Identification of an inhibitory Zn2+ binding site on the human glycine receptor α1 subunit

Authors


  • Robert J. Harvey and Philip Thomas made equal contributions to this study and should be considered as equal ‘first authors’.

Abstract

  • 1Whole-cell glycine-activated currents were recorded from human embryonic kidney (HEK) cells expressing wild-type and mutant recombinant homomeric glycine receptors (GlyRs) to locate the inhibitory binding site for Zn2+ ions on the human α1 subunit.
  • 2Glycine-activated currents were potentiated by low concentrations of Zn2+ (<10 μm) and inhibited by higher concentrations (>100 μm) on wild-type α1 subunit GlyRs.
  • 3Lowering the external pH from 7.4 to 5.4 inhibited the glycine responses in a competitive manner. The inhibition caused by Zn2+ was abolished leaving an overt potentiating effect at 10 μm Zn2+ that was exacerbated at 100 μm Zn2+.
  • 4The identification of residues involved in the formation of the inhibitory binding site was also assessed using diethylpyrocarbonate (DEPC), which modifies histidines. DEPC (1 mm) abolished Zn2+-induced inhibition and also the potentiation of glycine-activated currents by Zn2+.
  • 5The reduction in glycine-induced whole-cell currents in the presence of high (100 μm) concentrations of Zn2+ did not increase the rate of glycine receptor desensitisation.
  • 6Systematic mutation of extracellular histidine residues in the GlyR α1 subunit revealed that mutations H107A or H109A completely abolished inhibition of glycine-gated currents by Zn2+. However, mutation of other external histidines, H210, H215 and H419, failed to prevent inhibition by Zn2+ of glycine-gated currents. Thus, H107 and H109 in the extracellular domain of the human GlyR α1 subunit are major determinants of the inhibitory Zn2+ binding site.
  • 7An examination of Zn2+ co-ordination in metalloenzymes revealed that the histidine- hydrophobic residue-histidine motif found to be responsible for binding Zn2+ in the human GlyR α1 subunit is also shared by some of these enzymes. Further comparison of the structure and location of this motif with a generic model of the GlyR α1 subunit suggests that H107 and H109 participate in the formation of the inhibitory Zn2+ binding site at the apex of a β sheet in the N-terminal extracellular domain.

The divalent cation Zn2+ is concentrated within nerve terminals and packaged into synaptic vesicles in the mammalian central nervous system (Frederickson, 1989). Zinc ions can also be released into the synaptic cleft following nerve terminal stimulation (Assaf & Chung, 1984; Howell et al. 1984) resulting in multiple effects on neuronal excitability by inhibiting or potentiating current flow through ligand-gated and voltage-operated ion channels (Smart et al. 1994; Harrison & Gibbons, 1994). In particular, Zn2+ modulates inhibitory γ-aminobutyric acid type A (GABAA) and C (GABAC) receptors and also strychnine-sensitive glycine receptors (GlyRs) by binding to distinct site(s) on these receptor complexes. While some advances have recently been made in identifying determinants of Zn2+ binding sites on GABAA and GABAC receptor subunits (Wang et al. 1995; Wooltorton et al. 1997; Horenstein & Akabas, 1998; Fisher & McDonald, 1998), the corresponding site(s) on GlyRs remain unexplored. A complicating factor has been that Zn2+ exerts a biphasic effect on GlyRs (Bloomenthal et al. 1994; Laube et al. 1995). At low concentrations (0.5–10 μm) Zn2+ potentiates glycine-activated chloride currents, while at higher concentrations (>100 μm) Zn2+ inhibits the responses to glycine. Initial interpretations of this behaviour favoured Zn2+-induced changes in agonist binding affinity (Laube et al. 1995); however, a recent study (Lynch et al. 1998) has suggested that Zn2+ potentiation of GlyR function probably involves a complex allosteric process. For example, in GlyRs incorporating mutations in the first transmembrane domain (TM1)-TM2 or TM2-TM3 loops (which are known components of the agonist gating mechanism for GlyRs), potentiation of glycine-gated currents by Zn2+ was abolished, but Zn2+ potentiation of taurine currents was preserved. Interestingly, none of these mutations disrupted Zn2+ inhibition of glycine- or taurine-gated currents (Lynch et al. 1998). Lynch et al. (1998) also demonstrated that aspartate (D) 80, previously considered to contribute to part of the potentiating site (Laube et al. 1995), does not fulfil this role. Instead, it is likely to form part of the allosteric pathway linking Zn2+ binding to the agonist-gating mechanism. This was revealed by the mutant D80A, in which Zn2+ potentiation of glycine-gated currents was abolished, but the ability of Zn2+ to enhance glycine and taurine binding was unimpaired, and potentiation of taurine-gated currents by Zn2+ remained intact.

Taken together these studies suggest that the potentiating action and/or binding site for Zn2+ on GlyRs is complex and susceptible to numerous discrete mutations; however, this circumstance does not appear to pertain to the inhibitory Zn2+ binding site. It is also noteworthy that recent studies have implicated histidine residues in the formation of inhibitory Zn2+ binding sites on the GABAC receptor ρ1 subunit (Wang et al. 1995), and the GABAA receptor α6 (Fisher & MacDonald, 1998), β1 (Horenstein & Akabas, 1998) and β3 (Wooltorton et al. 1997) subunits. Interestingly, the exact position of the histidine residues involved shows some variation. In the human ρ1 subunit (Wang et al. 1995) H141 is extracellular, but in the rodent β1 and β3 subunits, the histidines are located at an equivalent position (H267) in the second membrane-spanning domain; however, H273 in the α6 subunit is found in the extracellular TM2-TM3 loop. We therefore decided to assess the possible role of histidines in the formation of the GlyR inhibitory Zn2+ binding site using a functional expression-site-directed mutagenesis approach. We conclude that H107 and H109 in the large extracellular domain are major determinants of the inhibitory Zn2+ binding site and/or inhibitory effect. Interestingly, these two histidine residues (H107 and H109) are separated by a hydrophobic residue in the GlyR α1 subunits which represents a characteristic motif of Zn2+ binding site(s) in a number of Zn2+-containing metalloenzymes.

