The α4β2 subtype is the most abundant nicotinic acetylcholine receptor (nAChR) in the brain and possesses the high-affinity binding site for nicotine. The α4 and β2 nAChR subunits assemble into two alternate stoichiometries, (α4)2(β2)3 and (α4)3(β2)2, which differ in their functional properties and sensitivity to chronic exposure to nicotine. Here, we investigated the sensitivity of both receptor stoichiometries to modulation by Zn2+. We show that Zn2+ exerts an inhibitory modulatory effect on (α4)2(β2)3 receptors, whereas it potentiates or inhibits, depending on its concentration, the function of (α4)3(β2)2 receptors. Furthermore, Zn2+ inhibition on (α4)2(β2)3 nAChRs is voltage-dependent, whereas it is not on (α4)3(β2)2 receptors. We used molecular modeling in conjunction with alanine substitution and functional studies to identify two distinct sets of residues that determine these effects and may coordinate Zn2+. Zn2+ inhibition is mediated by a site located on the β2(+)/α4(−) subunit interfaces on both receptor stoichiometries. α4H195 and β2D218 are key determinants of this site. Zn2+ potentiation on (α4)3(β2)2 nAChRs is exerted by a site that resides on the α4(+)/α4(−) of this receptor stoichiometry. α4H195 on the (−) side of the ACh-binding α4 subunit and α4E224 on the (+) side of the non-ACh-binding α4 subunit critically contribute to this site. We also identified residues within the β2 subunit that confer voltage dependency to Zn2+ inhibition on (α4)2(β2)3, but not on (α4)3(β2)2 nAChRs.
The α4β2 receptor is the most abundant nicotinic acetylcholine receptor (nAChR) in the brain where it forms the high-affinity binding site for nicotine (Picciotto et al., 2001). α4β2 receptors belong to the Cys-loop family of ligand-gated ion channels, which also includes the muscle nAChR and the GABAA receptor. When expressed in heterologous systems, α4 and β2 subunits produce two alternate stoichiometries, (α4)2(β2)3 and (α4)3(β2)2, which differ in functional pharmacology, desensitization kinetics, unitary conductance (Nelson et al., 2003; Moroni et al., 2006), calcium permeability (Tapia et al., 2007), and sensitivity to chronic exposure to nicotine (Nelson et al., 2003; Kuryatov et al., 2005; Moroni et al., 2006). A mixture of both receptors may be expressed by thalamic neurons (Butt et al., 2002), and their ratio is altered by an α4A529T polymorphism (Kim et al., 2003). By homology to the muscle α1γα1δβ1 nAChR (Unwin, 2005), it is presumed that the α4β2 nAChR harbors two functional agonist sites and that the order of subunits around the channel is α4β2α4β2(α4/β2). On both stoichiometries, the agonist sites reside on α4(+)/β2(−) interfaces, suggesting similar properties. That the fifth subunit can be either α4 or β2 leads to a different composition of non-agonist-binding interfaces. In the (α4)2(β2)3 receptor, there will be two β2(+)/α4(−) interfaces and a β2(+)/β2(−) interface, whereas in the (α4)3(β2)2 receptor there will be two β2(+)/α4(−) interfaces and an α4 (+)/α4(−) interface. Understanding how receptor stoichiometry influences the properties of α4β2 nAChRs is of general structural interest and has the potential to offer novel therapeutic strategies.
α4β2 nAChRs, like other Cys-loop ligand-gated ion channels, are modulated by Zn2+ (Hsiao et al., 2001). Although advances have been made in identifying Zn2+ sites on GABAA (Hosie et al., 2003) and glycine (Miller et al., 2005a,b) receptors, corresponding sites on α4β2 receptors have been relatively unexplored. A potentiating Zn2+-binding site has been proposed on α4β4/β2 nAChRs that involves Zn2+ coordination by a histidine residue (α4H162) and a less direct, probably structural, contribution from α4E59 (Hsiao et al., 2006). The site is likely to reside on the β2(+)/α4(−) subunit interfaces that alternate with the ACh-binding α4(+)/β2(−) subunit interfaces (Hsiao et al., 2006). Given that β2(+)/α4(−) interfaces are present on both receptor forms, receptor stoichiometry should not influence Zn2+ effects on α4β2 receptors (Hsiao et al., 2006). However, conceivably, because of structural homology, residues on non-agonist-binding subunits on (α4)3(β2)2 or (α4)2(β2)3 receptors could contribute to stoichiometry-specific Zn2+ effects.
We show that Zn2+ modulates (α4)2(β2)3 and (α4)3(β2)2 receptors differentially, which was not detected in previous studies (Hsiao et al., 2001, 2006). Whereas Zn2+ only inhibits (α4)2(β2)3 receptors, the function of (α4)3(β2)2 is either potentiated or inhibited, depending on concentration. Furthermore, Zn2+ inhibition on (α4)2(β2)3 is voltage dependent, whereas it is not on (α4)3(β2)2 receptors. By using mutagenesis with receptor modeling and functional studies, we have identified determinants of Zn2+ inhibition on both stoichiometries as well as residues implicated in Zn2+ potentiation on (α4)3(β2)2, providing evidence for a differential role of β2(+)/α4(−) and α4(+)/α4(−) interfaces in Zn2+ modulation.
Materials and Methods
Xenopus laevis were purchased from Horst Kaehler. The care and use of X. laevis frogs in this study were approved by the Oxford Brookes University Animal Research Committee and meet the guidelines of the Scientific Procedures Act, 1986, of the United Kingdom.
ACh, ZnCl2, tricine, and diethylpyrocarbonate (DEPC) were purchased from Sigma-Aldrich.
cDNA constructs and mutagenesis.
Human nAChR α4 and β2 subunit cDNAs were provided by Professor Jon Lindstrom (University of Pennsylvania, Philadelphia, PA) and subcloned into the pCI plasmid vector (Promega). Untranslated regions were eliminated by PCR to minimize differences in the transcription of the two cDNAs that may be brought about by untranslated regions flanking the coding sequence of the cDNAs (Briggs et al., 2006). Site directed mutagenesis was performed using the QuikChange mutagenesis kit (Stratagene). Unless otherwise mentioned, potential Zn2+ coordinating residues were replaced with alanine, a residue incapable of coordinating Zn2+. Mutants were tested, individually or in combinations, for their effects on Zn2+ modulation of α4β2 nAChRs. The full-length sequence of mutant subunit cDNAs was verified by DNA sequencing (Department of Biochemistry, University of Oxford, Oxford, UK). We present the numbering of the amino acid residues in terms of the full length, including the signal sequence. To obtain the position in the mature form, subtract 30 from the number for α4 and 25 for β2.
