Role of the Cys Loop and Transmembrane Domain in the Allosteric Modulation of α4β2 Nicotinic Acetylcholine Receptors*

Allosteric modulators of pentameric ligand-gated ion channels are thought to act on elements of the pathways that couple agonist binding to channel gating. Using α4β2 nicotinic acetylcholine receptors and the α4β2-selective positive modulators 17β-estradiol (βEST) and desformylflustrabromine (dFBr), we have identified pathways that link the binding sites for these modulators to the Cys loop, a region that is critical for channel gating in all pentameric ligand-gated ion channels. Previous studies have shown that the binding site for potentiating βEST is in the C-terminal (post-M4) region of the α4 subunit. Here, using homology modeling in combination with mutagenesis and electrophysiology, we identified the binding site for potentiating dFBr on the top half of a cavity between the third (M3) and fourth transmembrane (M4) α-helices of the α4 subunit. We found that the binding sites for βEST and dFBr communicate with the Cys loop, through interactions between the last residue of post-M4 and Phe170 of the conserved FPF sequence of the Cys loop, and that these interactions affect potentiating efficacy. In addition, interactions between a residue in M3 (Tyr309) and Phe167, a residue adjacent to the Cys loop FPF motif, also affect dFBr potentiating efficacy. Thus, the Cys loop acts as a key control element in the allosteric transduction pathway for potentiating βEST and dFBr. Overall, we propose that positive allosteric modulators that bind the M3-M4 cavity or post-M4 region increase the efficacy of channel gating through interactions with the Cys loop.

The ␣4␤2 nicotinic acetylcholine receptor (nAChR) 2 belongs to the superfamily of pentameric ligand-gated ion channels (pLGICs). In humans, this superfamily comprises the Cys loop receptors (including muscle and neuronal nAChRs, 5-HT 3 , GABA A , and glycine receptors), which mediate all fast central nervous system synaptic inhibition and much of fast peripheral excitation. Cys loop receptor subunits assemble as a pentamer of identical (homomeric channels) or different (heteromeric channels) subunits around a central ion channel. The individual subunits have a large extracellular N-terminal domain (ECD) that consists of 10 ␤-strands (␤1-␤10) folded into a ␤-sandwich and a transmembrane domain (TMD) with four transmembrane ␣ helices (M1-M4) connected by linkers (M1-M2, M2-M3, and M3-M4), as well as intracellular domains and a highly variable extracellular C-terminal (post-M4) region (1). There are 2-5 neurotransmitter binding sites at subunit interfaces within the ECD, and these sites are functionally coupled to the transmembrane ion channel located ϳ50 Å away. In the nAChR, the subunit contributing the principal face of the agonist binding site (an ␣ subunit) couples agonist-triggered agonist binding site movements to channel gating (2,3). The coupling is achieved at the ECD-TMD interface by a principal pathway that couples the pre-M1 region in the ECD to the M2-M3 linker through the ␤1-␤2 loop (4) and the canonical FPF motif of the ␤6-␤7 loop (the Cys loop) (5,6). More recently, it has been shown that gating is also affected by more peripheral pathways that couple M4 to M1 and M3 (7) and post-M4 to the Cys loop (8).
In common with all members of the pLGIC family, the current responses of ␣4␤2 nAChR can be enhanced by a variety of agents, including Zn 2ϩ (9,10), the endogenous steroid 17␤estradiol (␤EST) (11), desformylflustrabromine (dFBr) (12,13), and synthetic ligands CMPI (14), LY2087101 (15), and NS206 and NS9283 (16,17). The identification of positive allosteric modulator (PAM) binding sites in the TMD of the ␣4 subunit, e.g. ␤EST and NS206 binding sites (11,17), as well as on the signature ␣4/␣4 interface present in ␣4␤2 nAChRs composed of three ␣4 and two ␤2 subunits (potentiating Zn 2ϩ , CMPI, and NS9283 binding sites) (10,14,17), has revealed the critical role of this subunit in sensitivity to PAMs. Despite this insight, there is little understanding of how PAM binding to sites in the ␣4 subunit is transduced into receptor potentiation. Allosteric modulators are thought to induce the same global conformational transitions that promote channel opening by agonists by affecting gating elements near their binding sites (18). In this respect, it is significant that the ␤1-␤2 loop and the Cys loop in the ␣4 subunit have been identified as components of the transduction pathway linked to potentiation of ␣4␤2 nAChRs by the NS206 binding site in the TMD (17). Additionally, Trp 621 , the first residue in the N-terminal side of the post-M4 region of the ␣4 subunit, has been shown to affect ␤EST potentiating efficacy in a manner consistent with a role in transduction (19). A suitable model system to examine the role of post-M4 and ECD agonist-binding-gating coupling elements in positive allostery of ␣4␤2 nAChRs is that involving potentiation of this receptor type through PAM sites located in the ␣4 subunit. An example of this type of ␣4␤2 nAChR PAM is ␤EST, whose binding site is in the post-M4 region of the ␣4 subunit (11). To identify additional allosteric modulator binding sites on the ␣4 subunit, we performed docking of nicotinic PAMs on a homology model of the ␣4␤2 nAChR. These studies highlighted dFBr as a PAM that may bind the upper part of the TMD of the ␣4 subunit. By using mutagenesis together with the substituted cysteine accessibility method, we confirmed that the upper half of a cavity between M3 and M4 hosts a potentiating dFBr site. Here, we demonstrate that allosteric signals from the dFBr and ␤EST potentiating binding sites are propagated to the Cys loop through interactions between Phe 170 of the canonical FPF motif of the Cys loop and the final residue of the post-M4 region. We also found that interactions between Phe 167 in the Cys loop and Tyr 309 in M3 are also important for the transduction of dFBr binding into potentiation of the ␣4␤2 nAChR.

