miceli, F., Coppola, G., Del Giudice, E. M., Bellini, G., Pascotto, A., Trudell, J. R., Harrison, N. L., Annunziato, L., and Taglialatela, M. (2004) J. Biol. Chem. 279, 25598 –25604 27. Rajendra, S., Lynch, J. W., Pierce, K. D., French, C. R., Barry, P. H., and Schofield, P. R. (1994) J. Biol. Chem. 269, 18739 –18742 28. Elmslie, F. V., Hutchings, S. M., Spencer, V., Curtis, A., Covanis, T., Gardiner, R. M., and Rees, M. (1996) J. Med. Genet. 33, 435– 436 29. Pless, S. A., Dibas, M. I., Lester, H. A., and Lynch, J. W. (2007) J. Biol. Chem. 282, 36057–36067 30. Cashin, A. L., Torrice, M. M., McMenimen, K. A., Lester, H. A., and Dougherty, D. A. (2007) Biochemistry 46, 630 – 639 31. Pless, S. A., Galpin, J. D., Frankel, A., and Ahern, C. A. (2011) Nat. Commun. 2, 351 32. Nowak, M. W., Gallivan, J. P., Silverman, S. K., Labarca, C. G., Dougherty, D. A., and Lester, H. A. (1998) Methods Enzymol. 293, 504 –529 33. Kearney, P. C., Nowak, M. W., Zhong, W., Silverman, S. K., Lester, H. A., and Dougherty, D. A. (1996) Mol. Pharmacol. 50, 1401–1412 34. Pless, S. A., Galpin, J. D., Niciforovic, A. P., and Ahern, C. A. (2011) Nat. Chem. Biol. 7, 617– 623 35. De Saint Jan, D., David-Watine, B., Korn, H., and Bregestovski, P. (2001) J. Physiol. 535, 741–755


36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51.

52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63.

Taleb, O., and Betz, H. (1994) EMBO J. 13, 1318 –1324 Pless, S. A., and Lynch, J. W. (2009) J. Biol. Chem. 284, 15847–15856 Purohit, P., and Auerbach, A. (2007) J. Gen. Physiol. 130, 559 –568 Dougherty, D. A. (1996) Science 271, 163–168 Beene, D. L., Brandt, G. S., Zhong, W., Zacharias, N. M., Lester, H. A., and Dougherty, D. A. (2002) Biochemistry 41, 10262–10269 Lummis, S. C., L., Beene, D., Harrison, N. J., Lester, H. A., and Dougherty, D. A. (2005) Chem. Biol. 12, 993–997 Padgett, C. L., Hanek, A. P., Lester, H. A., Dougherty, D. A., and Lummis, S. C. (2007) J. Neurosci. 27, 886 – 892 Xiu, X., Puskar, N. L., Shanata, J. A., Lester, H. A., and Dougherty, D. A. (2009) Nature 458, 534 –537 Zhong, W., Gallivan, J. P., Zhang, Y., Li, L., Lester, H. A., and Dougherty, D. A. (1998) Proc. Natl. Acad. Sci. U.S.A. 95, 12088 –12093 Ahern, C. A., Eastwood, A. L., Dougherty, D. A., and Horn, R. (2008) Circ. Res. 102, 86 –94 Ahern, C. A., Eastwood, A. L., Lester, H. A., Dougherty, D. A., and Horn, R. (2006) J. Gen. Physiol. 128, 649 – 657 Santarelli, V. P., Eastwood, A. L., Dougherty, D. A., Horn, R., and Ahern, C. A. (2007) J. Biol. Chem. 282, 8044 – 8051 Limapichat, W., Lester, H. A., and Dougherty, D. A. (2010) J. Biol. Chem. 285, 8976 – 8984 Jha, A., Cadugan, D. J., Purohit, P., and Auerbach, A. (2007) J. Gen. Physiol. 130, 547–558 Bouzat, C., Gumilar, F., Spitzmaul, G., Wang, H. L., Rayes, D., Hansen, S. B., Taylor, P., and Sine, S. M. (2004) Nature 430, 896 –900 Vicente-Agullo, F., Rovira, J. C., Sala, S., Sala, F., Rodriguez-Ferrer, C., Campos-Caro, A., Criado, M., and Ballesta, J. J. (2001) Biochemistry 40, 8300 – 8306 Keramidas, A., Kash, T. L., and Harrison, N. L. (2006) J. Physiol. 575, 11–22 Mercado, J., and Czajkowski, C. (2006) J. Neurosci. 26, 2031–2040 Gallivan, J. P., and Dougherty, D. A. (2000) J. Am. Chem. Soc. 122, 870 – 874 Lynch, J. W., Rajendra, S., Pierce, K. D., Handford, C. A., Barry, P. H., and Schofield, P. R. (1997) EMBO J. 16, 110 –120 Pless, S. A., and Lynch, J. W. (2009) J. Biol. Chem. 284, 27370 –27376 Plested, A. J., Groot-Kormelink, P. J., Colquhoun, D., and Sivilotti, L. G. (2007) J. Physiol. 581, 51–73 Aldea, M., Castillo, M., Mulet, J., Sala, S., Criado, M., and Sala, F. (2010) J. Neurochem. 113, 1036 –1045 Sala, F., Mulet, J., Sala, S., Gerber, S., and Criado, M. (2005) J. Biol. Chem. 280, 6642– 6647 Chakrapani, S., Bailey, T. D., and Auerbach, A. (2004) J. Gen. Physiol. 123, 341–356 McLaughlin, J. T., Fu, J., and Rosenberg, R. L. (2007) Mol. Pharmacol. 71, 1312–1318 Lee, W. Y., Free, C. R., and Sine, S. M. (2008) J. Gen. Physiol. 132, 265–278 Crawford, D. K., Trudell, J. R., Bertaccini, E. J., Li, K., Davies, D. L., and Alkana, R. L. (2007) J. Neurochem. 102, 2097–2109

VOLUME 286 • NUMBER 40 • OCTOBER 7, 2011