Introduction

Top Summary Introduction Materials & Methods Results Discussion References

The GABA_A receptors are members of the ligand-gated ion channel gene superfamily and form anion-selective channels (Macdonald and Olsen, 1994; Karlin and Akabas, 1995; Sieghart, 1995). The functional receptor complex is formed as a pentamer of homologous subunits arranged pseudosymmetrically around the central channel axis (Unwin, 1993; Macdonald and Olsen, 1994; Nayeem et al., 1994). Numerous GABA_A receptor subunits have been cloned, including six , four , three , one , one , and three  subunits (Macdonald and Olsen, 1994; Sieghart, 1995). The subunits have a similar transmembrane topology with an ~200-amino acid extracellular amino terminus, three closely spaced membrane-spanning segments (M1, M2, M3), a cytoplasmic loop of variable length, a fourth membrane-spanning segment (M4), and an extracellular carboxyl terminus (Macdonald and Olsen, 1994). Using the substituted-cysteine-accessibility method, we have shown that the channel lining is formed, at least in part, by residues from the M2 segment (Xu and Akabas, 1996).

Functional receptors are formed by coexpression of  and  subunits, although the presence of the  subunit is essential for benzodiazepine potentiation (Schofield et al., 1987; Pritchett et al., 1989; Gorrie et al., 1997). In heterologous expression systems, the subunit stoichiometry for receptors formed by expression of the  and  subunits is uncertain; support has been provided for two  and three  subunits (Tretter et al., 1997) and for three  and two  subunits (Im et al., 1995; Kellenberger et al., 1996), as well as other stoichiometries (Gorrie et al., 1997). When the  subunit is included, the stoichiometry seems to be two , two , and one  subunit (Chang et al., 1996; McKernan and Whiting, 1996; Tretter et al., 1997). Some  subunits also form homomeric channels, but they tend to be constitutively open (Krishek et al., 1996).

The divalent cation Zn²⁺ blocks GABA-induced currents with variable affinity depending on the subunit composition of the receptors (Westbrook and Mayer, 1987; Draguhn et al., 1990; Legendre and Westbrook, 1991; Smart et al., 1991; Harrison and Gibbons, 1994; Saxena and Macdonald, 1994; Smart et al., 1994; Chang et al., 1995; Wang et al., 1995). It is likely that the effects of Zn²⁺ are mediated by interactions with different sites in receptors with different subunit compositions (Harrison and Gibbons, 1994; Smart et al., 1994; Saxena et al., 1997). Receptors formed by coexpression of  and  subunits display the highest affinity for Zn²⁺ block with an IC₅₀ of ~1 µM (Draguhn et al., 1990; Smart et al., 1991). In these receptors, Zn²⁺ block is slightly voltage dependent (Draguhn et al., 1990), implying that the binding site may be in the channel. The  homomeric channels also display high affinity Zn²⁺ block (Draguhn et al., 1990; Krishek et al., 1996). The addition of the 2 subunit to the functional complex markedly reduced Zn²⁺ block so that 100 µM Zn²⁺ caused only 17% inhibition of the GABA-induced currents (Draguhn et al., 1990; Smart et al., 1991). Zn²⁺ has been proposed to inhibit GABA-induced currents by stabilizing the closed state of the receptor (Smart et al., 1994). We sought to identify the residues that form the high affinity Zn²⁺ binding site in the 11 GABA_A receptor.

In proteins of known crystal structure, bound Zn²⁺ ions interact directly with two to four amino acids. The amino acids found at Zn²⁺ binding sites include histidine, cysteine, aspartate, or glutamate (Higaki et al., 1992; Regan, 1993; Berg and Shi, 1996). Because charged, water-soluble, sulfhydryl reactive reagents applied extracellularly have no effect on GABA_A receptors (Xu and Akabas, 1993, 1996) it is unlikely that a cysteine residue is available to interact with extracellularly applied Zn²⁺. In GABA receptors formed by the 1 subunit, a histidine residue in the extracellular domain was shown to mediate lower affinity Zn²⁺ block (IC₅₀ = 16 µM) (Wang et al., 1995); histidine, however, is not conserved at the aligned position in other subunits. In the aligned sequences of GABA_A receptor subunits, we noted that histidine was conserved at the position aligned with 1His267 in all  subunits but is not found at the aligned position in other subunit subtypes. This position aligns with 1Ser272 in the M2 membrane-spanning segment; we had previously shown that this residue was exposed in the channel lining (Fig. 1) (Xu and Akabas, 1996). We hypothesized that the high affinity Zn²⁺ binding site was formed by histidine residues from position 1 267 contributed by at least two  subunits.