METHODS

Vector construction and site-directed mutagenesis

The human GlyR α1 subunit cDNA (Grenningloh et al. 1990) was cloned into the vector pCIS2. Site-directed mutagenesis was achieved using 27-mer oligonucleotides and a primer-directed polymerase chain reaction method (QuikChange kit, Stratagene). DNAs for transfection were made using the Plasmid Maxi Kit (QIAGEN) and all mutant GlyR α1 subunit constructs were completely sequenced using the BigDye ready reaction mix (Perkin-Elmer/Applied Biosystems) and an ABI 310 automated DNA sequencer (Applied Biosystems).

Amino acid numbering

Wang et al. (1995) and Wooltorton et al. (1997) have previously reported that histidine residues H156 in the GABAC receptor ρ1 subunit and H292 in the GABAA receptor β3 subunit are determinants of inhibitory Zn2+ binding sites. In both papers, the residue numbering refers to the precursor rather than the mature polypeptide. To facilitate comparisons between receptor types, all numbering in this paper refers to mature polypeptide sequences (i.e. ρ1 H156 becomes H141 and β3 H292 becomes H267). Signal peptide cleavage sites were predicted using the SignalP server at: http://www.cbs.dtu.dk/services/SignalP.

Cell culture and transfection

For electrophysiology, human embryonic kidney (HEK) cells (ATCC CRL1573) were grown in Dulbecco's modified Eagle's medium supplemented with 10 % fetal calf serum, 2 mm glutamine, 100 u ml−1 penicillin G and 100 μg ml−1 streptomycin at 37°C in 95 % air-5 % CO2 (Smart et al. 1991). Exponentially growing cells were electroporated (400 V, infinite resistance, 125 μF; BioRad Gene Electropulser II) with plasmids containing wild-type or mutant GlyR α1 subunit cDNAs together with a reporter plasmid expressing the S65T mutant of jellyfish green fluorescent protein (GFP; Heim et al. 1995).

Electrophysiology

Whole-cell membrane currents were recorded from single cells using the patch-clamp technique in conjunction with an Axopatch-1C (Axon Instruments) or List EPC7 amplifier. Patch electrodes (resistance 1–5 MΩ) were made from thin-walled borosilicate glass and filled with a solution containing (mm): 140 KCl, 2 MgCl2, 1 CaCl2, 10 Hepes, 11 EGTA and 2 ATP, pH 7.2. Cells were continuously superfused with a Krebs solution containing (mm): 140 NaCl, 4.7 KCl, 1.2 MgCl2, 2.5 CaCl2, 10 Hepes and 11 glucose, pH 7.4. Cells used 24-72 h after transfection were voltage clamped at −40 mV with membrane currents filtered at 10 kHz (−3 dB, 6th pole Bessel, 36 dB octave−1). Drugs and Krebs solution were rapidly applied (exchange rate approximately 30 ms) to single cells using a modified U-tube (Wooltorton et al. 1997). All drugs applied to cells were made up in Krebs solution and corrected to pH 5.4 or 7.4. Diethylpyrocarbonate was made fresh immediately prior to use and discarded after 30 min.

Analysis of glycine-activated membrane currents

Peak (Ipeak) and desensitised (It4) whole-cell glycine-activated currents were measured for HEK cells in the absence and presence of Zn2+. It4 currents were defined as the whole-cell current routinely measured 4 s after the response onset and are expressed as a percentage of Ipeak for that particular response. The Zn2+ concentration-response curve data for the modulation of the glycine response were fitted according to the following equation designed to account for the potentiating and inhibitory effects of Zn2+ by assuming this ion binds to two distinct sites:

display math(1)

where I and Imax represent the potentiated peak or It4 glycine-activated currents by a Zn2+ concentration (A) and by a saturating concentration of Zn2+, respectively. Imin represents the control glycine current in the absence of the modulator and was set to 1. EC50 and IC50 define the ligand concentrations producing a half-maximal potentiating effect and 50 % inhibition of the maximally potentiated current, respectively. Both nH and mH represent the respective Hill coefficients.

Concentration-response curve data for glycine-induced receptor activation were fitted to the following equation:

display math(2)

Desensitisation time constants (τ) for whole-cell currents were determined from exponential curve fits using a Chebyshev non-linear fitting routine in Clampfit 8 (Axon Instruments). Data (means ±s.e.m.) were analysed using Origin 4.1 (MicroCal) and FigP (Biosoft). Student's two-way unpaired t test was applied to data where indicated to assess significance.