Injection of cDNAs constructs into nucleus of Xenopus oocytes.
Stage V and VI Xenopus oocytes were prepared as described previously (Moroni et al., 2006). Wild-type or mutant human α4 or β2 subunit cDNAs were injected into the nuclei of oocytes in a volume of 18.4 nl/oocyte, using a Nanoject Automatic Oocyte Injector (Drummond). To avoid the problem of stoichiometry heterogeneity, we used receptor expression strategies favoring the production of either (α4)2(β2)3 or (α4)3(β2)2 nAChRs (Moroni et al., 2006). Thus, to express the (α4)2(β2)3 nAChRs, α4 and β2 subunit cDNAs were combined in a ratio of 1:10, whereas a subunit ratio of 10:1 was used to produce the expression of (α4)3(β2)2 nAChRs (Moroni et al., 2006). The total amount of cDNA injected per oocyte was kept constant at 2 ng. After injection, oocytes were incubated at 18°C for 2–5 d in a modified Barth's solution containing 88 mm NaCl, 1 mm KCl, 2.4 mm NaHCO3, 0.3 mm Ca(NO3)2, 0.41 mm CaCl2, 0.82 mm MgSO4, 15 mm HEPES, and 50 μg/ml neomycin, pH 7.6.
Electrophysiology and data analysis.
Recordings were performed 2–5 d after injection. Oocytes were placed in a 0.1 ml recording chamber and perfused with modified Ringer solution [containing (in mm) 150 NaCl, 2.8 KCl, 10 HEPES, and 1.8 BaCl2, pH 7.2, adjusted with NaOH] at a rate of 15 ml/min. We chose a nominally Ca2+ free solution to minimize the contribution to the response of Ca2+-gated chloride channels which are endogenous to the Xenopus oocyte and may be activated by Ca2+ entry through the heterologously expressed nAChRs. The possible contribution of contaminant trace levels of Zn2+ in the external Ringer solution to the estimated potency of Zn2+ was investigated by adding 10 mm tricine to the following concentrations of Zn2+-Ringer solutions (in μm): 0.26, 0.78, 2.6, 7.8, 26, 77.5, and 254. Assuming that the KD for Zn2+ complexation is 10−5 m (Paoletti et al., 1997), the concentrations of free Zn2+ were estimated to be (in nm) 1, 3, 10, 30, 100, 300, and 1000, respectively. Oocytes were impaled by two agarose-cushioned microelectrodes filled with 3 m KCl (r = 0.3–1.0 MΩ) and voltage clamped at −60 mV, except for studies of voltage dependency, using a GeneClamp 500B amplifier and pClamp 8 software (Molecular Devices). Typically, traces were filtered at 1 kHz during recording and digitized at 1 kHz using the DigiData 1200 interface (Molecular Devices). All experiments were performed at room temperature.
To construct concentration–response curves for ACh, the peak current (I) was measured in the presence of each concentration of ACh applied at 4 min intervals to prevent desensitization. All peak currents were normalized to the peak amplitude of current induced by 1 mm ACh, the maximum effective concentration of ACh (Imax) at (α4)2(β2)3 or (α4)3(β2)2 nAChRs. Data were fitted with the Hill equation: I/Imax = [1/(1 + (EC50/[X])nHill)], where EC50 is the concentration of ACh that evokes 50% of the maximal response (Imax), I the current response at a given concentration of ACh and nHill is the Hill coefficient.
The sensitivity to Zn2+ was assessed by coapplying a range of Zn2+ concentrations with 1 μm ACh at (α4)2(β2)3 nAChRs [a concentration close to the ACh EC20 at (α4)2(β2)3 receptors] (see Fig. 1C), or 10 μm ACh at (α4)3(β2)2 receptors [the ACh EC10 at (α4)3(β2)2 receptors] (see Fig. 1D). For Zn2+to attain equilibrium around impaled oocytes, Zn2+ was preapplied for 30 s to the cell before coapplications of ACh and Zn2+. Concentration–response relationships for Zn2+ were obtained using this protocol and the peak responses elicited by ACh plus Zn2+ were normalized to the peak response of the appropriate ACh alone. Where a single component concentration–response relationship was evident, data were fitted to the Hill equation shown above. When Zn2+ produced both a potentiating and inhibiting effect, the data were fitted to the following equation designed to account for the potentiating and inhibitory effects of Zn2+ on α4β2 receptors assuming this cation binds to two distinct sites on the receptor: I = (1 + ((Imax − 1)/(1 + 10^((logEC50 − X) × nHill-pot))))/(1 + 10^((logIC50 − X) × nHill-inh)), where I represents the current responses at a given Zn2+ concentration (X), Imax, represents the maximally potentiated peak, EC50 and IC50 are the concentrations of Zn2+ inducing half-maximal potentiation or inhibition, respectively, and nHill-pot and nHill-inh are the Hill coefficients for potentiation and inhibition, respectively. F tests were always performed to assess the fitting of the data; the simpler one-component model was preferred unless the extra sum-of-squares F test had a value of p < 0.05.
To determine the effects of pH on Zn2+ modulation of the ACh responses of α4β2 nAChR, oocytes were superfused with Ringer solutions of pH 5.5 or 8.0. Zn2+ and ACh solutions applied onto oocytes were of the same pH as the superfusing Ringer solution. For experiments requiring the use of DEPC, oocytes, before recordings and one at a time, were incubated in Ringer solution containing 1 mm freshly prepared DEPC. Preliminary studies showed that higher concentrations of DEPC caused cell death; therefore, their effects on potentiating or inhibiting Zn2+were not examined further in this study. An incubation time of 10 min was sufficient for 1 mm DEPC to achieve maximal DEPC effects without decreasing cell survival or impairing cell functions. After DEPC incubation the oocyte was placed on a rotating platform and washed for an additional 10 min in DEPC-free Ringer solution. During washing, the solution was exchanged every 2–3 min to allow complete removal of DEPC.
