A Novel α2/α4 Subtype-selective Positive Allosteric Modulator of Nicotinic Acetylcholine Receptors Acting from the C-tail of an α Subunit*

Background: Nicotinic acetylcholine receptors (nAChRs) are involved in nicotine addiction and some neurological disorders. Results: A novel positive allosteric modulator potentiates activation through the C-tail of one α4 subunit but requires two α4 to reactivate desensitized nAChRs. Conclusion: Higher occupancy in allosteric sites promotes nAChR opening and alleviates desensitization. Significance: These α4 modulators may be useful for basic and clinical applications. Positive allosteric modulators (PAMs) of nicotinic acetylcholine receptors (nAChR) are important therapeutic candidates as well as valuable research tools. We identified a novel type II PAM, (R)-7-bromo-N-(piperidin-3-yl)benzo[b]thiophene-2-carboxamide (Br-PBTC), which both increases activation and reactivates desensitized nAChRs. This compound increases acetylcholine-evoked responses of α2* and α4* nAChRs but is without effect on α3* or α6* nAChRs (* indicates the presence of other nAChR subunits). Br-BPTC acts from the C-terminal extracellular sequences of α4 subunits, which is also a PAM site for steroid hormone estrogens such as 17β-estradiol. Br-PBTC is much more potent than estrogens. Like 17β-estradiol, the non-steroid Br-PBTC only requires one α4 subunit to potentiate nAChR function, and its potentiation is stronger with more α4 subunits. This feature enables Br-BPTC to potentiate activation of (α4β2)(α6β2)β3 but not (α6β2)2β3 nAChRs. Therefore, this compound is potentially useful in vivo for determining functions of different α6* nAChR subtypes. Besides activation, Br-BPTC affects desensitization of nAChRs induced by sustained exposure to agonists. After minutes of exposure to agonists, Br-PBTC reactivated short term desensitized nAChRs that have at least two α4 subunits but not those with only one. Three α4 subunits were required for Br-BPTC to reactivate long term desensitized nAChRs. These data suggest that higher PAM occupancy promotes channel opening more efficiently and overcomes short and long term desensitization. This C-terminal extracellular domain could be a target for developing subtype or state-selective drugs for nAChRs.


Positive allosteric modulators (PAMs) of nicotinic acetylcholine receptors (nAChR) are important therapeutic candidates as well as valuable research tools. We identified a novel type II PAM, (R)-7-bromo-N-(piperidin-3-yl)benzo[b]thiophene-2carboxamide (Br-PBTC), which both increases activation and reactivates desensitized nAChRs. This compound increases ace-
tylcholine-evoked responses of ␣2* and ␣4* nAChRs but is without effect on ␣3* or ␣6* nAChRs (* indicates the presence of other nAChR subunits). Br-BPTC acts from the C-terminal extracellular sequences of ␣4 subunits, which is also a PAM site for steroid hormone estrogens such as 17␤-estradiol. Br-PBTC is much more potent than estrogens. Like 17␤-estradiol, the non-steroid Br-PBTC only requires one ␣4 subunit to potentiate nAChR function, and its potentiation is stronger with more ␣4 subunits. This feature enables Br-BPTC to potentiate activation of (␣4␤2)(␣6␤2)␤3 but not (␣6␤2) 2 ␤3 nAChRs. Therefore, this compound is potentially useful in vivo for determining functions of different ␣6* nAChR subtypes. Besides activation, Br-BPTC affects desensitization of nAChRs induced by sustained exposure to agonists. After minutes of exposure to agonists, Br-PBTC reactivated short term desensitized nAChRs that have at least two ␣4 subunits but not those with only one. Three ␣4 subunits were required for Br-BPTC to reactivate long term desensitized nAChRs. These data suggest that higher PAM occupancy promotes channel opening more efficiently and overcomes short and long term desensitization. This C-terminal extracellular domain could be a target for developing subtype or state-selective drugs for nAChRs.