Molecular modelling

The construction of a model for the Zn2+ binding site on the GlyR α1 subunit was achieved by selecting structural motifs from the Brookhaven Protein Data Bank (PDB) containing the histidine- X-histidine sequence (where X represents a hydrophobic residue), and the corresponding co-ordinated Zn2+ ion. Three such structural motifs were extracted and compared by means of a rigid body fit of equivalent carbon and nitrogen backbone atoms, using carbonic anhydrase II as the target. Root mean square (RMS) values (Å)for the backbone deviation from the template structure were recorded. Molecular images were generated and displayed via the graphics program MOLMOL (Koradi et al. 1996). All enzyme structural information was obtained from the Brookhaven National Laboratory Protein Data Bank (http://www.pdb.bnl.gov). Individual codings for each protein are provided in Fig. 6.

Figure 6.

Molecular models of Zn2+ binding sites using the H-X-H motif

Ball-and-stick representation of the H-X-H peptide motif taken from the crystal structures of carbonic anhydrase II (PDB code 12ca; green, X = F), adenosine deaminase (1uio; red, X = V) and carbonic anhydrase IV (1znc; blue, X = L). The carbon and nitrogen atoms forming the peptide backbone, which in this orientation bisects the hydrophobic residue (X) on the left and the imidazole rings of the histidine pair on the right forming a bidentate complex with Zn2+, have been superimposed using the sequence analogous to the GlyR α1 subunit (i.e. H-F-H) as the template. The Zn2+ ions forming a complex with these residues are depicted as appropriately coloured spheres on the far right of the figure.

RESULTS

Zn2+ sensitivity of GlyR α1 subunit homomers

Glycine (50 μm: approximate EC50)-activated peak membrane currents, recorded from transfected HEK cells expressing human α1 subunit GlyRs, were potentiated by 0.01–10 μm Zn2+ up to a maximum of 139 ± 8 % at 10 μm Zn2+ relative to controls in the absence of Zn2+ (Fig. 1A). The EC50 for the potentiation was 0.08 ± 0.03 μm (n= 6). Exposure to greater concentrations of Zn2+ caused inhibition of the glycine-induced peak current to 48 ± 3 % of control in the presence of 3 mm Zn2+. The Zn2+ IC50 was 546 ± 163 μm. In the absence of Zn2+, It4 responses to glycine declined to 84 ± 2 % (n= 4) of the peak current reflecting some degree of desensitisation. In the presence of co-applied Zn2+ (>10 μm), the glycine-activated current declined rapidly and the It4 response was reduced to zero current with 300-3000 μm Zn2+ (n= 4; Fig. 1A, cf. Fig. 2A). Previous studies have indicated that for the GABAA and GABAC receptors, inhibitory Zn2+ binding sites are likely to involve histidine residues in the receptor complex (Fig. 1B; Wang et al. 1995; Wooltorton et al. 1997; Horenstein & Akabas, 1998; Fisher & Macdonald, 1998). Inspection of the amino acid sequence for the GlyR α1 subunit revealed five candidate external histidines (Fig. 1C). If histidine residues are responsible for a similar Zn2+ binding site on the GlyR then this site(s) should be modifiable by H+ ions which would be expected to affect the inhibition by Zn2+ (Smart & Constanti, 1982).

Figure 1.

Histidine residues in the human GlyR α1 subunit: modulatory effects of Zn2+ and sequence alignments

A, Zn2+ concentration-response curves for the biphasic potentiation and inhibition of the current elicited by 50 μm glycine on HEK cells expressing GlyR α1 homomers. Data points are normalised to the 50 μm glycine-activated response in the absence of Zn2+ and represent the peak (▪) and It4 (□) currents measured in the presence of increasing concentrations of Zn2+. EC50 and IC50 values, respectively, were determined as: ▪, 0.08 ± 0.03 and 546 ± 163 μm; □, 0.43 ± 0.02 and 52 ± 4.1 μm. Data are means ±s.e.m. of at least four cells. B, amino acid sequences incorporating the extracellular facing ends of TM2 and TM3 (continuous lines) and the connecting TM2-TM3 loop of the human GlyR α1 and α2 subunits, the rat GABAAα1 and α6 subunits, the mouse GABAA receptor β3 subunit and the human GABAC receptor ρ1 subunit. Histidine residues are indicated by bold lettering and those involved in the formation of inhibitory Zn2+ binding sites are also italicised. The asterisks indicate some of the residues that are predicted to line the anion-selective ion channel in the GABAA receptor α1 subunit (from Xu & Akabas, 1996). C, schematic representation of the membrane-spanning topology and amino acid sequence of the human GlyR α1 subunit. Histidine residues are numbered and depicted as black (predicted extracellular) or shaded (predicted intracellular) circles.

Figure 2.

Effect of H+ ions and DEPC on the modulatory actions of Zn2+ at homomeric α1 subunit GlyRs

