Additional Acetylcholine (ACh) Binding Site at α4/α4 Interface of (α4β2)2α4 Nicotinic Receptor Influences Agonist Sensitivity*

Nicotinic acetylcholine receptor (nAChR) α4 and β2 subunits assemble in two alternate stoichiometries to produce (α4β2)2α4 and (α4β2)2β2, which display different agonist sensitivities. Functionally relevant agonist binding sites are thought to be located at α4(+)/β2(−) subunit interfaces, but because these interfaces are present in both receptor isoforms, it is unlikely that they account for differences in agonist sensitivities. In contrast, incorporation of either α4 or β2 as auxiliary subunits produces isoform-specific α4(+)/α4(−) or β2(+)/β2(−) interfaces. Using fully concatenated (α4β2)2α4 nAChRs in conjunction with structural modeling, chimeric receptors, and functional mutagenesis, we have identified an additional site at the α4(+)/α4(−) interface that accounts for isoform-specific agonist sensitivity of the (α4β2)2α4 nAChR. The additional site resides in a region that also contains a potentiating Zn2+ site but is engaged by agonists to contribute to receptor activation. By engineering α4 subunits to provide a free cysteine in loop C at the α4(+)α4(−) interface, we demonstrated that the acetylcholine responses of the mutated receptors are attenuated or enhanced, respectively, following treatment with the sulfhydryl reagent [2-(trimethylammonium)ethyl]methanethiosulfonate or aminoethyl methanethiosulfonate. The findings suggest that agonist occupation of the site at the α4(+)/(α4(−) interface leads to channel gating through a coupling mechanism involving loop C. Overall, we propose that the additional agonist site at the α4(+)/α4(−) interface, when occupied by agonist, contributes to receptor activation and that this additional contribution underlies the agonist sensitivity signature of (α4β2)2α4 nAChRs.

The ␣4␤2 nAChR 2 is the predominant nAChR subtype in the brain where it constitutes one of the most important modula-tory receptor systems influencing activities such as cognition, mood, consciousness, and nociception (1). The ␣4␤2 nAChR is essential for nicotine self-administration (2) and has been implicated in autosomal dominant nocturnal frontal lobe epilepsy, depression, and age-related neurodegenerative diseases (1).
In this study, we focused our attention on the ␣4(ϩ)/␣4(Ϫ) interface of the (␣4␤2) 2 ␣4 nAChR and its possible role in the agonist sensitivity signature of this ␣4␤2 nAChR isoform. Using homology modeling, we first identified a putative agonist binding site at the ␣4(ϩ)/␣4(Ϫ) interface that is contributed by conserved agonist-binding aromatic residues. Obliteration of the site by exchanging the auxiliary ␣4 subunit for a chimeric subunit consisting of the N-terminal domain of the ␤2 subunit and the remaining part of the ␣4 subunit produced receptors with an agonist sensitivity comparable with that of the (␣4␤2) 2 ␤2 isoform. The putative site is capable of binding agonist as suggested by functional assays of mutant receptors engineered by alanine substitution of conserved aromatic residues contributing to the consensus agonist sites located at ␣4(ϩ)/ ␤2(Ϫ) subunit interfaces. Subsequently, using mutated receptors with a free cysteine residue in loop C together with covalent modifications with methanethiosulfonate (MTS) reagents, we demonstrated that the additional site contributes directly to channel gating. We propose that the additional site at the ␣4(ϩ)/␣4(Ϫ) defines the agonist sensitivity of the (␣4␤2) 2 ␣4 nAChR isoform.