Nicotinic acetylcholine receptors (nAChRs) 2 are critical for nicotine addiction and important for several neuropsychiatric disorders (1)(2)(3). They are ligand-gated ion channels formed from five homologous subunits whose subtypes are defined by their subunit composition. There are 12 neuronal types of subunits: ␣2-10 and ␤2-4. Homomeric nAChRs like ␣7 assemble from only ␣7 subunits, whereas heteromeric nAChRs usually require both ␣ and ␤ subunits (4,5). Both homomeric and heteromeric nAChRs form orthosteric agonist binding sites at interfaces between the subunits in the extracellular domain. Recently, various ligands have been identified that activate, inhibit, or potentiate activation of nAChRs from allosteric sites other than the agonist binding sites (4,6,7). These include positive allosteric modulators (PAMs), negative allosteric modulators, and allosteric agonists (6,8,9). These drugs bind to various places in nAChRs, including the extracellular domain, transmembrane domain, and the extracellular C terminus (e.g. C-tail) (6,7). There are interests in developing PAMs because agonists both activate and desensitize nAChRs and because subtype selectivity is hard to achieve with agonists due to similarity between ACh binding in different nAChR subtypes. By contrast, PAMs enhance nAChR function in an activity-dependent manner, potentially modulating the endogenous pattern of signaling rather than constantly activating or desensitizing nAChRs. PAMs also increase the potential for subtype specificity. This is because diversity of PAM binding sites in nAChRs provides better chances to develop selective therapeutics than does targeting the relatively similar ACh binding sites.
Based on pharmacology, there are two types of PAMs (10). Type I PAMs increase peak responses. Type II PAMs not only increase peak responses but also the duration of channel opening by delaying desensitization. This makes type II PAMs espe-* This work was supported by National Institute on Drug Abuse Grant DA030929. The authors declare that they have no conflicts of interest with the contents of this article. 1  cially efficacious. In some cases they can act as allosteric agonists (8). Understanding the pharmacology and potentiation mechanism of PAMs should facilitate design of more potent and selective modulators. There is no direct correlation between where a PAM binds and which type of PAM it is (6). Some PAMs bind in the transmembrane domain near the gate for the cation channel whose opening they influence (6). These transmembrane PAMs can be either type I or type II. Here we describe a novel type II PAM, Br-PBTC, 3 which acts from the C-tail of the ␣4 subunit. Discovering how this site, which is distant from both agonist binding sites and the channel gate, influences activation and desensitization should provide new insights on the structure and function of nAChRs.
Higher occupancy of agonist sites increases activation and speed of desensitization of both heteromeric and homomeric nAChRs (11)(12)(13). Knowledge of how binding of PAMs affects activation and desensitization is limited. Some estrogens act as PAMs through the C-tail of ␣4 (14). Their PAM effect increases with the number of ␣4 subunits with free C-tails in a nAChR (15). By contrast, Br-PBTC potentiates ␣4 concatemers and free ␣4 subunits. Using this novel PAM and various concatemers, we investigated how PAM site occupancy influences activation and desensitization of nAChRs expressed in Xenopus oocytes and mammalian cell lines. We found that occupying one PAM site is sufficient to potentiate nAChR activation, and higher PAM site occupancy promotes nAChR opening and alleviates short and long term desensitization more efficiently. This C-tail potentiation mechanism might be applicable to other nAChR subtypes and facilitate development of other subtypeselective drugs.

Experimental Procedures
Chemicals-Methodology for preparing reactants for synthesizing Br-PBTC is described as follows.
Ethyl 7-Bromobenzo[b]thiophene-2-carboxylate-3-Bromo-2-fluorobenzaldehyde (406.0 mg, 2.0 mmol), ethyl mercaptoacetate (242 l, 2.2 mmol), triethylamine (556 l, 4.0 mmol), and acetonitrile (10 ml) were added to a 50-ml round-bottom flask and stirred at 60°C overnight. The acetonitrile was removed in vacuo, and the residue was dissolved in ethyl acetate (30 ml) and water (10 ml). The layers were separated, and the aqueous layer was extracted with ethyl acetate (2ϫ). The combined organics were dried (MgSO 4 ) and concentrated to give the title compound (538 mg, 95%). 1 (6 ml), and water (8 ml) were added to a 50-ml round-bottom flask and stirred at room temperature until the starting material was consumed as judged by thin layer chromatography analysis. The majority of the tetrahydrofuran was removed in vacuo. The resulting crude mixture was acidified with aqueous hydrochlo-ric acid (ϳpH ϭ 3) and cooled in an ice bath. The solids were filtered and washed with cold water (about 6 ml) to give the title compound (392 mg, 90%).