Data analyses were performed using GraphPad Prism 5 software. Data were pooled from at least three different batches of oocytes. Statistical significance was assessed using a two-tailed unpaired t test, or one-way ANOVA followed by the Dunnett's post-test, as appropriate. p values <0.05 were considered significant.
Sequences of the human α4 and β2 nAChR subunits were obtained from the ExPASy proteomics server (Gasteiger et al., 2003) with accession numbers P43681 (α4) and P17787 (β2). The homopentameric Lymnaea stagnalis acetylcholine-binding protein (AChBP) structure (Protein Data Bank code 1UW6) (Celie et al., 2004) was used to generate models of the extracellular domain of the alternate stoichiometries of α4β2 nAChR, (α4)2(β2)3, and (α4)3(β2)2 (Nelson et al., 2003; Moroni et al., 2006). Although the sequence identity between AChBP and the subunits of the α4β2 nAChRs is only 18–20%, a similar fold and highly conserved binding-site residues (Brejc et al., 2001) make the AChBP structure suitable for modeling the extracellular domain of nAChRs. Additionally, and to model regions around the transmembrane domain, the 4 Å cryoelectron microscopy structure of the Torpedo marmorata nAChR (Protein Data Bank code 2BG9) (Unwin, 2005) was chosen as the template to generate models of the full receptor. In the Torpedo nAChR, agonist binding sites exist between α1-γ and α1-δ subunits. Hence, the Torpedo α1 subunit was used to model the human α4 subunit, γ and δ subunits were used to model human β2, and the remaining Torpedo β1 subunit was used to model the human α4/β2 interface. Multiple sequence alignment was performed using ClustalW (Higgins et al., 1994). One hundred models each of (α4)2(β2)3 and (α4)3(β2)2 receptors were generated using Modeler 8 (Sali and Blundell, 1993). In the case of the full receptor, the intracellular region, missing in the template, was not modeled. Procheck (Laskowski et al., 1993) was used to evaluate the stereochemical quality of the generated models. The best model for each receptor was energy-minimized in vacuo, using the molecular dynamics package Amber 9 (Case et al., 2006), with 500 cycles of the steepest descent method followed by another 500 cycles of the conjugate gradient method. The quality of the energy-minimized structures was checked using Procheck. Images were generated using PyMOL (DeLano, 2002).
Zn2+-binding site analysis.
We used information from previous studies (Hsiao et al., 2006) as well as performing a more general search to identify residues involved in Zn2+ binding. Because histidines are commonly associated with binding Zn2+ (Choi and Lipton, 1999; Low et al., 2000; Castagnetto et al., 2002; Dunne et al., 2002; Hosie et al., 2003; Miller et al., 2005a,b; Nagaya et al., 2005; Hsiao et al., 2006), we searched for histidine residues in our extracellular N-terminal domain models. In this domain, we found six histidines in the α4 subunit (H35, H94, H137, H142, H145, and H195) and four in the β2 subunit (H35, H71, H111, and H161). Next, we looked for aspartate and glutamate residues, which are known to form Zn2+-binding sites (Castagnetto et al., 2002), within 13 Å of a histidine residue to include regions that could be brought closer by protein flexibility. Cysteines were not considered because all cysteine residues in the extracellular N-terminal domain are disulfide bridged (Celie et al., 2004; Unwin, 2005) and, hence, not available for coordinating Zn2+. We then narrowed down the list by eliminating locations that were not accessible: intrasubunit or on inflexible regions of the receptor.
α4H195 was selected as a target because the equivalent histidine (α4H162) in the rat α4β4 receptor was identified previously as a site involved in Zn2+ potentiation (Hsiao et al., 2006) and also because this residue is located in loop F, which is a flexible loop (Szarecka et al., 2007) that may interact with the adjacent subunit in the non-ACh-binding interface. In the α4(+)/α4(−) interface of the (α4)3(β2)2 receptor, α4(−)E92 in loop D and α4(+)D217 and α4(+)D218 in loop C, which is a very flexible loop as evident from x-ray crystal structures (Hansen et al., 2005) and molecular dynamics studies (Henchman et al., 2003, 2005), were identified that could form part of the binding site. In the β2(+)/α4(−) interface of the (α4)2(β2)3 receptor, β2(+)E224 was selected using this approach.
β2H71 was chosen on the basis that it is within a region, near the transmembrane domain (from our model of the full-length receptor), with multiple residues (β2E74, β2D293, and β2E72) that could potentially bind Zn2+. β2E72 was excluded because it is buried and inaccessible.
A comparison of the sensitivity of the alternate stoichiometries of the α4β2 nAChR to Zn2+ modulation was initiated to examine the role of receptor stoichiometry on the effects of Zn2+ on α4β2 nAChRs. In a previous study, we showed that Xenopus oocytes injected with human α4 and β2 nAChR subunit cDNAs at a ratio of 1:10 or 10:1 produce homogeneous populations of (α4)2(β2)3 or (α4)3(β2)2 nAChRs, respectively (Moroni et al., 2006). (α4)2(β2)3 and (α4)3(β2)2 nAChRs differ significantly with respect to functional sensitivity to ACh (Tables 1, 2) and other α4β2-preferring nicotinic ligands, desensitization kinetics, sensitivity to long-term exposure to nicotine (Kuryatov et al., 2005; Moroni et al., 2006; Zwart et al., 2006), and Ca2+ permeability (Tapia et al., 2007). We determined the stoichiometries of the α4β2 nAChR using electrophysiological methods in combination with a reporter mutation within the transmembrane domain (TM) 2 of the receptor (Moroni et al., 2006). Human α4β2 nAChR expressed heterologously in human embryonic kidney tsA201 cells have also been shown to assemble as (α4)2(β2)3 and (α4)3(β2)2 receptors (Nelson et al., 2003).
