An Accessory Agonist Binding Site Promotes Activation of α4β2* Nicotinic Acetylcholine Receptors*

Background: In (α4β2)2α4 nicotinic acetylcholine receptors, there is an agonist binding site at the α4/α4 subunit interface. Results: α2, α3, and α6 accessory subunits can form an agonist site with α4. These promote activation upon agonist binding at α4/β2 agonist sites. Conclusion: Accessory subunit agonist sites greatly influence receptor function. Significance: These sites are promising drug targets. Neuronal nicotinic acetylcholine receptors containing α4, β2, and sometimes other subunits (α4β2* nAChRs) regulate addictive and other behavioral effects of nicotine. These nAChRs exist in several stoichiometries, typically with two high affinity acetylcholine (ACh) binding sites at the interface of α4 and β2 subunits and a fifth accessory subunit. A third low affinity ACh binding site is formed when this accessory subunit is α4 but not if it is β2. Agonists selective for the accessory ACh site, such as 3-[3-(3-pyridyl)-1,2,4-oxadiazol-5-yl]benzonitrile (NS9283), cannot alone activate a nAChR but can facilitate more efficient activation in combination with agonists at the canonical α4β2 sites. We therefore suggest categorizing agonists according to their site selectivity. NS9283 binds to the accessory ACh binding site; thus it is termed an accessory site-selective agonist. We expressed (α4β2)2 concatamers in Xenopus oocytes with free accessory subunits to obtain defined nAChR stoichiometries and α4/accessory subunit interfaces. We show that α2, α3, α4, and α6 accessory subunits can form binding sites for ACh and NS9283 at interfaces with α4 subunits, but β2 and β4 accessory subunits cannot. To permit selective blockage of the accessory site, α4 threonine 126 located on the minus side of α4 that contributes to the accessory site, but not the α4β2 sites, was mutated to cysteine. Alkylation of this cysteine with a thioreactive reagent blocked activity of ACh and NS9283 at the accessory site. Accessory agonist binding sites are promising drug targets.

Neuronal nicotinic acetylcholine receptors containing ␣4, ␤2, and sometimes other subunits (␣4␤2* nAChRs) regulate addictive and other behavioral effects of nicotine. These nAChRs exist in several stoichiometries, typically with two high affinity acetylcholine (ACh) binding sites at the interface of ␣4 and ␤2 subunits and a fifth accessory subunit. A third low affinity ACh binding site is formed when this accessory subunit is ␣4 but not if it is ␤2. Agonists selective for the accessory ACh site, such as 3-[3-(3-pyridyl)-1,2,4-oxadiazol-5-yl]benzonitrile (NS9283), cannot alone activate a nAChR but can facilitate more efficient activation in combination with agonists at the canonical ␣4␤2 sites. We therefore suggest categorizing agonists according to their site selectivity. NS9283 binds to the accessory ACh binding site; thus it is termed an accessory site-selective agonist. We expressed (␣4␤2) 2 concatamers in Xenopus oocytes with free accessory subunits to obtain defined nAChR stoichiometries and ␣4/accessory subunit interfaces. We show that ␣2, ␣3, ␣4, and ␣6 accessory subunits can form binding sites for ACh and NS9283 at interfaces with ␣4 subunits, but ␤2 and ␤4 accessory subunits cannot. To permit selective blockage of the accessory site, ␣4 threonine 126 located on the minus side of ␣4 that contributes to the accessory site, but not the ␣4␤2 sites, was mutated to cysteine. Alkylation of this cysteine with a thioreactive reagent blocked activity of ACh and NS9283 at the accessory site. Accessory agonist binding sites are promising drug targets. nAChRs 2 contain five homologous subunits organized to form a central cation channel whose opening is gated by the binding of ACh. Homomeric ␣7 nAChRs have five ␣7 subunits and five ACh binding sites at their extracellular interfaces (1). Heteromeric ␣4␤2 nAChRs assemble into (␣4␤2) 2 ␣4 or (␣4␤2) 2 ␤2 stoichiometries with three or two ACh binding sites (2,3). Binding of ACh to one site is able to activate, inefficiently, both homomeric and heteromeric nAChRs (4,5). Binding to two sites in an ␣7 nAChR is more efficient, and binding to three is most efficient for activation. By contrast, binding to four or five sites in ␣7 nAChRs promotes desensitization more rapidly than activation (6). In ␣4␤2 nAChRs, ACh can bind at the interface between ␣4 and ␤2 subunits (abbreviated as ␣4/␤2) where the ␣4 subunit on the left forms the plus face of an agonist site and the ␤2 subunit on the right forms the minus face. In the (␣4␤2) 2 ␣4 stoichiometry, a third low affinity ACh site is present at the ␣4/␣4 interface (2,3). This results in 4-fold larger responses evoked by ACh of (␣4␤2) 2 ␣4 stoichiometry than the (␣4␤2) 2 ␤2 stoichiometry. Because the high affinity ␣4/␤2 agonist sites contribute less than 25% of the total response of (␣4␤2) 2 ␣4 nAChRs, these nAChRs appear to be low affinity if the high affinity component is not clearly resolved (7,8). Actually, (␣4␤2) 2 ␣4 nAChRs can be activated by low concentrations of agonists from their intrinsic ␣4/␤2 sites to the same extent as the (␣4␤2) 2 ␤2 nAChRs (2).