Four of the five chimeras of ␣3 and ␣4 subunits were prepared previously (23). Chimeras were numbered according to the amino acid sequences of the mature subunit. ␣3 (1-440) / ␣4 (561-594) was prepared by ligating three pieces of DNA: a 0.6-kb fragment from the NcoI to BstEII site of the ␣3 subunit, a 1-kb fragment from the HidIII to BstEII site of the ␣3 subunit, and a 3.1-kb fragment from the NcoI to HidIII site of the ␣4 subunit in the pSP64 vector. The ligation mixture was transformed into XL10-Gold ultracompetent cells (Stratagene, La Jolla, CA), and the right clone was chosen from a restriction enzyme digest.
A C-tail mutant (noted as ␣4 AAC ) was obtained by mutating the last four amino acids of the ␣4 subunit, alanine-glycinemethionine-isoleucine, to alanine-alanine-cysteine followed by a stop codon. Mutations were introduced using the PfuUltra high-fidelity DNA polymerase (Agilent, Santa Clara, CA) following the manufacturer's instructions. All mutations were confirmed by sequencing.
After linearization and purification of cDNAs, cRNA transcripts were prepared in vitro using mMessage mMachine kits (Ambion, Austin, TX). Concentrations of cDNAs and cRNAs were calculated by spectrophotometry.
Cell Culture and Transfection-All cells were maintained as described previously (17). The human embryonic kidney tsA201 (HEK) cell lines stably expressing human ␣4␤2, ␣4␤4, ␣2␤2, ␣2␤4, ␣3␤2, and ␣3␤4 were described (13). The ␣4␤2 cell line expresses a mixture of (␣4␤2) 2 ␣4 and (␣4␤2) 2 ␤2 nAChRs (24). HEK cells that express only one stoichiometry, either (␣4␤2) 2 ␣4 or (␣4␤2) 2 ␤2, were obtained by transfecting a dimeric concatemer ␤2(QAP) n ␣4 cell line with ␣4 or ␤2 subunits. 4 FLEXstation Experiments-For functional tests of nAChRs expressed in HEK cells, we used a FLEXstation (Molecular Devices, Sunnyvale, CA) bench-top scanning fluorometer as described by Kuryatov et al. (25). To increase the expression level of ␣2␤3, ␣3␤2, and (␣4␤2) 2 ␤2 nAChRs, the plates were incubated at 29°C for 20 h before being tested. A membrane potential fluorescent indicator kit (Molecular Devices) was used according to the manufacturer's protocols. In PAM experiments, serial dilutions of Br-PBTC were manually added to the assay plate 15 min before the addition of agonists during recording unless otherwise noted. In short term desensitization experiments, 6 min after agonists were added to cell culture wells, Br-PBTC or dihydro-␤-erythroidine hydrobromide (DH␤E) was automatically added into the wells during recording. In long term desensitization experiments, nicotine or DH␤E was incubated with cells for 6 h before recording. Br-PBTC with or without DH␤E was added to the cell culture wells during recording. Each data point was averaged from three to four responses from separate wells. The potency and maximum efficacy of drugs were calculated by fitting the Hill equation to the concentration/response relationship using a nonlinear least squares curve fitting method (Kaleidagraph; Abelbeck/Synergy, Reading, PA): is the peak current measured at the drug concentration x, I max is the maximum current peak at the saturating concentration, EC 50 is the drug concentration required to achieve half of the maximum response, and n H is the Hill coefficient.
Oocyte injections were performed within 48 h after surgery. Oocytes were injected with 20 -40 ng of concatemer cRNA and free single subunit at a 1:1 ratio. A total of 2-20 ng of cRNA was injected for free wild type or chimeric ␣ and ␤ subunits at a 4:1 ratio to force expression of the (␣) 3 (␤2) 2 stoichiometry or at 1:4 ratio to force expression of the (␣) 2 (␤) 3 stoichiometry. To express homomeric ␣7 nAChRs, 70 ng of cRNA was injected to each oocyte. The function was assayed 3-7 days after injection.
