Molecular Basis for Zinc Potentiation at Strychnine-sensitive Glycine Receptors*

The divalent cation Zn2+ is a potent potentiator at the strychnine-sensitive glycine receptor (GlyR). This occurs at nanomolar concentrations, which are the predicted endogenous levels of extracellular neuronal Zn2+. Using structural modeling and functional mutagenesis, we have identified the molecular basis for the elusive Zn2+ potentiation site on GlyRs and account for the differential sensitivity of GlyR α1 and GlyR α2 to Zn2+ potentiation. In addition, juxtaposed to this Zn2+ site, which is located externally on the N-terminal domain of the α subunit, another residue was identified in the nearby Cys loop, a region that is critical for receptor gating in all Cys loop ligand-gated ion channels. This residue acted as a key control element in the allosteric transduction pathway for Zn2+ potentiation, enabling either potentiation or overt inhibition of receptor activation depending upon the moiety resident at this location. Overall, we propose that Zn2+ binds to a site on the extracellular outer face of the GlyR α subunit and exerts its positive allosteric effect via an interaction with the Cys loop to increase the efficacy of glycine receptor gating.

The glycine receptor is a major component of inhibitory neurotransmission in the spinal cord and brainstem (1). It forms part of the Cys loop receptor family, which includes acetylcholine, ␥-aminobutyric acid type A (GABA A ), 3 and serotonin type 3 receptors (2). Glycine receptors are pentameric assemblies of ligand binding ␣ (1)(2)(3)(4) subunits and the homologous structural ␤ subunit (3). Each subunit has an extracellular N-terminal domain followed by four transmembrane (TM) segments connected by two intracellular regions and an extracellular TM2-TM3 linker, which is important for receptor gating (4). The function of these receptors can be enhanced by a variety of agents, including alcohols, anesthetics, neurosteroids, and Zn 2ϩ (5)(6)(7)(8), but their exact binding sites and transduction pathways remain controversial.
However, Zn 2ϩ binding sites do provide realistic targets for identification, as these are traditionally compact, consisting of only 3-4 residues, and their coordination chemistry is understood (9). With regard to GlyRs, Zn 2ϩ exhibits biphasic activity, potentiating receptor activation at submicromolar Zn 2ϩ concentrations and causing inhibition at concentrations Ͼ10 M (8). A previous study demonstrates that the mutation D80A in the extracellular domain of the GlyR ␣ 1 subunit ablated Zn 2ϩ potentiation, and it is proposed that this residue partici-pates in the direct coordination of Zn 2ϩ (10,11). However, an additional report indicates that the Zn 2ϩ potentiation of responses to the partial agonist taurine, which binds to the same agonist site as glycine, is unaffected by mutating Asp-80. This suggests that either multiple Zn 2ϩ binding sites exist or that this mutation induces an indirect allosteric effect on receptor function that selectively disrupts Zn 2ϩ potentiation of responses to glycine rather than those to taurine (12). In accord with an indirect allosteric effect, mutations of several other residues in the TM2-TM3 linker are capable of disrupting Zn 2ϩ potentiation, though none of these residues are chemically suitable for the direct coordination of Zn 2ϩ .
The high sensitivity of the strychnine-sensitive GlyR to Zn 2ϩ potentiation makes this receptor an ideal substrate for modulation by basal levels of Zn 2ϩ . In a physiological context, Zn 2ϩ is released following neuronal stimulation (13,14) and can also modulate inhibitory neurotransmitter receptors at basal concentrations (15,16). Furthermore, Zn 2ϩ is concentrated into synaptic boutons that also contain either glutamate, GABA, or glycine in many areas of the brain, including the cortex, hippocampus, and spinal cord (17)(18)(19). Although the predicted concentration of Zn 2ϩ resulting from presynaptic release is probably Ͻ10 M (20), this is more than sufficient to modulate N-methyl-Daspartate receptors, certain GABA A receptor subtypes, and GlyRs (8,21). Indeed, low nanomolar basal Zn 2ϩ concentrations are adequate to prolong the decay phase of glycinergic inhibitory postsynaptic currents (22).
In this study, we accounted for the differential sensitivity to Zn 2ϩ potentiation of GlyR ␣ 1 and GlyR ␣ 2 by identifying the location of a single conserved residue in the N-terminal domain. Subsequently, by using structural homology modeling together with the identified residue underlying Zn 2ϩ sensitivity, we established the molecular determinants for the elusive Zn 2ϩ potentiation binding site. In doing so, we uncovered a prospective transduction residue for this site that is located in the Cys loop-gating domain, providing a plausible molecular pathway for Zn 2ϩ potentiation of glycine receptor gating.