The ACh current responses of (α4)2(β2)3 and (α4)3(β2)2 nAChR expressed heterologously in Xenopus oocytes were modulated differentially by Zn2+. Zn2+ inhibited the ACh responses of (α4)2(β2)3 with an IC50 value of 19 ± 6 μm (n = 10) (Fig. 1A,B). The Zn2+ concentration–response curve was best fitted with a single Hill equation (p = 0.001, F test; n = 10); however, the Hill slope of the curve was 0.45 ± 0.014, suggesting the likely presence of multiple inhibitory components in the Zn2+ concentration–response curve. The steepness of the curve did not change when the buffer tricine was used to remove potential low levels of Zn2+ contamination in the Ringer solution (data not shown). In contrast, (α4)3(β2)2 nAChR displayed a biphasic response to Zn2+ application (p = 0.003, F test; n = 10). Current responses activated by 10 μm ACh at (α4)3(β2)2 nAChRs were potentiated by 1–100 μm Zn2+, up to a maximum of 1.51 ± 0.5 at 100 μm Zn2+ relative to controls in the absence of Zn2+ (Fig. 1A,B). The EC50 value for potentiation was 50 ± 5 μm (n = 10). Higher concentrations of Zn2+ decreased the degree of potentiation until at concentrations >800 μm Zn2+ the amplitudes of the ACh responses elicited in the presence of Zn2+ were smaller than those mediated by applications of ACh alone (Fig. 1A,B). The estimated IC50 value for the inhibition of (α4)3(β2)2 nAChR by Zn2+ was 803 ± 43 μm (n = 10). Our findings contrast with those of Hsiao et al. (2001, 2006), who reported only Zn2+ biphasic effects on rat α4β2 nAChRs expressed heterologously in Xenopus oocytes. This is likely because of the use of different expression strategies. Hsiao et al. (2001, 2006) injected oocytes with equal amounts of α4 and β2 cRNAs, which is known to produce a mixture of diverse functional types of the α4β2 nAChR, whose predominant component (80–90%) is the low sensitivity form [i.e., (α4)3(β2)2] of the α4β2 nAChR (Zwart and Vijverberg, 1998; Houlihan et al., 2001; Exley et al., 2006; Zwart et al., 2006).
The mechanism of action of Zn2+ was subsequently investigated by determining both the effects of Zn2+ on the ACh concentration–response curve of the (α4)2(β2)3 and (α4)3(β2)2 nAChRs and the voltage dependency of the Zn2+ effects. We chose to investigate Zn2+ potentiation on (α4)3(β2)2 nAChR and Zn2+ inhibition on the (α4)2(β2)3 nAChR because such distinct effects are likely to be mediated by different Zn2+-binding sites. At (α4)2(β2)3 nAChR, 10 μm Zn2+, a concentration slightly lower than the IC50 value for Zn2+ inhibition, significantly decreased the amplitude of the current responses elicited by saturating ACh concentrations (p < 0.05; n = 6), but had no significant effect on the apparent functional ACh sensitivity of the receptors (EC50 ACh = 4.04 ± 0.3 μm; EC50 ACh plus Zn2+ = 3 ± 0.4 μm; n = 6) (Fig. 1C). These results indicate that Zn2+ inhibits the (α4)2(β2)3 nAChR in a noncompetitive manner, which is in keeping with previous findings that residues coordinating Zn2+ do not contribute to the agonist binding site in ligand-gated ion channels structurally related to nAChRs (Low et al., 2000; Hosie et al., 2003; Miller et al., 2005; Hsiao et al., 2006). The inhibition of the ACh responses of the (α4)2(β2)3 nAChR by 10 μm Zn2+ was examined at several holding potentials ranging from −120 to −30 mV. The degree of inhibition of the ACh responses of the (α4)2(β2)3 nAChR by Zn2+ decreased with hyperpolarization (Fig. 1E). Voltage-dependent effects of Zn2+ have been reported previously for α2β4 nAChRs expressed heterologously in Xenopus oocytes (García-Colunga et al., 2001) and for native nAChRs in rat parasympathetic ganglia (Nutter and Adams, 1995).
At the (α4)3(β2)2 nAChR, we examined the effects of the maximally potentiating concentration of Zn2+ (100 μm) to minimize possible counteracting effects by inhibitory Zn2+, which would reduce the degree of Zn2+ potentiation. Zn2+ decreased significantly the EC50 value for ACh activation from 89 ± 5 μm to 34 ± 5 μm (n = 8; p < 0.05) (Fig. 1D). However, although Zn2+ increased the amplitude of responses to nonmaximal ACh concentrations, it did not increase the amplitude of the response to saturating ACh concentrations. This, as suggested by the concentration–response curve for Zn2+ effects on (α4)3(β2)2 nAChRs (Fig. 1B), could reflect masking of Zn2+ potentiation by counteracting Zn2+ inhibition. The degree of potentiation was similar at all holding potentials tested (Fig. 1E). The degree of inhibition of the (α4)3(β2)2 nAChR by 2 mm Zn2+, a concentration that decreased the ACh responses of (α4)3(β2)2 nAChRs by almost 70% the amplitude of the ACh responses of the (α4)3(β2)2 nAChR, was also independent of the holding potential from −120 to −30 mV (Fig. 1E).
Histidine residues are involved in the effects of Zn2+
Histidine is the most common residue that participates in Zn2+ binding by metalloenzymes (McCall et al., 2000) and by all ligand-gated ion channels examined so far (Choi and Lipton, 1999; Low et al., 2000; Dunne et al., 2002; Hosie et al., 2003; Miller et al., 2005a,b; Nagaya et al., 2005; Hsiao et al., 2006). If histidine residues participate in similar Zn2+ site(s) on the alternate forms of the human α4β2 nAChR, then these site(s) should be impaired by DEPC, which reacts with neutral imidazole groups converting the residues to N-carbethoxyhistidyl derivatives. Exposure to 1 mm DEPC significantly reduced the extent of Zn2+ inhibition on the current responses activated by ACh at (α4)2(β2)3 nAChRs (IACh+Zn2+ = 0.57 ± 0.009; after DEPC treatment, IACh+Zn2+ = 0.77 ± 0.012, relative to control IACh; p < 0.05; n = 5). In contrast, DEPC treatment abolished both the voltage dependency of Zn2+ inhibition on (α4)2(β2)3 nAChRs and Zn2+ potentiation on (α4)3(β2)2 receptors (IACh+Zn2+ = 1.51 ± 0.05; after DEPC treatment, IACh+Zn2+ = 0.99 ± 0.3, relative to control IACh; p < 0.05; n = 5) (Fig. 2). Because the ability of Zn2+ to react with Zn2+-coordinating amino acids is influenced by pH (Miles, 1977), one possible explanation for the failure of DEPC to eliminate completely voltage-independent Zn2+ inhibition is that at pH 7.2 DEPC does not react effectively with all the residues implicated in Zn2+ inhibition and that partial disruption does not completely abolish the functionality of the site(s). Nevertheless, because DEPC effectively reduces the ability of Zn2+ to bind to the imidazole ring of histidines (Dunne et al., 2002; Hosie et al., 2003; Miller et al., 2005b; Nagaya et al., 2005), these findings implicated histidines in the coordination of inhibitory and potentiating Zn2+ (Fig. 2).