View larger version (11K): [in this window] [in a new window]   Fig. 1.   Aligned sequences in and flanking the M2 membrane-spanning segments of the rat GABAA receptor 1, 1, and 2 subunits. Shaded, residues aligned with 1His267; *, channel-lining residues identified in the 1 subunit by Xu and Akabas (1996).

	Materials and Methods

Top Summary Introduction Materials & Methods Results Discussion References

Oligonucleotide-mediated mutagenesis. The cDNAs encoding the rat 1 and 2 subunits in the pBluescript SK() plasmid (Stratagene, La Jolla, CA) were obtained from Dr. P. Seeburg (Max-Planck Institute for Medical Research, Heidelberg, Germany), and the 1 subunit in the pBluescript SK vector was from Dr. A. Tobin (University of California, Los Angeles). The subunits were excised from the pBluescript clones by restriction digestion with the enzymes 1 XhoI, 1 XbaI and HindIII, and 2 EcoRI. The 1 and 2 subunits were ligated into the pGEMHE vector (Liman et al., 1992) digested with the corresponding enzymes. For the 1 subunit, the XhoI cut fragment was blunt ended and ligated into pGEMHE, which had been digested with SmaI. Subcloning and orientation were confirmed by restriction digestion and DNA sequencing. The Altered-Sites Mutagenesis procedure (Promega, Madison, WI) was used to generate mutations as described previously (Xu et al., 1995). Mutations were confirmed by DNA sequencing.

Preparation of mRNA and oocytes. Plasmids were linearized with NheI for in vitro mRNA transcription with T7 RNA polymerase. In vitro mRNA transcription and the preparation and injection of Xenopus laevis oocytes were performed as described previously (Xu et al., 1995). Oocytes were injected with 10 ng of mRNA encoding the 1 and 1 subunits in a 1:1 ratio or with 1, 1, and 2 subunits in a 1:1:1 ratio.

Electrophysiology. GABA-induced currents were recorded from individual oocytes under two-electrode voltage-clamp, at a holding potential of 50 mV. Electrodes were filled with 3 M KCl and had a resistance of <2 M. The ground electrode was connected to the bath via a 3 M KCl/Agar bridge. Data were acquired and analyzed on a 486/33 MHz computer using a TEV-200 amplifier (Dagan Instruments, Minneapolis, MN), a Digidata 1200 data interface (Axon Instruments), and pCLAMP 6 software (Axon Instruments, Foster City, CA). The oocyte was perfused at 5 ml/min with CFFR (115 mM NaCl, 2.5 mM KCl, 1.8 mM MgCl₂, 10 mM HEPES, pH 7.5, with NaOH) at room temperature. The recording chamber had a volume of ~0.25 ml.

Experimental protocol. In all experiments, GABA was applied at a concentration 10 times the GABA EC₅₀ value of the mutant or wild-type unless otherwise indicated. Applications of GABA were separated by 3-5-min washes with CFFR. In all experiments used for analysis, the magnitude of the GABA-induced current changed by <10% between two consecutive applications of GABA.

Curve fitting. The concentration dependence of the inhibition of the GABA-induced currents by Zn²⁺ was fit with the Hill equation, I/I_max = 1/[1 + (IC₅₀/Zn)ⁿ], where IC₅₀ is the concentration of Zn²⁺ that causes half-maximal inhibition, Zn is the concentration of Zn²⁺, and n is the Hill coefficient, using either Prism 2.0 (GraphPAD, San Diego, CA) or custom software kindly provided by Dr. Juan Pascual (Columbia University, New York, NY).