EXPERIMENTAL PROCEDURES
Mutagenesis and Expression in Oocytes-Mutant ␣4 subunits were created as described previously (7,8). The fully concatenated form of the (␣4␤2) 2 ␣4 isoform, construct ␤2_␣4_␤2_␣4_␣4, was engineered as described previously (9). Briefly, the signal peptide and start codon were removed from all the subunits but the first (a ␤2 subunit), and the subunits were bridged by Ala-Gly-Ser linkers. Only the last subunit in the construct (an ␣4 subunit) contained a stop codon. The subunits were subcloned into a modified pCI plasmid vector (Promega) using unique restriction enzyme sites flanking the N and C termini of each subunit. Constructs were assayed for integrity by determining the ACh sensitivity of constructs co-expressed with an excess of ␤2 or ␣4 monomers carrying the LT reporter mutation (L9ЈT in the second transmembrane domain). No changes were observed in comparison with constructs expressed alone. This indicates that the constructs did not degrade into lower order concatamers or monomers as such degradation products would incorporate the ␤2LT or ␣4LT monomers into receptors of higher sensitivity to ACh than the intact constructs (13). Henceforth, concatenated receptors will be referred to as ␤2_␣4_␤2_␣4_␣4, whereas (␣4␤2) 2 ␣4 or (␣4␤2) 2 ␤2 will be used to denote receptors assembled from loose ␣4 and ␤2 subunits when ␣4 subunits are in excess or shortage, respectively, over ␤2 subunits. For clarity, mutations in the concatenated receptors are shown as superscript posi-tioned in the (ϩ) or (Ϫ) face of the mutated subunit (e.g. in ␤2_ W182A ␣4_␤2_␣4_␣4 the mutation W182A is located in the (ϩ) face of the ␣4 subunit occupying the second position of the linear sequence of the concatamer; see supplemental Fig. 1 for experimental evidence showing the spatial orientation of the subunits in the concatamer). A subunit made of the N terminus of the ␤2 subunit and the remaining part of the ␣4 subunit was ligated to ␤2_␣4_␤2_␣4 to construct the chimera ␤2_␣4_␤2_␣4_␤2/␣4. Chimeric ␤2/␣4 subunit was synthesized by GeneArt (Regensburg, Germany) and comprised the N terminus of ␤2 (1 to Arg-231) and the remaining part of the ␣4 subunit (Arg-241 to Ile-628). The residue numbering we use below includes the signal peptide sequence. To obtain the position in the mature form, subtract 28 for ␣4 and 26 for ␤2. Nonlinked or concatenated ␣4␤2 nAChRs were expressed in Xenopus oocytes as described previously (8,9).
Oocyte Electrophysiology-Oocyte isolation and two-electrode voltage clamp recordings on oocytes were carried out as described previously (7)(8)(9). Concentration-response curves (CRCs) for agonists were obtained by normalizing agonist-induced responses to the control responses induced by a nearmaximum effective agonist concentration as described previously (7)(8)(9)(10). A minimum interval of 5 min was allowed between agonist applications to ensure reproducible recordings. The agonist CRC data were first fitted to the one-component Hill equation, where EC 50 represents the concentration of agonist inducing 50% of the maximal response (I max ), x is the agonist concentration, and n H is the Hill coefficient. When agonists induced biphasic receptor activation, the CRC data were fitted with a two-component Hill equation, where Top and Bottom are the plateaus at the right and left ends of the curve in the same units as I; logEC 50_ 1 and logEC 50_ 2 are the concentrations that give half-maximal high sensitivity or low sensitivity stimulatory effects, respectively; n H1 and n H2 are the Hill coefficients; Frac is the proportion of maximal response due to the higher sensitivity component; and Span is a fitted coefficient between 0 and 1 that gives the weight of the first component. To determine the effects of the ␣4␤2selective competitive antagonist dihydro-␤-erythroidine (Dh␤E) on the ACh responses of wild type (WT) or mutant receptors, the antagonist was included in the ACh and perfusing (Ringer's) solution. The responses to ACh obtained in the presence of the antagonists were normalized to control ACh responses (responses to ACh evoked in the absence of the antagonist).
Modification of Substituted Cysteines by MTS Reagents-MTS reagents aminoethyl methanethiosulfonate (MTSEA) and [2-(trimethylammonium)ethyl]methanethiosulfonate (MTSET) (Toronto Research Chemicals) were used to modify covalently a free Cys residue in loop C. The free Cys residue was created by dismantling the disulfide bond between vicinal Cys-225 and Cys-226 at the tip of loop C through serine substitution of Cys-226. The effect of the MTS reagents was assessed as follows. Oocytes expressing receptors with a free cysteine in loop C or WT receptors were first challenged with a control ACh concentration every 5 min until a stable response was obtained. Oocytes were then perfused with Ringer's solution containing MTSEA (2.5 mM) or MTSET (1 mM) for 180 s after which time the impaled cells were washed with Ringer's solution for 60 s. After washing, ACh was applied every 5 min until the amplitude of the responses was constant. The average of the current amplitudes prior to application of MTS reagents was the control response current (I initial ), and the average of current amplitudes after rinsing was the average response after MTS application (I after MTS ). The effect of the MTS reagents was estimated using the following equation: Percent change ϭ ((I after MTS / I initial ) Ϫ 1) ϫ 100. The covalent effects of MTS were confirmed by exposing the oocytes to the reducing reagent dithioerythritol (DTT) at 5 mM for 60 s (MTSET-treated oocytes) or at 1 mM for 180 s (MTSEA-treated oocytes). In all cells tested, the response after DTT treatment was comparable with the one prior to the MTSET or MTSEA application, confirming that the functional changes observed were due to the covalent modification of Cys-225 by the MTS reagents. MTS reagents had no functional effects on WT ␤2_␣4_␤2_␣4_␣4 (see Fig. 6E).
