Identification of a Zn2+ binding site on the marine gABAA receptor complex: Dependence on the Second transmembrane domain of β subunits


  • This manuscript was accepted as a Short Paper for rapid publication.

    Author's email address
    T. G. Smart:


  • 1Whole-cell currents were recorded from Xenopus laevis oocytes expressing wild-type and mutant recombinant GABAA receptors to locate a binding site for Zn2+ ions in the β3 subunit.
  • 2The Cl-selective current, spontaneously gated by β3 subunit homomers, was enhanced by pentobarbitone and inhibited by picrotoxinin. The potencies of these agents were minimally affected by mutating histidine (H) 292 to alanine (A) in the second transmembrane domain (TM2).
  • 3Zn2+ inhibited the β3 subunit-gated conductance (IC50, 0.31 μm); the inhibition was voltage insensitive. The H292A mutation in β3 subunits caused a 1000-fold reduction in Zn2+ potency (IC50, 307 μm).
  • 4GABA-activated responses recorded from heteromeric α1β3 GABAA receptors were also inhibited by Zn2+ (IC50, 0.11 μm). This inhibition was reduced by mutating H292A in the β3 subunit (IC50, 22.8 μm).
  • 5H292 in TM2 of the β3 subunit is an important determinant of a Zn2+ binding site on the GABAA receptor. Its location in the presumed ion channel lining suggests that Zn2+ can penetrate into an anion-selective channel and that the ionic selectivity filter and channel gate are located beyond H292.

Molecular cloning and electrophysiological studies indicate that γ-aminobutyric acidA (GABAA) receptors are composed of numerous subunits that influence the functional properties of the expressed receptor. These subunits are selected from five different families, three of which contain multiple members and are designated as: α(1–6), β(1–4), γ(1–4), δ(1) and ɛ(1) (Rabow, Russek & Farb, 1995; Davies, Hanna, Hales & Kirkness, 1997), to form a presumed pentameric assembly.

GABAA receptors underlie synaptic inhibition in the central nervous system (CNS) and are sensitive to a variety of drugs (Rabow et al. 1995) in addition to possessing binding sites for divalent (Smart, Xie & Krishek, 1994) and polyvalent cations (Ma & Narahashi, 1993) that will regulate receptor function. Of the ions demonstrated to interact with GABAA receptors, Zn2+ is of importance since it is concentrated in the CNS and can be released upon neural stimulation into the synaptic cleft (Frederickson, 1989). Zn2+ has been demonstrated to be an inhibitor of Cl -selective GABA receptors of invertebrate and vertebrate origin (Smart et al. 1994; Harrison & Gibbons, 1994). Moreover, the binding site(s) for Zn2+ on GABAA receptors is discrete from ligand binding sites associated with GABA, the benzodiazepines, barbiturates, bicuculline, picrotoxinin and the neurosteroids (Celentano, Gyenes, Gibbs & Farb, 1991; Smart, 1992). Quite where Zn2+ is binding in the receptor complex is unknown although histidine residues are implicated from earlier pH titration studies of the GABA-activated conductance on invertebrate muscle (Smart & Constanti, 1982).

The inhibitory activity of Zn2+ on GABA-activated responses is dependent upon the receptor subunit composition. The most dramatic effect occurs with γ subunit-containing receptors that are less sensitive to inhibition compared with corresponding α,β constructs (Draguhn, Verdoorn, Ewert, Seeburg & Sakmann, 1990; Smart, Moss, Xie & Huganir, 1991). In addition, the identity of the α-subunit in αβγ subunit complexes can also influence Zn2+ inhibition (White & Gurley, 1995).