Reagents. A 1 M stock solution of ZnCl₂ was prepared by adding sufficient HCl to eliminate all visible precipitates. All working solutions of ZnCl₂ were prepared daily by diluting the 1 M stock solution in CFFR. The addition of ZnCl₂ to CFFR did not change the pH of the solution.

	Results

Top Summary Introduction Materials & Methods Results Discussion References

Characterization of the mutants. We expressed all of the mutants in X. laevis oocytes as either 11 or 11 2 combinations where the mutant subunit or subunits replaced the corresponding wild-type subunit or subunits. GABA-induced currents were observed in oocytes expressing each of the combinations. For all of the subunit combination, we determined the EC₅₀ value for GABA and the Hill coefficient of the GABA dose-response relationship (Table 1). For wild-type 11, the GABA EC₅₀ value was 3.4 ± 0.7 µM and the Hill coefficient was 0.97 ± 0.12. The EC₅₀ values of the mutants ranged from 5-fold smaller than wild-type for 11H267S to 1.6-fold larger than wild-type for the 1S272H1 mutant. The EC₅₀ value is a function of both the intrinsic affinity of the binding sites for GABA and the isomerization rate constants for transitions between the various closed, open, and desensitized states (Akabas et al., 1992). Because channel gating involves conformational changes in the membrane-spanning segments to transduce ligand binding in the extracellular domain to opening of the gate near the cytoplasmic end of the channel (Xu and Akabas, 1996), mutations of residues in membrane-spanning segments can alter the isomerization rate constants. Thus, the observed changes in EC₅₀ probably arise from effects of the mutations on the transduction process.

Residues forming the high affinity Zn²⁺ binding site. Zn²⁺ blocked GABA-induced currents arising from wild-type 11 GABA_A receptors with an IC₅₀ value of 0.54 ± 0.02 µM (four experiments) (Fig. 2) similar to the IC₅₀ values reported by other investigators for _xy receptors (Draguhn et al., 1990; Smart et al., 1991). 1His267 aligns with 1Ser272, a residue that we had previously shown to be a channel-lining residue (Xu and Akabas, 1996). To determine whether 1His267 was involved in forming the high affinity Zn²⁺ binding site, we mutated the histidine to the amino acids found at the aligned positions in the 1 and 2 subunits, serine and isoleucine, respectively (Fig. 1). The affinity for Zn²⁺ block of both mutants was reduced by >300-fold (Fig. 2); for 11H267S, the IC₅₀ value was 161 ± 40 µM (three experiments), and for 11H267I, the IC₅₀ value was 654 ± 113 µM (nine experiments) (Table 2). The residual inhibition likely arises from an interaction of Zn²⁺ with a different site or sites on the receptor; removal of the high affinity site has unmasked this previously unrecognized low affinity site or sites. The location of this low affinity binding site is unknown.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   High affinity Zn2+ inhibition of wild-type 11 GABAA receptors is reduced in the mutants 11H267S and 11H267I. A, GABA-induced currents recorded by two-electrode voltage clamp from an oocyte expressing wild-type 11 receptors in the presence of varying concentrations of ZnCl2. B, Concentration dependence of the Zn2+ inhibition of GABA-induced currents for wild-type 11 (), 11H267S (), and 11H267I (). Mean ± standard error values are plotted; for some points, the error bars are smaller than the symbol. The data were fit by the Hill equation. Solid lines, calculated from the fitted equation.