Results
The TMD of the ␣4 nAChR Subunit Houses a Potentiating dFBr Binding Site-We examined the role of post-M4 and agonist binding-gating coupling elements in the ␣4 subunit in the propagation of allosteric signals generated by binding of the PAM compounds ␤EST and dFBr to ␣4␤2 nAChRs. The potentiating efficacy of ␤EST and dFBr is greater at ␣4␤2 nAChRs containing three copies of ␣4 subunit (Fig. 1), suggesting that ␣4 encodes the binding and transduction elements for ␤EST and dFBr potentiation. The ␣4␤2 nAChR can assemble in two functional stoichiometries: one with three copies of ␣4 and two copies of ␤2 [(␣4␤2) 2 ␣4] and the other with two copies of ␣4 and three copies of ␤2 [(␣4␤2) 2 ␤2] (20). The alternate receptors differ in pharmacological properties, including sensitivity to agonists and allosteric modulators (20,21).
Previous studies have shown that ␤EST binds the post-M4 region of the ␣4 subunit (11). The site for dFBr binding has not been identified so far, although it has been suggested to lie within the ECD (22). Hence, to aid the identification of the potentiating dFBr binding site in the ␣4␤2 nAChR, we performed docking stimulations with dFBr on a homology model of the ␣4␤2 nAChR. Some docking positions were found in the pore, but the majority were found in the cavity between M3 and M4 of both the ␣4 subunit ( Fig. 2A) and the ␤2 subunit (not shown). The position of dFBr in the latter tended to be located more superficially to the cavity. The M3-M4 cavity is a common allosteric site in pLGICs and is the target of a wide variety of allosteric modulators such as general anesthetics in anionic pLGICs (23)(24)(25), neurosteroids in GABA A receptors (26), and PNU-120596 and LY-2087101 in ␣7 nAChRs (27).
Note that during revision of this manuscript, the X-ray structure of the (␣4␤2) 2 ␤2 nAChR in a presumed desensitized conformation was published (28). We overlaid our homology model and the X-ray structure of the (␣4␤2) 2 ␤2 receptors. This showed that the proposed dFBr binding region is fairly similar in both the homology model and the X-ray structure of the (␣4␤2) 2 ␤2 receptor (Fig. 2B) and demonstrates that our homology model is a valid tool to predict interactions between the ␣4␤2 nAChR and ligands.
To elucidate which residues in the TMD of the ␣4 subunit might contribute to the putative dFBr binding site, we priori- ACh + dFBr or βEST dFBr βEST FIGURE 1. Effects of ␤EST and dFBr on alternate ␣4␤2 nAChRs. A, structure of ␤EST. B, concentration response effects of ␤EST on the ACh EC 10 current responses of (␣4␤2) 2 ␣4 and (␣4␤2) 2 ␤2 nAChRs. The data points represent the means Ϯ S.E. of at least four experiments. The data were fit to the monophasic Hill equation, as described under "Experimental Procedures." The effects of ␤EST were determined on ACh currents evoked by EC 10 (3 M for (␣4␤2) 2 ␤2) nAChRs and 10 M for (␣4␤2) 2 ␣4 nAChRs. C, structure of dFBr. D, potentiating effects of dFBr on the ACh responses of alternate (␣4␤2) 2 ␣4 and (␣4␤2) 2 ␤2 nAChRs. The concentration-responses curves were obtained as for those of ␤EST. The data points represent the means Ϯ S.E. of five experiments. For B and D, functional expression of (␣4␤2) 2 ␣4 or (␣4␤2) 2 ␤2 nAChRs was achieved by expressing the concatenated forms of the alternate stoichiometries of the ␣4␤2 nAChRs in Xenopus oocytes, as described under "Experimental Procedures." tized residues with their side chain pointing toward the cavity and individually substituted them with alanine. These studies highlighted ␣4 M3 Tyr 309 , Phe 312 , Thr 313 , and Phe 316 and M4 Leu 617 and Phe 606 as residues that may interact with dFBr in the putative dFBr binding site to enhance, allosterically, the ACh responses of ␣4␤2 nAChRs (Fig. 2, A and B).
The consequences of the alanine substitutions on dFBr potentiation were assessed on ␣4␤2 nAChRs assembled from equal ratios of ␣4 and ␤2 cDNA. This was necessary to obviate potential problems in data analysis because of low levels of functional expression, which may occur when working with mutants of homogeneous populations of ␣4␤2 nAChRs. The receptors obtained using equal ratios of cDNA consist of ϳ80 -90% of (␣4␤2) 2 ␣4 and 10 -20% of (␣4␤2) 2 ␤2 receptors (20).