Recently, it has become apparent that β1 and β3 subunits form functional homomeric ion channels exhibiting a pento-barbitone-modulated, spontaneously gated Cl current that can be inhibited by Zn2+. The search for the residue(s) forming the Zn2+ binding site was conducted by utilizing these unique properties of βsubunit homomers because their sensitivity to Zn2+ is high with IC50 values in the nanomolar range (Krishek, Moss & Smart, 1996; Wooltorton, Moss & Smart, 1997b), suggesting that the β subunit may be essential in forming the Zn2+ binding site that exists on α1βj constructs (where j=1–3). Moreover, GABAA receptors composed of a single subunit are presumed to form ion channels with a symmetrical distribution of amino acid residues. This is advantageous when searching for a ligand binding site using site-directed mutagenesis which should have clearly discernable effects on receptor function. Furthermore, the involvement of a single type of subunit removes many uncertainties regarding receptor heterogeneity when two or more subunits are co-expressed.

The present study concludes that a major determinant of the Zn2+ binding site is contained in the second transmembrane domain of the β subunit. A preliminary report of some of these results has been published previously (Wooltorton, McDonald, Moss & Smart, 1997a).


cDNA and expression vectors

Murine β3 cDNA was cloned as EcoR1 fragments into the mammalian expression vector pGW1 and site-directed mutagenesis was performed as described previously (Connolly, Krishek, McDonald, Smart & Moss, 1996).

Cell preparation: extraction of oocytes and microinjection

Oocytes were removed from anaesthetized Xenopus laevis following immersion in 0.5% tricaine as described previously (Smart & Krishek, 1995) and stored in modified Earth's medium (MBM) containing (mm): 110 NaCl, 1 KCl, 2.4 NaHCO3, 7.5 Tris-HCl, 0.33 Ca(NO3)2, 0.41 CaCl2, 0.82 MgSO4, and 50 μg ml−1 gentamicin; pH 7.6. Stage IV and V oocytes were separated and centrifuged (700–1100 g for 7–8 min at 10 °C) for nuclear microinjection with 10 nl of 1 mg ml−1 DNA solution, encoding either for the murine β3, β3H292A, αlβ3 (ratio of 1:1) or αlβ3H292A (1:1) GABAA receptor subunits. Injected oocytes were incubated at 18 °C for 24 h then subsequently stored at 10 °C and replenished with fresh MBM every 2–3 days.

Electrophysiology: intracellular recording

Membrane currents were recorded from Xenopus oocytes using a two-electrode voltage clamp technique. Oocytes were superfused with an amphibian Ringer solution containing (mm): 110 NaCl, 2 KCl, 5 Hepes and 1.8 CaCl2 (pH 7.4) at 8–10 ml min−1 (bath volume, 0.5ml). Voltage and current microelectrodes were filled with 0.6 M K2SO4 (1–2 MΩ). Currents were recorded using an Axoclamp 2B amplifier in conjunction with a Gould 2200S pen recorder and pCLAMP (Axon Instruments) for storage and analysis of data on a Viglen Pentium P90 computer using a DigiData 1200 interface.

Analysis of ligand-modulated membrane conductances

Conductances were determined by applying hyperpolarizing voltage commands (1 s duration, −10 mV amplitude, 0.2 Hz frequency) from a holding potential of −25 mV in the absence and presence of a ligand. To construct equilibrium concentration–response relationships for GABA and pentobarbitone, the ligand-induced conductance change (▵G) was calculated by subtracting the resting conductance from the conductance measured in the presence of each ligand. All the conductances were normalized (▵GN) to the maximum conductance change (▵GN,max) and subsequently fitted with the following equation:

display math

where EC50 represents the concentration of ligand ([A]) inducing 50% of the maximal conductance evoked by a saturating concentration of ligand and nH is the Hill coefficient.

The reductions in the resting membrane conductance by picrotoxinin and Zn2+ were used to construct antagonist concentration–inhibition relationships. The antagonist-sensitive conductance (equivalent to the β3 subunit-gated spontaneous membrane conductance) was defined as 100% after the addition of a saturating concentration of picrotoxinin. The inhibition of this conductance by intermediate concentrations of antagonists were fitted with the equation:

display math

where ▵GN and ▵GN represent the normalized GABA-induced (at a given GABA concentration) or β3 subunit-gated conductance in the presence and absence of antagonist, respectively. [B] represents the antagonist concentration and IC50 defines the concentration of antagonist producing a 50% inhibition of the GABA-induced or β3 subunit-gated conductance.

