α4β2 Nicotinic Acetylcholine Receptors

Acetylcholine receptors comprising α4 and β2 subunits are the most abundant class of nicotinic acetylcholine receptor in the brain. They contribute to cognition, reward, mood, and nociception and are implicated in a range of neurological disorders. Previous measurements of whole-cell macroscopic currents showed that α4 and β2 subunits assemble in two predominant pentameric stoichiometries, which differ in their sensitivity to agonists, antagonists, and allosteric modulators. Here we compare agonist-elicited single channel currents from receptors assembled with an excess of either the α4 or β2 subunit, forming receptor populations biased toward one or the other stoichiometry, with currents from receptors composed of five concatemeric subunits in which the subunit stoichiometry is predetermined. Our results associate each subunit stoichiometry with a unique single channel conductance, mean open channel lifetime, and sensitivity to the allosteric potentiator 3-[3-(3-pyridinyl)-1,2,4-oxadiazol-5-yl]benzonitrile (NS-9283). Receptors with the composition (α4β2)2α4 exhibit high single channel conductance, brief mean open lifetime, and strong potentiation by NS-9283, whereas receptors with the composition (α4β2)2β2 exhibit low single channel conductance and long mean open lifetime and are not potentiated by NS-9283. Thus single channel current measurements reveal bases for the distinct functional and pharmacological properties endowed by different stoichiometries of α4 and β2 subunits and establish pentameric concatemers as a means to delineate interactions between subunits that confer these properties.

Nicotinic acetylcholine receptors (AChRs) 2 are pentameric ligand-gated ion channels formed from homologous subunits of which there are many different subtypes (1). The ability to combine different types of subunits into an individual penta-mer enables a wide diversity of functional properties and, as a consequence, the ability to meet a range of physiological needs (2). In the brain, the vast majority of [ 3 H]nicotine binding sites arise from AChRs comprising ␣4 and ␤2 subunits (3)(4)(5)(6). These subunits assemble into pentamers in which the stoichiometry of ␣4 and ␤2 subunits varies (7,8). The two predominant subunit stoichiometries differ in their sensitivity to acetylcholine (ACh), unitary current amplitude, selectivity for different agonists and antagonists, potentiation by ions or drugs, and agonist concentration dependence of desensitization (8 -15). Because the agonist binding sites form at interfaces between the principal face of an ␣4 subunit and the complementary face of either a ␤2 or ␣4 subunit, changes in subunit stoichiometry alter the number of agonist binding sites per pentamer (15,16). In addition, at interfaces that do not bind agonist, changes in subunit pairing can alter receptor function via allosteric effects (17). Thus understanding the functional consequences of changes in the stoichiometry of ␣4 and ␤2 subunits is of paramount significance toward the design of drugs to treat nicotine dependence and a variety of neurological and neurodegenerative diseases.
Insight into the relationship between the stoichiometry of ␣4 and ␤2 subunits and receptor function has been achieved using concatemeric receptors in which five subunits are covalently linked head to tail in a predetermined order and stoichiometry (18). Concatemeric receptors containing two non-consecutive ␣4 and three ␤2 subunits activate in response to low concentrations of ACh and mimic the high agonist sensitivity of receptors assembled with an excess of ␤2 over ␣4 subunits. In contrast, concatemeric receptors containing three ␣4 and two non-consecutive ␤2 subunits activate in response to high concentrations of ACh and mimic the low agonist sensitivity of receptors assembled with an excess of ␣4 over ␤2 subunits (18). Likewise, the agonist sazetidine A activates concatemeric receptors containing two non-consecutive ␣4 and three ␤2 subunits but not receptors containing three ␣4 and two non-consecutive ␤2 subunits, in agreement with studies of receptors assembled with an excess of either the ␤2 or ␣4 subunit, respectively (15,18,19). Finally, zinc potentiates agonist-elicited responses of concatemeric receptors containing three ␣4 and two non-consecutive ␤2 subunits but not those of receptors containing two non-consecutive ␣4 and three ␤2 subunits, in agreement with studies of receptors assembled with an excess of either the ␣4 or ␤2 subunit, respectively (10,18).