Electrophysiology-Currents in oocytes were measured using the OpusXpress 6000A (Molecular Devices, Union City, CA), an automated two-electrode voltage clamp amplifier that enables recording up to eight oocytes in parallel (13). Oocytes were voltage-clamped at a holding potential of Ϫ50 mV. 200 l of drugs were delivered on top of oocytes for 4 s (s) through the sidewall of the bath to minimize disturbance to oocytes. Between drug applications, oocytes received a 30-s prewash and 223-s post-wash of ND96 solution (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.6) with 0.5 M atropine perfused through the bath at a rate of 3 ml/min unless otherwise noted.
Peak amplitudes of experimental responses were calculated relative to ACh responses to normalize the data and compensate for variable expression levels among oocytes. The PAM effect of Br-PBTC was calculated by comparing increased responses with Br-PBTC relative to responses to ACh alone. Mean and S.E. were calculated from normalized responses. Statistical analyses were performed using Student's t test. More than four oocytes were tested for each experiment.
Pre-application of Br-PBTC gave slightly higher PAM effects on wild type and chimeric nAChRs than co-application with agonists. But conclusions were the same for both application methods. To save time, we thereafter used the co-application method to evaluate the PAM effects in experiments performed in oocytes. In short term desensitization experiments, 1 mM ACh was applied to oocytes for 4 s at the rate of 3 ml/min followed by another 56 s at 0.75 ml/min. Then the oocytes were incubated for an additional 5 min in a static bath before a coapplication with 1 mM ACh plus 3 M Br-PBTC for 4 s at 3 ml/min. Control experiments were performed on the same oocytes before Br-PBTC applications following the same protocol in which 3 M Br-PBTC was replaced with 0.1% (v/v) DMSO. Reactivation by Br-PBTC was calculated by normalizing the response of 3 M Br-PBTC to the response of ACh applied before Br-PBTC. In antagonist inhibition experiments, ␣-conotoxin MII was applied for 4 s at the rate of 3 ml/min followed by another 56 s at 0.75 ml/min. To fully block ACh activation, the oocytes were then incubated for an additional 16 min in a static bath before co-application of ACh (3 M) or ACh together with Br-PBTC (3 M).

Results
Br-PBTC Selectively Affects ␣2 and ␣4 Subunits-To investigate the subtype selectivity of compound Br-PBTC (Fig. 1A), we tested its ability to potentiate ACh activation of various subtypes of nAChRs stably expressed in HEK cells (Fig. 1B). Br-PBTC increased the activation by EC 20 ACh of ␣2and ␣4-containing nAChRs by 119 -560% (Table 1). EC 50 values for Br-PBTC ranged from 0.261 to 0.660 M (Table 1), equal to the most potent nAChR PAMs (6,7). At Ͼ3 M, Br-PBTC inhibited its own potentiation effect, perhaps because it behaved as an open channel blocker like some other nAChR PAMs and ACh itself (26). Br-PBTC did not alter activation by ACh of ␣3␤2 or ␣3␤4 nAChRs (Fig. 1B). We also evaluated the effect of Br-PBTC (up to 3 M) on activation on homomeric ␣7 nAChRs expressed in Xenopus oocytes. The maximum increased peak response by Br-PBTC was 33.9 Ϯ 33.8%. Because the potentiation is small with large error and only appeared at 1 M Br-PBTC, we do not think that Br-PBTC potentiates activation of ␣7 nAChRs. Br-PBTC did not activate these heteromeric or homomeric nAChR subtypes by itself (data not shown). Therefore, Br-PBTC is an ␣2 and ␣4 nAChR subtype-selective PAM.
In summary, Br-PBTC is an ␣2/␣4 subtype-selective PAM. It increases potencies of nAChRs with three ␣4 subunits and efficacies of nAChRs with two or three ␣4 subunits. Its potentiation effect is larger on ␤2-containing than ␤4-containing nAChRs. The following pharmacology study mainly focused on ␣4␤2* nAChRs, which are the most prevalent nAChRs in brain.