Functional assays revealed Tyr 309 , Phe 312 , and Leu 617 as pivotal determinants of dFBr potentiation. Y309A and F312A abolished dFBr potentiation and increased dFBr EC 50 , compared with wild type (Fig. 2C and Table 1). L617A reduced the maximal potentiating effect of dFBr (dFBr efficacy) by 9.5-fold and dFBr EC 50 by 7-fold (p Ͻ 0.001; n ϭ 5), as compared with wild type (Fig. 2C and Table 1). Increased polarity in these positions significantly affected dFBr potentiation. Thus, L617C decreased both dFBr efficacy (3.5-fold) and dFBr EC 50 (6.7fold), whereas F312C and Y309C obliterated the potentiating effects of dFBr (Table 1). To determine whether aromaticity at these positions is essential for dFBr potentiation, the effects of F312Y, Y309F, and L617F on dFBr potentiation were examined. F312Y, Y309F, and L617F reduced dFBr efficacy to similar levels. However, although L617F decreased dFBr EC 50 , F312Y and Y309F increased it, compared with wild type ( Table 1). None of the Tyr 309 , Phe 312 , and Phe 167 mutants affected ACh EC 50 (Table 1). Alanine or cysteine substitutions on Thr 313 , Phe 316 , or Phe 606 , which are predicted to lie below Phe 312 , also decreased dFBr efficacy and increased dFBr EC 50 values, although their effects were less pronounced than those of Tyr 309 , Phe 312 , or L617 (Table 1). Substitutions on Thr 313 did not affect sensitivity to activation by ACh (Table 1), but all substitutions engineered on Phe 316 and Phe 606 did ( Table 1), suggesting that the effects of mutations on Phe 316 and Phe 606 on dFBr potentiation were a consequence of mutant-induced perturbation of receptor function.
To further confirm the role of the ␣4 subunit on dFBr potentiation of ␣4␤2 nAChRs and further establish that M3 encodes binding residues for dFBr, we introduced ␣4F312A sequentially into concatenated (␣4␤2) 2 ␣4 and (␣4␤2) 2 ␤2 receptors. Concatenated receptors have fixed stoichiometry and subunit arrangement, and these constructs have been shown to replicate the functional properties of the receptors assembled from loose ␣4 and ␤2 subunits (21).
Effects of dFBr on MTSET Reaction Rates-Demonstrating that dFBr can access residues in the top half of the cavity between M3 and M4 of the ␣4 subunit would further support the conclusion that this region houses a dFBr potentiation site. We determined this by assessing the ability of dFBr to protect L617C from covalent modification by the thiol compound MTSET (protection assays). L617C decreased the efficacy of dFBr but had no effects on ACh EC 50, indicating that the observed changes in dFBr potentiation are not linked to changes in ACh EC 50 (Table 1). Neither F312C nor Y309C were FIGURE 2. M3 and M4 residues in the ␣4 nAChR subunit impact dFBr potentiation. A, full model of (␣4␤2) 2 ␣4 with ␣4 in yellow and ␤2 in blue on the left, and a zoom on the potential dFBr site is on the right. Residues that might be involved in binding dFBr are shown as sticks. dFBr is shown in light blue. B, superimposition of the X-ray structure of (␣4␤2) 2 ␤2 nAChR onto the homology model used in this study to predict the binding site for dFBr in the ␣4␤2 nAChR. The homology model is shown in yellow, and the X-ray structure is in gray. Relevant residues are shown as sticks (dark pink in the homology model and gray in the X-ray structure). C, representative current responses elicited by ACh EC 10 in the absence or presence of dFBr from oocytes expressing mutant ␣4Y309A␤2, ␣4F312A␤2, or ␣4L617A␤2 nAChRs. D, maximal dFBr potentiation of ACh EC 10 current responses from wild type or mutant (␣4␤2) 2 ␣4 or (␣4␤2) 2 ␤2 nAChRs. The alternate stoichiometries of the ␣4␤2 nAChR were expressed using concatemeric constructs, as described under "Experimental Procedures." Maximal potentiation by dFBr of ACh EC 10 current responses (I max pot ) was calculated as (I ACh EC 10 ϩ dFBr)/I ACh EC 10 from concentration response curve data, as described under "Experimental Procedures." The values represent the means Ϯ S.E. of at least five independent experiments. Asterisks indicate that the change in I max pot is statistically significant (**, p Ͻ 0.001, ***, p Ͻ 0.0001), as measured by one-way ANOVA with Dunnett's correction. The dotted line indicates a potentiation ratio of 1 (no potentiation). The cartoon underlying each column show how many copies of F312A (black dots) are present in the receptors. The red dotted line indicates a potentiation ratio of 1 (no potentiation).
used for dFBr protection assays because these mutants abolished dFBr potentiation (Table 1). We first determined the accessibility of L617C to covalent modification by MTSET by exposing ␣4L617C␤2 receptors to MTSET for 2 min following stabilization of ACh EC 10 responses. ACh-elicited responses and dFBr potentiation of ACh responses were measured before and after MTSET exposure (Fig. 3A). MTSET had no effect on the ACh responses or dFBr potentiation (Fig. 3B) of wild type receptors, indicating that any change measured in ACh responses or dFBr potentiation after MTSET exposure is due to MTSET-modification of L617C. Potentiation of ACh responses was abolished by MTSET, indicating that ␣4L617C is fully accessible to MTSET (Fig. 3, A and B). This effect is selective because MTSET treatment of ␣4F316C, which we propose perturbs dFBr potentiation of ␣4␤2 receptors indirectly, reduced dFBr potentiation by only 30% (Fig. 3B), indicative of limited accessibility to MTSET at this position. For the protection assays, we first stabilized the currents elicited by an EC 10 concentration of ACh at ␣4L617C␤2 receptors and then co-applied ACh EC 10 with EC 100 dFBr to test the maximal dFBr potentiation on ␣4L617C␤2 receptors. Next, a sequence of applications (10 s) of 20 M MTSET in the presence or absence of dFBr EC 100 (10 M) was tested for a total time of 40 s (Fig. 3C). After each application ACh EC 10 ϩ dFBr EC 100 responses were tested for changes in the amplitude of the responses (Fig. 3C). The reaction rate of MTSET with ␣4L617C in the absence of dFBr was 12-fold faster (k 1 ϭ 0.05 Ϯ 0.01 s Ϫ1 ) than in the presence of dFBr (k 1 ϭ 0.004 Ϯ 0.0 03 s Ϫ1 ) (p Ͻ 0.0001; n ϭ 4) (Fig. 3D). Together, the findings show that dFBr protects L617C from reacting with MTSET, which would occur if dFBr binds the dFBr binding site in the top half of the cavity between M3 and M4 in the ␣4 subunit.