MATERIALS AND METHODS
cDNA Constructs-Human (h) wild-type cDNA constructs were used for hGlyR ␣ 1L , hGlyR ␣ 2A , and hGlyR ␤. Site-specific mutant cDNAs were prepared using the Stratagene Quikchange mutagenesis kit. The mutated sequences were confirmed by complete sequencing of the cDNA insert using an ABI sequencer.
Cell Culture and Transfection-Human embryonic kidney (HEK) cells (American Type Culture Collection CRL1573) were grown and transfected as previously documented (23). Plasmids of hGlyR cDNA clones were co-transfected in a ratio of 1:1 with enhanced green fluorescent protein (24). To co-express GlyR ␣␤ heteromers, ␤ subunit cDNA was added in excess at a ratio of 20:1 to ␣ subunit cDNA. HEK cells were plated onto poly-L-lysine-coated coverslips (100 g/ml) sufficient to achieve 20% confluence and used for recording on the next day.
Solutions-The internal pipette solution contained (mM): 140 KCl, 2 MgCl 2 , 1 CaCl 2 , 10 HEPES, 11 EGTA, and 2 ATP, pH 7.2 (Ϸ 300 mosM). The external Krebs solution consisted of (mM): 140 NaCl, 4.7 KCl, 1.2 MgCl 2 , 2.5 CaCl 2 , 10 HEPES, and 11 D-glucose, pH 7.4 (Ϸ 300 mosM). For those experiments requiring control over the basal levels of Zn 2ϩ (see Fig. 1 Electrophysiology-An Axopatch 200B amplifier (Axon Instruments) recorded whole-cell currents from single HEK cells using the patch clamp technique. HEK cells exhibited resting potentials between Ϫ10 and Ϫ40 mV and were voltage-clamped at a Ϫ40 mV holding potential. The cells were visualized with differential interference contrast optics using a Nikon Optiphot microscope with an epifluorescence attachment to identify green fluorescent protein-transfected cells. A Y-tube was used to rapidly apply drugs and Krebs solutions (exchange rate ϳ50 -100 ms) to the recorded cells. Patch electrodes were fabricated using a Narashige PC-10 puller with resistances, after polishing, of 4 -5 megohms. All recordings were performed in constantly perfusing Krebs-Ringer solution at room temperature (20 -22°C).
Data Acquisition and Analysis-Recorded currents were filtered using a high pass Bessel filter at 3 kHz (Ϫ36 db/octave), and series resistance compensation was achieved up to 70%. Data were recorded in 20-s acquisition epochs directly to a Pentium IV, 1.8 GHz computer into Clampex software, version 8.0, via a Digidata 1322A (Axon Instruments) sampling at 200-s intervals. Zn 2ϩ was co-applied with the agonist to attenuate any delayed onset of Zn 2ϩ -mediated inhibition. Strychnine and picrotoxin were pre-incubated for 15 s, sufficient to attain equilibrium, and then also co-applied with the agonist. The digitized membrane current records were analyzed off-line using Axoscope, version 8.2. Biphasic (potentiation and inhibition) Zn 2ϩ concentration response curves were fitted according to a modified Hill equation as previously described (26). Where a single component to the concentration response relationships was evident, it was fitted with a form of the Hill equation. For the agonist and Zn 2ϩ concentration potentiation curves, I ϭ I min ϩ (I max Ϫ I min )([1/(1 ϩ (EC 50 /A) nH )]), and for the strychnine and Zn 2ϩ concentration-inhibition curves, The EC 50 represents either the concentration of agonist-inducing or Zn 2ϩ -potentiating (A) 50% of the maximal current (I max ) evoked or potentiated by a saturating concentration of agonist or Zn 2ϩ , and n is the Hill coefficient. For Zn 2ϩ potentiation, I min represents the control glycine current in the absence of Zn 2ϩ and was set to 100%. For inhibition, the IC 50 defines the antagonist concentration (B) producing a 50% inhibition of the current, and m H represents the Hill coefficient. When Zn 2ϩ induced a biphasic inhibition, the concentration response data were fitted with a two-component inhibitory curve using I/I max ϭ 1 Ϫ (aB mH /(B mH ϩ IC 50 mH )) ϩ (bB mH /(B mH ϩ IC 50 mH )), where a and b represent the relative proportions of each inhibitory component. All statistical comparisons used an unpaired t test.
Structural Homology Modeling-The mature N-terminal extracellular domain of the hGlyR ␣ 1 subunit was modeled on the crystal structure of the acetylcholine-binding protein (27) using SwissProt DeepView, version 3.7, in accordance with a ClustalW protein alignment. All three-dimensional images were subsequently rendered using the freeware program POV-Ray.