So far, relationships between the stoichiometry of ␣4 and ␤2 subunits and the functional properties of a receptor have been inferred from measurements of agonist-elicited macroscopic currents that yield the average response of the overall receptor population. In addition, functional properties such as unitary conductance, open channel lifetime, and kinetics of activation and deactivation are largely inaccessible to macroscopic current recording. Here, using patch clamp electrophysiology, agonist-elicited single channel currents were recorded from receptors formed in the presence of an excess of either the ␣4 or ␤2 subunit. These unitary currents were then compared with those from concatemeric receptors in which the stoichiometry and positioning of the subunits is predetermined. The results reveal relationships between the stoichiometry of ␣4 and ␤2 subunits and functional properties at the level of single receptor channels.

Results
Identification of Single Channel Currents from ␣4␤2 AChRs-Patch clamp recordings, before and after addition of agonist, are required to unambiguously demonstrate single channel currents from ␣4␤2 AChRs. Using an agonist-free recording pipette, a gigaohm seal was established to a BOSC 23 cell transfected with equal amounts of unlinked ␣4 and ␤2 subunit cDNAs. The baseline current was recorded to verify the absence of single channel currents, and then the membrane-permeable agonist nicotine was added to the extracellular solution (Fig. 1A). After a short delay, single channel openings appeared, initially at a high frequency followed by a slow decline due to desensitization. Inspection of individual channel openings reveals two distinct unitary current amplitudes of ϳ3 and 4 pA, in agreement with previous studies (9,20). Applying the same procedure to a BOSC cell transfected with cDNA encoding a pentameric concatemer composed of three non-consecutive ␣4 and two ␤2 subunits reveals nicotine-activated single channel currents that again decline in frequency due to desensitization (Fig. 1B). Inspection of individual channel openings reveals one unitary current amplitude of ϳ4 pA, in contrast to the dual amplitudes observed with unlinked subunits. These observations establish the identity of agonist-elicited single channel currents from ␣4␤2 AChRs formed by unlinked as well as concatemeric subunits. Furthermore, they suggest unitary current amplitude depends on subunit stoichiometry.
Subunit Stoichiometry and Single Channel Current Amplitude-To establish that the different unitary current amplitudes arise from receptors with different subunit stoichiometry, agonist-elicited single channel currents were recorded following transfection of an excess of either the ␣4 or ␤2 subunit cDNA, as opposed to the equal amounts in Fig. 1A. During receptor assembly, an excess of one subunit over the other is expected to bias the receptor population toward pentamers  with an excess of the surplus subunit. To determine subunit stoichiometry, recordings from receptors composed of unlinked subunits were compared with receptors composed of concatemeric subunits containing either two or three ␣4 subunits plus the pentameric complement of ␤2 subunits. For these determinations, single channel currents were recorded in the cell-attached patch configuration with a concentration of 10 M ACh in the recording pipette and a holding potential of Ϫ70 mV. Under these conditions, the concentration of ACh exposed to the receptors is held constant, which establishes an equilibrium between activatable and desensitized receptors. When unlinked ␣4 and ␤2 subunits were transfected in a 1:1 ratio, the single channel current traces, as well as a histogram of single channel current amplitudes, again reveal two distinct current amplitudes of 3.4 and 4.2 pA ( Fig. 2A); the histogram is well described by the sum of two Gaussian components. However, when ␣4 and ␤2 subunits were transfected in a 1:10 ratio, a condition expected to promote assembly of receptors with two rather than three ␣4 subunits, the single channel traces, as well as the histogram of current amplitudes, reveals one current amplitude of 3.4 pA (Fig. 2B). Conversely, when the ␣4 and ␤2 subunits were transfected in a 10:1 ratio, a condition expected to promote assembly of receptors with three rather than two ␣4 subunits, the single channel current traces and the histogram of current amplitudes reveal one current amplitude of 4.1 pA (Fig.  2C). Thus the unitary current amplitudes obtained from receptors formed with biased ratios of ␣4 and ␤2 subunits recapitulate those of receptors formed with an equal ratio of the subunits.