Tyr 309 and Phe 312 are conserved in the ␤2 subunit (Fig. 4A), and docking studies indicate that dFBr may bind in a narrow cavity between the top of M3 and M4 in ␤2 (not shown). We therefore tested whether Tyr 300 and Phe 303 in the ␤2 subunit, equivalents to ␣4Y309 and ␣4F312, respectively, altered the potentiating effects of dFBr on ␣4␤2 receptors. Incorporation

TABLE 1 Concentration effects of ACh and dFBr on wild type and mutant ␣4␤2 nAChRs
The concentration effects of dFBr were determined on ACh responses elicited by EC 10 ACh concentrations. The data points were used to generate concentration response curves from which EC 50 and Hill coefficient (n H ) (not shown) were estimated, as described under "Experimental Procedures." Maximal potentiation by dFBr of ACh EC 10 current responses (I max pot ) was calculated as (I ACh EC10 ϩ dFBr) /I ACh EC10 . The values represent the means Ϯ S.E. of a number (n) of experiments. Statistical differences were determined using one-way ANOVA with Dunnett's correction.

Receptor ACh
DFBr of ␤2F303A or ␤2Y300A decreased both dFBr efficacy and potency (Table 1). However, because ␤2Y300A and ␤2F303A markedly decreased ACh EC 50 (p Ͻ 0.0001; n ϭ 5), their effects on dFBr efficacy likely reflect perturbations to receptor function rather than direct effects on dFBr potentiation. The TMD of pLGICs play a pivotal role in gating, and structural integrity in some regions is critically important for this function (32)(33)(34).
Finally, it has been previously suggested that the potentiating binding site of dFBr is located in the ␤2 ϩ /␣4 Ϫ interface of the ␣4␤2 nAChR and that key contributors to this site are ␤2 subunit residues Trp 176 , Tyr 120 , Asp 217 , and Asp 218 (22). We performed alanine substitutions on these residues but found that none disturbed dFBr potentiation significantly ( Table 2).
The Cys Loop Affects dFBr Potentiation-Surprisingly, Phe 312 , Tyr 309 , and Leu 617 are conserved in the ␣3 nAChR subunit (Fig. 4A), yet ␣3␤2 nAChRs were found to be only inhibited by dFBr (IC 50 118 Ϯ 5 M; n ϭ 6) (Fig. 4B). Inhibition of ␣3␤2 nAChRs by dFBr is not mediated by a site in the M3-M4 cavity because alanine substitution on ␣3F310, the ␣3 residue equivalent to the crucial Phe 312 in the ␣4 subunit, has no effect on dFBr-mediated inhibition (IC 50 ϭ 115 Ϯ 7 M; n ϭ 6) (Fig. 4B). In addition, inhibition of ␣3␤2 nAChRs is voltage-dependent (not shown), and the current responses to ACh EC 10 in the presence of concentrations of dFBr greater than 10 M rebound are strongly suggestive of ion channel blockade (Fig. 4B, inset). Together, these findings suggest that the ␣3 subunit may lack structural elements required for transducing the positive allosteric signals generated by binding of dFBr to the M3-M4 cavity. The agonist binding-gating coupling elements, the ␤1-␤2 loop and the Cys loop, have been shown to contribute to the transduction of positive allosteric signals in ␣4␤2 nAChRs (17). Importantly, ␣3 and ␣4 differ in five amino acid positions in these regions, two in loop ␤1-␤2, and three in the Cys loop (Fig.  4A). To determine whether these residue differences are defining factors for sensitivity to potentiation by dFBr, we mutated the ␣4 residues to their ␣3 counterparts. Neither mutating ␣4S162 in the Cys loop to lysine, its ␣3 equivalent, nor changing ␣4 loop ␤1-␤2 into an ␣3 ␤1-␤2 loop (␣4(␣3␤1-␤2 loop)) had any significant effect on dFBr potentiation ( Fig. 4C and Table  3). In contrast, substituting Phe 167 in the ␣4 subunit Cys loop with its ␣3 equivalent (a tyrosine) reduced the efficacy of dFBr by 5-fold without changes in dFBr EC 50 ( Fig. 4C and Table 3). In addition, alanine substitution of Phe 167 had no effect on dFBr EC 50 but almost abolished potentiation by decreasing dFBr efficacy by 9.6-fold ( Fig. 4C and Table 3).