RESULTS
GlyR ␣ 1 and ␣ 2 Exhibit Distinct Sensitivities to Zn 2ϩ Potentiation-The sensitivity of the GlyR ␣ 1 and ␣ 2 subtypes to Zn 2ϩ potentiation was assessed from whole-cell recordings of half-maximal (EC 50 ) responses to glycine obtained from transfected HEK cells maintained at Ϫ40 mV FIGURE 1. Zn 2؉ potentiation of glycine-activated currents for GlyR ␣ 1 and GlyR ␣ 2 . Zn 2ϩ concentration response curves were determined for the modulation of EC 50 responses to glycine recorded in the presence of 10 mM tricine from HEK cells. A, typical glycine (EC 50 )-activated currents in the presence and absence of 30 M Zn 2ϩ for GlyR ␣ 1 , ␣ 2 , and ␣ 2 E201D. Zn 2ϩ concentration curves for homomeric GlyR ␣ 1 , ␣ 2 , or ␣ 2 E201D (B) or heteromeric GlyR ␣ 1 ␤ or ␣ 2 ␤ (C) (n ϭ 4 -6). D, amino acid sequence alignment of GlyR ␣ 1 and GlyR ␣ 2 showing the region of the N-terminal extracellular domain that contains differences in potential Zn 2ϩ binding residues (bold). A gray filled box highlights the residue responsible for the differential sensitivity to Zn 2ϩ -mediated potentiation between GlyR ␣ 1 and ␣ 2 . All numbering is for the mature protein.
holding potential (Fig. 1A). To accurately compare the modulation of GlyR ␣ 1 and GlyR ␣ 2 at submicromolar concentrations of Zn 2ϩ , the buffer tricine (25) was used to remove Zn 2ϩ contamination in the external Krebs solution (22). The ␣ 1 subtype, considered to be an adult form of the GlyR and highly expressed throughout the spinal cord and brainstem (1), exhibited a very high sensitivity to Zn 2ϩ potentiation with an EC 50 of only 37 Ϯ 10 nM (n ϭ 5) (Fig. 1B). This sensitivity is comparable with the high affinity inhibitory Zn 2ϩ site found on the N-methyl-Daspartate receptor NR2A subunit (25). By comparison, the principal embryonic subtype GlyR ␣2 (1) also displayed a high nanomolar sensitivity to Zn 2ϩ potentiation with an EC 50 of 540 Ϯ 180 nM (n ϭ 5) but was nevertheless 15-fold less sensitive than the GlyR ␣ 1 isoform (p Ͻ 0.05). This difference in sensitivity was unaffected by co-expression with the GlyR ancillary ␤ subunit (p Ͻ 0.05) (Fig. 1C), which assemble to form ␣␤ heteromeric receptors, as confirmed by an approximate 20-fold shift in the sensitivity to picrotoxin (data not shown) (28). The Hill slopes for Zn 2ϩ -mediated potentiation varied between 1.2 and 1.5, suggesting more than one Zn 2ϩ ion was probably coordinated by each receptor. To identify the structural determinant(s) responsible for differential Zn 2ϩ sensitivity, the extracellular domains were scanned for the classical Zn 2ϩ binding residues, Cys, Asp, Glu, and His (9), focusing particularly on differences between GlyR ␣ 1 and ␣ 2 . On this basis, four residues were prioritized (Fig. 1D). Upon individual substitution of the GlyR ␣ 1 variants into GlyR ␣ 2 , no influence on Zn 2ϩ sensitivity was observed for S179E, D180Q, or E187D mutated receptors (data not shown). However, a single conservative E201D substitution was found to be sufficient and necessary to enable GlyR ␣ 2 to exhibit a similar sensitivity to GlyR ␣ 1 toward potentiation by Zn 2ϩ (Fig. 1, B and D).
A Cluster of GlyR ␣ 1 Residues Are Essential for Zn 2ϩ Potentiation-The discovery that GlyR ␣ 1 Asp-194 (which corresponds to GlyR ␣ 2 Glu-201) is capable of influencing the sensitivity to potentiating Zn 2ϩ , suggested that this moiety might be a direct contributor to Zn 2ϩ binding, forming part of the potentiation site. This hypothesis was examined by mutating ␣ 1 Asp-194 to alanine, a residue incapable of coordinating Zn 2ϩ . However, because of the biphasic nature of Zn 2ϩ action at the GlyR, where Zn 2ϩ potentiates at nanomolar to low micromolar concentrations and inhibits at doses Ͼ10 M (8), any attempt to measure a reduced sensitivity to Zn 2ϩ potentiation might be occluded by the onset of Zn 2ϩ mediated inhibition. To obviate this problem, all experiments to identify the Zn 2ϩ potentiation binding site were performed using an H107N "background" mutation (hereafter referred to as "reduced inhibition" (RI)). This mutation dramatically attenuated the GlyR sensitivity to Zn 2ϩ inhibition, increasing the Zn 2ϩ IC 50 from 15 M to Ͼ3 mM without affecting other macroscopic properties of the receptor, particularly the sensitivity of the receptor to Zn 2ϩ potentiation (see below next paragraph and TABLE ONE).