An additional test for the involvement of histidine residues on the effects of Zn2+ on α4β2 receptors is the assessment of the effect of H+ ions on the responses to ACh in the absence and presence of Zn2+. This test would ascertain whether histidine residues contribute to Zn2+ effects because H+ will compete with Zn2+ for binding to the imidazole group on histidine residues. At pH 5.5, the extent of inhibition of the ACh-evoked currents on the (α4)2(β2)3 nAChR by 10 μm Zn2+ was decreased from 0.57 ± 0.03 to 0.84 ± 0.01, whereas at pH 8.0 inhibition increased to 0.49 ± 0.04, compared with inhibition at pH 7.2 (n = 4). At pH 5.5, the degree of potentiation of the (α4)3(β2)2 nAChR by 100 μm Zn2+ was reduced from 1.48 ± 0.5 (n = 6) to 1.1 ± 0.1 (n = 4), whereas at pH 8.0 potentiation increased to 1.65 ± 0.09 (n = 4). Because histidine is the only amino acid with a pKa that would be affected by these pH changes (pKa 6.1), it clearly contributes to the effects of Zn2+ on α4β2 nAChRs.
Structural determinants for Zn2+ effects on (α4)2(β2)3 and (α4)3(β2)2 nAChRs
Previous studies have suggested that α4H162 within the F loop might be a direct contributor to Zn2+ binding forming part of a potentiating site on rat α4β4 or α4β2 nAChRs (Hsiao et al., 2006). We examined this proposal on human α4β2 nAChRs by mutating α4H195, the human equivalent of rat α4H162, to alanine. (Note that amino acid numbering is from the initiating methionine of the unprocessed subunit. To obtain the position in the mature form, subtract 30 from the number for α4 and 25 for β2.) However, because Zn2+-binding sites require at least three amino acid residues and an activated water molecule for complete tetrahedral coordination (Auld, 2001), we searched for further residues capable of coordinating Zn2+ and that were in close structural proximity to α4H195. In our model of the extracellular N-terminal domain of both forms of the human α4β2 nAChR, α4H195 could interact with β2D218 in loop C to form a Zn2+-binding site at the non-ACh-binding β2(+)/α4(−) interfaces on both α4β2 nAChR forms (Fig. 3A). An additional residue that could contribute to the proposed Zn2+-binding sites is α4E92, a residue within loop D (Fig. 3A).
Introducing either α4H195A or β2D218A in the (α4)2(β2)3 nAChR decreased sensitivity to Zn2+inhibition significantly by approximately eightfold (p < 0.05; n = 6) (Fig. 3B, Table 1). By comparison, α4E92A reduced the Zn2+ IC50 value by only threefold (Table 1). Pairing α4H195A with β2D218A further reduced the potency of Zn2+ inhibition to 1144 ± 54 μm (p < 0.05; n = 5) (Fig. 3B, Table 1). However, pairing α4H195A with α4E92A did not increase the Zn2+ IC50 value beyond that observed with the single mutations of α4H195 or β2D218 (Table 1). By comparison, pairing β2D218A with α4E92A significantly increased the Zn2+ IC50 value relative to that of the single mutants (p < 0.05; n = 5) (Table 1). Introducing simultaneously α4H195A, α4E92A, and β2D218A did not increase the Zn2+ IC50 value beyond that observed with the pairing of α4H195A and β2D218A (Table 1). For all mutants, individually or in combinations, the functional sensitivity to ACh was indistinguishable from that of the wild-type (α4)2(β2)3 nAChR, suggesting that these mutations selectively affected Zn2+ inhibition and did not induce a general perturbation on (α4)2(β2)3 function (Table 1). These studies suggested that α4H195, β2D218, and α4E92 might contribute to Zn2+ binding, forming part of an inhibitory site predicted to reside at the β2(+)/α4(−) interfaces on both α4β2 nAChR forms.
Although α4H195A, α4E92A, or β2D218A decreased the magnitude of Zn2+ inhibition, they did not, singularly or in combinations, abolished it (Table 1), indicating the likely presence of an additional Zn2+ site or, conceivably, Zn2+ binding was not completely disrupted around the environment of α4H195, α4E92, and β2D218. Further examination of our model of the extracellular N-terminal domain of (α4)2(β2)3 and (α4)3(β2)2 receptors revealed that β2D217 is a residue that could coordinate Zn2+ as part of a binding site formed by α4H195, β2D218, and α4E92at β2(+)/α4(−) interfaces (Fig. 3A). However, as shown in Table 1, introducing β2D217A decreased sensitivity to Zn2+ inhibition only when paired with α4E92A or β2D218A, which suggests an indirect role on Zn2+ inhibition for this residue. Although these findings did not answer the question of whether alanine substitutions of α4H195, β2D218, or α4E92 fully disrupted the inhibiting Zn2+ locus predicted to reside at β2(+)/α4(−) interfaces, the observation that none of these mutations, singularly or combined, affected the voltage dependency of Zn2+ inhibition suggested additional inhibitory Zn2+ sites.