Phe 167 is next to the canonical FPF motif carried by the Cys loop in pLGICs, which interacts with neighboring pre-M1 in the ECD and the M2-M3 linker in a well established pathway   linking agonist binding to channel gating (5,6). We examined the relevance of FPF for dFBr potentiation by individual mutations on the ␣4 Cys loop motif (Phe 168 , Pro 169 , and Phe 170 ). Substitutions on proline (P169A, P169F, and P169L) did not yield functional expression. Alanine substitution on Phe 170 was well tolerated and decreased dFBr potentiation efficacy by 5.9fold ( Fig. 4C and Table 3). Alanine substitution on Phe 168 did not yield functional expression (not shown), and F168L, which did not affect the functional expression of ␣4␤2 nAChRs, had no effect on dFBr potentiation ( Fig. 4C and Table 3). The discovery that Phe 167 and Phe 170 are capable of influencing dFBr potentiation of ␣4␤2 nAChRs suggests that the Cys loop may be part of the transduction mechanism for potentiation by dFBr. Interactions between the Cys loop and post-M4 are thought to be important for efficient gating of muscle nAChRs (8). To determine the relevance of ␣4 post-M4 (Fig.  4A) for dFBr potentiation of ␣4␤2 nAChRs, we first determined the effect of dFBr on receptors containing ␣4 subunits devoid of post-M4 (␣4 Ϫ PM4). In accord with previous studies (35), we found that ␣4 Ϫ PM4␤2 nAChRs were functional, albeit with reduced sensitivity to ACh, in comparison with wild type (Table  3). Significantly, removal of post-M4 reduced dFBr efficacy by 8.8-fold (Fig. 5A and Table 3). To confirm the importance of post M4 for dFBr potentiation, we examined dFBr effects on ␣4LMAREDA␤2 nAChRs. ␣4LMAREDA comprises the ECD and TMD from the ␣4 subunit linked to ␣3 post-M4 (LMAREDA) (Fig. 4A). If post-M4 in the ␣4 subunit is a key determinant of dFBr potentiation, ␣4LMAREDA␤2 nAChRs should not be potentiated by dFBr. As shown in Fig. 5A (Table  3), dFBr potentiation was ablated in ␣4LMAREDA␤2 nAChRs. To elucidate which ␣4 post-M4 amino acid residues (WLAGMI) are important for dFBr potentiation of ␣4␤2 nAChRs, we examined the effects of dFBr on alanine mutants of the ␣4 post-M4 region. Individual alanine substitutions on the sequence WLAGM reduced potentiation but not significantly ( Fig. 5A and Table 3). In contrast, alanine substitution of the final residue of ␣4 subunit post-M4 (Ile 626 ) abolished dFBr potentiation ( Fig. 5A and Table 3).
␣4 M4 carries a double proline (PP) motif that precedes post-M4 (Fig. 4A) that is conserved only in the ␣4 and ␣2 nAChR subunits, both of which are sensitive to dFBr potentiation (11,36). The PP motif may introduce considerable restrictions on the orientation and mobility of post-M4, which could be pivotal for dFBr potentiation. To determine the relevance of the PP motif for dFBr potentiation, the prolines were mutated to alanine individually, as well as simultaneously. As shown in Fig. 5A (Table 3), AAWLAGMI obliterated the potentiating effects of dFBr, and APWLAGMI or PAWLAGMI reduced potentiation by 9-fold. Furthermore, exchange of the PP motif for QP, the motif preceding post-M4 in the ␣3 subunit (Fig. 4A), decreased dFBr potentiation by 8.2-fold ( Fig. 5A and Table 3).
To further confirm the relevance of post-M4 and the double PP motif for dFBr potentiation, we tested the effect of changing the ␣3 subunit QPLMAREDA region to the equivalent ␣4 region (PPWLAGMI). We had shown in this study that ␣3␤2 nAChRs are inhibited by dFBr (Fig. 4B), even though residues pivotal for potentiating dFBr binding in the ␣4 subunit are conserved in the ␣3 subunit (Fig. 4A). If dFBr potentiation requires the binding site in the top half of the cavity between M3 and M4 and the PPWLAGMI sequence, ␣3PPWLAGMI␤2 receptors should be sensitive to potentiation by dFBr. As shown in Fig. 5  (B and C), ␣3PPWLAGMI␤2 nAChRs were potentiated by  (Fig. 5C).
Interactions between post-M4 and Cys Loop Are Necessary for dFBr Potentiation-The atomic structure of post-M4 in pLGICs has not been resolved (3,28). However, because it has been proposed that post-M4 in muscle nAChRs may interact with the Cys loop to modulate gating (8), we hypothesized that post-M4 extends toward the Cys loop and that binding of dFBr to the M3-M4 cavity promotes interactions between Phe 170 and Ile 626 (Fig. 6A), leading to potentiation of receptor function.
To test this possibility, we disrupted any possible functional interdependence by mutating the residues individually (F170I and I626F) and also in pairs (F170I/I626F) to potentially restore functional interdependence. Mutants F170I, I626F, and F170I/ I626F were well tolerated, all yielding functional expression. F170I and I626F reduced dFBr potentiation by 4.6-and 4.8-fold, respectively, as compared with wild type (Fig. 6, B and E, and Table 3). When both mutations were present (F170I/I626F), dFBr potentiation was increased by 2.4-fold, in comparison with the single mutants ( Fig. 6B and Table 3). Moreover, dFBr efficacy was 1.2-fold higher than the sum of the dFBr efficacy at the individual mutants (p Ͻ 0.01, n ϭ 4). If mutants F170I and I626F acted independently, the effect of the double mutation should be additive. We also examined whether the double proline could interact with Phe 170 , but mutants F170P, F170P/ P620F, and F170P/P626F did not yield functional expression.