Recordings from concatemeric receptors comprising two non-consecutive ␣4 subunits and three ␤2 subunits reveal one current amplitude of 3.4 pA (Fig. 2D), in close accord with recordings from receptors formed using a 1:10 ratio of unlinked ␣4 and ␤2 subunits (Fig. 2B). In contrast, recordings from concatemeric receptors comprising three ␣4 subunits and two non-consecutive ␤2 subunits reveal one current amplitude of 4.2 pA (Fig. 2E), in close accord with recordings from receptors formed using a 10:1 ratio of unlinked ␣4 and ␤2 subunits (Fig.  2C). Thus, based on measurements of unitary current amplitude, the subunit stoichiometries inferred for receptors formed using biased ratios of unlinked subunits coincide with the known subunit stoichiometries in concatemeric receptor pentamers.
To quantify the unitary conductance for receptors of each subunit stoichiometry, single channel currents were recorded at different holding potentials, and histograms of single channel current amplitudes were analyzed (Fig. 2F). For each receptor, a plot of the mean current amplitude against holding potential reveals a straight line, the slope of which gives the unitary conductance. Receptors containing two ␣4 and three ␤2 subunits, whether formed from a 1:10 ratio of unlinked subunits or a concatemeric receptor composed of two non-consecutive ␣4 and three ␤2 subunits, exhibit a unitary conductance of 48 -49 pS. By contrast, receptors containing three ␣4 and two ␤2 subunits, whether formed from a 10:1 ratio of unlinked subunits or a concatemeric receptor composed of three ␣4 and two non-consecutive ␤2 subunits, exhibit a unitary conductance of 60 pS. Thus measurements of single channel current amplitude associate a defined stoichiometry and positioning of subunits within the pentamer with a distinct unitary channel conductance.
To confirm the association between unitary current amplitude and subunit stoichiometry, single channel currents elicited by the agonist 4-(5-ethoxy-3-pyridinyl)-N-methyl-(3E)-3buten-1-amine difumarate (TC-2559) were recorded in the cell-attached patch configuration at a holding potential of Ϫ70 mV (Fig. 2G). TC-2559 was chosen because it is a selective agonist for ␣4␤2 AChRs (21). In all cases, the current amplitude from each concatemeric receptor pentamer concurs with that from the corresponding receptors formed from biased ratios of unlinked ␣4 and ␤2 subunits. Thus, based on recordings with multiple agonists (nicotine, ACh, and TC-2559), receptors with greatest unitary conductance correspond to pentamers with three ␣4 and two non-consecutive ␤2 subunits, whereas receptors with lowest unitary conductance correspond to receptors with two non-consecutive ␣4 and three ␤2 subunits.
Subunit Stoichiometry and Open Channel Lifetime-To determine whether receptors with different subunit stoichiometry differ in their open channel lifetimes, single channel currents elicited by a concentration of 10 M ACh were recorded in the cell-attached patch configuration at a holding potential of Ϫ70 mV (Fig. 3). The concentration of ACh was designed to minimize ACh-mediated channel block while still being sufficient to activate receptors with either high or low sensitivity to ACh. Receptors formed with a 1:10 ratio of unlinked ␣4 to ␤2 subunits exhibit four exponential components of channel openings with mean durations that differ by approximately 1 order of magnitude ( Fig. 3A and Table 1). The mean duration of the briefest component is ϳ40 s, whereas that of the longest component is ϳ10 ms. Analysis of single channel current amplitudes from the same recording reveals one amplitude of ϳ3 pA, verifying that the stoichiometry of these receptors is two ␣4 and three ␤2 subunits.