NS9283 is representative of a new class of selective agents targeting ␣4␤2 nAChRs that have proven useful in aiding nAChR agonists in reducing neuropathic pain and improving cognition (9 -15). It has been termed a positive allosteric modulator (PAM) because it cannot activate nAChRs by itself but enhances ␣4␤2 nAChR activity in response to agonist stimulation (9,10,12). However, it was recently established that NS9283 is neither allosteric nor a modulator. It is not allosteric because NS9283 acts as a selective agonist at the ACh binding site formed at the ␣4/␣4 interface (16,17). NS9283 cannot activate through its action on the ␣4/␣4 site alone. In combination with agonists at ␣4/␤2 sites, it produces the high probability of channel opening resulting from increased binding site occupancy (12,17). Because NS9283 achieves its effect by occupying a third ACh site, just as any other full agonist would at that site, it is not a modulator. NS9283 does not exceed the maximum activation efficiency of (␣4␤2) 2 ␣4 nAChRs by ACh (18). As expected, NS9283 is without effect on (␣4␤2) 2 ␤2 nAChRs for lack of the ␣4/␣4 binding site (17,19).
To avoid misleading nomenclature and provide nomenclature that reflects the mechanism of action, we suggest designating sazetidine and NS9283 as ACh binding site site-selective agonists (SSAgs). There may also be site-selective antagonists. Based on specific site selectivity, SSAgs can be further divided into two groups: accessory SSAgs, such as NS9283, and primary SSAgs, such as sazetidine. Here we confirm and extend what is known about accessory site activation of ␣4␤2* nAChRs (which contain ␣4, ␤2, and possibly other subunits).
Because ␣4 and ␤2 subunits can form conventional ACh binding sites, they are referred to as structural subunits. The fifth subunit of ␣4␤2* nAChRs, which may or may not form an ACh site depending on the subunit, is referred to as an accessory subunit. Some subunits like ␣4 and ␤2 can function as both structural and accessory subunits, whereas others like ␣5 and ␤3 usually function only as accessory subunits. Here we investigate the ability of (␣4␤2) 2 * nAChRs to form ACh and NS9283 binding sites with the accessory subunits ␣2, ␣3, ␣4, ␣6, ␤2, and ␤4. Functional impairment by blocking accessory sites suggests that accessory agonist sites exist and that they promote channel activation from the interface between several ␣ subunits and ␣4.

Experimental Procedures
Chemicals-2-Aminoethyl methanethiosulfonate (MTSEA) was purchased from Toronto Research Chemicals Inc. (North York, Ontario, Canada). NS9283 was synthesized as described previously (27). A 10 mM stock of NS9283 was prepared in dimethyl sulfoxide. Dilutions of NS9283 and MTSEA were prepared daily in testing buffer before use. All other chemicals were purchased from Sigma-Aldrich unless otherwise noted.
cDNAs and cRNAs-Human ␣3, ␣4, ␣6, ␤2, and ␤4 were cloned in this laboratory (28 -31). The human ␣2 subunit was obtained from OriGene Technologies, Inc. (Rockville, MD). The ␣2 sequence was cut out with the restriction enzymes SmaI and XhoI to shorten the untranslated region and improve functional expression. The 2.0-kb DNA fragment coding for ␣2 was subcloned into the pSP64 vector for RNA preparation or into pcDNA3.1/Zeo(ϩ) (Invitrogen) for human cell transfection.