Structure Homology Modeling-Sequences of the human ␣4 and ␤2 nAChR subunits were obtained from the ExPASy proteomics server with accession numbers P43681 (␣4) and P17787 (␤2). The homopentameric Lymnaea stagnalis AChBP structure (Protein Data Bank code 1UW6) was used to generate models of the extracellular domain of the (␣4␤2) 2 ␣4 receptor as described previously (8).
Statistical Analysis-Data analyses were performed using GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA). Data were pooled from at least three different batches of oocytes. An F-test determined whether the one-site or biphasic model best fit the data; the simpler one-component model was preferred unless the extra sum-of-squares F-test had a value of p less than 0.05. log EC 50 values for ACh or ACh plus Dh␤E and changes in current-response amplitudes in response to mutations or MTS application were analyzed using one-way analysis of variance with a Dunnett or Bonferroni post hoc correction for the comparison of all mutated receptors to determine significance between WT and mutant receptors. Significance levels between mutant receptors were determined using unpaired t tests. Data are plotted as mean Ϯ S.E. Fit parameter values are the best fitting values with the S.E. values estimated from the fit.

RESULTS
The conservation of aromatic residues contributing to the agonist binding site in nAChR subunits together with homology modeling of the extracellular domain of the (␣4␤2) 2 ␣4 nAChR suggested an additional agonist site at the ␣4(ϩ)/␣4(Ϫ) interface ( Fig. 1, B and C). The primary (ϩ) face of this putative agonist site would be contributed by the auxiliary ␣4 subunit, whereas the complementary (Ϫ) face would be contributed by an ␣4 subunit that simultaneously provides the primary face of the consensus agonist site at an ␣4(ϩ)/␤2(Ϫ) interface. The putative site is homologous to the agonist site at the ␣4(ϩ)/ ␤2(Ϫ) interfaces, having the key conserved agonist-binding residues ␣4Tyr-126 (loop A), ␣4Trp-182 (loop B), on the primary face of the binding site. Importantly, the (ϩ) site also contains a segment of amino acids containing a pair of vicinal cysteines (␣4Cys-225 and ␣4Cys-226) flanked by the conserved aromatic residues ␣4Tyr-223 and ␣4Tyr-230. This region is homologous to loop C in the (ϩ) face of the agonist site at ␣4(ϩ)/␤2(Ϫ) interfaces. This segment is not present in the ␤2(ϩ)/␣4(Ϫ) interface as neither the vicinal cysteines nor the residue equivalent to ␣4Tyr-223 are conserved in ␤2 (Fig. 1B) (11). Presently, it is believed that agonist binding triggers a capping motion of loop C that traps bound agonist in the binding site, which then leads to channel gating via molecular interactions in the coupling interface (14 -16). On its complementary side, ␣4Trp-88 (loop D) is in a position homologous to ␤2Trp-82 in the agonist sites at ␣4(ϩ)/␤2(Ϫ) interfaces (Fig. 1C). The homologous residues in Torpedo and muscle nAChR are ␣1Tyr-93, ␣1Trp-149, ␣1Tyr-190, ␣1Tyr-198, and residues Trp-55 and Trp-57 located in the complementary ⑀/␥ or ␦ subunits of the binding site (11).