Br-PBTC Potentiates nAChRs through a Single ␣4 Subunit-PAM efficacy of 17␤-estradiol increases with more free ␣4 C-tails in a nAChR (15). Because Br-PBTC also binds to this C-tail site, we investigated the effect of the number of ␣4 subunits on the potentiation profile of Br-PBTC. Br-PBTC does not affect functions of ␣3* nAChRs; thus, we expressed the free ␣3 subunit with various concatemers of ␣4 and ␤2 subunits in Xenopus oocytes to decrease the numbers of Br-PBTC potentiation sites in a nAChR (Fig. 6A). Another benefit of using ␣3 to replace the ␣4 subunit is that then these nAChRs have the same numbers of agonist binding sites. They all have at least one high ACh affinity ␣4/␤2 site. A low affinity ACh site can be formed at ␣4/␣4 and ␣3/␣4 interfaces (13, 27, 28). (␣3␤4) 2 ␣3 nAChRs showed lower ACh sensitivity than (␣3␤4) 2 ␣4 nAChRs (30). Therefore, a low affinity ACh site is likely to be present at the ␣3/␣3 interface.
We tested the effect of 3 M Br-PBTC on peak currents evoked by ACh, a feature shared by both type I and type II PAMs (Fig. 6B). This concentration is enough to evoke a max-

TABLE 2 Effect of 3 M Br-PBTC on potencies and efficacies of ACh to activate nAChRs
ACh concentration/response curves were determined on oocytes or HEK cell lines expressing defined stoichiometries. The maximum efficacy was defined as 100% for ACh without PAMs. In oocytes, defined stoichiometries were obtained by injecting ␤2-␣4-␤2-␣4 concatemers with a free subunit. Each data point was collected from more than four oocytes or more than three wells of cells. The (␣4␤2) 2 ␣4 cell line exhibits a two-component concentration/response curve due to a high sensitivity component reflecting its two ␣4/␤2 ACh binding sites and a low affinity component reflecting activation in combination with the low sensitivity ␣4/␣4 site. The concentration/response data for (␣4␤2) 2 ␣4 obtained from oocytes were too noisy to fit a biphasic curve, so were approximated with a monophasic curve. imal PAM effect for nAChRs containing two or three ␣4 subunits (Fig. 3A). Br-PBTC potentiated activation of (␣4␤2) 2 ␣4 nAChRs expressed from the ␤2-␣4-␤2-␣4 concatemer and free ␣4 similarly to those expressed from only free subunits (Figs. 5A and 6). This is different from estrogens, which do not potentiate concatemers in which the ␣4 subunit C-tail is linked to another subunit such as ␣4-␤2 (29,31). Therefore, the concatemer linker has no effect on potentiation by Br-PBTC. Br-PBTC increased activation of nAChRs by both medium (100 M) and maximal (3000 M) concentrations of ACh as long as there was more than one ␣4 subunit (Fig. 6B). This is consistent with the property of 17-␤ estradiol, another known C-tail binding PAM (15). Moreover, the potentiation effect of (␣4␤2) 2 ␣3 nAChRs with two ␣4 subunits was larger, and the effect on (␣4␤2) 2 ␣4 with three ␣4 subunits was the largest. These data suggest that higher PAM occupancy increases the efficiency of channel opening. Br-PBTC can increase channel activation by a maximal concentration of ACh. This is similar to what was observed with ␣4␤2 nAChRs expressed in HEK cells (Fig. 3B). At higher concentrations of agonists, nAChRs desensitize more rapidly. The potentiation by Br-PBTC on 3000 M ACh could be due to increasing channel conductance or increased open state probability or destabilizing or slowing entry into the desensitized state.
This method is not as sensitive to the kinetics of channel function as the two-electrode voltage clamp method performed on oocytes. Therefore, we did not observe a higher PAM effect on desensitized (␣4␤2) 2 ␣4 than (␣4␤2) 2 ␤2 (Fig. 8), as we expected from oocyte experiments (Fig. 7). Another factor contributing to this discrepancy is that because (␣4␤2) 2 ␤2 desensitizes slower than (␣4␤2) 2 ␣4, more (␣4␤2) 2 ␤2 nAChRs were still in the open state (represented by the portion blocked by DH␤E applied alone in Fig. 8) when Br-PBTC was applied. This was not the case for the experiments we performed on ␣4* nAChRs expressed in oocytes (Fig. 7).