A, whole-cell membrane currents recorded from a single HEK cell transfected with the wild-type GlyR α1 cDNA construct (VH=−40 mV). Successive traces represent the currents elicited at pH 7.4 (left-hand records) by a control (C1) dose of glycine (50 μm) co-applied with either a potentiating (10 μm) or inhibitory (100 μm) concentration of Zn2+. At pH 5.4 (middle records) the revised control EC50 glycine response (C2, 200 μm) was potentiated by co-application with 10 and 100 μm Zn2+. These effects were reversible in Krebs solution within 2 min. In the presence of 1 mm DEPC (right-hand records) control responses to 50 μm glycine (C3) were relatively unaffected and subsequently unaltered in the presence of co-applied 10 or 100 μm Zn2+. The effect of DEPC was irreversible. A typical current profile for a saturating dose (100 μm) of glycine (100 G) obtained at pH 7.4 is also shown on the far left of the panel for comparison with the effects of the modulators. The filled (▪) and open (□) squares indicated on the same trace demonstrate the peak and ‘It4′ current levels, respectively (see Methods), used to derive the histogram bars in B, which are represented by the same symbols. B, bar chart representing normalised glycine-activated whole-cell currents collated from experiments performed as in A at pH 7.4, at pH 5.4 and with DEPC (1 mm). ▪, peak whole-cell currents, upper error bars; □, It4 currents, lower error bars - data are normalised to the peak response (C1) elicited by 50 μm glycine at pH 7.4, except for the 10 and 100 μm doses of zinc applied at pH 5.4 or with DEPC which were normalised against C2 and C3 (200 and 50 μm glycine response at pH 5.4 and with DEPC, respectively). C, concentration-response curves obtained from whole-cell peak glycine-activated currents in HEK cells superfused with pH 7.4 (▪) and pH 5.4 (□) Krebs solution. Currents are normalised to the 50 μm response at pH 7.4 and the EC50 values and Hill coefficients (in parentheses) determined from the fitted curves were: pH 7.4, 32.7 ± 4.2 μm (2.2 ± 0.6), and pH 5.4, 94.7 ± 5.5 μm (2.2 ± 0.25). Data are means ±s.e.m. from at least three cells.

pH titration of glycine-evoked currents

The effect of H+ ions on responses to 50 μm glycine in the absence and presence of 10 or 100 μm Zn2+ was assessed in HEK cells expressing α1 subunit GlyRs in Krebs solutions of pH 5.4 and 7.4. At pH 7.4, 10 and 100 μm Zn2+ both enhanced the peak glycine-activated current, but with 100 μm Zn2+ the It4 response was markedly inhibited (Fig. 2A and B). On reducing the external pH from 7.4 to 5.4, peak and It4 responses to the control 50 μm glycine concentration were reduced to 12 ± 4 % with no apparent desensitisation (data not shown). This behaviour was explained by a 2.9 ± 0.3-fold (Fig. 2C, n= 6) lateral competitive shift in the glycine concentration-response curve. This reduction by H+ ions was compensated for by increasing the concentration of glycine to 200 μm in order to be comparable to the EC50 glycine concentration applied to the wild-type receptor; however, even with this compensation at pH 5.4, 10 μm Zn2+ caused a similar potentiation of glycine-activated peak currents (pH 7.4, 138 ± 10 %; pH 5.4, 131 ± 1 %; n= 5 and 3, respectively) and the inhibition by 100 μm Zn2+ was abolished revealing an increased level of potentiation (pH 7.4, 112 ± 3 %; pH 5.4, 205 ± 34 %; n= 5 and 3, respectively) (Fig. 2A and B). This relief of inhibition was most apparent from the It4 currents in the presence of 100 μm Zn2+, which represented 14 ± 3 and 89 ± 6 % of the peak current at pH 7.4 and 5.4, respectively. These reversible effects, resulting from a 100-fold increase in the H+ ion concentration, suggested that increased H+ ion binding may competitively exclude the binding of Zn2+ from a proton-sensitive inhibitory site on GlyRs. The increased potentiation that was noted with 100 μm Zn2+ at pH 5.4 could then reflect the temporary removal of concurrent inhibition by Zn2+ at discrete sites. The sensitivity of Zn2+ inhibition to pH over the range 5.4 to 7.4 is a characteristic of histidine residues, since the pKa for the imidazole group is approximately 6.1. Thus, depending on the local environment, at pH 5.4 potentially 83 % of histidines will now be positively charged in comparison to only 4.8 % at pH 7.4. In this way the co-ordination sites normally available to Zn2+ are now occluded at pH 5.4.

Effects of the histidine modifying reagent diethylpyrocarbonate (DEPC)

An additional test for the involvement of histidine residues in the inhibitory effect of Zn2+ on GlyRs requires the use of the histidine modifying reagent, diethylpyrocarbonate, which reacts with neutral imidazole groups converting the residues to N-carbethoxyhistidyl derivatives (Miles, 1977; Lundblad & Noyes, 1984). Treatment of α1 subunit GlyRs with 1 mm DEPC did not affect the amplitude of 50 μm glycine-induced currents (114 ± 5 % of Ipeak at pH 7.4; Fig. 2A and B) compared to the effect observed by reducing external pH. However, relative to control (100 %), the potentiation induced by 10 μm Zn2+ (110 ± 4.2 %) and the inhibition induced by 100 μm Zn2+ (64 ± 2 %) were abolished after 2 min in DEPC. These effects were not reversible on reverting to control Krebs solution.

Type of Zn2+ inhibition

The mode of antagonism by which Zn2+ inhibited glycine-activated currents on the GlyR α1 subunit was investigated. The previous co-application protocols were designed to reproduce the co-release of neurotransmitter (glycine) and neuromodulator (Zn2+) at the synaptic cleft. However, preincubation for 1–2 min with an inhibitory concentration of zinc (100 μm) slowed the activation rate of the receptor by glycine (50 μm) reducing peak glycine-activated currents to the It4 levels seen in the co-application protocol (Fig. 3A). Repeated application of 50 μm glycine in the presence of 100 μm Zn2+ revealed no evidence of any use dependence. Clearly Zn2+ is able to access its binding site(s) without the need for receptor activation. Construction of glycine concentration-response curves also revealed that Zn2+ inhibits the action of glycine in a competitive manner (4.7-fold increase in the EC50 in the presence of 50 μm Zn2+; data not shown).

Figure 3.