To determine whether the ␣4(ϩ)/␣4(Ϫ) interface influences the agonist sensitivity of (␣4␤2)␣4 nAChR, we first exchanged the N-terminal region of the ␤2 subunit for the N-terminal domain of the ␣4 subunit at the auxiliary subunit position (i.e. changed from an extracellular ␣4(ϩ)/␣4(Ϫ) interface to an extracellular ␤2(ϩ)/␣4(Ϫ) interface; Fig. 2A) and then tested the sensitivity of the chimeric receptor to ACh. To obviate uncertainties about subunit assembly, stoichiometry, and positions especially concerning placement of mutant or chimeric subunits in the additional as opposed to the consensus agonist binding positions when engaging heterologous expression from loose subunits, we conducted our studies using as a template the ␤2_␣4_␤2_␣4_␣4 nAChR construct, a fully concatenated form of the (␣4␤2) 2 ␣4 nAChR that replicates the functional properties of (␣4␤2) 2 ␣4 nAChRs (9). Importantly, ␤2_␣4_␤2_␣4_␣4 nAChRs are stable and produce homogeneous channels when expressed in Xenopus oocytes (9). We have shown previously that the ␣4 subunit is involved either in the inhibitory effects of Zn 2ϩ (due to ion binding at ␤2(ϩ)/␣4(Ϫ) interfaces) or in the potentiating effects of Zn 2ϩ (due to ion binding at the ␣4(ϩ)/␣4(Ϫ) interfaces) (8). Using a mutation that impairs sensitivity of (␣4␤2) 2 ␣4 receptors to inhibition or potentiation by Zn 2ϩ (␣4H195A) (8), we determined that the ␣4 subunits located in the fourth and fifth positions of the linear sequence of the concatenated receptor contribute to the Zn 2ϩ potentiating site (supplemental Fig. 1). The (ϩ) face of the site is contributed by the ␣4 subunit in the fifth position, which is therefore the auxiliary subunit in the concatamer ␤2_␣4_␤2_␣4_␣4 receptor. The inhibitory Zn 2ϩ sites are located at the interface between the second (an ␣4 subunit) and third (a ␤2 subunit) positions and between the fifth (the auxiliary ␣4 subunit) and first (a ␤2 subunit) subunits of the concatamer receptor (supplemental Fig. 1). From these findings, we inferred that the consensus agonist sites at ␣/␤ interfaces are located at the interface between the first ␤2 and second ␣4 subunits and between the third ␤2 and fourth ␣4 subunits of the concatamer receptor.
Alanine substitution of Trp-182 in the fifth subunit in ␤2_␣4_␤2_␣4_␣4 receptors should decrease the sensitivity of the receptor to ACh. Based on the finding that obliteration of the additional site in chimeric ␤2_␣4_␤2_␣4_␤2/␣4 nAChRs increased ACh sensitivity, we anticipated that impairment of the site at the ␣4(ϩ)/␣4(Ϫ) by alanine substitution of ␣4Trp-182 would produce sufficient separation between the ACh EC 50 values of the intact and impaired binding sites to generate a biphasic ACh CRC comprising a high sensitivity and a low sensitivity component. The high sensitivity component would be   3A and Table  1). The second CRC component became evident at ACh concentrations higher than 500 M and displayed an ACh sensitivity that was 45-fold lower than that for WT ␤2_␣4_␤2_␣4_␣4 receptors (p Ͻ 0.0001; n ϭ 10) ( Table 1). We next explored whether impairment of the binding sites at the ␣4(ϩ)/␤2(Ϫ) interfaces produced similar effects. Fig. 3, B and C, show that the effects of ACh on ␤2_ W182A ␣4_␤2_␣4_␣4 or ␤2_␣4_␤2_ W182A ␣4_␣4 receptors were biphasic (p Ͻ 0.001; n ϭ 6). The relative abundance of the CRC component with the highest sensitivity for ACh was similar for all three mutant receptors. However, the ACh sensitivity of the high sensitivity component in the CRC for ␤2_ W182A ␣4_␤2_␣4_␣4 or ␤2_␣4_␤2_ W182A ␣4_␣4 receptors was ϳ78-fold greater than that for WT, making it comparable with the ACh sensitivity of the isoform (␣4␤2) 2 ␤2 or chimeric ␤2_␣4_␤2_␣4_␤2/␣4 receptors, both of which lack the ␣4(ϩ)/␣4(Ϫ) interface (Table  1). In contrast, when the mutation was in the fifth subunit (i.e. ␤2_␣4_␤2_␣4_ W182A ␣4), the ACh sensitivity for the equivalent CRC component was increased by only 3-fold relative to WT ( Table 1). The ACh CRC component with the lowest ACh sensitivity also differed in both types of mutant receptors. For receptors with mutant ␣4(ϩ)/␤2(Ϫ) interfaces, the ACh sensitivity of this component was 8-fold lower than WT (Table 1), whereas for ␤2_␣4_␤2_␣4_ W182A ␣4 nAChRs it was 45-fold lower than that for WT (Table 1). For all three mutant receptors, the Hill coefficient of the component with lowest ACh sensitivity increased to 2-3 relative to WT, whereas changes in the Hill coefficient of the CRC component with the highest ACh sensitivity were not significant in comparison with WT.