To investigate the high potency Zn 2ϩ potentiation phenomenon on the background of a low sensitivity Zn 2ϩ inhibition component, it was necessary to fully characterize the modulatory curves over a wide concentration range from 0.01 M to 3 mM. As tricine can only effectively buffer Zn 2ϩ concentrations below 1 M, it was not included in these comparisons. In accord with previous studies (10,22), the apparent sensitivities to potentiating Zn 2ϩ of the wild-type receptor and RI␣ 1 were lower in the absence of tricine because of the competing background Zn 2ϩ present in the external solution, although the relative EC 50 values remained indistinguishable between the two receptors, (Zn 2ϩ EC 50 values: To assess the significance of prospective binding site residues for Zn 2ϩ potentiation, the consequences of their replacement were compared for two GlyR agonists, glycine and taurine. This is necessary, as previously identified residues, which have been postulated to be part of a potentiating Zn 2ϩ binding site, ablated enhancement of responses to one agonist but not the other presumably due to indirect affects on downstream transduction mechanisms (12). This emphasized the potential importance of Asp-194, because the Zn 2ϩ potentiation of both glycine-and taurine-activated (EC 50 ) responses was ablated in the RI␣ 1 D194A receptor (Fig. 2, A and B). As expected, removing Zn 2ϩ potentiation led to an apparent increase in sensitivity to Zn 2ϩ inhibition, with the IC 50 shifting from Ͼ3000 to 270 Ϯ 50 M (n ϭ 3). This is probably a consequence of Zn 2ϩ potentiation and inhibition having overlapping concentration ranges; therefore, by removing potentiation, the inhibitory component appears to have a lower threshold. We cannot discount the possibility that there is some allosteric interaction between the two Zn 2ϩ sites, but this would seem unlikely, as disruption of the inhibitory site does not affect Zn 2ϩ potentiation (29). Furthermore, although not a guarantee of independence, Asp-194 of the putative potentiation site is predicted to reside on the external face of the GlyR N terminus (Fig. 2C), far away from the inhibitory Zn 2ϩ site located on the opposite face of the subunit (30).
To elucidate which residues were capable of interacting with GlyR ␣ 1 Asp-194 in forming the putative Zn 2ϩ potentiation binding site, classical Zn 2ϩ binding residues were selected from motifs predicted to lie structurally close to Asp-194, according to a GlyR homology model based on the acetylcholine-binding protein (Fig. 2C). The selected residues Asp, Cys, Glu, His, and also a Thr (threonine residues have been previously implicated in Zn 2ϩ inhibition of GlyRs) (10,29) were sequentially substituted with alanine. This strategy identified Glu-192 and His-215 as potential contributors to the Zn 2ϩ potentiation site (Fig. 2C). The mutated GlyR RI␣ 1 E192A was unresponsive to Zn 2ϩ potentiation, whereas RI␣ 1 H215A exhibited a markedly reduced sensitivity to Zn 2ϩ potentiation by ϳ30-fold (Fig. 2, A and B). The EC 50 for Zn 2ϩ potentiation of glycine responses was increased from 0.8 Ϯ 0.2 M for RI␣ 1 to 22 Ϯ 4 M for RI␣ 1 H215A (n ϭ 4; p Ͻ 0.05). This reduction in Zn 2ϩ sensitivity was directly comparable with that observed when the GlyR was activated by taurine (EC 50 , 0.9 Ϯ 0.3 to 21 Ϯ 4 M, respectively; n ϭ 4; p Ͻ 0.05). As previously, removing or reducing Zn 2ϩ potentiation increased the apparent sensitivity to Zn 2ϩ inhibition for both of these mutants. Moreover, substituting other acidic candidate residues for alanines (highlighted in Fig. 2C) had no effect on the potency of Zn 2ϩ potentiation. For all three RI mutants ␣ 1 E192A, ␣ 1 D194A, and ␣ 1 H215A, the maximal responses evoked by the agonists glycine and taurine were indistinguishable from the wild-type receptor, and only RI␣ 1 D194A exerted a modest 3-fold increase in the agonist EC 50 values (TABLE ONE), suggesting that these mutations selectively affected the Zn 2ϩ potentiation binding site and did not exert a general perturbation on GlyR function. Most importantly, the GlyR homology model (Fig.  2C) predicts that Glu-192, Asp-194, and His-215 reside in close proximity to one another on the outside face of the N-terminal extracellular domain. Typically, functional groups involved in direct coordination of Zn 2ϩ lie within 2-5 Å of the divalent ion (9), which is easily accommodated by the predicted distances between the three residues identified on the GlyR model (Fig. 2C). To highlight the localized specific role this domain plays in Zn 2ϩ potentiation, scanning alanine mutagenesis was performed on other residues that lie immediately to either side of the putative Zn 2ϩ binding site and that are also predicted to have externally orientated side chains. These mutated GlyRs (RI mutants ␣ 1 K190A, ␣ 1 R196A, ␣ 1 R213A, and ␣ 1 E217A) did not affect the glycine, taurine, or Zn 2ϩ EC 50 values, the maximal Zn 2ϩ potentiation, or the maximal glycine-activated current (supplemental Fig. 1). In addition, co-expression of a GlyR RI␣ 1 E192A with the ␤ subunit did not recover Zn 2ϩ -mediated potentiation, suggesting the ␤ subunit is unable to compensate or provide a Zn 2ϩ potentiation site of its own (n ϭ 3) ( Fig. 2A).