Zn2+ inhibition is not use dependent and does not affect receptor desensitisation

A, whole-cell membrane currents recorded from a single HEK cell transfected with the wild-type GlyR α1 cDNA construct (VH=−40 mV). The first two records represent the currents elicited by an EC50 (C1) concentration of glycine (50 μm) co-applied for 20 s with (C1 + 100) or without (C1) 100 μm Zn2+. After pre-incubation with Zn2+ (100 μm, hatched bar) for 2 min, successive (every 3 min) responses to glycine in 100 μm Zn2+ applied for 5–20 s were reduced but exhibited no evidence of any use dependence. The It4 levels of inhibition attained using the co-application and pre-incubation protocols were similar. Inhibition by Zn2+ was readily reversible (C1, final record). B, whole-cell 30 μm glycine-activated currents recorded prior to (C1) and after a 2 min pre-incubation with 50 μm Zn2+ (C1 + 50, hatched bar). Increasing the glycine concentration to 300 μm in the presence of 50 μm Zn2+ restored the peak current level (Cnew+ 50) to that of the control. Note the similar desensitisation rate of this compensated response to C1. The inhibitory Zn2+ effect was reversible. C, left panel, bar charts depicting the reduction in whole-cell glycine-induced current (approximate EC50) following pre-incubation with Zn2+ (50 μm) and restoration to control levels following an increase in glycine concentration (‘Zn2+ compensation’ Cnew; n= 5–6). Right panel, desensitisation time constants (τ) for control glycine responses prior to (14.29 ± 4.54 s) and following pre-incubation with 50 μm Zn2+ (14.79 ± 6.56 s), and after increasing the glycine concentration (‘Zn2+ compensation’ Cnew; 19.97 ± 6.22 s; P > 0.05) as detailed above (n= 4–5).

The co-application experiments suggested that Zn2+ was accelerating the decline of the glycine-activated currents. This could be due to the onset of Zn2+ block during activation of the GlyR and/or Zn2+-induced accentuation of GlyR desensitisation. To distinguish between these possibilities, we examined GlyR desensitisation rates following activation by glycine alone (control, 30-50 μm) and after pre-application of Zn2+ (50 μm) with increased concentrations of glycine to produce matched inward peak currents compared to the control. The control glycine responses (30-50 μm: approximate EC50) were reduced in the presence of 50 μm Zn2+ to 13 ± 1 %. This Zn2+-induced reduction was surmounted (98 ± 6 % of control) in the presence 100-500 μm glycine (Fig. 3B and C). A comparison of the desensitisation time constants (τ) for the control (30-50 μm: 14.29 ± 4.54 s) glycine responses and those in the presence of 50 μm Zn2+ (14.79 ± 6.56 s) revealed no significant differences (P > 0.05; Fig. 3C).

Binding sites for Zn2+ ions on the GlyR receptor α1 subunit: site-directed mutagenesis

Based on current topological models of GlyRs (Becker, 1995), five extracellular histidine residues can be identified as potential binding sites for Zn2+, including H107, H109, H201, H215 and H419 (Fig. 1C). Any role in Zn2+ inhibition for the intracellular histidines H311, H323 and H324 was discounted since the onset and recovery from Zn2+ were rapid. Histidine 107 is only present in the human and rodent GlyR α1 subunits (Grenningloh et al. 1990). For GlyR α2, α3 and α4 subunits, an asparagine (N) residue occupies the equivalent position (Kuhse et al. 1990; Matzenbach et al. 1994). In contrast, H109 is completely conserved in all known GABAA, GABAC and GlyR subunits (Wang et al. 1995). Moreover, histidines 201, 215 and 419 are conserved in all GlyR α subunits (Grenningloh et al. 1990; Kuhse et al. 1990; Matzenbach et al. 1994). We therefore systematically mutated all accessible histidines to alanine using site-directed mutagenesis. Our initial hypothesis was that H109, H201, H215 and/or H419 are possibly involved with the inhibitory action of Zn2+ since these residues can be found in both GlyR α1 and α2 subunits and previous studies indicated that Zn2+ was equipotent as an inhibitor on these two α subunit homomers (Bloomenthal et al. 1994; Laube et al. 1995).

Histidines 107 and 109

Mutating H109 to alanine (H109A) in the GlyR α1 subunit revealed its potential involvement with the Zn2+ binding site because the inhibitory effects of Zn2+ were abolished (up to 1 mm Zn2+) compared to the wild-type receptor (Ipeak, 89 ± 3 % (10 μm Zn2+) and It4, 85 ± 2 % (100 μm Zn2+)). Intriguingly, the potentiating effect of Zn2+ was also abolished by this mutation suggesting that H109A may have caused allosteric disruption to this particular action of Zn2+. The nearby histidine at position 107 is located sufficiently proximal to H109 to contribute by co-ordination towards part of the Zn2+ binding site. Mutating H107 to alanine (H107A) in the α1 subunit had no effect on the degree of potentiation of 50 μm glycine-activated currents by 10 μm Zn2+ compared to that seen with the wild-type receptor (Ipeak, 142 ± 8 %; It4, 83 ± 4 %, Fig. 4A). However, the H107A mutation substantially relieved the 100 μm Zn2+-induced inhibition observed with wild-type GlyR α1 subunit homomers. The most profound effect was apparent from the It4 currents which for H107A attained 79 ± 2 % of Ipeak, compared to 14 ± 3 % for the wild-type α1 subunit GlyR.

Figure 4.