To test whether the biphasic ACh CRC reflected the ACh sensitivity of a homogeneous population of receptors with two intact agonist sites and one impaired agonist site, we introduced the mutation simultaneously into all three ␣4 subunits of the concatamer to produce triple mutant ␤2_ W182A ␣4_␤2_ W182A ␣4_ W182A ␣4 receptor. We expected that the ACh CRC for this receptor would be monophasic and that the estimated ACh EC 50 would be similar to that for non-linked (␣4 W182A ␤2) 2 ␣4 W182A nAChRs. Fig. 3D shows that when the W182A was simultaneously introduced into the ␣4(ϩ)/␣4(Ϫ) and the two ␣4(ϩ)/␤2(Ϫ) interfaces (triple  mutant receptor) in the ␤2_␣4_␤2_␣4_␣4 receptor the ACh CRC was monophasic (p Ͻ 0.001; n ϭ 10) and had an ACh EC 50 value that was no different from that obtained for nonlinked (␣4 W182A ␤2) 2 ␣4 W182A nAChRs (Tables 1 and 2). Next, we introduced ␣4W182A simultaneously into the (ϩ) face of two subunit interfaces to create mutant ␤2_␣4_␤2_ W182A ␣4_ W182A ␣4, ␤2_ W182A ␣4_␤2_␣4_ W182A ␣4, or ␤2_ W182A ␣4_␤2_ W182A ␣4_␣4 receptors. As shown in Fig. 4, introducing W182A simultaneously into two ␣4 subunits of ␤2_␣4_␤2_␣4_␣4 receptors still produced biphasic ACh CRC (p Ͻ 0.007; n ϭ 6 -8) (Fig. 4) regardless of the combination of binding sites mutated. In comparison with single mutant receptors, the relative abundance of the CRC component with the highest sensitivity to ACh was reduced in all three mutants (p Ͻ 0.001; n ϭ 6 -12) (Figs. 3 and 4 and Table 1). For all three double mutant receptors, the ACh sensitivity of the high sensitivity component was enhanced as compared with WT but decreased as compared with the equivalent ACh EC 50 for single mutant receptors (p Ͻ 0.001; n ϭ 6 -8) ( Table 1). The relative abundance of the CRC fraction with the highest ACh sensitivity was higher when the mutation was incorporated into the second and fifth subunits, but this difference was not statistically significant ( Fig. 4 and Table 1). The ACh EC 50 estimated for the component with the lowest ACh sensitivity was similar for all three double mutant receptors and was not different from the ACh EC 50 for (␣4 W182A ␤2) 2 ␤2 receptors (EC 50 ϭ 500 Ϯ 78 M; Table 2). Overall, therefore, the findings with the single and double W182A mutant ␤2_␣4_␤2_␣4_␣4 receptors together with the effects of ␣4W182A on non-linked ␣4␤2 nAChR function (Table 2) indicate that the ACh CRC component with the highest sensitivity for ACh is contributed predominantly by unaltered binding sites, whereas the mutated sites contribute predominantly to the CRC component with the lowest sensitivity for ACh.
Loop C in Fifth Subunit Affects Receptor Function-Loop C in the agonist site of Cys loop ligand-gated ion channels does not bind agonists directly, but agonist binding elicits loop C closure, and this is considered to lead to channel gating via molecular interactions in the coupling interface (14,(21)(22)(23). To assess the significance of loop C on the function of the putative site at the ␣4(ϩ)/␣4(Ϫ) interface, we introduced ␣4Tyr-230 into the (ϩ) face of the site and then assayed its effects on the function of ␤2_␣4_␤2_␣4_␣4 receptors. ␣4Tyr-230 is equivalent to muscle ␣1Tyr-198 in the C terminus of loop C. It has been shown to affect gating but not agonist binding affinity in muscle nAChR (19), and in the (␣4␤2) 2 ␣4 receptors, it decreases the sensitivity to ACh and functional expression ( Table 2). We hypothesized that if the site at the ␣4(ϩ)/␣4(Ϫ) interface had the capability of contributing to receptor activation, alanine substitutions of Tyr-230 should impair channel gating, leading to a biphasic ACh CRC comprising a fraction contributed by agonist sites with unaltered loop C and a component contributed by mutant loop C. As shown in Table 1 (see also supplemental Fig. 2), ␤2_␣4_␤2_␣4_ Y230A ␣4 nAChRs produced a biphasic ACh CRC, comprising a component with an ACh EC 50 comparable with the ACh EC 50 of the (␣4␤2) 2 ␤2 nAChRs and a component with reduced ACh sensitivity as compared with the high sensitivity fraction. The high sensitivity component represented ϳ15% of the CRC (Table 1). Incorporating ␣4Y230A into either of the ␣4(ϩ)/␤2(Ϫ) ACh binding sites produced comparatively similar effects except that the high sensitivity fraction was 2 times greater (p Ͻ 0.001; n ϭ 10) ( Table 1 and supplemental Fig.  2, A-C).