Unlike ACh, which is quickly hydrolyzed by esterase, nicotine can persist in brains for hours (32). This causes long term desensitization of nAChRs in smokers. Because Br-PBTC can reactivate short term desensitized nAChRs with more than two ␣4 subunits (Figs. 7 and 8), we studied its effect on nAChRs expressed in HEK cells after 6 h of exposure to 0.5 M nicotine (Fig. 9). This concentration of nicotine is found in smokers (32). After 6 h with nicotine, nAChRs were all desensitized because application of DH␤E to these nAChRs showed no blockage of activation (black traces in Fig. 9, A and B). Interestingly, Br-PBTC (4 M) efficiently reactivated nicotine long term desensitized (␣4␤2) 2 ␣4 nAChRs but only weakly reactivated desensitized (␣4␤2) 2 ␤2 nAChRs (Fig. 9, A and B). The desensitized (␣4␤2) 2 ␤2 could be less sensitive to reactivation by Br-PBTC. Therefore, we determined the dependence on Br-BPTC concentration of reactivation of desensitized nAChRs (Fig. 9C).  To desensitize nAChRs, ACh (1000 M) was applied to oocytes for 6 min before its co-application with Br-PBTC (3 M). Each data point was collected from more than five oocytes. A, Br-PBTC requires two or more ␣4 subunits to reactivate short term desensitized nAChRs. The efficacy of reactivation increases with more ␣4 subunits in a nAChR. B, response kinetics from representative oocytes. Results are the mean Ϯ S.E. (error bars).
Competitive Antagonists Block Potentiation by Br-PBTC-Competitive antagonists block activation by agonists because they bind to the same sites as agonists but do not activate nAChRs. We investigated whether competitive antagonists affect potentiation of the allosteric ligand Br-BPTC. One of the important native nAChR subtypes that contain only one potential Br-PBTC site is (␣6␤2)(␣4␤2)␤3 (33). This is the subtype that regulates nicotine addiction because knock-out of ␣4, ␣6, or ␤2 abolished nicotine self-administration in rodents (34). Br-PBTC did not potentiate activation of (␣6␤2) 2 ␤3 expressed in oocytes (data not shown), but it increased ACh (3 M) activation of (␣6␤2)(␣4␤2)␤3 by 99.0 Ϯ 13.6% (representative kinetics shown in Fig. 10A). This is consistent with the finding in Fig. 5 that only one ␣4 subunit is required for Br-PBTC potentiation. One feature of (␣6␤2)(␣4␤2)␤3 is that the competitive antagonist ␣-conotoxin MII selectively blocks its activation from the ␣6/␤2 interface. This antagonist site is far away from the ␣4 C-tail where Br-PBTC acts. ␣-Conotoxin MII (50 nM) completely blocked activation by ACh and potentiation by Br-PBTC (Fig. 10B). This is consistent with the idea that activation is a cooperative event involving conformational change in the whole nAChR, and antagonist inhibition of any one ACh site is sufficient to prevent activation (35,36). Blockage by competitive antagonists also applies to potentiation of Br-PBTC on other ␣4* nAChRs. The competitive antagonist DH␤E selective for ␤2 nAChRs blocked activation of HEK cell lines expressing (␣4␤2) 2 ␤2 and (␣4␤2) 2 ␣4 nAChRs in the presence of Br-PBTC (Fig. 11). DH␤E also inhibited reactivation of both short term and long term desensitized nAChRs by Br-PBTC (gray traces in Figs. 8 and 9).