Interactions with residues in other subunits. To investigate whether a histidine at the aligned position in the 1 subunit would interact with 1His267, we generated the mutant 1S272H and expressed it with the wild-type 1 subunit. With five histidines in the channel lining, the IC₅₀ value of the 1S272H1 receptor was 0.025 ± 0.007 µM (three experiments) (Fig. 3); that the affinity was significantly higher than wild-type suggests that the aligned residues in different subunits are in close proximity in the channel lumen. To ensure that the higher affinity was not solely due to placement of the histidine in the 1 subunit, we examined the effect of Zn²⁺ on the receptor that at this position only contained histidine in the 1 subunit, 1S272H1H267S; the IC₅₀ value was 26 ± 1 µM (eight experiments) (Fig. 3, Table 2). The affinity of 1S272H1H267S for Zn²⁺ was lower than that in wild-type but higher than that in the 11H267S receptor, a receptor with no histidines at this level of the channel.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   The affinity for Zn2+ inhibition is increased in the 1S272H1 receptor and decreased in the 1S272H1H267S receptor relative to the wild-type 11 receptor. Concentration dependence of the Zn2+ inhibition of GABA-induced currents for wild-type 11 (), 1S272H1 (), and 1S272H1H267S (). Mean ± standard error values are plotted; for some points, the error bars are smaller than the symbol. The data were fit by the Hill equation. Solid lines, calculated from the fitted equation.

Cysteine substitution for 1His267. It was hypothesized that Zn²⁺ inhibits GABA-induced currents by binding to and stabilizing the closed state of the GABA_A receptor (Smart et al., 1994). Therefore, we sought to determine whether 1His267 was accessible to interact with ions in the closed state of the channel. We used the substituted-cysteine-accessibility method (Akabas et al., 1992; Xu and Akabas, 1993) to probe the accessibility of a cysteine residue substituted at position 267 in the closed state of the channel. In this approach, a cysteine residue is engineered into the protein. The cysteine-substitution mutant is expressed, and the ability of charged, sulfhydryl-specific reagents to react with the engineered cysteine residue is tested. The magnitude of the GABA-induced current is determined; then, the sulfhydryl reagent is applied for 1 min in the absence of GABA (i.e., in the closed state of the channel) and washed out, and the magnitude of the GABA-induced current is determined again. If the magnitude of the GABA-induced currents after application of the sulfhydryl reagent is significantly different than the magnitude of the GABA-induced currents before application of the sulfhydryl reagent, we infer that the sulfhydryl reagent has covalently modified the engineered cysteine. The sulfhydryl reagents we used are pCMBS, MTSES, and MTSET⁺ (Xu and Akabas, 1996). Because these reagents react more rapidly with the thiolate anion (S) than with the neutral thiol (SH) (Hasinoff et al., 1971; Roberts et al., 1986) and because only cysteines on the water-accessible surface of the protein are likely to ionize to a significant extent, we assume that these reagents will only react with cysteine residues on the water-accessible surface of the protein at an appreciable rate. Covalent modification of a channel-lining cysteine may alter the single-channel conductance and/or the gating kinetics (Akabas et al., 1994). To detect the effects of covalent modification, we used two different test concentrations of GABA: one at the GABA EC₅₀ value and the other at 10 times the EC₅₀ value. Zhang and Karlin (1997) have shown that applying the agonist at the EC₅₀ value for the test responses before and after the reagent provides a more sensitive test for detecting covalent modification because it allows one to detect reaction when the effect of modification is a change in gating alone with little or no change in single-channel conductance. In contrast, application at 10 times the EC₅₀ value will detect changes in conduction but should be insensitive to changes in gating.

View larger version (19K): [in this window] [in a new window]   Fig. 4.   The engineered cysteine in the mutant 11H267C is accessible to covalent modification by charged, sulfhydryl-specific reagents applied extracellularly in the closed state of the receptor. A and B, Currents recorded from oocytes expressing the mutant 11H267C by two-electrode voltage-clamp. Bars over the traces, periods during which reagents were applied to the oocytes. Holding potential was 50 mV. Note the increase in the magnitude of the two current responses after the application of MTSET+ relative to the initial two responses. Traces, separated by a 5-min wash with CFFR. The GABA concentration was 5 µM in A and 50 µM in B. C, Effect of a 1-min application of 0.5 mM pCMBS, 10 mM MTSES, or 1 mM MTSET+ applied extracellularly in the absence of GABA on wild-type () and 11H267C (). The test concentrations of GABA were at the EC50 value (left) and at 10 times the EC50 value (right). Mean ± standard error values are plotted. Parentheses, number of oocytes tested. Positive effect, potentiation of the subsequent GABA-induced currents; negative effect, inhibition of the subsequent GABA-induced currents.