Current–voltage relationships

Current–voltage (I–V) relationships for the responses induced by Zn2+ were determined under voltage clamp. The membrane potential was changed by ramping from −100 mV to 50 mV over 600 ms. Ramp I–V plots were obtained in control Ringer solution and after achieving a steady-state reduction in membrane conductance to an approximate IC50 concentration and a saturating concentration of Zn2+. The I–V plots were fitted with fifth-order polynomials. The Zn2+I–V relationship was determined by subtracting the I–V relationship determined in the presence of a saturating concentration of Zn2+ from the control I–V plot and the I–V plot determined for the IC50 concentration of Zn2+.


Rationale for site-directed mutagenesis

Binding sites for Zn2+ on metalloenzymes and metallothioneins typically involve histidine (H) and cysteine (C) residues (Vallee & Falchuk, 1993). Given that Zn2+ is a potent inhibitor of Cl currents gated by β1 (IC50, 0.23 μM; Krishek et al. 1996) and β3 (0.33 μM; Wooltorton et al. 1997b) homomeric receptors, the subunit amino acid sequences were screened for likely binding sites involving histidine and cysteine residues that are unique to the β subunits compared with α and γ subunits (Fig. 1). Only extracellular domains of the GABAA receptor were selected for scrutiny since Zn2+ has an extracellular site of action (Celentano et al. 1991). Figure 1 reveals that a number of histidine residues are contained in the N-terminus and second transmembrane domain (TM2), the latter purportedly forming the ion channel pore. The four external cysteines are unlikely to be involved since Zn2+ inhibition can be pH sensitive (Smart & Constanti, 1982); moreover two of the external cysteine residues are highly conserved and could form a disulphide bridge (Barnard, Darlison & Seeburg, 1987; Pan, Bahring, Grantyn & Lipton, 1995) that would preclude their participation in Zn2+ binding. This deduction limited the search to histidine residues, many of which are common to α, β and γ subunits; however, one histidine residue (position 292 in β3) is unique to the β subunits and is located near the postulated external entrance to TM2, a region thought previously to be inaccessible to cations by virtue of the Cl-selective permeability of GABAA receptor ion channels. This histidine was mutated to alanine (A) in β3 subunits and expressed in homomeric and heteromeric forms in oocytes.

Figure 1.

Selected amino acid sequences of GABAA receptor α, β and γ subunits

Transmembrane topology and alignments of the relative positions of external histidine (H) and cysteine (C) residues for α1–3, β1–3 and γ2S subunits. Only the last 9 residues in TM2 are shown. The N-terminal domain details the positions of H and C residues for the β3 subunit only although the 2 cysteines forming the putative disulphide bridge are conserved for α1–3, β1–3 and γ2S subunits.

β3 homomers: effect of mutating H292 in TM2 on ion channel properties

The expression of β3 homomers in Xenopus oocytes resulted in higher resting membrane conductances (23.7 ± 6.9 μS, mean ±s.e.m., n=9) when compared with control oocytes from the same donor frogs (1.71 ± 0.34 μS, n= 10, P < 0.05). The resting conductance was increased further by application of 100 μm pentobarbitone (Fig. 2A). In contrast, the GABAA receptor antagonist picrotoxinin (10 μM), inhibited the resting conductance to values approaching those of control uninjected oocytes (Fig. 2C; Wooltorton et al. 1997b). Expression of the mutant β3 subunits (β3H292A) resulted in the formation of functional ion channels similar to the wild-type β3 subunits, since they could not be activated by GABA (data not shown), possessed increased membrane conductances (16.8 ± 2.8 μS, n=20, P < 0.05) compared with uninjected oocytes, and were positively modulated by 100 μm pentobarbitone and inhibited by 10 μm picrotoxinin (Fig. 2A and C).