Concatemeric receptors comprising two non-consecutive ␣4 and three ␤2 subunits also exhibit four exponential components of openings with mean durations similar to those of receptors formed by unlinked subunits (Fig. 3B and Table 1). Previous measurements of macroscopic currents from the same concatemeric receptors reveal an EC 50 of ϳ1 M (18), suggesting that the ACh concentration of 10 M ACh used in the single channel recordings approaches full occupancy of the ligand binding sites.
In contrast, receptors formed with a 10:1 ratio of unlinked ␣4 to ␤2 subunits exhibit three exponential components of channel openings with mean durations similar to those of the three briefest components by receptors formed with a 1:10 ratio of subunits. Analysis of single channel current amplitudes from the same recording reveals one amplitude of ϳ4 pA, verifying   Table 1. For each panel, a histogram of detected single channel currents that reached full amplitude is shown along with a representative trace filtered at 2 kHz. that the stoichiometry of these receptors is three ␣4 and two ␤2 subunits. Concatemeric receptors comprising three ␣4 and two non-consecutive ␤2 subunits also exhibit three exponential components of openings, although the mean durations are somewhat longer than those of the corresponding receptors formed by unlinked subunits. Previous measurements of macroscopic currents from these concatemeric receptors reveal an EC 50 of ϳ100 M, indicating that the ACh concentration of 10 M in the single channel recordings achieves only partial occupancy of the ligand binding sites. Thus the two different subunit stoichiometries give rise to receptors with distinct distributions of open channel lifetimes in which a component with long mean duration is observed for receptors with two ␣4 and three ␤2 subunits, whereas receptors with three ␣4 and two ␤2 subunits lack this component. However, this distinction could arise either from differences in the inherent stability of the open channel or from differences in the number of bound agonists.
Subunit Stoichiometry and Drug Potentiation-The drug 3-[3-(3-pyridinyl)-1,2,4-oxadiazol-5-yl]benzonitrile (NS-9283) potentiates ␣4␤2 receptors and is selective for receptors in which the number of ␣4 subunits is greater than the number of ␤2 subunits. Potentiation arises via a left shift of the agonist concentration-response relationship rather than an increase of the maximum response. Thus measurements of potentiation of ␣4␤2 receptors by NS-9283 were designed to further distinguish between receptors with different subunit stoichiometry (Fig. 4). A gigaohm seal was formed to a cell transfected with a 10:1 ratio of ␣4 to ␤2 subunit cDNAs, and ACh was included in the recording pipette without or with NS-9283. The ACh concentration of 10 M is below the EC 50 for receptors with three ␣4 and two ␤2 subunits, and the NS-9283 concentration of 10 M approaches that which maximally potentiates macroscopic currents and is below that which produces channel block. In the presence of ACh alone, channel openings appear as random isolated events, whereas in the presence of both ACh and NS-9283, channel openings appear in clearly defined clusters of several events in quick succession (Fig. 4, A and B). Analyses of single channel current amplitudes from the same recordings reveal one amplitude of ϳ4 pA (data not shown), verifying that the subunit stoichiometry is three ␣4 and two ␤2 subunits. For recordings in either the absence or presence of NS-9283, histograms of open durations exhibit three exponential components with similar mean durations for each condition (Fig. 4, A and B, and Table 2). However, after defining the mean intracluster closed duration, histograms of cluster durations comprise three components in the presence of ACh alone and four components in the presence of ACh and NS-9283 (Fig. 4, A and B, and Table 3). The mean durations of the two briefest components are similar without and with NS-9283 and likely correspond to channel openings by unpotentiated receptors. Compared with the third component in the presence of ACh alone, the third and fourth components in the presence of NS-9283 are markedly prolonged and thus correspond to channel openings by potentiated receptors.