Discussion
Developing subtype-selective therapeutics is challenging for nAChRs. Efforts have been made to develop allosteric modulators binding to non-conserved regions to achieve subtype selectivity (6,7). Only the human ␣2 and ␣4 subunits share similar sequences at the C-tail. Other human subunits differ in length and the amino acid sequences in this region. This makes the C-tail a promising target for selective PAMs. Here we showed that, besides steroids, non-steroid structures could act from the ␣4 C-tail as PAMs and exhibit submicromolar affinity. Engineering the ␣4 C-tail onto ␤2 subunits enabled estrogens to potentiate through this mutant ␤2 subunit (15). A suitable  PAM to bind the C-tail of ␤2 and interact with the end of its M4 might produce a ␤2-selective effect. Perhaps in this way PAMs could be found that would be selective for any subunit. These ligands might behave similarly to type II PAMs like Br-PBTC, but they might also be negative allosteric modulators or allosteric agonists, depending on their structures. There is not clear guidance for how to design or select such ligands, but 17␤estradiol and Br-PBTC illustrate examples of structurally different compounds with similar PAM properties. Suitable selection approaches using stoichiometry-specific nAChR cell lines might allow for the discovery of PAMs, negative allosteric modulators, and allosteric agonists for many nAChR subunits that would be useful tools for studying nAChRs and as drugs both in vitro and in vivo.
The C-tail PAM site is stereoselective. Neither the enantiomer of Br-PBTC 5 nor estrogens (14) potentiate ␣4* nAChRs. Stereoselectivity suggests that the PAM and the C-tail of the ␣4 subunit interact with protein rather than membrane lipid. PAM bound to the short ␣4 C-tail must interact stereospecifically with a nearby region, probably on the same subunit, which is capable of influencing the channel gate. There are prolines at the extracellular end of the M4 transmembrane domains. These prolines may contribute to a stereoselective site that interacts with the PAM bound to the C-tail to mediate the PAM effects. Several other important subunit structural elements are close to this stereoselective site: the end of ␤10 strand to M1, the cys-loop, M2-M3 loop, ␤1-␤2 loop, and ␤8-␤9 loop. Movement of these structural elements contributes to passing the conformational changes from binding of the agonist in the extracellu-lar domain to opening of the transmembrane channel pore (37,38). Ivermectin, a compound acting directly on these structural elements, is a PAM on ␣7 (6) and an allosteric agonist on glutamate-gated chloride channels where its binding site has been localized in receptor crystals near this region (38). A lipid molecule binds competitively to the same region as ivermectin (39). Both ivermectin and this lipid induce channel pore opening but through slightly different conformation changes. These suggest that the transmembrane domains are quite flexible. The subtleties involved in channel opening are small (39); thus, pulling or pushing a bit on the extracellular end of M4 might be enough to mediate action of a PAM.
Although Br-PBTC acts from ␣ subunits, ␤ subunits also play a role. Br-PBTC has greater effects on nAChRs with ␤2 subunits over those with ␤4 subunits (Fig. 2). It is not clear whether this is because of the greater sensitivity to activation of ␤2* nAChRs or because ␤ subunits directly contribute to the Br-PBTC PAM effect.
It is not evident how a PAM binding to an ␣4 C-tail affects activation or desensitization. Estrogens do not potentiate ␣4 subunits whose C-terminal end is linked into another subunit such as the concatemer ␣4-␤2 (29). However, Br-PBTC potentiated the C-tail-linked ␣4 subunit equally efficiently as the free subunit (Figs. 5 and 6). This allowed us to express various concatemers with a free ␣3 subunit to reduce binding sites for Br-PBTC (Fig. 6). We cannot rule out the possibility that Br-BPTC can bind to the ␣3 C-tail but cannot potentiate nAChR activation. Occupancy by agonists affects nAChR activation and desensitization (11,12). Using an ␣3 subunit to replace ␣4 maintained the number of agonist sites among nAChRs. Therefore, Br-PBTC is a better tool than estrogens to study the relationship between occupancy and potentiation of PAMs acting at the C-tail. Moreover, given the activity of estrogens at nuclear receptors, Br-PBTC would be a better ligand to study effects of nAChRs in vivo. That Br-PBTC potentiated linked ␣4 C-tails in concatemers indicates that a free tip of the C-tail is not required for potentiation from this site. The linker in the ␣4-␤2 concatemer might have prevented the entrance of estrogens into the C-tail site.