We next examined the possible role of Phe 167 in dFBr potentiation. Phe 167 is close to Leu 305 in the M2-M3 linker of the ␣4 subunit (Fig. 6A), a residue that in the muscle nAChR is energetically coupled to the Cys loop and pre-M1 region to contribute to gating (6). ␣4F167L had no effect on dFBr potentiation, whereas L305F abolished it (Fig. 6C and Table 3). When both mutations were present, dFBr potentiation was abolished ( Fig.  6C and Table 3), indicating that the effect of Leu 305 on dFBr potentiation is not dependent on Phe 167 . The side chain of Phe 167 is also predicted to orientate toward Tyr 309 in M3 (Fig.  6A). We established earlier that Tyr 309 (Fig. 2B) and Phe 167 (Fig.  4C) are critically important for dFBr potentiation. To determine whether Phe 167 and Tyr 309 contribute to dFBr potentiation interdependently, we mutated Phe 167 to tyrosine and Tyr 309 to phenylalanine and examined the effects of these The post-M4 domain of the ␣3 subunit was changed to that of the ␣4 subunit, first keeping the ␣3 PQ motif preceding post-M4 (␣3PQWLAGMI) and then substituting PQ for the PP motif found in the ␣4 subunit (␣3PPWLAGMI). F310A effects on the effects of dFBr on ␣3PPWLAGMI␤2 receptors was also determined. The data were fit by non-linear regression as described under "Experimental Procedures." mutants on dFBr potentiation individually or in pairs. F167Y and Y309F decreased dFBr efficacy by 4.8-and 3.8-fold, respectively, compared with wild type (Fig. 6, D and E, and Table 3). When the mutations were introduced simultaneously (F167Y/ Y309F), dFBr efficacy was restored to near wild type values (Fig.  6, D and E, and Table 3). Phe 170 and Ile 626 Interactions Are Necessary for ␤EST Potentiation-␤EST is an established positive allosteric modulator of ␣4␤2 nAChRs (11,19,35). Previous studies have shown that residues in ␣4 post-M4 are involved in both binding (AGMI, the last four residues of post-M4) and transduction (Trp 621 , the first post-M4 residue) of potentiating ␤EST (11,35). As for dFBr potentiation, the potentiating efficacy of ␤EST is greater at (␣4␤2)2␣4 receptors (Fig. 1A). To assess whether the Cys loop-post-M4 interactions that affect dFBr potentiation are relevant for ␤EST potentiation, we assessed the effects of F170I and I626F on ␤EST potentiation, individually and in pairs. Individually, F170I and I626F abolished ␤EST potentiation ( Fig. 7A and Table 4). When both mutations were present (F170I/I626F), ␤EST efficacy was restored to near wild type values ( Fig. 7A and Table 4).
We also examined whether Phe 167 and Tyr 309 were relevant for ␤EST potentiation. Mutants F167Y, Y309F, and F167Y/ Y309F had no impact on ␤EST potentiation (Fig. 7B). These results suggested that Phe 167 and Tyr 309 may affect selectively potentiation by compounds that bind the M3-M4 cavity. To examine this possibility, we assayed the effect of F167A, Y309A, and F3212A on ␤EST potentiation. ␤EST potentiation was not affected by F167A, Y309A, or F312A (Table 4). In contrast, alanine substitution on Leu 617 abolished ␤EST potentiation (Table 4).

Discussion
This study provides the first demonstration that interactions between the Cys loop and the post-M4 region of the ␣4 subunit affect potentiation of ␣4␤2 nAChR function by PAMs that bind the TMD region of the ␣4 subunit. Our data reveal that in the ␣4 subunit Phe 170 of the Cys loop and Ile 626 , the final residue of post-M4, are functionally coupled and show that this coupling is essential to transduce binding of ␤EST and dFBr into potentiation of ␣4␤2 nAChR function. This conclusion is based on the observation that individual residue swaps, F170I and I626F, ablated (␤EST) or attenuated (dFBr) potentiation, whereas double mutant (F170I/I626F) restored potentiation to wild type  values (␤EST) or increased it above the levels expected from simple additivity (dFBr). In addition, individual alanine mutations on Phe 170 and Ile 626 also affected ␤EST and dFBr potentiation, in a manner consistent with annulment of the functional link between them. Significantly, transfer of the ␣4 subunit post-M4 region to the ␣3 subunit, which conserves the key dFBr binding residues in the TMD, conferred sensitivity to dFBr potentiation to ␣3␤2 nAChRs. A similar phenomenon has been reported for the potentiation of ␣4␤2 nAChR by ␤EST (35). Without structural data for ␣4 post-M4, it is not possible to infer any specific details about the structural framework that could account for Phe 170 /Ile 626 functional interdependence and its role in dFBr and ␤EST potentiation. In the muscle nAChR structure, post-M4 extends beyond the lipid bilayer toward the Cys loop to seemingly interact with Phe 137 (Phe 170 in ␣4) (8). Moreover, the post-M4 residue that appears to interact with Phe 137 is Gln 435 , a residue that aligns with ␣4I626 (Fig.  4A). The functional consequences of putative Phe 137 /Gln 435 interactions have not been examined so far, but it has been suggested that they may contribute to coupling agonist binding to channel gating, as well as transducing the allosteric effects of lipid binding to M4 (8). In regard to the ␣4␤2 nAChR, localized motion transitions triggered by binding of PAMs to the TMD could promote Phe 170 -Ile 626 gating interactions, leading to more efficacious agonist-triggered receptor activation. Given that M4 is covalently linked to post-M4, local structural changes induced by PAM binding to the M3-M4 cavity could be relayed to post-M4 through M4 motions. In accord with this possibility, we found that dFBr and ␤EST may engage in binding interactions with M4 Leu 617 , and these binding interactions are likely to trigger M4 motions that could subsequently alter the orientation of post-M4. Significantly, NMR structures of the TMD of the ␣4␤2 nAChR bound to allosteric inhibitors halothane or ketamine show that binding of these modulators to their binding sites in the TMD elicits changes in protein dynamics beyond the binding sites (37). Also, molecular dynamics simulations show the M4 as a structurally dynamic element undertaking substantial motion during muscle nAChR gating (38). Additionally, M4 is known to affect gating of the muscle nAChR (32), and more recently, it has been proposed that enhanced interactions between M4 and M1-M3 promotes channel function in GLIC, a prokaryotic pLGIC (7).