View larger version (7K): [in this window] [in a new window]   Fig. 5.   Covalent modification of the engineered cysteine in the 11H267C mutant reduces the inhibition by Zn2+. The reagents were applied during the periods indicated (bars above the traces) at 2 µM GABA, 5 mM MTSET+, and 30 µM Zn2+. Zn2+ inhibited the GABA-induced current by 63% before modification by MTSET+ and by 14% after modification by MTSET+. All other conditions were the same as for Fig. 4. Note that the rate of GABA-induced desensitization is increased after modification.

Effect of the 2 subunit on Zn²⁺ affinity. It has previously been shown that the IC₅₀ value for Zn²⁺ inhibition of GABA_A receptors containing the 2 subunit is much higher than that for receptors formed from only the  and  subunits (Draguhn et al., 1990; Smart et al., 1991). Consistent with this, we found that the Zn²⁺ IC₅₀ value for wild-type 112 was >1 mM (three experiments). In the 2 subunit, the residue Ile282 aligns with 1His267 (Fig. 1). We mutated this residue to histidine, 2I282H, and expressed the mutant with wild-type 1 and 1 subunits, but the Zn²⁺ IC₅₀ value also was >1 mM (three experiments) (Table 2). Expression of the other combinations of histidines at this position 1S272H1H267S2I282H, and 1S272H12I282H, which should contain five histidines, also gave IC₅₀ values of >1 mM (Table 2). Thus, inclusion of the 2 subunit prevents the ability of Zn²⁺ to inhibit GABA-induced currents regardless of the number of histidines present at this level.

	Discussion

Top Summary Introduction Materials & Methods Results Discussion References

Zn²⁺ is a high affinity inhibitor of 11 GABA_A receptors (Draguhn et al., 1990; Smart et al., 1991). We have shown that the Zn²⁺ affinity of 11 GABA_A receptors is reduced by >300-fold by mutation of 1His267 in the M2 membrane-spanning segment to either serine or isoleucine, the amino acids at the aligned positions in the 1 and 2 subunits (Fig. 1). Furthermore, we have shown that 1His267 is exposed in the channel lining (Fig. 4) as we had previously shown for the aligned residue in the 1 subunit, 1Ser272 (Xu and Akabas, 1996). Histidine is conserved at the position aligned with 1His267 in all GABA_A receptor  subunits but is not found at the aligned position in other subunit subtypes. Histidine residues frequently form part of metal ion binding sites (Higaki et al., 1992; Regan, 1993; Berg and Shi, 1996). Thus, we infer that 1His267 participates in the formation of the high affinity Zn²⁺ site in 11 GABA_A receptors. The position of this residue near the extracellular end of the M2 channel-lining segment is consistent with the slight voltage dependence reported for Zn²⁺ block of 11 GABA_A receptors (Draguhn et al., 1990). A different histidine in the extracellular amino-terminal domain of the GABA 1 subunit was implicated in lower affinity Zn²⁺ inhibition of the GABA_C receptor (Wang et al., 1995).

Because the GABA_A receptor is formed as a pentamer of subunits arranged pseudosymmetrically around the central channel axis (Unwin, 1993; Nayeem et al., 1994), one would hypothesize that the aligned channel-lining residues from each subunit should be at approximately the same distance into the channel, thereby forming a ring of residues at a given level of the channel. Insertion of histidine into the aligned position in the 1 subunit and expression with wild-type 1, which should give five histidines at this level, increases the affinity for Zn²⁺ by 20-fold. This suggests that the aligned residues in different subunits are in close proximity in the channel lumen and that the presence of more histidines at this level allows for the formation of a higher affinity binding site.