Figure 2.

Minimal effect on the pentobarbitone- and picrotoxinin-induced modulation of β3 homomeric function by the H292A mutation

A, membrane currents recorded from oocytes expressing β3 (a) or β3H292A (b) subunits before and after application of 100 μm pentobarbitone (PB). The membrane current steps were evoked by hyperpolarizing voltage commands (see Methods). B, normalized concentration–response relationships for PB obtained from β3 (▪) and β3H292A (▴) homomers; data were fitted as described in Methods. C, membrane currents before and during application of 10 μm picrotoxinin (PTX) for β3 (a) and β3H292A (b) subunits. Calibration bars in A and C are 100 nA and 10 s. D, PTX inhibition curves constructed for the antagonism of the spontaneous conductance using β3 (▪) and β3H292A (▴) homomers. Curve fitting was performed as described in Methods. All points are means ±s.e.m.

Concentration–response curves for the pentobarbitone enhancement of the spontaneous conductance over the range 1–2000 μm revealed that the H292 A mutation laterally shifted the curve, lowering the EC50 for pentobarbitone from 138.7 ± 9.5 μm (β3; nH 1.34 ± 0.1) to 46.0 ± 3.2 μM(β3H292A; nH 1.20 ± 0.08; n= 5–33; Fig. 2B). Analysis of the concentration–inhibition curves for picrotoxinin (1 nm–10 μM) indicated a similar 3- to 4-fold shift in potency (Fig. 2D) with a reduction in IC50 from 84 ± 4.4 nm (β3, nH, 0.84 ± 0.13) to 23.5 ± 0.7 nm (β3H292A; nH 1.2 ± 0.04; n= 3–5). Thus the functional properties of β3 homomers were not substantially affected by the H292A mutation.

However, mutation of H292 produced a profound reduction in the potency of Zn2+ antagonism of the β3 subunit-gated conductance. For β3 wild-type subunits, the Zn2+-sensitive conductance was substantially inhibited by 10–100 μm Zn2+, reducing the resting leak conductance (2.49 ± 0.26 μS, n=6) to values expected for uninjected cells. The level of inhibition (approximately 100%) caused by 100 μm Zn2+ was abolished by the H292A mutation (Fig. 3A). Interestingly, the concentration–inhibition curves revealed that even for the mutant β3H292A homomer, raising the Zn2+ concentration up to 2 mm still caused substantial inhibition (Fig. 3BB). The inhibition curve for the wild-type β3 homomer was laterally shifted by mutating H292 with a substantive 1000-fold increase in the Zn2+ IC50 from 0.31 ± 0.02 μm (β3; nH, 0.91 ± 0.04) to 307.2 ± 32.6 μm (β3H292A; nH l.6 ± 0.26; n=4–10). Zn2+ also produced a mixed/non-competitive block of the pentobarbitone concentration–response curves for both β3 (+1 μm Zn2+) and β3H292A (+300 μm Zn2+) subunits (n=3; data not shown).

Figure 3.

Sensitivity of Zn2+ inhibition to H292 in TM2 of the β3 subunit: β homomeric receptors

A, membrane currents for β3 (a) and β3H292A (b) before and after the application of 100 μm Zn2+. Calibration bars are 100 nA and 10s. B, concentration–inhibition curves for Zn2+ were obtained for β3 (▪) and β3H292A (▴) homomers. The H292A mutation shifted the curve to the right by 1000-fold. C, I–V relationships for the resting membrane conductance in the absence and presence of 300 nn and 100 μm Zn2+ for an oocyte expressing wild-type β3 subunits. The reversal potentials from 4 cells are -27.9 ± 2.5 mV (control), -28.0 ± 2.9 mV (+300 nm Zn2+) and -32.3 ± 5.1 mV (+100 μm Zn2+). D, the percentage inhibition by 300 nm Zn2+ of the wild-type β3 subunit-gated conductance over the membrane potential range −100 to 50 mV was calculated from the I–V plots obtained from 4 cells.