A previous study used concatemers to achieve different numbers of ␣4 subunits with free C-tails and showed that more ␣4 subunits increased potentiation efficacy by estrogens (15). Using Br-PBTC, we confirmed and extended this C-tail potentiation mechanism. Upon agonist binding, nAChRs go through various conformational changes from the resting state (R) to the open state (O) and or desensitized state (D) (Fig. 12A). There are different types of desensitized states (6,40,41). Some have lower energy barriers and are favored soon after ligand binding, i.e. short term desensitization (D S ). Some have lower energy levels and are preferred after long term incubation with agonists (D L ). When an antagonist binds, nAChRs go into an inactive state (I) or are forced to remain in a resting state that prevents activation. When a PAM binds to the C-tail of ␣4, it increases the probability of channel opening (42). The increase of channel open probability only requires one C-tail site, and its extent is proportionate to the number of C-tail PAM sites in a nAChR (Figs. 6B and 12B) (15). PAMs reactivate short term desensitized nAChRs from the C-tail also in an occupancy-de- pendent manner but require occupying two or more C-tail sites (Figs. 7). This suggests that Br-PBTC increases exit rates from the D S state to the O state, thus destabilizing the D S state (Fig.   12B). Occupying three C-tail sites is required to efficiently reactivate long term desensitized nAChRs (Figs. 9 and 11B). The D L state is favored over time because it has the lowest energy level. Br-PBTC likely needs to bind to three ␣4 subunits to initiate sufficient conformational change to compensate for the energy loss from leaving the D L state. The cooperative effect of Br-PBTC binding to three sites also enables Br-PBTC to increase agonist sensitivity of (␣4␤2) 2 ␣4 nAChRs. PAMs at the ␣4 C-tail cannot activate antagonist-bound nAChRs, and antagonists block their potentiation (Figs. 8 -11). This is consistent with the concerted conformational change model for activation, i.e. any one ACh site being held in a resting conformation through an antagonist blocking closing of its C-loop prevents activation (43).
The unique potentiation profile of Br-PBTC makes it a good research tool for differentiating nAChR subtypes. The nAChR subtype expression pattern differs between brain areas (5,33). The ␣6-selective antagonist ␣-conotoxin MII helps distinguish ␣6 and non-␣6 nAChRs. Br-PBTC selectively potentiates ␣6␣4* nAChRs. This differentiates them from ␣6(non␣4) nAChRs. In combination with ␣-conotoxin MII, Br-PBTC can further distinguish ␣4␣6* from ␣4(non␣6) nAChRs. Because  When an antagonist binds to nAChRs, nAChRs go into an inactive state (I) or is held in a resting state that prevents further activation by agonists. nAChRs may pass through various transitional states, which are not displayed in the figure. B, hypothetical PAM effects on probability of nAChR states. * indicates that the position can be occupied either by an ␣ or a ␤ subunit. An agonist site can form at the ␣/␣ and ␣/␤ subunit interface but not the ␤/␣ subunit interface. Therefore, a question mark for agonist binding is annotated at those undefined interfaces. Higher PAM occupancy increases the probability of nAChRs being in the open state and decreases the probability in the D S or D L states. Therefore, ␣4-selective PAMs showed the greatest potentiation effect on (␣4␤2) 2 ␣4 nAChRs with three ␣4 subunits.
In summary, using the novel type II PAM Br-PBTC, we learned more about potentiation from the C-tail PAM site. We found that activation and reactivation increase with higher PAM occupancy at the C-tail site. It remains to be determined in vivo how chronic exposure to ACh or agonist drugs in the presence of Br-PBTC will influence smoldering activation of nAChRs largely desensitized by agonists. It also remains to be determined how ligands bound to the ␣4 C-tail interact with the channel to influence its opening, whether negative allosteric modulators or allosteric agonists can act from this site, whether similar sites can be found on other subunits, and whether ligands for them will prove to be useful drugs.
Author Contributions-J. W. and J. L. designed the study and wrote the paper. J. W. and A. K. designed and constructed plasmids and cell lines. J. W. and J. N. performed the FlexStation and electrophysiology assays. Z. J. and T. M. K. provided the chemical tools. P. J. K. contributed to discussions of the in vivo use of the PAM. P. J. K., J. L., and T. M. K. acquired funding to support this study. All authors analyzed data, revised, and approved the final version of the manuscript.