Modification of Loop C with MTS Reagents-Demonstrating that conformational changes in loop C in the fifth subunit alters the ACh responses of the ␤2_␣4_␤2_␣4_␣4 nAChR would strengthen the conclusion that the putative agonist binding site in the ␣4(ϩ)/␣4(Ϫ) interface contributes directly to channel gating. Loop C conformational transitions can be inferred from changes in the agonist responses of receptors brought about by covalent modification of substituted (22) or free (24) Cys residues in loop C by MTS reagents. Accordingly, we substituted serine for Cys-226 at the tip of loop C to dismantle the disulfide bond between Cys-225 and Cys-226 to make Cys-225 accessible to MTS reagents (MTSET or MTSEA; Fig. 6A). Substitution of serine for Cys-226 in any of the three ␣4 subunits of the ␤2_␣4_␤2_␣4_␣4 nAChR yielded biphasic ACh CRCs that were comparable with those produced by the ␣4Y230A mutation (Fig. 6B, Table 1, and supplemental Fig. 2, D-F). These findings are in accord with recent studies of muscle nAChR showing that removal of the Cys bridge in loop C decreases agonist sensitivity without abolishing loop C function (24). To assess the effects of MTS reagents on the function of intact and mutated sites, we performed the experiments using 800 or 3 M ACh. 800 M ACh produced near maximal activation of the ACh CRC fraction with the lowest ACh sensitivity, whereas 3 M ACh produced almost maximal activation of the ACh CRC fraction with the highest ACh sensitivity (Table 1 and supplemental Fig. 2, D-F). We have shown above that ACh produces a biphasic CRC at ␤2_␣4_␤2_␣4_␣4 receptors with one mutant agonist site and that the CRC component with the lowest ACh sensitivity is contributed predominantly by the mutated site, whereas the component with the highest sensitivity component reflects predominantly the activation of intact binding sites. As shown in Fig. 6C, when the current responses of ␤2_␣4_␤2_␣4_ C226S ␣4 nAChRs were activated by 800 M ACh, MTSET decreased the amplitude of the ACh responses by ϳ30%. In accord with a covalent interaction between ␣4Cys-225 and MTSET, exposure to DTT reversed the effects of MTSET (Fig. 6C). By comparison, MTSET had no effects on the amplitude of the responses evoked by 3 M ACh (Fig.  6C). These results were expected because 3 M ACh activated predominantly intact agonist sites, and MTSET produced similar effects on the ACh of ␤2_ C226S ␣4_␤2_␣4_␣4 or ␤2_␣4_␤2_ C226S ␣4_␣4 mutant nAChRs (Fig. 6E). Exposure to MTSEA did not have any apparent effect on the current responses elicited by 800 M ACh in ␤2_␣4_␤2_␣4_ C226S ␣4 nAChRs (Fig. 6D). However, when the responses were elicited by lower concentrations of ACh (3 M) after MTSEA treatment, the amplitude of the ACh responses was enhanced markedly, and this effect was reversed by DTT (Fig. 6D). This result demonstrated that MTSEA enhanced the ACh responses of the mutant receptor by covalently modifying Cys-225. Comparable effects were observed when ␤2_ C226S ␣4_␤2_␣4_␣4 or ␤2_␣4_␤2_ C226S ␣4_␣4 were exposed to MTSEA (Fig. 6E). Although we observed differences in the effects of MTS at each agonist site, the differences were not significant.

DISCUSSION
The ␣4 and ␤2 subunits of the nAChR assemble into alternate forms that differ markedly in sensitivity to activation by agonists. Because ␣4␤2 nAChRs receptors contain two identical agonist binding sites at the ␣(ϩ)/␤(Ϫ) interfaces, a long standing question has been what structural features confer agonist sensitivity to the alternate forms of the ␣4␤2 nAChR. Here, we provide evidence that an additional site at the ␣4(ϩ)/␣4(Ϫ) interface in the (␣4␤2) 2 ␣4 nAChR underlies the agonist sensitivity signature of the isoform (␣4␤2) 2 ␣4 nAChR. The additional site binds ACh and when engaged by ACh contributes to channel gating through a coupling pathway that includes loop C. The findings have important structural and functional implications about the role of subunit composition in the structuralfunctional properties of the ␣4␤2 nAChR and the manner by which agonists activate heteromeric Cys loop ligand-gated ion channels.