To form a high affinity binding site, Zn²⁺ must interact with more than one residue. The affinity of a site will depend on the number of chelating residues, the relative position of the residues, and the local electrostatic environment (Higaki et al., 1992; Regan, 1993; Berg and Shi, 1996). It is likely that at least two histidines form the high affinity Zn²⁺ binding site in the 11 GABA_A receptor. The subunit stoichiometry of the 11 GABA_A receptor is uncertain, and evidence has been reported for both two  and three  subunits (Tretter et al., 1997) and three  and two  subunits (Im et al., 1995; Kellenberger et al., 1996; Gorrie et al., 1997). We are uncertain why the affinity for Zn²⁺ is 50-fold higher when the histidine is in the  subunit in wild-type receptor compared with when the histidine is in the  subunit in the 1S272H1H267S mutant (Table 2). There are several potential explanations for this difference in affinity. If the subunit stoichiometry is 2:3, then the high affinity site could be formed between histidines in the adjacent  subunits that would be present (Fig. 6A); alternatively, all three histidines might interact with the Zn²⁺ (Fig. 6A). The lower affinity observed when the histidine was in the  subunit would arise because there would be two histidines in nonadjacent subunits that might be less favorable for Zn²⁺ binding (Fig. 6B). Alternatively, neighboring channel-lining residues, such as 1Glu270, also might influence the interaction between the histidines and Zn²⁺, thereby resulting in a higher affinity interaction when the histidine is in the  subunit compared with the  subunit, where the adjacent channel-lining residue is 1Asn275. Finally, the position of the  and  subunits relative to the channel may not be symmetrical, and the difference in the affinity is due to the asymmetry and not to the relative number of subunits.

View larger version (15K): [in this window] [in a new window]   Fig. 6.   Position of subunits and the histidine residues in the wild-type 11 receptor and in the 1S272H1H267S double mutant relative to the channel and the bound Zn2+ ion. A, Top view of the channel formed by two 1 and three 1 subunits. B, Top view of the channel formed by two 1S272H and three 1H267S subunits.

In X-ray crystal structures of proteins containing Zn²⁺ bound through histidine residues, the separation between the Zn²⁺ and the -amino group is ~2 Å (Higaki et al., 1992). This distance and the size of histidine constrain the maximum separation of the  carbons of two histidine residues bound to a Zn²⁺ ion to <13 Å (Higaki et al., 1992). Thus, in the Zn²⁺ bound state of the GABA_A receptor, at the level of 1His267, the  carbons of the aligned residues in both adjacent and nonadjacent subunits must be <13 Å apart (Fig. 7B, distances A-B and A-C).

View larger version (32K): [in this window] [in a new window]   Fig. 7.   Two views of the M2 membrane-spanning segments of the GABAA receptor, the positions of the Zn2+ and picrotoxin binding sites, and the inferred distances between subunits and within the channel. A, Channel-lining residues in the M2 segments of the 1 and 1 subunits. The channel-lining residues () in the 1 subunit were identified by the substituted-cysteine-accessibility method (Xu and Akabas, 1996). , Aligned residues in the 1 subunit. PTX, inferred location of the picrotoxin binding site (Xu et al., 1995). Zn2+, location of the Zn2+ binding site determined in this report. B, Three-dimensional representation of the M2 segments lining the GABAA receptor channel. Three of the five M2 segments are shown; the front two have been removed. Top, extracellular end. The M2 segments are shown as kinked helices based on their inferred structure from the 9-Å resolution structure of the homologous ACh receptor (Unwin, 1993). Points A-C, level of 1His267, the Zn2+ binding site. AB, distance between adjacent subunits. AC, distance between nonadjacent subunits. The distances between the C carbons of the residues at these positions must be <13 Å. Points D and E, at the level of 1Val257, the picrotoxin binding site (Xu et al., 1995). DE, must be 9 Å, the diameter of picrotoxin. Points F and G, at the narrowest region of the channel, the functional diameter of which was inferred to be ~5.6 Å based on the size of the largest permeant anion (Bormann et al., 1987).