Asymmetry of Function at the Putative Zn 2ϩ Potentiation Binding Site-Demonstrating that this discrete domain is accessible to water is a vital requirement for any dynamic Zn 2ϩ binding site and would strengthen the conclusion that the identified residues may act as direct coordinators of Zn 2ϩ . We determined this by individual cysteine substitutions of GlyR ␣ 1 Glu-192, Asp-194, and His-215, which were then exposed to the cysteine-modifying reagent MTSEA. If MTSEA covalently binds to the potentiation site, it will replace a Zn 2ϩ -coordinating Cys moiety with a positively charged amine group, which should attenuate the Zn 2ϩ potentiation of glycine-activated currents. Pre-application of 3 mM MTSEA for 1 min to the control GlyR RI␣ 1 did not affect Zn 2ϩ potentiation (Fig. 3A) or the glycine EC 50 and glycine maximal responses (data not shown). Upon individual replacement of Glu-192, Asp-194, and His-215 with cysteine, GlyRs were generated that retained high sensitivities to Zn 2ϩ potentiation, with EC 50 values for Zn 2ϩ within 10-fold of the wild-type receptor (n ϭ 4) (Fig. 3, B-D). This was not surprising, as cysteine is quite capable of coordinating Zn 2ϩ . However, following exposure to MTSEA, RI␣ 1 D194C and RI␣ 1 H215C were rendered unresponsive to Zn 2ϩ potentiation, suggesting that the side chains of both of these residues are surface-exposed and important for Zn 2ϩ potentiation. The sensitivity of RI␣ 1 E192C, however, was largely unaffected FIGURE 2. GlyR ␣ 1 subunit residues that affect Zn 2؉ potentiation. Zn 2ϩ concentration response curves for the modulation of EC 50 responses to glycine (A) and taurine (B) constructed for mutant homomeric GlyRs RI␣ 1 , RI␣ 1 E192A, RI␣ 1 D194A, and RI␣ 1 H215A and one heteromeric GlyR, RI␣ 1 E192A␤. All experiments were performed on a background mutant receptor, GlyR ␣ 1 H107N, that exhibited a reduced inhibition (RI) to Zn 2ϩ . Note the absence of any Zn 2ϩ potentiation for most of the mutant receptors, apart from RI␣ 1 and RI␣ 1 H215A. The insets show typical glycine and taurine (EC 50 )-activated currents in the absence (continuous line) and presence of 10 M Zn 2ϩ (dotted line). C, colorcoded amino acid motifs identified in the N-terminal domain of the GlyR ␣ 1 subunit and homologous GlyR ␣ and ␤ subunits that reside in close proximity to the previously identified ␣ 1 D194A. The single extracellular domain in the GlyR structural model illustrates the side chains of three putative Zn 2ϩ binding residues. The ␣-helical section at the start of the mature protein is shown in pink. The inset is a plan view of the GlyR pentamer (red circles represent the extracellular subunit N-terminal domains), and the arrow denotes the viewing angle. The protein alignments show potential Zn 2ϩ binding residues in bold in each motif, whereas important residues for Zn 2ϩ potentiation are on color-coded backgrounds. A divergent ␣␤ residue in the Cys loop is also highlighted in gray.