Recordings from cells expressing concatemeric receptor pentamers containing three ␣4 and two non-consecutive ␤2 subunits mirror those from cells expressing receptors formed with an excess of ␣4 over ␤2 subunits. In particular, channel openings appear as random isolated events in the presence of ACh alone, whereas openings appear in clusters in the presence of ACh and NS-9283 (Fig. 4, C and D). Open duration histograms comprise three exponential components of channel openings either without or with NS-9283 (Fig. 4, C and D, and Table 2). Histograms of cluster durations comprise two components with similar mean durations either without or with NS-9283, thus corresponding to unpotentiated receptors. However, a third component, common to both conditions, is prolonged in the presence of NS-9283 compared with that in its absence (Fig. 4, C and D, and Table 3). Thus NS-9283 potentiates receptors with three ␣4 and two non-consecutive ␤2 subunits. Furthermore, potentiation by NS-9283 arises through a marked change in channel gating kinetics where channel openings with longest mean duration coalesce into clusters of events in rapid succession.
To further quantify potentiation by NS-9283, the fraction of channel openings that were followed by a reopening is plotted against the number of reopenings (Fig. 5). A reopening is defined as an opening preceded by a closing shorter than the critical closed time used to define clusters. In the presence of ACh alone, the distribution is well described by a single exponential decay, and the mean number of reopenings, 0.4 -0.5, is similar whether the receptors comprise unlinked or linked subunits. In the presence of ACh and NS-9283, a second exponential component is observed corresponding to potentiated receptors. For this second component, the mean number of reopenings increases markedly to 3.5 for receptors formed from unlinked subunits and to 9.9 for receptors formed from linked subunits. Thus potentiation by NS-9283 arises from a marked increase in the fraction of channel openings that are followed by a reopening, as well as an increase in the number of reopenings per activation episode.
Whether NS-9283 potentiation depends on subunit stoichiometry was evaluated by applying the experimental procedures described in Fig. 4 to cells transfected with either a 1:10 ratio of ␣4 to ␤2 subunit cDNAs or concatemeric receptor pentamers with two non-consecutive ␣4 subunits and three ␤2 subunits. For a recording from receptors formed from unlinked subunits, the open duration histogram comprises four exponential com-

TABLE 1 Open durations of unlinked and linked receptors activated by ACh
Mean open times (O) in units of milliseconds and fractional areas (A) are given along with ϮS.E. for the indicated number of patches, n. In cases in which S.E. is not given, the frequency of channel openings in individual patches was low, and the indicated number of patches was combined prior to fitting.

Single Channel Properties of ␣4␤2 nAChR
ponents in the presence of ACh alone, whereas the accompanying closed duration histogram contains two brief and two long duration components ( Fig. 6A and Table 2). The two brief closed time components represent infrequent reopening of the same receptor channel, generating a minicluster, and the intersection of the second brief with the succeeding long closed time component allows definition of these clusters. The histogram of cluster durations comprises five components, four of which have mean durations similar to those of the open duration histogram, thus representing distinct kinetic classes of isolated channel openings, whereas the fifth component represents successive openings of the same channel (Table 3). For a recording obtained in the presence of both ACh and NS-9283, the open duration histogram comprises four exponential components, and the accompanying closed duration histogram contains two brief and two long duration components ( Fig. 6B and Table 2). The cluster duration histogram comprises three exponential components, the means of which are briefer, rather than longer, than those in the absence of NS-9283, indicating lack of potentiation by NS-9283 and possibly block of the channel (Table 3).
Single channel data from concatemeric receptor pentamers containing three non-consecutive ␤2 and two ␣4 subunits mirror those from receptors formed from biased ratios of unlinked subunits, showing briefer rather than longer channel openings in the presence of NS-9283 (Fig. 6, C and D, and Tables 2 and 3). Thus NS-9283 does not potentiate receptors with two ␣4 and three ␤2 subunits and is selective for receptors with three ␣4 and two ␤2 subunits.