Role of histidines H107 and H109 in the formation of the inhibitory Zn2+ binding site on the human GlyR α1 subunit

A, bar chart of normalised whole-cell currents recorded from HEK cells (VH=−40 mV) transfected with cDNA for either the wild-type (WT) GlyR α1 subunit or the histidine mutant receptors, H107A or H109A. Bar clusters for each receptor type represent the normalised peak current (filled bar; error bars as in Fig. 2) of the control 50 μm glycine response (C1-3) and the effects of co-application of 10 and 100 μm Zn2+. It4 currents appear as open bars. All experiments were performed at pH 7.4. B, glycine concentration-response curves for the wild-type (▪; interrupted line) and histidine mutant GlyRs (□, H107A; ○, H109A). Whole-cell currents are normalised to the maximum response of each curve and the EC50 values and Hill coefficients (in parentheses), were: wild-type, 32.2 ± 1.4 μm (2.3 ± 0.2); H107A, 48.0 ± 1.4 μm (3.1 ± 0.2); H109A, 24.4 ± 1.3 μm (1.8 ± 0.2). EC50 values for H107A and H109A were significantly different from that of control data (P < 0.05). Data are means ±s.e.m. of at least four cells.

The glycine concentration-response curves were also compared for the wild-type GlyR α1 subunit, H107A and H109A mutants (Fig. 4B). Removal of these extracellular histidine residues did not unduly perturb glycine binding and activation of these receptors. The largest shift of only 1.5-fold in the glycine EC50 occurred between wild-type (32 ± 1.4 μm; n= 6) and the H107A mutant (48 ± 1.4 μm; n= 4).

Histidines 201, 215 and 419

Of the remaining extracellular histidine residues in both the N- and C-terminal domains of the GlyR α1 subunit, it is apparent that they play little or no role in the formation of the inhibitory Zn2+ binding site. Low concentrations of Zn2+ (10 μm) enhanced the 50 μm glycine-activated peak current by between 111 and 138 % in all but the H215A mutant, which appeared to be insensitive (Ipeak, 97 ± 2 %, Fig. 5A). The 100 μm Zn2+-induced inhibition of the It4 glycine-activated current was similar to wild-type for all mutants including H215, attaining only 14 ± 4 % of the Ipeak for all the mutants compared to 14 ± 3 % for the equivalent It4 current in the wild-type α1 subunit GlyR. Analysis of the glycine concentration-response curves revealed that, as for the H107A and H109A mutants, there was no significant shift in the curves as a result of the mutations H201A or H419A. The largest change in the glycine EC50 occurred between control and the H201A (EC50, 50.4 ± 5.4 μm, n= 4) mutant involving a 1.6-fold change (Fig. 5B).

Figure 5.

N-terminal histidines H201 and H215, and the C-terminal H419 histidine are not involved in the formation of the inhibitory Zn2+ binding site

A, bar chart of normalised whole-cell glycine-activated currents elicited from HEK cells (VH=−40 mV) transfected with cDNA for either the wild-type (WT) GlyR α1 receptor, or the histidine mutants H201A, H215A or H419A. Bar clusters for each receptor type are normalised to the peak current (filled bar) of the 50 μm glycine response (C1-4) and the effects of 10 and 100 μm Zn2+ co-application are shown. It4 currents are represented by open bars. All experiments were performed at pH 7.4. B, glycine concentration-response curves for wild-type (▪; interrupted line) and histidine mutant GlyRs (▴, H201A; ▾, H419A). Whole-cell currents are normalised to the maximum of each curve and the EC50 values and Hill coefficients (in parentheses) were: wild-type, 32.2 ± 1.4 μm (2.3 ± 0.2); H201A, 50.4 ± 5.4 μm (2.6 ± 0.6); H419A, 30.3 ± 0.5 μm (2.2 ± 0.1). EC50 values for H201A, but not H419A, were significantly different from control (P < 0.05). Data are means ±s.e.m. of at least four cells.

Model of the putative Zn2+ binding site on the human GlyR α1 subunit

To construct a potential model for the Zn2+ binding site on the GlyR α1 subunit, a search was conducted of a crystallographic database for proteins incorporating a bound Zn2+. Three proteins were selected with a common H-X-H motif that is capable of co-ordinating Zn2+. In the GlyR α1 subunit, H107 and H109 straddle a hydrophobic phenylalanine (F) residue. The hydrophobic character of residue X between the two histidines was also a conserved feature of the three selected crystallised proteins: carbonic anhydrase II contained a phenylalanine at this position, and was deemed to be most analogous to the GlyR α1 subunit; adenosine deaminase incorporated a valine (V); and, carbonic anhydrase IV contained a leucine (L) residue. Rigid body superposition of all three crystal structures, using carbonic anhydrase II as the template, revealed a high degree of geometrical similarity (Fig. 6). The RMS deviation from the template peptide backbone atoms was 0.214 Å for adenosine deaminase and 0.048 Å for carbonic anhydrase IV, and the histidine nitrogen to Zn2+ ion distance for the three proteins was 2.29 ± 0.3 Å. It is evident from the three crystal structures that the H-X-H motif is part of a β strand that is characterised by the respective hydrophobic residue (F, V or L) lying on the opposite side of the protein backbone to its adjacent histidines.

DISCUSSION

The transition metal ion Zn2+ can interact with a number of ligand-gated ion channels generally causing inhibition, with the exception of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and glycine receptors where Zn2+ can exert a biphasic effect. For glycine receptors, the potentiating and inhibitory effects of Zn2+ have been postulated to occur by Zn2+ binding at two separate sites on α subunits (Bloomenthal et al. 1994; Laube et al. 1995; Lynch et al. 1998). Our present study validates this concept by demonstrating that two histidine residues, at positions 107 and 109 in the human GlyR α1 subunit, are essential for Zn2+ inhibition and differentially affect the ability of Zn2+ to enhance glycine-activated currents.