Voltage dependence of Zn2+ inhibition

The location of H292 within TM2 suggests that Zn2+ binding to this residue will be subject to the membrane electric field, and the inhibition should exhibit voltage sensitivity. A control I–V plot for β3 homomers exhibited outward rectification as expected (Wooltorton et al. 1997b), and interestingly, similar I–V analysis for the mutant β3H292A also resulted in outward rectification. Application of Zn2+ reduced membrane current without any overt voltage sensitivity; the I–V relationship for β3 subunits exhibited only outward rectification (Fig. 3C and D).

α1β3 GABAA receptor constructs: relevance of H292 to inhibition by Zn2+

The inhibition of recombinant GABAA receptors by Zn2+ was first noted for α1β1 (Smart et al. 1991) and α1β2 (Draguhn et al. 1990) constructs, which are more likely to exist in vivo than GABA-insensitive, spontaneously gated β3 homomers. Whether H292 in TM2 of β3 subunits is also critical in supporting the inhibition by Zn2+ of GABA-activated responses was addressed using α1β3 receptors. The potency of Zn2+ on α1β3 receptor subunits was comparable to that for β3 homomers (Fig. 4A), with the inhibition curve for antagonism of GABA-activated responses providing an IC50 of 0.11 ± 0.01 μm (nH 0.92 ± 0.1; n= 3–5; Fig.4D). This IC50 is in accordance with values reported previously for α1β1 and α1β2 constructs (Draguhn et al. 1990; Smart et al. 1991). Co-expression of α1β3H292A subunits produced a receptor with comparable sensitivity to GABA compared with the wild-type α1β3 with EC50 values of 3.24 ± 0.2 μm (α1β3; bH, 1.14 ± 0.04) and 5.7 ± 0.36 μm (α1β3H292A; nH, 1.01 ± 0.03; n= 3–5; Fig. 4C). However, the Zn2+ sensitivity of the α1β3H292A construct was significantly reduced compared with the wild-type receptor with an IC50 of 22.8 ± 4.2 μM (nH, 0.49 ± 0.04; n= 3–5; (Fig. 4B and D). Thus H292 is also critical for the functional inhibition induced by Zn2+ on heteromeric GABAA receptors.

Figure 4.

Zn2+ inhibition of α1β3 heteromers is also affected by H292 in the β3 subunit

A, GABA-activated (2 μm) responses recorded from oocytes expressing α1β3 subunits in the absence (a) and presence (b) of 3 μm Zn2+. Calibration bars are 100 nA and 10s. B, membrane currents recorded in oocytes expressing α1β3H292A after application of equipotent GABA (5 μm) compared with the wild-type α1β3 receptor in the absence (a) and presence (b) of 3 μm Zn2+. Calibration bars are 200 nA and 10s. C, concentration–response curves for GABA for α1β3 (▪) and α1β3H292A (▴) heteromers. Responses were normalized to the maximum GABA-evoked response. D, Zn2+ inhibition curves for the antagonism of the EC50 (2 or 5 μm) GABA-activated response on α1β3 (▪) and α1β3H292A (▴).


Locating the Zn2+ binding site on recombinant GABAA receptors has proved elusive. This is because of the heterogeneous nature of these pentameric receptors and the possibility that more than one subunit may contribute to the binding site, as shown by the complex effects on inhibition following the substitution of γ2 and variation of α subunits (Smart et al. 1994; White & Gurley, 1995). The observation that β1 and β3 subunit homomers formed Cl-selective, spontaneously gated ion channels sensitive to picrotoxinin and Zn2+ proved experimentally useful, particularly since the potency of Zn2+ was higher than previously reported for many heteromeric GABAA receptor subunit combinations (Krishek et al. 1996).