Discussion
Previous work, based on measurements of agonist-elicited macroscopic currents, established that ␣4 and ␤2 subunits assemble with variable subunit stoichiometry, producing at least two populations of receptor pentamers (7-9). Evidence included a biphasic agonist concentration-response relationship in which the magnitudes of the high and low affinity components varied, depending on the ratios of either the mRNA or cDNA encoding the two subunits. Furthermore, analyses of purified receptors showed that the molar ratio of ␣4 to ␤2 subunits depended on the ratio of the mRNAs encoding the two  Tables 2 and 3.

TABLE 2 Open durations of unlinked and linked receptors activated by ACh ؎ NS-9283
Mean open times (O) in units of milliseconds and fractional areas (A) are given along with ϮS.E. for the indicated number of patches, n. In cases in which S.E. is not given, the frequency of channel openings in individual patches was low, and the indicated number of patches was combined prior to fitting.

O1 (A1) O2 (A2) O3 (A3) O4 (A4)
Unlinked   Single Channel Properties of ␣4␤2 nAChR FEBRUARY 17, 2017 • VOLUME 292 • NUMBER 7 subunits (9). Studies of pentameric concatemers, in which the stoichiometry and positioning of subunits is predetermined, confirmed that agonist sensitivity depended on the stoichiometric ratio of the ␣4 and ␤2 subunits and identified two constructs that recapitulated the high and low agonist sensitivities of receptors formed from unlinked subunits (15,18). The present results, based on agonist-elicited single channel currents, establish that receptors with different subunit stoichiometry differ in their unitary current amplitudes, kinetics of channel closing, and susceptibility to drug potentiation, as summarized in Table  4. The overall results provide evidence for independent as well interdependent subunit contributions to receptor function and pharmacology. Previous single channel recordings from receptors comprising ␣4 and ␤2 subunits showed multiple unitary current amplitudes, in accord with multiple receptor populations (9,22,23). Recently, a study of receptors assembled from biased ratios of unlinked subunits revealed either uniformly high or low unitary conductance, depending on which subunit was present in excess (20). However, the question of subunit stoichiometry remained. Herein, by comparing receptors formed from biased ratios of unlinked subunits with receptors formed from concatemeric subunits, we associate a defined subunit stoichiometry and positioning with a particular unitary conductance; receptors with two non-consecutive ␣4 and three ␤2 subunits exhibit a conductance of 48 pS, whereas receptors with two non-consecutive ␤2 and three ␣4 subunits exhibit a conductance of 60 pS ( Table 4). The conductance differences likely originate from differences in charged residues in the extracellular ring that borders the second transmembrane domain (M2) where the ␤2 subunit contains a lysine residue and the ␣4 subunit contains a glutamate (24). The recent crystal structure of the human (␣4␤2) 2 ␤2 receptor revealed that these residues project into the path of ion flow at the entrance to the pore domain (31). Previous studies showed that these two residues confer differences in calcium permeability between receptors with different stoichiometric ratios of ␣4 and ␤2 subunits (12). However, the present studies were done in the absence of calcium, so the conductance difference arises from differences in permeability of monovalent alkali ions. The residues in the inner and intermediate rings of charge, between M1 and M2 at the cytoplasmic side of the membrane, are negatively charged in both the ␣4 and ␤2 subunits (24). Thus, as concluded in classical work that unveiled three rings of charge that contribute to ion conductance, the ␣4 and ␤2 subunits likely contribute independently of each other according to the sign of the charge in their extracellular ring (25).