(n ϭ 4) (Fig. 3, B-D). The binding of MTSEA alone was insufficient to induce potentiation of glycine-activated responses at RI␣ 1 D194C or RI␣ 1 H215C (Fig. 3, B and C). Surprisingly, MTSEA did increase glycine potency in the absence of Zn 2ϩ for RI␣ 1 E192C, potentiating the glycine response by 61 Ϯ 16% (n ϭ 4, p Ͻ 0.05) (Fig. 3E). Thus, Glu-192 appears to be accessible to MTSEA, although this covalent modification did not affect the ability of Zn 2ϩ to bind to the receptor. If the carboxyl side chain of Glu-192 was not directly coordinating Zn 2ϩ , then perhaps substitution of this residue with alanine would perturb the ␤-strand backbone, which then would either indirectly disrupt the Zn 2ϩ potentiation site or disrupt a possible contribution of the polar peptide backbone at this locus for Zn 2ϩ coordination. To determine the relevance of the backbone at Glu-192, this residue was mutated to proline (RI␣ 1 E192P), an amino acid associated with placing conformational restraints upon peptide backbones (31). Even though proline cannot coordinate Zn 2ϩ , potentiation in this mutated receptor was retained, although at a 5-fold reduced sensitivity (0.8 Ϯ 0.2 M for RI␣ 1 and 4.2 Ϯ 0.8 M for RI␣ 1 E192P (n ϭ 4, p Ͻ 0.05) (Fig. 3F). This suggested that, to retain Zn 2ϩ -mediated potentiation, this region must retain a specific conformation of the backbone at Glu-192, which can be accom-modated by the introduction of a proline but not by insertion of an alanine. Of course, without precise structural data, it is not possible to infer any specific details about the structural organization at this locus other than that the backbone is particularly sensitive to perturbation in a fashion that affects Zn 2ϩ potentiation. As a control for this strategy, the mutant GlyR RI␣ 1 D194P ablated Zn 2ϩ -mediated potentiation, an outcome that is expected if, indeed, the side chain moiety is contributing to Zn 2ϩ coordination at this particular position (data not shown).
Although the current data most strongly support a role for GlyR ␣ 1 Asp-194 and His-215 in direct Zn 2ϩ coordination, the GlyR RI␣ 1 H215A substitution attenuated the sensitivity to Zn 2ϩ , whereas GlyR RI␣ 1 D194A entirely removed Zn 2ϩ potentiation (Fig. 2, A and B). To examine the actual extent by which Zn 2ϩ -mediated potentiation was affected by perturbation of Asp-194, it was necessary to further disrupt the competing inhibitory Zn 2ϩ site such that inhibition does not occlude any remaining sensitivity to Zn 2ϩ potentiation. A number of GlyRs were generated to ablate Zn 2ϩ -mediated inhibition based on previously identified targets (10,26,29). Of these, the one that most effectively attenuated Zn 2ϩ -mediated inhibition while still retaining mostly "normal" receptor function in terms of agonist specificity, sensitivity, Shown are Zn 2ϩ concentration response curves for the modulation of EC 50 glycine responses measured before and after exposure to 3 mM MTSEA for 1 min for the receptors RI␣ 1 (A) and for those receptors mutated to incorporate cysteines at putative Zn 2ϩ binding residues RI␣ 1 D194C (B), RI␣ 1 H215C (C), and RI␣ 1 E192C (D) (n ϭ 4). The insets present example glycine (EC 50 )-activated currents in the absence (solid line) and presence (broken line) of 10 M Zn 2ϩ before and after 3 mM MTSEA. E, glycine (EC 50 ) current amplitudes for RI␣ 1 and RI␣ 1 E192C before and after 3 mM MTSEA. The inset depicts typical glycine-activated currents for the same concentration of glycine before and after exposure to MTSEA. F, Zn 2ϩ concentration response curves for the modulation of EC 50 glycine responses on RI␣ 1 and RI␣ 1 E192P to explore whether inducing specific conformational restraints on the peptide backbone at this location affects Zn 2ϩ potentiation (n ϭ 4). * denotes significance at p Ͻ 0.05 and maximal activation was the RI␣ 1 H109F receptor (TABLE ONE). On this new mutant background, we introduced the mutation D194K to prevent any Zn 2ϩ binding by replacing aspartate with a positively charged amine group. The resultant GlyR RI␣ 1 H109F,D194K was unaffected by 0.1-100 M Zn 2ϩ (n ϭ 4), a concentration range over which the wild-type GlyR ␣ 1 underwent its full Zn 2ϩ modulatory profile. This demonstrated that the potentiating Zn 2ϩ site was effectively ablated up to 100 M Zn 2ϩ , a concentration that is 125-fold greater than the Zn 2ϩ EC 50 value of 0.8 Ϯ 0.3 M on the wild-type GlyR ␣ 1 (Fig. 4).