External histidine residues were targeted since studies on invertebrate GABA receptors indicated that H+ and Zn2+ modulated receptor function by binding to the same site and that the pKa for H+ modulation was 6.1, implicating histidine residues in the binding site for Zn2+ (Smart & Constanti, 1982). Although Zn2+ inhibition of GABA-activated responses on sympathetic neurones was unaffected by external pH (Smart, 1992), Zn2+ modulation of recombinant α1β1 GABAA receptors is sensitive to pH (B. J. Krishek & T. G. Smart, unpublished observations) suggesting that external histidines were forming the binding site. The histidine in TM2 was selected for mutagenesis since this is unique to the β subunit isoforms and may be the cause of the high sensitivity to Zn2+. The mutation of H292 in the β3 subunit had such a profound effect on Zn2+ potency that it is likely to be directly involved in the binding site rather than influencing binding in an allosteric manner. The location of H292 within the TM2 domain also suggests that Zn2+ is gaining access to a region of the channel thought to be the exclusive preserve of anions. However, the access of positively charged Zn2+ indicates that ionic selection is probably enforced nearer to the cytoplasmic entrance of the ion channel. This view is supported by substituted cysteine mutagenesis of residues in TM2 that are accessible to cysteine-selective reagents (one of which is positively charged) by apparently penetrating into TM2 as far as threonine 261 (rat α1 subunit; Xu & Akaras, 1996), indicating the selectivity filter must be deep in the ion channel. It may also be deduced from the location of H292 that Zn2+ inhibition should be voltage sensitive and clearly it is not, as noted for other neuronal and recombinant GABAA receptors (Smart et al. 1994). To explain this apparent anomaly and clarify further how Zn2+ may gain access to the presumed anion channel lumen requires Zn2+ to form the complex ion (ZnCl4)2- (Cotton & Wilkinson, 1988), within or near the ion channel mouth. By virtue of the net negative charge, presumably (ZnCl4)2- would gain access to the ion channel and following Cl ion (and water) stripping within the channel, yield Zn2+ which then binds to the site of action. In this way we would not expect the Zn2+ block to exhibit any degree of voltage sensitivity. In addition, the topographic view of the GABAA receptor ion channel is still only a model; it is therefore possible that H292 is close to the surface membrane and perhaps not subject to the membrane electric field expected for an ion channel location. This could also account for a lack of voltage dependence to Zn2+ inhibition.

Both the mutant β3H292A subunit and the wild-type β3 homomer exhibited similar whole-cell membrane conductances and apparent levels of receptor expression, which is in accordance with the histidine residues having little effect on Cl conductance per se; however, the sensitivities to pentobarbitone and picrotoxin were slightly increased by the H292A mutation suggesting some effect on ion channel gating and/or an allosteric effect on the ligand binding sites which, for picrotoxinin (Xu, Covey & Akabas, 1995; Gurley, Amin, Boss, Weiss & White, 1995) and pentobarbitone (Birnir, Tierney, Dalziel, Cox & Gage, 1997; Belelli, Lambert, Peters, Wafford & Whiting, 1997), may involve residues in TM2. Interestingly, the Hill coefficient for the block of the β3 Cl-selective conductance by Zn2+ is less than the theoretically predicted 5, assuming the channel is a pentamer. This is not unusual for homomers and may indicate that only one Zn2+ ion is sterically capable of blocking the channel. Alternatively, five Zn2+ ions could bind but inhibition ensues after the binding of only a single Zn2+ ion. The lack of open channel block by Zn2+ of the pento-barbitone-gated currents, as might be expected for the relative position of the Zn2+ binding site, could also be explained by assuming that not only the anion selectivity filter, but also the ion channel gate is located deep into the ion channel. Thus blockers acting at proximal sites to the gate would not be expected to display a dependence on the agonist/modulator concentration. This notion of a near-cytoplasmic location for the ion channel gate is also in accordance with cysteine modifying reagents penetrating deep into a closed ion channel in the absence of GABA (Xu & Akabas, 1996). The block of GABA-mediated currents by Zn2+ in α1β1 heteromers also failed to exhibit an open channel block profile (Draguhn et al. 1990; Smart et al. 1991). The relative position of the anion selectivity filter and channel gate beyond the Zn2+ binding site into TM2 may also account for the lack of effect of internal Zn2+ (Celentano et al. 1991) which probably cannot penetrate past these structures from the inside to gain access to H292 in the β subunit.