The agonist site at the ␣4(ϩ)/␣4(Ϫ) is a major determinant of the agonist sensitivity of the (␣4␤2) 2 ␣4 nAChR. Its removal increases ACh sensitivity and sazetidine-A efficacy and obliterates cytisine efficacy, all of which are functional signatures of the isoform (␣4␤2) 2 ␤2 nAChR. Moreover, when agonist binding by the ␣4(ϩ)/␣4(Ϫ) site is weakened by ␣4W182A, ACh produces biphasic agonist responses in accord with the functional behavior of ␤2_␣4_␤2_␣4_␣4 receptors containing unaltered and mutated agonist sites. The high sensitivity component of the curve represents the contribution of unaltered agonist sites at the ␣4(ϩ)/␤2(Ϫ) interfaces, whereas the mutated site contributes to the low sensitivity component as demonstrated by the differential sensitivity of these components to ACh or the antagonist Dh␤E. To our knowledge, this is the first time that biphasic agonist CRCs have been obtained from receptors with fixed stoichiometry. This is possible because in concatenated pentamers the agonist sites can be selectively impaired or ablated. A similar phenomenon occurs when the ACh binding sites of muscle nAChR are impaired individually. In muscle nAChR, the two agonist sites are located at the ␣/␥ and ␣/␦ interfaces and are thus structurally different (11). Selective mutagenesis of the sites is therefore possible in muscle nAChR assembled from non-linked subunits, and when the mutant receptors are expressed heterologously, exposure to the nicotinic competitive antagonist tubocurarine produces a biphasic 125 I-bungarotoxin CRC (25).
The additional agonist site clearly contributes to the activation of the receptor by agonist in a manner comparable with that of the consensus agonist site located at each of the ␣4(ϩ)/ ␤2(Ϫ) interfaces. When loop C in the primary face of the ␣4(ϩ)/␣4(Ϫ) was impaired through alanine substitution of Tyr-230 or serine substitution of Cys-226 or by covalent modification of a free Cys-225 by MTS reagents, the amplitude of the ACh responses was reduced (alanine substitution and MTSET) or enhanced (MTSEA) Although this site resides in the same interface where the signature potentiating Zn 2ϩ site of the (␣4␤2) 2 ␣4 isoform is located (8), analogous to that of the benzodiazepine site in GABA A receptors (26), it is not a classic allosteric site. Neither Zn 2ϩ nor benzodiazepines are capable of channel gating but appear to enhance receptor responses to agonist by increasing channel opening frequency (benzodiazepines; Ref. 27) or by increasing burst duration (Zn 2ϩ ; Ref. 28).
MTS modification of a free cysteine engineered at the tip of loop C in the fifth subunit provided strong support for the idea that the putative site at the ␣4(ϩ)/␣4(Ϫ) interface contributes to channel gating. Interestingly, MTSET and MTSEA affected loop C function differentially. It is possible that MTS reagents lock loop C into different conformations depending on how the Cys-MTS moiety is oriented within the binding pocket. MTSET could attenuate ACh responses by stabilizing the mutated loop C into an extended, antagonist-bound-like conformation, thus reducing ACh efficacy. By comparison, MTSEA could enhance ACh responses by locking the mutated loop C into a capped, agonist-bound-like conformation, which would increase agonist efficacy, leading to the activation of the mutated sites at submaximal ACh concentrations. High resolution x-ray crystal structures of an MTS-Y53C mutant of the AChBP support this view (29). Loop C in MTSET-Y53C AChBP is in an extended, antagonist-bound conformation, which is consistent with the finding that ␣7W55C-MTSET nAChRs are unresponsive to ACh (29). Another reagent, methyl methanethiosulfonate, enhances the ACh responses of ␣7Y53C nAChR and locks loop C in methyl methanethiosulfonate-Y53C AChBP into a closed, agonist-bound-like confor-mation (29). Although MTSET and MTSEA may affect the contribution of loop C to receptor activation differentially, it is clear from their effects on the ACh responses of the mutant agonist sites that all three sites contribute to activation of (␣4␤2) 2 ␣4 nAChRs.