Our studies show that receptors with different stoichiometric compositions of subunits differ in their distributions of open channel durations. This observation is significant because each type of receptor contains two ligand binding sites formed at interfaces between ␣4 and ␤2 subunits, and occupancy of these sites is coupled to channel opening. Furthermore, an ACh concentration of 10 M is sufficient to occupy both binding sites on receptors in which the fifth subunit is ␤2, as these receptors are highly sensitive to ACh. However, when the fifth subunit is ␣4, a third ligand binding site forms at the interface between ␣4 and ␣4 subunits, and the resulting receptors exhibit relatively brief channel openings as shown by the absence of an exponential component of openings with long mean duration. Thus the presence of an ␣4-␣4 ligand binding site has an apparent allosteric effect on the ␣4-␤2 ligand binding sites, either reducing their affinity for ACh or reducing the ability of the channel to open in response to occupancy by ACh. This allosteric effect could originate from a change in symmetry of the ␣4-␤2 binding sites. In particular, for high sensitivity, low conductance receptors, the subunits that form one ␣4-␤2 site are flanked by ␤2 and ␣4 subunits, whereas those that form the other site are flanked by ␤2 and ␤2 subunits (Table 4). Conversely, for low sensitivity, high conductance receptors, the subunits that form one ␣4-␤2 site are flanked by ␤2 and ␣4 subunits, whereas those that form the other site are flanked by ␣4 and ␣4 subunits.
Of further importance is our observation that receptors with both stoichiometric compositions of subunits exhibit multiple exponential components of channel openings: four components for high sensitivity, low conductance receptors and three   Tables 2 and 3. components for low sensitivity, high conductance receptors.
Assuming that the ␣4-␤2 binding sites are non-equivalent, because of asymmetry in their neighboring subunits, for high sensitivity, low conductance receptors, three components could arise from two species of singly occupied receptors and one component could arise from the doubly occupied receptor (Table 4). However, four components are observed, suggesting that openings may arise from a closed state intermediate between the resting and open channel states (26,27). For low sensitivity, high conductance receptors, the three observed exponential components could arise from three species of singly occupied receptors, which is plausible given that the agonist concentration is low. With higher concentrations of agonist, additional exponential components are expected due to doubly and triply occupied receptors. Further insights into kinetic mechanism could thus emerge from studies of the distribution of channel open times as a function of agonist concentration. Previous studies demonstrated that the drug NS-9283 potentiates receptors with an excess of ␣4 over ␤2 subunits but not receptors with an excess of ␤2 over ␣4 subunits (11). Our findings confirm that potentiation depends on subunit stoichiometry, but moreover they define the stoichiometry and positioning of subunits that confer potentiation, and reveal the kinetic basis for potentiation. In the absence of NS-9283, individual channel openings appear predominantly in isolation, and the number of reopenings per activation episode follows a single exponential decay. However, in the presence of the drug, a second exponential component is observed comprising clusters of closely spaced events. This change in kinetics is reminiscent of the clustering of channel openings that occurs as the concentration of agonist is increased (28). Thus, although NS-9283 does not elicit channel opening by itself, when it is present together with a low concentration of agonist, the kinetics of channel opening resemble those at high concentrations of agonist. It is possible that NS-9283 acts as a co-agonist of ACh by binding to the ␣4-␣4 binding site, whereas ACh binds to the two ␣4-␤2 sites (29). However, the site for NS-9283 binding could be elsewhere, such as the ␤2-␣4 interface, which does not bind agonist. In this scenario, agonist would bind to three sites, two ␣4-␤2 and the ␣4-␣4, whereas the drug binds to one or both ␤2-␣4 interfaces. Alternatively, NS-9283 may bind at a non-subunit interface, whereas the agonist binds to the three ligand binding sites. Future studies with concatemeric receptors could potentially distinguish between these mechanisms.
Our findings not only identify three functional attributes of ␣4␤2 receptors that depend on subunit stoichiometry, but they also identify a positioning of subunits within the pentamer that confers these attributes (Table 4). In particular, in both stoichiometric arrangements, one pair of ␣4 subunits is separated by a ␤2 subunit as opposed to being consecutive. This non-consecutive arrangement of ␣4 subunits is reminiscent of the Torpedo AChR in which the two ␣1 subunits are non-consecutive (30), and is identical to that in the crystal structure of the (␣4␤2) 2 ␤2 receptor (31).