We have shown that α4H195, α4E92, and β2D218 residues are implicated in non-voltage-dependent Zn2+ inhibition of (α4)2(β2)3 receptors. Given that these residues are predicted to reside on the non-ACh-binding β2(+)/α4(−) interfaces (Fig. 3A), which are common to both α4β2 stoichiometries, we expected these residues to contribute to Zn2+ inhibition on the (α4)3(β2)2 stoichiometry as well. This implied that replacement of any of these residues with alanine could result in an increase in sensitivity to Zn2+ potentiation and maximal potentiation, if, as suggested by the concentration–response curve for Zn2+ inhibition on (α4)2(β2)3, Zn2+ potentiation and inhibition had overlapping concentration ranges. In accord with our prediction, β2D218A increased maximal Zn2+ potentiation on (α4)3(β2)2 and significantly increased the sensitivity of (α4)3(β2)2 to Zn2+ potentiation from 50 to 31 μm (p < 0.05; n = 6) (Fig. 3C, Table 2). Moreover, β2D218A caused potentiation of maximal ACh currents by Zn2+ (1.34 ± 0.09) (see Fig. 6B), which did not occur on wild-type (α4)3(β2)2 (Fig. 1D). However, unexpectedly, α4H195A abolished Zn2+ potentiation on its own and increased sensitivity to Zn2+ inhibition, and when paired with β2D218A, it reduced maximal potentiation from 1.51 to 1.18 (Fig. 3C, Table 2). α4E92A reduced potentiation, albeit not significantly, and did not increase or reduce the individual or combined effects of α4H195A or β2D218A (Table 2), suggesting that it might not form part of the Zn2+ potentiating site. As for Zn2+ inhibition on (α4)2(β2)3 nAChRs, β2D217A did not impact on the potentiating effects of Zn2+ on (α4)3(β2)2 nAChRs (Table 2).
The α4(+)/α4(−) subunit interface bears a potentiating Zn2+ site in (α4)3(β2)2 nAChRs
The discovery that Zn2+ potentiation on (α4)3(β2)2 nAChRs is abolished by α4H195A but enhanced by β2D218A suggested that the signal transduction pathway associated with the Zn2+ inhibiting site located at the β2(+)/α4(−) subunit interfaces of the α4β2 nAChR may differ in the (α4)2(β2)3 and (α4)3(β2)2 receptors. An alternative explanation, given the structural homology of the subunits of the nAChR family, is that α4H195 contributes to two independent Zn2+-binding sites on the (α4)3(β2)2 nAChR. One site is inhibitory and resides on the β2(+)/α4(−) subunit interfaces of both receptors; α4H195, β2D218, and α4E92 contribute to this site that is represented twice in both receptor stoichiometries (Fig. 4A). The other site is potentiating and resides on the α4(+)/α4(−) interface of the (α4)3(β2)2 nAChR (Fig. 4A). Conceivably, H195 on the (−) side of the ACh-binding α4 subunit could contribute to the proposed Zn2+ potentiating site together with residues homologous to β2D218 on the (+) side of the non-ACh-binding α4 subunit. Because D218 on the β2 subunit is within loop C, this hypothesis was examined by introducing α4(+)E224A or α4(−)E228A in (α4)3(β2)2 (Fig. 4B). Because of their location within loop C, α4(+)E224 or α4(+)E228A could contribute to a Zn2+-binding site together with α4H195on the (−) side of the adjacent ACh-binding α4 subunit. Other residues in loop C such as C225 and C226 were not considered as likely contributors to a Zn2+-binding site because these residues are disulfide bridged (Celie et al., 2004; Unwin, 2005) and, therefore, unable to coordinate Zn2+. Figure 4C shows that introducing mutant α4E224A abolished Zn2+ potentiation. Furthermore, pairing α4E224A with β2D218A (Table 2) or with α4H195A and β2D218A (Fig. 4C) also eliminated Zn2+ potentiation. In contrast, introducing α4E228A did not affect Zn2+ potentiation (Table 2), supporting a role for α4E224, but not for α4E228, in potentiating Zn2+. Abolition of Zn2+ potentiation by α4E224A could reflect a perturbation of the general properties of the receptor or involvement in downstream signal transduction pathways. As shown in Table 2, α4E224A, individually or in combination with other residues, did not change the sensitivity to activation by ACh, indicating that the effect of α4E224A was specific on Zn2+ potentiation.
To test whether α4E224 might be a direct contributor to Zn2+ binding, we substituted E224 with histidine and then examined whether DEPC treatment abolished Zn2+ potentiation on receptors containing this mutation. The more common strategy of cysteine substitution in combination with exposure to the cysteine-modifying reagent methyl thiosulfonate ethylammonium (MTSEA) was not considered appropriate for these studies because of the presence of neighboring C225 and C226. For these studies, we also mutated α4H195 to glutamate to eliminate possible ambiguities that could have arisen from modification of α4H195 by DEPC. Although introducing α4H195E abolished potentiation (Table 2), pairing of α4E224H with α4H195E produced levels of Zn2+ potentiation comparable with those obtained on wild receptor (Fig. 4D, Table 2), indicating that the mutations introduced maintained the functionality of the Zn2+ potentiating site. α4E224H on its own did not eliminate Zn2+ potentiation, although it reduced its potency (Table 2). Exposure to 1 mm DEPC completely abolished Zn2+ potentiation on α4H195E,E224Hβ2 nAChRs (Fig. 4D). Furthermore, introducing α4E224H and α4H195E into (α4)3(β2)2 nAChRs did not influence the sensitivity of (α4)3(β2)2 nAChRs for activation by ACh (Table 2), or indeed that of (α4)2(β2)3 nAChRs (Table 1). These data therefore are consistent with our view that α4E224 and α4H195 directly contribute to Zn2+ binding, forming part of a potentiating site at the α4(+)/α4(−) subunit interface on the (α4)3(β2)2 nAChR. Of importance, Zn2+ inhibition on (α4)2(β2)3 nAChRs was not influenced by α4E224H or α4E224A, further indicating that this residue specifically influences Zn2+ effects on (α4)3(β2)2 nAChRs (Table 1).
Interestingly, the presence of a Zn2+ potentiating site on the α4(+)/α4(−) interface reduced the apparent sensitivity of (α4)3(β2)2 nAChRs to Zn2+ inhibition by 42-fold compared with the sensitivity of the (α4)2(β2)3 stoichiometry to Zn2+ inhibition (Table 2). This may reflect allosteric interactions between the Zn2+ potentiating and Zn2+ sites present on (α4)3(β2)2 nAChRs; however, we cannot discount the possibility of subtly different properties between the Zn2+ inhibiting sites on both stoichiometries. At least one of these is flanked by different subunits in both receptor stoichiometries (Fig. 4A).