After linearization and purification of cDNAs, RNA transcripts were prepared in vitro using mMessage mMachine kits (Ambion, Austin, TX). Concentrations of cDNAs and cRNAs were calculated by spectrophotometry.
Oocyte Removal and Injection-Oocytes were removed surgically from Xenopus laevis as described except that a higher concentration (0.26 mg/ml) of collagenase type IA (Sigma) was used to obtain oocytes optimal for the automated voltage clamp instrument described later.
Electrophysiology-Currents in oocytes were measured using a manual two-electrode voltage clamp amplifier setup (oocyte clamp OC-725, Warner Instrument, Hamden, CT) or Opus-Xpress 6000A (Molecular Devices, Sunnyvale, CA) (30,35). OpusXpress is an integrated system that provides automated impalement, voltage clamp, and drug delivery for up to eight oocytes in parallel (35). Electrodes were filled with 3 M KCl and had resistances of 0.5-10 megaohms for the voltage electrode and 0.5-3 megaohms for the current electrode. Oocytes were voltage-clamped at a holding potential of Ϫ50 mV. Data were collected and filtered at 50 Hz. 200 l of drugs were delivered on top of oocytes for 4 s through the sidewall of the bath to minimize disturbance. Between drug applications, oocytes received a 30-s prewash and 223-s postwash with ND96 solution (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.6) plus 0.5 M atropine perfused through the bath at a rate of 3 ml/min unless otherwise noted.
In concentration/response experiments, each oocyte received two initial control applications of ACh (300 M for concatamer expressed with free ␣ subunits and 30 M for concatamer expressed with free ␤ subunits) followed by application of various concentrations of ACh (from low to high). In MTSEA experiments, each oocyte received 2 mM MTSEA for 60 s at the rate of 0.9 ml/min and then was incubated with MTSEA for an additional 5 min while the ND96 wash was stopped to retain the reagent in the bath before being washed with buffer solution for 287 s. Oocytes were discarded after experiments involving MTSEA because they had been covalently modified. In NS9283 experiments, 10 M NS9283 was preapplied to oocytes expressing ␤2-␣4-␤2-␣4 and free subunits for 30 s at the rate of 1.8 ml/min and incubated for an additional 80 s in a static bath before its co-application with ACh for 4 s at 3 ml/min at EC 20 -40 concentrations (30 M for ␣ subunits, 1 M for ␤2 subunit, and 3 M for ␤4 subunit). Preincubation with NS9283 eliminates kinetic effects of NS9283 binding, leaving only the kinetics of ACh responses in nAChRs at equilibrium with NS9283. Control experiments were performed on the same oocytes before NS9283 applications following the same protocol in which 10 M NS9283 was replaced with 0.1% (v/v) DMSO. Potentiation by NS9283 was calculated by increased response to ACh with 10 M NS9283 versus response of ACh co-applied with 0.1% DMSO. In the ACh concentration/response curve experiment with NS9283, 10 M NS9283 was coapplied with different concentrations of ACh without preapplication to shorten the experiment duration.
The peak amplitudes of experimental responses were calculated relative to the maximum ACh response or the average of the first two control ACh responses to normalize the data and compensate for variable expression levels among oocytes. Mean and S.E. were calculated from normalized responses. The numbers of oocytes tested are listed. The Hill equation was fit to the concentration/response relationship using a nonlinear least squares curve fit method (Kaleidagraph, Abelbeck/Synergy, Reading, PA): is the peak current measured at the agonist concentration x, I max is the maximum current peak at the saturating concentration, EC 50 is the agonist concentration required to achieve half of the maximum response, and n H is the Hill coefficient.
FLEXstation Experiments-For functional tests of nAChRs expressed in HEK cells, we used a FLEXstation (Molecular Devices) bench top scanning fluorometer as described by Kuryatov et al. (36). To increase the expression level of ␣2␤2 and ␣3␤2 nAChRs, the plates were incubated at 29°C for 20 h before being tested. The membrane potential kit (Molecular Devices) was used according to the manufacturer's protocols. Serial dilutions of NS9283 were manually added to the assay plate 15 min prior to recording. ACh dilutions were prepared in V-shaped 96-well plates (Fisher Scientific Co.) and added in cell culture wells at 20 l/s during recording. Each data point was averaged from three to four responses from separate wells. The potency and efficacy of drugs were calculated from the Hill equation described above.