Extensive mutagenesis studies of ␣4 post-M4 have led to the view that the final four residues of the ␣4 subunit bind ␤EST and that the C terminus of the tail likely plays a role in transduction (35). In contrast, our data strongly suggests that the last residue of post-M4 (Ile 626 ) functionally couples to Phe 170 to transduce binding of ␤EST into receptor potentiation. Without structural data for post-M4 of the ␣4␤2 receptor or adequate probes to photo-label the binding site for ␤EST, we cannot easily reconcile our findings with the view that ␤EST binds the tail of post-M4. However, our observation that substitutions of M4 Leu 617 ablated ␤EST potentiation suggests that Leu 617 may contribute to the ␤EST binding site, locating this site to the top part of M4. This would be in accord with the observation that neither Phe 167 nor Tyr 309 had any effects on ␤EST potentiation. Moreover, a binding site on the top of M4 would place the site nearby to Phe 170 and Ile 626 , the crucial transduction components for this site.
An important finding of this study is that a cavity between the top half of the M3 and M4 of the ␣4 nAChR subunit hosts the potentiating binding site of dFBr. M3 (Phe 312 and Thr 313 ) and M4 (Leu 617 ) residue side chains are predicted to project toward the M3-M4 cavity and reside sufficiently close to one another to all be able to bind dFBr. When their capacity to bind dFBr was annulled by alanine substitution, the sensitivity to dFBr potentiation was either abolished (as for F312A) or drastically reduced (as for L617A or T313A). Moreover, protection assays with MTSET demonstrated that dFBr impeded accessibility of L617C by MTSET, suggesting that dFBr occupies the cavity toward which the side chain of residues Leu 617 , Phe 312 , and Thr 313 orientates. This cavity is a common allosteric site in pLGICs and is the target of a wide variety of allosteric modulators such as general anesthetics in anionic pLGICs (23)(24)(25), small neurosteroids in GABA A receptors (26), and PNU-120596 and LY-2087101 in ␣7 nAChRs (27). Alanine or cysteine substitutions on M3 Phe 316 or M4 Phe 606 also attenuated dFBr potentiation, but their effects on ACh EC 50 suggest that these residues may affect dFBr potentiation indirectly by affecting channel gating. In support of this possibility, it has been reported that Phe 316 , a residue highly conserved in the ␣ subunits of the nAChR family, affects gating in the muscle nAChR (33,34), and M4 is a well established gating modulator domain (8,34).
In addition to Phe 170 and Ile 626 , which we suggest to be transducing elements linked to the binding sites of ␤EST and dFBr, Phe 167 and Tyr 309 were also identified as important components of the transduction of dFBr binding. Individually, mutants F167Y and Y309F drastically attenuated dFBr potentiation, but the double mutant F167Y/Y309F restored potentiation to wild type levels. Furthermore, individual alanine substitutions on these residues affected dFBr potentiation similarly, supporting the possibility of these two residues being involved functionally. Importantly, Phe 167 is predicted to project downwards toward M3 to meet the side chain of Tyr 309 that extends upwards, thus positioning the transduction complex nearby the dFBr site. Furthermore, Phe 167 precedes the critical FPF gating motif. Thus, conceivably, Phe 167 -Tyr 309 interactions could optimize the gating conformations of the Cys loop FPF motif, leading to enhanced receptor function.
Because Tyr 309 is not predicted to project toward Phe 170 , Phe 167 -Tyr 309 interactions are unlikely to be part of the Phe 170 -Ile 626 transducing pathway. Thus, analogously to the gating of the muscle nAChR, in which the ␤1-␤2 and Cys loops appear to act jointly on the M2-M3 linker to gate the ion channel (6), dFBr binding is transduced by two independent pathways that act on the Cys loop to produce receptor potentiation. Compared with ␤EST, dFBr is ϳ7-fold more efficacious, thus suggesting that a functional consequence of Phe 167 -Tyr 309 and Phe 170 -Ile 626 dFBr-dependent interactions is greater potentiating efficacy.
Finally, given that the FPF motif of the Cys loop and M3 are highly conserved in the nAChR subunits that form heteromeric nAChRs (Fig. 4A), the structural element that is pivotal in determining sensitivity to potentiation by dFBr and ␤EST is post-M4. In comparison with other ␣ subunits that form het-eromeric receptors insensitive to dFBr potentiation (e.g. muscle nAChR and ␣3␤2 nAChR), post-M4 in ␣4 is more hydrophobic, suggesting that it could extend more easily toward the hydrophobic ECD/TMD coupling region than less hydrophobic M4 tails. Furthermore, the presence of the signature double proline motif preceding ␣4 post-M4 likely places considerable constraint on the orientation of post-M4, perhaps anchoring post-M4 to the membrane underneath the Cys loop FPF motif, thereby aiding interactions of this domain with the F170 in the Cys loop. In accord with this view, double or individual alanine substitutions on PP severely attenuated (AP or PA) or ablated (AA) dFBr potentiation.
The transduction mechanism presented here for PAM binding sites in the TMD of the ␣4␤2 nAChR provides strong experimental evidence that interactions between the TMD and the Cys loop are critical for allosteric modulation of pLGICs by PAMs. Thus, in addition to its role in coupling agonist binding to channel gating, the Cys loop also couples TMD to the channel gate. The Cys loop is thus a hub that conveys gating signals to the channel gates from both the ECD and the TMD.