Voltage-dependent inhibition of (α4)2(β2)3 by Zn2+
The discovery that the voltage dependency of Zn2+ inhibition was abolished by DEPC treatment but not by substitution of α4H195, α4E92, or β2D218 by alanine suggested an additional inhibitory Zn2+ site. As the suggested site should be close to the ion channel to exert voltage-dependent effects, we searched our model of the (α4)2(β2)3 receptor for histidine residues close to the ion channel and that could be accessible to external Zn2+ ions. According to our model, β2H71 and β2E74 both within the β1–β2 loop and β2D293 within the TM2–TM3 extracellular linker could form a Zn2+-binding site located near the ion channel region (Fig. 5A).
Introducing β2H71A, β2E74A, or β2D293A individually or in combinations abolished the ability of Zn2+ to inhibit the ACh responses of (α4)2(β2)3 nAChRs in a voltage-dependent manner without affecting the functional sensitivity of (α4)2(β2)3 nAChRs to activation by ACh (Fig. 5B, Table 1). β2H71A, β2E74A, or β2D293A, individually or in combinations, had no effects on the estimated IC50 value for Zn2+ inhibition on wild-type (α4)2(β2)3 nAChRs, but significantly increased the nHill coefficient (Fig. 5C, Table 1). Furthermore, when β2H71A, β2E74Aand β2D293Awere introduced with α4H195A or α4H195A and β2D218A, the nHill coefficient was further increased, whereas the sensitivity to Zn2+ inhibition, compared with that obtained with α4H195A alone or in combination with β2D218A, remained unchanged (Table 1).
Introduction of β2H71A, β2E74A, or β2D293A had little effect on the ACh sensitivity of (α4)3(β2)2 nAChRs (Table 2), but doubled maximal potentiation by Zn2+ compared with wild-type (α4)3(β2)2 nAChRs (Fig. 6A, Table 2). Furthermore, the stepwise introduction of these mutants into (α4)3(β2)2 nAChRs progressively and significantly increased the degree of Zn2+ potentiation to levels greater than those produced by the individual mutations (Fig. 6A, Table 2). These results suggested that the site within the β2 subunit producing voltage-dependent Zn2+ inhibition on the (α4)2(β2)3 receptor is also present on (α4)3(β2)2 nAChRs. Further evidence for the involvement of β2H71A, β2E74A, and β2D293A in the inhibitory effects of Zn2+ on (α4)3(β2)2 nAChRs was obtained by determining the effects of these mutations on the concentration–response curve for ACh at this receptor type. As shown in Figure 6B, Zn2+ potentiated the maximal ACh responses of (α4)3(β2H71A,E74A,D293A)2 receptors. Furthermore, when β2D218A was introduced against the background of β2H71A,E74A,D293A, potentiation of maximal ACh responses increased additively from 1.17 ± 0.2 (α4)3(β2H71A,E74A,D293A)2 and 1.34 ± 0.03 at (α4)3(β2D218A)2 to 1.52 ± 0.02 at (α4)3(β2H71A,E74A,D293A,D218A)2 (Fig. 6B).
The most important conclusions from this study are (1) that the effects of Zn2+ on α4β2 nAChRs are heavily influenced by the receptor stoichiometry, (2) that Zn2+ potentiation on the (α4)3(β2)2 stoichiometry is determined by a site located at the α4(+)/α4(−) subunit interface, and (3) that the β2(+)/α4(−) interfaces at both receptor types house a Zn2+ inhibitory site. These findings have important structural and functional implications about the role of subunit composition on the structural-functional properties of the α4β2 nAChR and the manner by which Zn2+ influences α4β2 nAChR function. Indeed, this study shows for the first time that non-ACh-binding interfaces confer distinct functional signatures to the alternate stoichiometries of the α4β2 nAChR.
Effects of Zn2+ on the α4β2 nAChR are determined by subunit composition
Zn2+ exerts an inhibitory effect on (α4)2(β2)3 receptors, whereas it potentiates or inhibits, depending on its concentration, the function of (α4)3(β2)2 receptors. The finding that subunit composition influences the effects of Zn2+ on heteromeric receptors is not limited to the α4β2 nAChR. Glycine α1 and α2 receptors, which are homologous to the α4β2 nAChR, differ markedly in sensitivity to both Zn2+ inhibition (Miller et al., 2005a) and Zn2+ potentiation (Miller et al., 2005b), the differences reflecting divergence in key residues (Miller et al., 2005a,b). Subunits adjacent to subunits harboring Zn2+ sites may also alter the sensitivity of ligand-gated ion channels to Zn2+ effects. For example, incorporation of the γ2 subunit to αβ GABAA receptors markedly decreases sensitivity to Zn2+ inhibition by disrupting the high-affinity Zn2+-binding sites that reside at the α–β subunit interface (Hosie et al., 2003). Regarding the α4β2 nAChR, subunit composition not only affects the sensitivity to Zn2+ modulation, but also the manner by which Zn2+ modulates the function of the receptors.
Residues on β2(+)/α4(−) interfaces contribute to Zn2+ inhibition on (α4)2(β2)3 and (α4)3(β2)2 receptors
From our model of the α4β2 nAChRs and the findings of our alanine substitution studies, we conclude that α4H195 and β2D218 critically contribute to an inhibiting Zn2+-binding site predicted to reside at the β2(+)/α4(−) interfaces on both α4β2 nAChR forms. Despite the lack of structural information on the α4β2 nAChR, the findings are compatible with α4H195 and β2D218 binding inhibiting Zn2+. First, the identified residues are all capable of coordinating Zn2+ and when they were replaced with alanine, the sensitivity to Zn2+ decreased markedly. Second, mutation of either residue specifically disrupted Zn2+ inhibition without significant changes in the sensitivity of the receptor to activation by ACh. Third, β2D218 is within loop C, which, as we highlight above, can undergo large conformational changes. Such motions would allow β2D218 to come into close enough proximity to α4H195 to form part of a discrete Zn2+-binding site with other suitably located Zn2+ coordinating residues. Although α4E92 is close to α4H195 and β2D218, the effects of alanine substitutions at this position have considerably less impact than what is observed with α4H195 or β2D218, suggesting that α4E92 might not contribute directly to Zn2+ binding. A similar view is held by Hsiao et al. (2006), who showed that mutagenesis or MTSEA treatment are less effective on the rat equivalent of α4E92 than on α4H162 (the rat equivalent of α4H195).