Inhibitory binding site for Zn2+

Three pieces of evidence suggested the involvement of histidine residues in forming the GlyR α1 subunit inhibitory Zn2+ binding site. Firstly, lowering external pH from 7.4 to 5.4, which spans the pKa (6.1) for histidine, markedly reduced Zn2+ inhibition even after adjusting the glycine concentration to account for the competitive inhibitory action of H+ ions on glycine receptors. This change in the pH will have a dramatic effect on the proportion of charged imidazole groups potentially affecting the ability of Zn2+ to bind to the receptor. Secondly, the histidine-modifying reagent, DEPC, abolished the inhibition by Zn2+. Finally, mutation of selected histidines profoundly reduced the inhibitory activity of Zn2+. Site-directed mutagenesis alone does not prove the involvement of any particular residue(s) in the formation of the ligand/ion binding site, as any given substitution could merely disrupt an allosteric transduction mechanism. However, the ability of increased H+ concentrations and DEPC to directly affect in a differential manner histidine residues thereby abolishing the inhibition by Zn2+ is compelling evidence implicating H109 and/or H107 in the formation of the Zn2+ binding site.

Current topology models based on hydropathy analyses of the GlyR suggest that H107 and H109 in the α1 subunit are located in the N-terminal extracellular domain some 110 residues from the start of TM1. This location shares some similarity with H141 in the GABAC receptor ρ1 subunit which plays a pivotal role in Zn2+ inhibition of GABA-gated responses (Fig. 1B; Wang et al. 1995), and also with H42 and H44 which underlie the high-affinity Zn2+ inhibition associated with the NMDA receptor (Choi & Lipton, 1999). However, this region is spatially quite distinct from histidines involved in Zn2+ binding to GABAA receptors. For β1 and β3 subunits, a single histidine located at position 267, just inside the postulated ion channel lining TM2, forms the basis of an inhibitory Zn2+ binding site (Fig. 1B; Wooltorton et al. 1997; Horenstein & Akabas, 1998). For the GABAA receptor α6 subunit, H273, located just outside TM2 in the TM2-TM3 extracellular loop, appears responsible for the greater sensitivity of α6 subunit-containing receptors to Zn2+ compared to their α1-containing counterparts, which lack this residue (Fisher & Macdonald, 1998). Currently, only single histidine residues have been identified as being important for Zn2+ binding to GABAA and GABAC receptor subunits. Interestingly, mutation of the highly conserved H148 (the equivalent of H109 in the GlyR α1 subunit) to asparagine or tyrosine in the ρ1 subunit prevented the expression of functional GABAC receptors (Wang et al. 1995). It is likely that other amino acids participate in forming the Zn2+ binding sites in the GABAA and GABAC receptor subunits and these could be contributed by other subunits in the mature receptor structure or by duplicate copies of subunits known to contain the critical histidines. The GlyR α1 subunit is unusual in that there appear to be two proximal histidines that could co-ordinate Zn2+ forming part of a polyhedral complex, obviating the need for other subunit involvement in forming the Zn2+ binding site. This could indicate the presence of multiple (up to three) Zn2+ binding sites, if the GlyR α1 subunit is assumed to form a pentameric heteromer with the β subunit assuming a 3α:2β stoichiometry. Interestingly, mutation of either H107 or H109 is sufficient to abolish Zn2+ inhibition on α1 subunit GlyRs supporting the concept that a single ion binding site exists and is reliant on dual co-ordination from these histidines in the α1 subunit. The spacing of the two histidines suggests that only a single Zn2+ ion need bind at this site to produce the degree of inhibition of glycine-gated currents observed.

The proximity of residues H107 and H109 and a co-ordinated Zn2+ ion does allow some prediction of the local topology of the peptide backbone in this part of the N-terminus. In order to gain some structural insight into the potential Zn2+ binding site, we considered how Zn2+ is incorporated into metalloenzymes. Typically Zn2+ is tetrahedrally co-ordinated and can form part of the catalytic site, or part of a co-active or co-catalytic site, and also, but less frequently, fulfill a structural role (Vallee & Falchuk, 1993). Catalytic and co-active Zn2+ sites are co-ordinated by histidines in preference to aspartate and glutamate residues whereas cysteines appear to predominate at structural Zn2+ sites (Vallee & Falchuk, 1993). To obtain a close structural match to the GlyR we scanned a crystallographic database for Zn2+-containing proteins that may have similar structural motifs to that likely to be involved in the inhibitory effect of Zn2+ on the GlyR α1 subunit, i.e. H-X-H. The three enzymes selected (see below) all possessed this characteristic motif with carbonic anhydrase II offering the closest example incorporating phenylalanine as the hydrophobic moiety (NB all the carbonic anhydrase family with the exception of carbonic anhydrase IV contain the H-F-H motif). The co-ordination of Zn2+ may also be augmented by additional residues such as cysteine or aspartate, or by activated water molecules. However, for the inhibitory site on the GlyR α1 subunit, we expect the binding of Zn2+ to be relatively weak since inhibition is easily reversible, suggesting that Zn2+ is not as tightly bound compared to metalloenzymes where Zn2+ forms part of the protein structure and/or catalytic site, and cannot be removed by chelating agents (cf. Frederickson, 1989). Comparison of the two histidines pivotal to the inhibitory site in the GlyR α1 subunit with a generic model for the GlyR (Gready et al. 1997) places H109 at the apex of a β sheet formed by the protein prior to entering into a loop structure. Thus H109, and presumably H107, are ideally placed to interact with adjacent flanks of the same or juxtaposed subunits in the receptor protein and this may be crucial to the inhibitory activity of Zn2+. Interestingly, the H-X-H motif in a number of Zn2+-containing enzymes also occupies an apical position at the end of a β sheet, which was an additional reason prompting their selection for comparison with the GlyR α1 subunit. A similar Zn2+ binding site could exist on the NMDA receptor 2A subunit where an H-X-H motif can be identified that is important for voltage-independent inhibition of NMDA-activated currents. In this case X is represented by serine possessing a polar hydroxyl in the side chain (Choi & Lipton, 1999).