The smaller shift in Zn2+ sensitivity caused by the H292A mutation in α1β3H292A heteromers (200-fold) compared with β3H292A homomers (1000-fold) may relate to the formation of a putative ‘histidine ring’ in homomeric β3 subunit ion channels. Mutation to alanines will cause disruption and possibly complete loss of Zn2+ binding. In contrast, for α1β3 constructs, amino acids contributed by the α1 subunit(s) could help to stabilize Zn2+ binding in the mutant receptor resulting in a smaller shift in sensitivity. Alternatively, Zn2+ binding might rely on juxtaposed β subunits each supplying H292 to co-ordinate the binding site, but in the heteromer α1β3, if only two or three β subunits are present, they may not be juxtaposed being largely or completely separated by α subunits resulting in a less significant shift in sensitivity upon β subunit mutation. For αβγ subunit constructs, the number of β subunits present might be reduced further to at most two or even one per receptor. Zn2+ binding could be disrupted further by the insertion of γ2 subunits between either α1 and β3, or between juxtaposed β3 subunits in the receptor complex. Moreover, the presence of a conserved positively charged lysine residue in the extracellular TM2 flanking zone of γ2 subunits might also hinder the access of Zn2+ to H292 in the ion channel (Smart, 1992; Smart et al. 1994). Either of these features would serve to reduce the potency of Zn2+ in γ2 subunit-containing receptors. Determination of receptor stoichiometry and subunit positions in GABAA receptors will clearly help to answer these questions.

Recently, using GABA-activated ρ1 subunit homomers, the inhibition caused by Zn2+ was abolished by mutating a histidine residue at position 156 located in the presumed N-terminus of the subunit (Wang, Hackam, Guggino & Cutting, 1995a). This is clearly different from the TM2 location proposed for GABAA receptors suggesting that Zn2+ is acting at two quite distinct sites on GABAA and ρ1 subunit-containing GABAC receptors, which is quite unlike picrotoxin which has a similar TM2 binding location on ρ1 receptors (Wang, Hackam, Guggino & Cutting, 1995b; Enz & Bormann, 1995). Despite the profound shift in the inhibition curve for Zn2+ on β3 homomers by mutating H292, higher concentrations of Zn2+ were still capable of inhibiting the conductance suggesting that another, possibly lower affinity, site exists elsewhere on the receptor. Complexation of Zn2+ is unlikely to play a role in this decline in sensitivity since Zn2+ is inhibiting a spontaneous conductance for which no ligand is necessary. Interestingly, for the ρ1 subunit, the Zn2+ inhibition curves for both the wild-type receptor and the H156Y mutant were displaced to the right of comparable curves reported in the present study for β3 and β3H292A subunits. Intriguingly, at high Zn2+ concentrations (> 5 mm), the ρ1 subunit GABA response exhibits threshold inhibition, indicative of another potential low affinity Zn2+ binding site.

In conclusion this represents the first description of a Zn2+ binding site on the GABAA receptor, demonstrating that such a site can exist within TM2, an area thought previously to be accessible only to anions. We propose that access of cations to the ion channel indicates that structures involved in the gate and selectivity filter are located deep within the channel. This site of action for Zn2+ would account for its modulation of susceptible (γ2 subunit-lacking) GABAAreceptors.

Note added in proof

Following submission of this paper, J. Horenstein & M. H. Akabas (Society for Neuroscience Abstracts 23, 51.19 (1997)) have noted in the rat β1 subunit that mutating histidine 267 (comparable to H292 in the murine β3 subunit) to serine also markedly reduces the sensitivity of the α1β1 receptor to inhibition by Zn2+.


This work was supported by the MBC. J.B.A.W. is supported by an MBC Collaborative Studentship with Merck, Sharp & Dohme.