GlyR ␣ 1 Thr-151 Is a Critical Control Element for Zn 2ϩ Potentiation-Besides the classical Zn 2ϩ binding moieties of the potentiating site, we also identified a nearby polar residue at position 151 located in the Cys loop called L7 (27). Alanine substitution of threonine 151 generated an RI␣ 1 T151A GlyR that was insensitive to the potentiating effect of Zn 2ϩ (Fig. 5, A and D). Although the lack of potentiation was clear, there was also an unusual additional effect revealed in the form of a novel biphasic sensitivity to Zn 2ϩ inhibition, with high (IC 50 ϭ 1.6 Ϯ 0.6 M, n ϭ 5) and low potency components (IC 50 ϭ 1040 Ϯ 290 M, n ϭ 5). This biphasic inhibitory profile was also apparent when taurine was the agonist (IC 50 ϭ 3.2 Ϯ 0.8 M and 2840 Ϯ 820 M, n ϭ 5) (Fig. 5B). Intriguingly, the IC 50 value for the high sensitivity inhibitory component is directly comparable with the original Zn 2ϩ EC 50 value of 0.8 Ϯ 0.2 M for potentiation at RI␣ 1 . Conceivably, the ␣ 1 T151A substitution may have converted the Zn 2ϩ potentiation site to a high sensitivity Zn 2ϩ inhibitory site. In support of this hypothesis, incorporating the mutation E192A (producing GlyR RI␣ 1 T151A,E192A), designed to disrupt the Zn 2ϩ potentiating site, mostly removed the high potency inhibitory component ( Fig. 5A and TABLE TWO). Furthermore, the lower potency component was attributed to the previously identified Zn 2ϩ inhibitory site (26), because restoration of this site by reinstating His-107 (producing GlyR ␣ 1 T151A) increased the relative sensitivity to inhibition (Fig. 5, A and B). The effects of similar experiments, introducing D194A or H215A onto the RI␣ 1 T151A-mutated GlyR, were not possible, as these receptors (RI␣ 1 T151A,D194A , n ϭ 24; and RI␣ 1 T151A,H215A, n ϭ 16) were effectively non-functional with very low maximal currents (TABLE TWO).
To further characterize the role of Thr-151, this residue was mutated to two variants selected from the Cys loop receptor family, including an arginine from the GlyR ␤ subunit, which previously appeared incapable of supporting Zn 2ϩ potentiation, and an Asn from the serotonin type 3A receptor, which displays a comparable sensitivity profile to the GlyR ␣ subunit with regard to Zn 2ϩ potentiation (32,33). These mutated GlyRs, RI␣ 1 T151R and RI␣ 1 T151N, failed to support Zn 2ϩ potentiation and instead revealed biphasic inhibitory profiles for glycine-activated responses (Fig. 5C). The sensitivity of the high potency inhibitory component was comparable with that seen for RI␣ 1 T151A (RI␣ 1 T151R IC 50 ϭ 3.6 Ϯ 0.7 M and RI␣ 1 T151N IC 50 ϭ 1.6 Ϯ 0.7 M; n ϭ 4; p Ͼ 0.05). However, the maximal contribution of the high potency inhibitory component was reduced to 39 Ϯ 3% for RI␣ 1 T151R and to just 6.5 Ϯ 0.3% for RI␣ 1 T151N from 73 Ϯ 8% for RI␣ 1 T151A (Fig. 5C), suggesting the nature of the residue at position 151 was important in determining both the direction and the extent of the Zn 2ϩ effect.
GlyR ␣ 1 Thr-151 Influences Apparent Agonist Gating-As Thr-151 resides in the Cys loop, a domain that is important for agonist gating in this receptor superfamily (34), each of these mutations was assessed for their effect on agonist potencies. All of the mutations ␣ 1 T151A, RI␣ 1 T151A, RI␣ 1 T151R, RI␣ 1 T151A,E192A, and RI␣ 1 T151N caused a progressive reduction in sensitivity to both glycine and taurine with RI␣ 1 T151N demonstrating 11-and 23-fold increases in the EC 50 values for glycine and taurine, respectively (n ϭ 5; p Ͻ 0.05) (Fig. 6, A and B; TABLE TWO). In accord with the possibility of Thr-151 being involved in ion channel gating, the percentage of maximum current evoked by the lower potency agonist taurine compared with maximal glycine responses in the same cell decreased significantly from 100% and 98 Ϯ 2% in RI␣ 1 and RI␣ 1 T151A, respectively, to 90 Ϯ 3% for RI␣ 1 T151R, 65 Ϯ 4.3% for RI␣ 1 T151A,E192A and 46 Ϯ 6% for   Fig. 2), and its proximity to another important determinant in gating, the TM2-TM3 linker region (maroon), which rests above the transmembrane helices that form the receptor ion channel (shaded blue). The inset depicts the viewing angle (blue arrow) of the GlyR pentamer. RI␣ 1 T151N (n ϭ 4 -7; p Ͻ 0.05) (Fig. 6, B and E; TABLE TWO). In addition, the maximum glycine-evoked currents were also significantly reduced from 4.5 Ϯ 0.4 nA for wild-type GlyR ␣ 1 to 1.9 Ϯ 0.5 nA for RI␣ 1 T151A,E192A and to 2.9 Ϯ 0.5 nA for RI␣ 1 T151N (n ϭ 5-12, p Ͻ 0.05) (Fig. 6D). These mutated receptors did not substantially distort the region of agonist binding, as sensitivities to the competitive antagonist strychnine were directly comparable with the wild-type GlyR (Fig. 6C). The non-functional nature of the receptors RI␣ 1 T151A,D194A and RI␣ 1 T151A,H215A precluded their study in this experiment (TABLE TWO). Additional mutations at GlyR ␣ 1 Thr-151 to Cys, Asp, Glu, Ser, and Phe revealed no obvious relationship between side chain polarity and the volume requirements of a residue occupying position 151 for receptor activation (supplemental Table 1; supplemental Fig. 2).