Experimental Procedures
Materials-DFBr was purchased from Tocris Chemicals and ␤EST from Sigma-Aldrich. The cationic methanethiolsulfonate reagent [2-(trimethylammonium) ethyl] methanethiosulfonate (MTSET) was purchased from Toronto Chemicals. 100 mM stocks were prepared and stored at Ϫ80°C. Just before use, MTSET stocks were diluted to the appropriate concentration in Ringer's solution and were applied immediately to the oocytes. Xenopus laevis were purchased from Portsmouth University (Portsmouth, UK) or Xenopus-One (Chicago, IL). Ovaries were dissected from the toads using procedures in accordance with the UK Home Office regulations.
Molecular Biology-Human cDNA of the ␣4 and ␤2 subunits were cloned into the expression vector pCI (Promega), whereas the ␣3 nAChR subunit cDNA (kindly provided by Prof. L. Sivilotti from UCL, London, UK) was cloned into pcDNA3.1 (Invitrogen). Site-directed mutagenesis was performed using the QuikChange mutagenesis kit (Stratagene). The full-length sequence of mutant subunit cDNAs was verified by DNA sequencing (BioSource Sequencing and Eurofins-MWG). We present the numbering of the residues in terms of the full length, including the signal sequence. To obtain the position in the mature form, subtract 28 from the number for ␣4, 25 for ␤2, and 31 for ␣3.
Expression of nAChR in Xenopus Oocytes-Stage V and VI Xenopus oocytes were prepared as previously described (10). Wild type or mutant human ␣4 or ␣3 subunit cDNAs were co-injected with ␤2 subunit cDNA into the nuclei of oocytes in a volume of 18.4 nl/oocyte at equal ratios, using a Nanoject Automatic Oocyte Injector (Drummond, Broomall, PA). The total amount of cDNA injected per oocyte was kept constant at 2 ng. After injection, the oocytes were incubated at 18°C for 2-5 days in a modified Barth's solution containing 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.3 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 0.82 mM MgSO 4 , 15 mM HEPES, and 5 mg/liter neomycin (pH 7.6). Equal ratios of ␣4 and ␤2 subunits yield a mixture of two functional stoichiometries: (␣4␤2) 2 ␣4 and (␣4␤2) 2 ␤2. These two receptor types have distinct pharmacological properties, including sensitivity to activation by ACh (20). To test the effects of dFBr in the alternate ␣4␤2 receptors, we expressed fully concatenated receptors. We have shown previously that concatenated ␣4␤2 receptors replicate the properties of their non-linkedcounterparts (21).Concatenatedreceptorswereconstructed as ␣4␤2 receptors as previously described (21). To introduce mutations into the subunits of the concatemers, the mutations were first introduced in free subunits cloned into a modified pCI plasmid (Promega), and following sequence verification by double-stranded DNA sequencing (SourceBioscience, Oxford, UK), the subunits were ligated to the desired position in the concatemer using unique enzyme restriction sites (21). To verify that the mutated subunits were incorporated into the concatenated receptors, following ligation and DNA amplification, the mutated subunit were excised enzymatically from the concatemer for sequence verification by double-stranded DNA sequencing (SourceBioscience, Oxford, UK).
Electrophysiology and Concentration Response Curves-Recordings were performed 2-5 days post-injection, as previously described (10). Briefly, oocytes were placed in a 0.1-ml recording chamber and perfused with modified Ringer's solution (150 mM NaCl, 2.8 mM KCl, 10 mM HEPES, 1.8 mM CaCl 2 , pH 7.2, adjusted with NaOH) at a rate of 15 ml/min. Note that it has been reported that (␣4␤2) 2 ␤2 nAChRs are potentiated by HEPES, possibly by binding a site in the signature ␤2/␤2 interface of this receptor type (39). We have tested the effects of HEPES on the function of (␣4␤2) 2 ␤2 nAChRs and found no effects. Moreover, we compared the sensitivity of (␣4␤2) 2 ␤2 and (␣4␤2) 2 ␣4 nAChRs to ACh, Zn 2ϩ , dFBr, and ␤EST using Ringer solutions buffered by HEPES or phosphate buffer and found no differences (not shown). In accord with these findings, the recently published X-ray structure of the (␣4␤2) 2 ␤2 nAChRs suggests that ligands may not access the ␤2/␤2 interface easily (28). Current responses were obtained by two-electrode voltage-clamp recording at a holding potential of Ϫ60 mV using an oocyte clamp OC-725C amplifier (Warner Instruments) and Labscribe software (iWorx, Dover, NH). Electrodes contained 3 M KCl and had a resistance of Ͻ1 M⍀. ACh, dFBr, and ␤EST were prepared daily in Ringer's solution from frozen stocks (10 mM). All experiments were carried out at room temperature. For wild type and mutant ␣4␤2 or ␣3␤2 nAChR, a 6 -7-point concentration-response curve was generated for ACh alone or with allosteric modulators (dFBr or ␤EST). Peak currents for ACh were normalized to the currents elicited by 1 mM (ACh EC 100 ). Allosteric modulators were co-applied with ACh EC 10 for the receptor under study, and the peak current responses were normalized to the responses elicited by ACh EC 10 alone. The oocytes were superfused with Ringer's solution for 5 min between all drug applications. To eliminate data interpretation or analysis arising from run-up or run-down of current responses over the course of experiments, all oocytes were initially stabilized with ACh EC 100 . The oocytes were discarded if the response to ACh EC 100 varied by more than Ϯ 10%.
Substituted Cysteine Accessibility Method-The accessibility of introduced cysteines to MTSET was determined by exposing the cysteines to a maximal concentration of MTSET (1 mM)