NS9283 increases activation of nAChRs by partial agonists, but its potentiation profiles differ between the ␣4/␤2 primary SSAg sazetidine and the non-selective partial agonist cytisine. NS9283 increased the efficacy of both sazetidine and cytisine (Fig. 2, B and C). Interestingly, NS9283 made sazetidine a full agonist at ␣4␤2 nAChRs without changing its potency. This is most likely because NS9283 activates the ␣4/␣4 site (the critical aspect of potentiation), a site that is not bound by sazetidine. NS9283 (10 M) also potentiated activation of ␣4␤2 nAChRs by cytisine up to 10 M but not at higher concentrations of cytisine (Fig. 2C). Because NS9283 is an agonist at the ␣4/␣4 site and cytisine is a partial agonist at both ␣4/␤2 and ␣4/␣4 sites, NS9283 likely augments activation by cytisine at ␣4/␤2 sites, whereas increasing concentrations of cytisine compete with NS9283 for binding to the ␣4/␣4 site, thereby preventing the full agonist effect of NS9283 at this site. Channel block by cytisine could also contribute to reduced responses at high concentrations of cytisine. In summary, NS9283, which binds only to the ␣4/␣4 site, increases efficacies of SSAgs that bind only to the two ␣4/␤2 sites and the potencies of agonists that bind to all three sites. Its effect is lost if it is displaced from the ␣4/␣4 site by the partial agonist cytisine but not by the full agonist ACh; i.e. NS9283 behaves like a full agonist for the ␣4/␣4 site.
Activation and Potentiation of (␣4␤2) 2 * nAChRs with Different Accessory Subunits-Residues on the minus side of the ␣4 subunit determine agonist sensitivity and selectivity of the ␣4/␣4 agonist site (2,3,19). To investigate the pharmacological requirement for the plus side of this interface, we expressed (␣4␤2) 2 * nAChRs with various accessory subunits by injecting oocytes with mRNAs of ␤2-␣4-␤2-␣4 concatamers and a free accessory subunit (Fig. 3A). The concatamer alone gives minimal 300 M ACh-evoked current, less than 40 nA on average (n ϭ 17) (Fig. 3B). Concatamer expressed with a free subunit produces maximum ACh-evoked responses ranging from 673 nA to 3.17 A (Fig. 3B). Higher total current and responses dominated by the low affinity agonist site at the ␣4/␣4 interface characterize (␣4␤2) 2 ␣4 nAChRs (2). Like ␣4, other ␣ structural subunits except ␣6 also show higher absolute currents than ␤ subunits when assembling into the accessory position (Fig. 3B). Because of expression variation between individual oocytes, the standard errors of absolute currents are large. Because of assembly, trafficking, and functional issues, it is difficult to express ␣6 in heterologous systems (37). The gain-of-function mutant ␣6 L250S can, however, be more readily expressed with the concatamer ␤2-␣4-␤2-␣4. This produced an average maximum ACh-evoked current of 3.00 Ϯ 0.50 A, which is as high as the other ␣ subunits tested. These data suggest that an additional ACh site can form at interfaces of an accessory ␣ with the ␣4 subunits to increase activation. This additional accessory ␣6/␣4 site in (␣4␤2) 2 ␣6 L250S resulted in 64.8 Ϯ 6.4% inhibition of ACh (300 M) responses by the ␣6-selective antagonist conotoxin MII (100 nM) (38).
Absolute current differences could be due to different protein expression levels, assembly or maturation efficiencies, and/or channel electrophysiological properties. However, besides absolute current, we observed another characteristic of the agonist site at the accessory ␣/␣4 sites, i.e. low sensitivity to ACh activation ( Fig. 3C and Table 1). ␤2 and ␤4 subunits produced nAChRs highly sensitive to ACh when expressed with ␤2-␣4-␤2-␣4 with EC 50 values of 1.02 Ϯ 0.10 M for (␣4␤2) 2 ␤2 and 4.96 Ϯ 1.53 M for (␣4␤2) 2 ␤4 nAChRs. ␣ subunits (␣2-4 and ␣6 wild type and mutant) as accessory subunits display low sensitivities to ACh with EC 50 values higher than 100 M. Note that (␣4␤2) 2 ␣ nAChRs have two high affinity agonist sites at ␣4/␤2 interfaces. This usually results in ACh concentration/ response curves fitting better to a biphasic curve with a high and a low EC 50 value (2, 3). Unfortunately, our data set could not resolve the two EC 50 values very well for all of the nAChR subtypes; thus all dose/response curves were fit to a monophasic curve for comparison. The Hill coefficient values of (␣4␤2) 2 ␣2, (␣4␤2) 2 ␣4, and (␣4␤2) 2 ␣6 calculated in this way are less than 1 (0.625-0.762), consistent with the presence of agonist sites with different affinities.