Importance of histidine residues for Zn2+-induced potentiation

It was notable that some of the histidine mutations also affected the ability of Zn2+ to potentiate the glycine-activated current. In particular, H109A and H215A both abolished potentiation by Zn2+; however, unlike H109A, Zn2+ inhibition was unaffected by the H215A mutation. In contrast, the H107A mutation abolished inhibition by Zn2+ but appeared to leave Zn2+-induced potentiation unaffected. The labile nature of the Zn2+-induced potentiation of glycine-activated responses is perhaps not so surprising given the number of TM1-TM2 and TM2-TM3 loop mutations which can disrupt Zn2+ potentiation of glycine-gated currents (Lynch et al. 1998). It appears that potentiation by Zn2+ is more difficult to locate in terms of a binding site(s) because of diverse allosteric components. This conclusion is unlikely to apply to the inhibition by Zn2+ because H107A and H109A are, to date, the only mutations that have affected Zn2+ inhibition (cf. Lynch et al. 1998) suggesting these residues actually form part of the inhibitory binding site. This is further supported by the actions of H+ ions and DEPC affecting Zn2+ inhibition on wild-type GlyR α1 subunit homomers.

Variable inhibition by Zn2+ on native neuronal GlyRs: a potential role for the α subunit isoform?

Interestingly, only the GlyR α1 subunit contains H107. The equivalent position in the GlyR α2 subunit is an asparagine (N114); however, the conserved position H109 in the GlyR α1 subunit is still occupied by a histidine residue in α2 (H116; Grenningloh et al. 1990). Previous studies have suggested that glycine-activated currents mediated by human GlyR α2 subunit homomers are also inhibited by Zn2+ exhibiting a similar potency to that on α1 subunit receptors (Bloomenthal et al. 1994; Laube et al. 1995). This suggests that either H116 alone is sufficient to support Zn2+ binding in α2 subunits, which seems unlikely, or additional, as yet unidentified, residues are important for co-ordination with this H109 equivalent. It is noteworthy that the equivalent positions to H107 in the GlyR α3 and α4 subunits are also occupied by an asparagine residue (Kuhse et al. 1990; Matzenbach et al. 1994) but whether Zn2+ can act as a modulator on GlyRs containing these subunits has not yet been addressed.

Examination of native neuronal GlyRs has revealed differential regulation by Zn2+ of glycine-gated currents. Rat spinal cord neurones exhibited both potentiation to low concentrations of Zn2+ (<10 μm) and inhibition at higher concentrations (>50-100 μm; Bloomenthal et al. 1994; Laube et al. 1995) in accordance with the recombinant data obtained with GlyR α1 or α2 subunits. In contrast, rat septal neurones appeared to be only potentiated by Zn2+ (Kumamoto & Murata, 1996); however, inspection of the glycine-activated currents reveals that the It4 responses are inhibited by Zn2+ at concentrations greater than 100 μm despite the peak currents exhibiting modest potentiation. Hence, these results probably reflect the speed of drug application and the method chosen to measure the glycine-activated currents. The spinal cord could be relevant for the physiological regulation of GlyRs by Zn2+. Using sodium selenite staining, Velazquez et al. (1999) have recently demonstrated that Zn2+ is concentrated in primary afferent neurones in lamina V of the spinal cord and is also found in dorsal root ganglion cell bodies co-localised with the Zn2+ binding protein, metallothionein III. Similarly, the olfactory bulb also offers a physiological prospect of Zn2+ modulating glycine receptors as Zn2+ is concentrated in this part of the central nervous system (Frederickson, 1989). Using low concentrations of glycine (30 μm), Trombley & Shepherd (1996) demonstrated that 100 μm Zn2+ produced a potentiation of the glycine-activated response on olfactory bulb neurons. In comparison to our study, this concentration of Zn2+ would be expected to inhibit the It4 response to glycine but this apparently did not occur (Trombley & Shepherd, 1996). Our current investigation reinforces and extends these studies and suggests that physiological Zn2+ may play a role in the regulation of glycine receptor activity. This could have potential implications for the control of neuronal excitability. We have established that Zn2+, besides potentiating responses transduced via GlyR α1 subunits, is able to directly inhibit its function through interactions that involve key histidine residues in the N-terminal domain. The effectiveness of the action of Zn2+ may well also depend on the predominant isoform of α subunit expressed at individual glycinergic synapses.

Acknowledgements

This work was supported by the Medical Research Council. We thank Heinrich Betz (Max-Planck-Institut für Hirnforschung, Frankfurt, Germany) for kindly providing the human GlyR α1 subunit cDNA.

Ancillary