Our previous work demonstrated that picrotoxin, a rigid, roughly spherical molecule 9 Å in diameter, binds near the cytoplasmic end of the channel in the region of 1Val257 (Fig. 7, distance D-E) (Xu et al., 1995). Because it reaches that site from the extracellular end of the channel, we inferred that the channel lumen must be 9 Å in diameter to the level of 1Val257 (Fig. 7) (Xu et al., 1995). Our current results constrain the maximum separation of the  carbon atoms on nonadjacent subunits at the level of 1His267 to <13 Å. It should be noted, however, that these distances may be measured in different states of the receptor. Picrotoxin binds in the open state of the channel (Newland and Cull-Candy, 1992), and Zn²⁺ may bind in the closed state of the channel (Smart et al., 1994). We do know, however, that charged sulfhydryl reactive reagents can react with engineered cysteine residues in the 1 M2 segment in the closed state of the channel (Xu and Akabas, 1996). These reagents would fit into a right cylinder 6 Å in diameter and 10 Å in length. Thus, in the closed state of the channel, the lumen must be 6 Å in diameter to allow these reagents to reach the engineered cysteine residues at positions more cytoplasmic than 1His267 (Xu and Akabas, 1996), but the  carbons of the residues aligned with 1His267 must be closer than 13 Å. The narrowest region in the channel was inferred, based on the size of the largest permeant anion, to be ~5.6 Å (Bormann et al., 1987), but this must be at a position that is more cytoplasmic than the picrotoxin binding site (Fig. 7B, distance F-G).

If the channel were lined by five  helical M2 segments arranged perpendicular to the membrane, the separation between the aligned position on the surface of adjacent  helices would be ~4 Å and that between nonadjacent -helices would be ~7 Å: The channel, however, is unlikely to be lined by five helices perpendicular to the membrane. In the 9 Å resolution structure of the ACh receptor reported by Unwin (1993), the putative M2 segments are not perpendicular to the membrane but rather angle out from the central channel axis toward the extracellular end of the channel (as illustrated in Fig. 7B). In the ACh receptor, we showed that residues near the extracellular end of the M1 membrane-spanning segment also were exposed in the channel lining, and we suggested that in the closed state, the M1 segments may intercalate between the M2 segments at the extracellular end of the channel (Akabas and Karlin, 1995). Thus, the 13 Å constraint on the separation of the  carbons seems to be reasonable given our current structural picture of the channel.

The GABA_A receptor forms a nearly ideally anion-selective channel (Bormann et al., 1987). The ability of a divalent cation, Zn²⁺, to penetrate from the extracellular end of the channel to the level of 1His267 in the M2 membrane-spanning segment indicates that the charge-selectivity filter that discriminates between anions and cations must be located at a more cytoplasmic position than 1His267. This is consistent with our previous results (Xu and Akabas, 1996) that showed that cationic sulfhydryl reagents could react with cysteines substituted for residues as cytoplasmic as 1Thr261, which aligns with 1Thr256 (Fig. 7A). We inferred that the charge-selectivity filter is located at a more cytoplasmic position than these residues, perhaps near the cytoplasmic end of the channel where the channel seems to narrow and form a picrotoxin binding site (Xu et al., 1995).

As other investigators had shown (Draguhn et al., 1990; Smart et al., 1991), the presence of the 2 subunit markedly reduces the affinity for Zn²⁺. The mechanism of this inhibition is not solely due to the lack of a histidine at the aligned position in the channel because substitution of a histidine at that position did not increase the affinity for Zn²⁺ (Table 2). Thus, some other aspect of the 2 subunit prevents Zn²⁺ inhibition of GABA-induced currents. Potential explanations include that (1) the 2 subunit may sterically restrict the conformations of the other subunits, and particularly 1His267, from adopting the conformation to which Zn²⁺ binds; (2) the putative adjacent channel-lining residue 2Lys285 may electrostatically interfere with Zn²⁺ binding; and (3) Zn²⁺ might still bind to the receptor complex containing the 2 subunit, but the presence of the 2 subunit prevents the conformational change induced by Zn²⁺ in the 11 receptor.