MTSEA Blocks Activation from Many Accessory ␣/␣4
Sites-If an agonist site is present at an accessory site, MTSEA should selectively block this interface of (␣4 T126C ␤2) 2 * nAChRs, leaving only the ␣4/␤2 sites (Fig. 5A), i.e. decreasing the efficacy and leaving only the high affinity responses. To test this idea, we expressed the mutant concatamer ␤2-␣4 T126C with free subunits in oocytes and investigated the effect of MTSEA on activation by ACh of these nAChRs. Fig. 6 and Table 2 summarize the effects of MTSEA on the potency and efficacy of ACh of these nAChRs.

Discussion
There is interest in developing PAM drugs for nAChRs because agonists both activate and desensitize nAChRs (24,40). PAMs bind away from agonist binding sites to enhance function when agonists are bound. Benzodiazepines are PAMs for ␥-aminobutyric acid (GABA) receptors and have proven clinically useful in modulating their function (41). There is optimism that the same principle can be applied to nAChRs. PAMs offer the potential to modulate endogenous patterns of signaling rather than constantly activating or desensitizing as results from sustained exposure to agonist drugs. Here we have defined another group of drugs, accessory SSAgs, such as NS9283, that functionally behave like a PAM but adopt an agonist-like mechanism to potentiate nAChR activation.
NS9283 was understandably designated as a PAM initially because it appeared not to have intrinsic activity at nAChRs or to compete with agonist for binding to nAChRs (9, 10, 12). We confirm that NS9283 does not act as a PAM but instead poten-tiates activation of nAChRs by binding to a low affinity ACh binding site at the ␣4/␣4 subunit interface (17,19) (Figs. 4, 5A,  and 7). The low affinity of this site and the site selectivity of NS9283 explain why this compound demonstrated PAM-like activity during its initial characterization (12,17) (Figs. 1 and 3). Demonstration of agonist activation by NS9283 (16,17) and identification of selectivity for a particular ACh site (17,19) (Figs. 3 and 7) disqualify NS9283 as a PAM because it is neither a modulator nor allosteric.
We define NS9283 and other agonists selective for this accessory ACh binding site as accessory SSAgs. NS9283 binds to the extracellular domain of nAChRs (17,19) differently from PAMs  and allosteric agonists like PNU-120596 and 4-(4-bromophenyl)-3a,4,5,9b-tetrahydro-3H-cyclopenta[c]quinoline-8-sulfonamide, which bind to transmembrane sites near the channel gate whose opening they influence (24 -26). Both NS9283 and benzodiazepines bind to the extracellular domain of accessory subunits. However, the accessory SSAg NS9283 binds to the opposite side of the accessory subunit relative to the benzodiazepine PAM site in GABA A receptors, and the benzodiazepine PAM site is not a GABA binding site (42). Through binding to the ␣4/␣4 agonist site, NS9283 cannot activate wild type nAChRs by itself because single agonist site occupancy is not sufficient to efficiently activate nAChRs (6). Neither can ACh activate from binding only to the accessory ␣4/␣4 agonist site (Fig. 5C). Hence, the accessory agonist site appears to serve as a novel potentiation site that binds agonists and potentiates activity in response to ACh binding at the other agonist sites. Binding only to the accessory site is not sufficient to induce desensitization nor can it prevent or reverse desensitization resulting from agonist binding to ␣4/␤2 sites (18). However, NS9283 increases sensitivity of ACh desensitization because it occupies the ␣4/␣4 site as an agonist (18). Binding only to the accessory site cannot up-regulate nAChRs. Agonists and antagonists at ␣4/␤2 sites are thought to increase the amount of nAChRs by promoting a conformation of (␣4␤2) 2 that assembles more efficiently with accessory subunits to form mature nAChRs (36). A binding site for NS9283 does not exist in the (␣4␤2) 2 assembly intermediate. Therefore, NS9283 did not upregulate ␣4␤2 nAChRs expressed in HEK cells (data not shown). The potentiation mechanism of accessory SSAgs like NS9283 provides a target mechanism for development of novel therapeutics.