DISCUSSION
This study reports the first molecular description of a Zn 2ϩ potentiation site on a Cys loop ligand-gated ion channel. The residues Asp-194, His-215, and the peptide backbone located at Glu-192 in GlyR ␣ 1 are all predicted to reside in close structural proximity to one another, and they all influence Zn 2ϩ potentiation in accord with a role in binding. Each of these residues is chemically adept at coordinating Zn 2ϩ , and when this capacity is annulled, through alanine substitution, the sensitivity to Zn 2ϩ was either attenuated (as for RI␣ 1 H215A) or entirely ablated (as for RI␣ 1 E192A and RI␣ 1 D194A). Moreover, experiments using MTSEA demonstrated that the residues in this putative site are accessible to this water-soluble compound and therefore must also be accessible to dynamic Zn 2ϩ binding. Additionally, as Zn 2ϩ potentiation of both glycine-and taurine-activated currents was similarly affected by these mutations, it is likely these residues are either part of a universal Zn 2ϩ binding site or participate in the process of allosteric signal transduction from such a binding site. Finally, the analysis of the ␣ 1 E192P receptor suggested that this location might contribute structurally to the site in a manner dependent upon the restraints of the peptide backbone. If the backbone itself can contribute to Zn 2ϩ binding, then the site isolated here requires only a fourth coordinating ligand for completion, and in the case of reversible Zn 2ϩ binding catalytic sites, this is predominantly provided by an activated water molecule (9).
In addition to those residues thought to line the Zn 2ϩ binding site, Thr-151 was also identified as an important transduction component for this site because of its nearby location. The exact nature of the residue introduced at position 151 did not alter the potency of Zn 2ϩ , precluding an involvement in binding, but it was able to determine the "direction of output" from the Zn 2ϩ site to be either potentiating (for threonine) or inhibitory (for alanine, arginine, and asparagine). Furthermore, the type of residue at position 151 also controlled the efficacy of inhibition from the Zn 2ϩ site. As this site is quite distinct in its structure and location, Thr-151 is most unlikely to be associated with the previously reported Zn 2ϩ inhibitory site on GlyR ␣ 1 , which resides on the other side of the subunit (10,26,30). In accordance with the location of Thr-151 being in the critical Cys loop-gating domain (34 -36), this residue was also shown to be an important determinant of agonist potency for receptor activation. Thus, from a molecular perspective, it provides a potential connection between the Zn 2ϩ potentiation binding site and the Cys loop-gating domain to possibly increase the efficacy of agonistinduced channel opening. In accordance with Zn 2ϩ potentiation being mediated via an increase in agonist efficacy, the partial agonist taurine is converted to a full agonist by Zn 2ϩ concentrations that cause potentiation at GlyR ␣ 1 expressed in oocytes (10). Furthermore, residues located in the TM2-TM3 linker, which are predicted to be closely apposed to, and capable of interacting with, the Cys loop-gating domain (27,37), also affect Zn 2ϩ -mediated potentiation (12), as would be expected if there is direct communication between these two domains.
The reversal of signal output, exemplified here by changing Zn 2ϩ potentiation to inhibition following the mutation of Thr-151, is not a unique observation for ligand-gated ion channels. Mutation of isoleucine 307 in the TM2 domain of the GABA C receptor to glutamine reverses the inhibition caused by the neurosteroid 5␤-pregnane-3␣-ol-20-one on the wild-type receptor to potentiation (38). This suggests that whether a modulator exerts a positive or negative effect on receptor activation may be highly dependent on the nature of the allosteric transduction pathway leading away from the binding site.

CONCLUSIONS
We report here the identification of the Zn 2ϩ potentiation binding site on the GlyR and a possible molecular route of action for Zn 2ϩ potentiation via the Cys loop channel-gating domain. The Zn 2ϩ potentiation site is a novel site on the GlyR clearly distinct from the previously identified interfacial inhibitory Zn 2ϩ binding sites for GlyRs (30) and GABA A receptors (39), which reside on the inside faces of these pentamers. The potentiation site has a high sensitivity to Zn 2ϩ , with the EC 50 estimated at Ͻ100 nM, well within the range of current estimates for physiological levels of basal and released Zn 2ϩ (40). The location of this site away from the intrasubunit interfaces of the GlyR pentamer may have implications for the manner in which Zn 2ϩ achieves its effect, because when Zn 2ϩ is bound at the potentiation site, it is predicted to interact with the gating apparatus of the receptor. In contrast, when Zn 2ϩ is bound at the interfacial inhibitory site, it acts in an apparent competitive manner, stabilizing the closed agonist-unbound state of the receptor (1).