Our findings establish concatemeric pentamers combined with single channel measurements as a means to clarify mechanisms of activation and potentiation of ␣4␤2 receptors. By subunit-selective mutagenesis, one could determine whether agonist occupancy of the ␣4-␣4 site alone is responsible for the kinetic differences between receptors with the two stoichiometric compositions of subunits or whether the presence of a third ␣4 subunit is responsible. Analogously, one could determine whether potentiation by NS-9283 is mediated by a co-agonist mechanism in which the drug binds to the ␣4-␣4 site or whether potentiation arises through binding to non-canonical ␤2-␣4 sites or a site distinct from subunit interfaces. Both investigations require single channel measurements as a functional end point. Finally, our findings offer a means to clarify mechanisms of disease-causing mutations of ␣4␤2 receptors by determining whether the functional consequences of the mutations depend on subunit stoichiometry.
Patch Clamp Recordings-Single channel recordings were obtained in the cell-attached patch configuration (38) at a membrane potential of Ϫ70 mV and a temperature of 20°C as described previously (39,40). For all experiments, the extracellular bathing solution contained 142 mM KCl, 5.4 mM NaCl, 1.8 mM CaCl 2 , 1.7 mM MgCl 2 , and 10 mM HEPES adjusted to pH 7.4 with NaOH. The pipette solution contained 80 mM KF, 20 mM KCl, 40 mM potassium aspartate, 2 mM MgCl 2 , 1 mM EGTA, and 10mM HEPES adjusted to pH 7.4 with KOH (41,42). Concentrated stock solutions of ACh and TC-2559 were made in pipette solution and stored at Ϫ80°C until the day of each experiment. A concentrated stock solution of NS-9283 was prepared in DMSO, stored at Ϫ80°C, and added to the pipette solution the day of each experiment. Pipette solution without NS-9283 contained an equivalent volume of DMSO to achieve a final concentration of 0.2% (v/v). Patch pipettes were pulled from glass capillary tubes (number 7052, Garner Glass) and coated with Sylgard (Dow Corning).
For experiments in which nicotine was applied to the extracellular bathing solution, a gigaohm seal was initially established to a cell using a patch pipette filled with agonist-free pipette solution. A baseline current was recorded for 1-5 min, and then nicotine was added dropwise to the extracellular solution to establish a final extracellular concentration of 342.5 M. Continuous recordings were obtained before, during, and after nicotine addition.
Data Analysis-Single channel currents were recorded using an Axopatch 200B patch clamp amplifier (Molecular Devices) with the gain set at 100 mV/pA and the internal Bessel filter at 10 kHz. Currents were sampled at intervals of 20 s with a PCI-6111E acquisition card (National Instruments) and recorded to hard disk using the program Acquire (Bruxton Corp.). Channel opening and closing transitions were determined using the program TAC 4.2.0 (Bruxton Corp.), which digitally filters the data (Gaussian response; final effective bandwidth, 5 kHz), interpolates the digitized points using a cubic spline function, and detects events using the half-amplitude threshold criterion, as described previously (43). To determine single channel current amplitudes, the variable amplitude option in TAC was used, whereas to determine open and closed dwell times, the fixed amplitude option was used. Dwell time histograms were plotted using a logarithmic abscissa and square root ordinate (44), with a uniformly imposed dead time of 40 s, and fitted by the sum of exponentials by maximum likelihood using the program TACFit 4.2.0 (43). Clusters of channel openings were identified as a series of closely spaced openings preceded and followed by closed intervals longer than a specified critical time ( crit ). This duration was taken as the point of intersection between consecutive components in the closed time histogram and ranged between 1 and 30 ms. Cluster duration therefore comprises the total time of a series of openings plus that of the intervening closings briefer than crit .