Developing a new Food and Drug Administration-approved drug can take more than 7 years on average to pass through clinical trials, not to mention their high cost, low success rate, and high preclinical expenses (43). Drugs like NS9283 could be cost-effective because they offer unique properties when combined with already approved agonists and partial agonists. Indeed, by enhancing agonist activation and desensitization affinities, such drugs have the potential to reduce side effects of high doses of nAChR agonists such as nausea and motor or cardiovascular impairment, which result from nonspecific interaction with other receptors (9 -11). This property could also enhance older drugs to treat new diseases or syndromes by enhancing their activity without the need to increase their dose (10). Our in vitro characterization of NS9283 has provided insights in guiding design of combined therapy for SSAgs and other known nAChR agonists. NS9283 increases agonist site occupancy by binding to accessory agonist sites (17) (Figs. 7 and  8), which results in gain of function of both full and partial agonists (Fig. 2). Because of different binding site selectivity, NS9283 binds to the accessory ␣4/␣4 site independently from binding of sazetidine to ␣4/␤2 sites, thus turning this primary SSAg into a full agonist without altering its potency. Conversely, ACh and the partial agonist cytisine bind to both the ␣4/␤2 and ␣4/␣4 agonist sites (20). NS9283 primes the ␣4/␣4 agonist sites for these non-selective agonists binding to the ␣4/␤2 sites but competes with them binding to the ␣4/␣4 sites, thus increasing their potencies but retaining partial agonism of cytisine and maximum efficacy of ACh.
Like ACh, NS9283 binds to the interface between the minus face of ␣4 subunit and plus face of ␣2, ␣3, ␣4, and ␣6 (Fig. 4). These ␣ subunits share a C-loop and several aromatic residues that are critical for agonist binding. In the crystal structure of NS9283 and ACh-binding protein (17), these residues from the plus sides interact with the pyridine ring of NS9283 that is partially protonated under physiological pH and resembles the charged nitrogen of an agonist. Therefore, when MTSEA prevents agonist binding at accessory ␣/␣4 interfaces via electric repulsion (39), it can also disrupt potentiation of NS9283 binding to these agonist sites (Figs. 7 and 8).
Here we showed the relationship between activation of the two kinds of ACh binding sites. 1) The ␣4/␤2 sites are necessary for activation by accessory ␣/␣4 sites. 2) The accessory ␣/␣4 sites promote activation of nAChRs from ␣4/␤2 agonist sites. ACh cannot activate (␣4 L121C ␤2)␣4 nAChRs after MTSEA blocks their ␣4/␤2 agonist sites (Fig. 5). This is consistent with the observation that occupying one agonist site does not efficiently activate nAChRs (6). It is possible that MTSEA induces an inactive conformation of nAChRs after reacting with (␣4 L121C ␤2)*, preventing further activation of nAChRs. However, other ligands selective for the ␣4/␣4 sites, such as NS9283, cannot activate wild type nAChRs (Fig. 4), which also supports the low occupancy theory. Therefore, accessory SSAgs like NS9283 that bind only to one site are unlikely to activate nAChRs by themselves and will behave pharmacologically like a PAM in vivo. Maximum activation by ACh is about equal to NS9283 with other agonists and much better than cytisine with or without NS9283 because cytisine is a very low efficacy partial agonist at both ␣4/␣4 and ␣4/␤2 agonist sites (Fig. 2). Perhaps that is because the bridged bicyclic ring of cytisine precludes the C-loop from closing properly to fully activate nAChRs (48). A partial agonist structure like cytisine should be avoided in designing accessory SSAgs with high efficacies.
In conclusion, we confirm and extend the mechanism proposed by others (2, 3, 16, 17) that there is a binding site for ACh and NS9283 at many ␣/␣4 subunit interfaces that promotes activation by providing a third agonist site in addition to ACh binding sites at the ␣4/␤2 subunit interfaces in (␣4␤2) 2 * nAChRs. Low affinity accessory ACh binding sites may not encounter sufficient ACh concentrations to activate them during volume transmission in the brain. Postsynaptic brain nAChRs like those in muscle may transiently be exposed to ACh concentrations sufficiently high to fully activate low affinity ␣/␣4 sites. It is unknown whether in the brain different stoichiometry nAChRs such as (␣4␤2) 2 ␤2 and (␣4␤2) 2 ␣4 are colocalized to provide broad ACh concentration/response curves or localized in different circuits to mediate different functions. High affinity accessory SSAgs and site-selective antagonists could help to sort this out.