The α4(+)/α4(−) subunit interface houses a Zn2+ potentiating site
Here, we provide evidence that is consistent with α4H195, a residue located on the (−) side of ACh-binding α4 subunit, playing a determinant role on Zn2+ potentiation on (α4)3(β2)2 nAChRs. This finding is in agreement with those of Hsiao et al. (2006), who suggested α4H162, the rat equivalent of human α4H195, as a key determinant of Zn2+ potentiation on α4β4/β2 nAChRs. However, our study also demonstrated that α4H195 is critically involved in Zn2+ inhibition on both α4β2 nAChR stoichiometries. Given these conflicting data, we investigated whether the α4(−)/α4(+) subunit interface of the (α4)3(β2)2 nAChRs could harbor a Zn2+ potentiating site partly formed by α4H195 on the (−) face of an ACh-binding α4 subunit and α4E224, on the (+) side of the nonbinding α4 subunit. Our findings are in accord with this possibility. When these residues were substituted by alanine, the sensitivity to Zn2+ potentiation was abolished, without perturbations of the general functional properties of the receptor. Furthermore, alanine substitution of α4E224 did not affect Zn2+ modulation on (α4)2(β2)3 or the overall function of this receptor type, confirming that α4E224 specifically confers a Zn2+ potentiation site on the (α4)3(β2)2 stoichiometry. The fact that the Zn2+ inhibiting and Zn2+ potentiating sites are located at homologous positions arises because of subunit homology, and such a phenomenon has been observed previously in other pseudosymmetric proteins. For example, it has been shown that some of the residues contributing to the benzodiazepine site of GABAA receptors are homologous to residues implicated in agonist binding (Sigel and Buhr, 1997) and recent work on the ACh-binding protein suggests that aromatic residues found in the binding pocket are conserved at non-α interfaces, where they contribute to noncompetitive galanthamine-binding sites (Hansen and Taylor, 2007).
Do α4H195 and α4E224 contribute to a Zn2+-binding site?
Our findings make a strong case for a role in binding. Both α4H195 and α4E224 are chemically able to coordinate Zn2+ and when substituted by alanine or covalently modified by DEPC, only the sensitivity to Zn2+ potentiation was altered. Furthermore, α4H195 is sensitive to MTSEA (Hsiao et al., 2006), which indicates that it is accessible to Zn2+. In the α4(−)/α4(+) interface, and in terms of a potential Zn2+-binding site, α4E224 is situated on what would normally be considered the principal (+) interface. The distance between these residues in the model is somewhat larger (∼12 Å) than would be expected for a normal Zn2+-binding site (typical values between coordinating groups is of the order of 6–7 Å) (Auld, 2001). Thus, to form a competent binding site, it seems likely that some movement at the interface is required. There is a precedent for this in that loop C has been shown (in AChBP at least) to adopt a variety of different conformations (Hansen et al., 2005). Furthermore, molecular dynamics simulations (Henchman et al., 2003, 2005) have suggested that this loop is highly flexible. Thus, the formation of a Zn2+-binding site by these residues is a distinct possibility. In addition, analysis of the metalloprotein database (Castagnetto et al., 2002) suggests that water molecules are often involved in the coordination of Zn2+ ions. Water molecules are not explicitly considered in our modeling here, but could help to stabilize potential Zn2+ coordination. However, we cannot at this stage rule out alternative modes of action.
Voltage-dependent Zn2+ inhibition is mediated by β2 subunit residues
The relative positioning of the β2H71, β2E74, and β2D293 in the predicted quaternary structure of α4β2 nAChR is in accord with these residues contributing to a site that may exert voltage-dependent Zn2+ inhibition on (α4)2(β2)3 nAChRs. Clearly, this presumed site is independent of the inhibiting site located at β2(+)/α4(−) interfaces; β2H71 and β2E74 are in a loop tilting toward the channel, whereas β2D293 is on the pore lining the M2 helix (Fig. 6A). Although mutating β2H71, β2E74, or β2D293 eliminated voltage-dependent inhibition on (α4)2(β2)3 nAChRs, it did not significantly alter the IC50 value for Zn2+ inhibition. Moreover, although alanine substitutions increased maximal Zn2+ potentiation on (α4)3(β2)2 nAChRs, the estimated EC50 value for Zn2+ potentiation was not significantly altered. These findings suggest that β2H71, β2E74, and β2D293 may coordinate Zn2+ as part of a site that may influence the downstream pathways associated with inhibiting and/or potentiating Zn2+.
It should be noted that although β2H71A, β2E74A, or β2D293A increased the degree of Zn2+ potentiation on (α4)3(β2)2 receptors, no evidence was found of voltage-dependent Zn2+ effects on (α4)3(β2)2 receptors lacking the Zn2+ potentiating site (Table 2). However, because the functional properties of the alternate forms of the α4β2 nAChR differ substantially, a Zn2+-binding site that modulates Zn2+ signal pathways would be expected to influence receptor function in a stoichiometry-specific manner.
Overall, our findings show that the effects of Zn2+ on α4β2 nAChRs are heavily influenced by subunit stoichiometry. Among the nAChR assembled from two different subunits, this is the first time that it has been shown that the non-ACh-binding subunit drastically changes the responses to modulators and that subunit homology and stoichiometry might result in one residue contributing to two independent modulatory sites. Although more work, including structural data, will be needed to ultimately understand the effects of Zn2+ on α4β2 receptors and how modulatory sites may be formed, our studies have highlighted a crucial role played by non-ACh-binding interfaces on receptor modulation.
This work was supported by the Oxford Brookes University (M.M., A.C.), the University of Oxford Clarendon Fund (R.V.), the United Kingdom Overseas Research Students Awards Scheme (R.V.), and the Wellcome Trust (P.C.B.). P.C.B. is a Research Councils UK Fellow. We thank J. Connolly and J. Dempster from Strathclyde University, Glasgow, UK and H. Vijverberg from Utrecht University, Utrecht, The Netherlands, for helpful advice on data analysis.
- Correspondence should be addressed to Isabel Bermudez, Gipsy Lane, Oxford OX3 0BP, UK.