The sites at the ␣4(ϩ)/␤2(Ϫ) and ␣4(ϩ)/␣4(Ϫ) interfaces respond differentially to alanine substitutions in respect to sensitivity to ACh or Dh␤E. The sites also appeared to interact differently with the agonists cytisine and sazetidine as suggested by the remarkable changes in the efficacy of these ligands brought about by ablation of the ␣4(ϩ)/␣4(Ϫ) interface. This could arise from differences in the overall architecture of the agonist sites and the coupling regions. Although the primary face in the agonist site at ␣4(ϩ)/␣4(Ϫ) and ␣4(ϩ)/␤2(Ϫ) interfaces is contributed by identical amino acids, the complementary face is not. In the case of the agonist site at the ␣4(ϩ)/ ␣4(Ϫ), the complementary face is contributed by the (Ϫ) face of the adjacent ␣4 subunit, providing a likely structural determinant for differential agonist binding site interactions. Moreover, the coupling pathway leading to gating may also differ in both types of agonist sites, and this could affect the interfacial interactions implicated in binding and gating. Loops Cys, ␤1␤2, ␤8␤9, the end of ␤10, pre-M1 region, M2-M3 linker, and post-M4 have been shown to contribute to gating in the Cys loop receptor family (30,31). Although these regions are conserved in the fifth subunit, the interactions of these regions with the complementary face of the binding sites are likely to vary, thus altering or introducing coupling steps that may lead to profoundly different effects on agonist-induced gating. In support of this possibility, it has been shown previously that single point mutations in the coupling regions of Cys loop receptors alter the effects of ligands on receptor function (for example, see Refs. [32][33][34][35]. How does the additional agonist site impact (␣4␤2) 2 ␣4 receptor function? The additional site appears to increase the efficacy of ACh. The biphasic nature of the ACh CRC of ␤2_␣4_␤2_␣4_ W182A ␣4 suggests that the three binding sites must be fully engaged by ACh to attain maximal receptor activation, although occupancy of two agonist sites allows receptor activation. This possibility is consistent with our previous findings that the maximal ACh current responses of (␣4␤2) 2 ␤2, which has only two agonist sites, are smaller than those of the (␣4␤2) 2 ␣4 receptors (7,9,10). Interestingly, the W182A mutation produced similar effects when incorporated in any of the three agonist sites of the ␤2_␣4_␤2_␣4_␣4 receptor, suggesting that ACh efficacy in (␣4␤2) 2 ␣4 receptors is determined by the number of agonist sites engaged by ACh. This is comparable with the activation of homomeric ␣7/5HT3 chimeric receptors that requires occupancy of three binding sites to maximally stabilize the open channel as long as one site is at a subunit separated from the other two (36). In the case of the ␤2_␣4_␤2_␣4_␣4 receptor, the agonist sites are arranged nonconsecutively as well ( Fig. 1A and supplemental Fig. 1).
Sensitivity to activation by ACh was also affected by the number of agonist sites contributing to ␤2_␣4_␤2_␣4_4 activation. Sensitivity to ACh increased through either ablation of the ␣4(ϩ)/␣4(Ϫ) interface or through alanine substitution in the ␣4(ϩ)/␣4(Ϫ) or either of the ␣4(ϩ)/␤2(Ϫ) sites. In contrast, the ACh sensitivity of homomeric ␣7/5HT3 chimeric receptors appears to increase by occupancy of three non-consecutive agonist sites (36). It may be that the agonist site at the ␣4(ϩ)/ ␣4(Ϫ) interface has less affinity for ACh than the sites at the ␣4(ϩ)/␤2(Ϫ) interfaces, which could decrease the overall ACh sensitivity of the receptor. Unlike the agonist sites in homomeric receptors, the ACh sites at (␣4␤2) 2 ␣4 receptors are not structurally identical. This could lead to each type of site having different gating effects, leading to overall changes in agonist sensitivity, in comparison with ␣4␤2 nAChR activated by two identical agonist sites. In accord with this possibility, we found that the Trp-182 mutation affected the site at the ␣4(ϩ)/␣4(Ϫ) and ␣4(ϩ)/␤2(Ϫ) interfaces differentially. Furthermore, the pattern of changes in receptor function of the single or double agonist binding mutants brought about by alanine substitution of agonist site residues as observed in terms of relative abundance of the components of the ACh CRC, sensitivity to ACh, and changes in the n H coefficient was influenced by the subunit interface location of the mutated agonist sites. This suggests complex interactions between the agonist sites that may also impact the overall ACh sensitivity of the (␣4␤2) 2 ␣4 receptor. Studies of the microscopic currents of the (␣4␤2) 2 ␣4 nAChR should aid the understanding of how the additional agonist site impacts the activation of this (␣4␤2) nAChR isoform in comparison with (␣4␤2) 2 ␤2 as well as homomeric Cys loop receptors.