Subtype-Selective Inhibition of Neuronal Nicotinic Acetylcholine Receptors by Cocaine Is Determined by the α4 and β4 Subunits

Materials and Methods

Chemicals.

Acetylcholine stock solutions were made daily and diluted to the working concentration in oocyte saline solution. Cocaine hydrochloride was provided by the National Institute on Drug Abuse and concentrated stock solutions were stored frozen in aliquots. All other chemicals were purchased from Sigma (St. Louis, MO) unless otherwise noted.

Production of Chimeric cDNAs and Mutagenesis.

Rat nAChR cDNA clones were provided by Drs. Steve Heinemann and Jim Boulter (Salk Institute, La Jolla, CA). The production and functional evaluation of chimeric β subunits has been described previously (Papke et al., 1993). For these chimeras (β2₂₂₈/β4 and β4₂₃₀/β2), the numeric subscript refers to the number of N-terminal amino acids contributed by a given β subunit. Chimeric nicotinic receptor α subunit cDNAs were created using homologous restriction sites conserved between α subunit sequences. Specifically, a BstXI site located in a stretch of conserved sequence that codes for the extracellular loop region between M2 and M3 was used to create chimeric subunits that exchange sequence between α3 and α4 (α3₂₆₇/α4 and α4₂₆₇/α3). Similarly, a homologousPstI site located immediately after sequence coding for the first of the two cysteines in the N-terminal extracellular domain (which form the cysteine loop characteristic of the nicotinic gene family) was used to create chimeric subunits which exchange sequence between α2 and α4 (α2₁₂₈/α4 and α4₁₂₈/α2). In each case, the numeric subscript refers to the amino acid immediately preceding the site at which the coding sequence is exchanged. α Subunit amino acids are numbered according to the system adopted for the muscle α subunit, in which the vicinal cysteine residues characteristic of nicotinic receptor α subunits are numbered 192 and 193. Restriction fragments were isolated from an agarose gel using the Geneclean III kit (Bio 101,Vista, CA) and subsequently ligated by homologous end ligation. After bacterial transformation, chimeric DNAs were evaluated by restriction digest and subsequent DNA sequencing.

Single mutations were introduced using the Quickchange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Briefly, antiparallel mutagenic oligonucleotides were synthesized containing the base(s) coding for the mutation(s) of interest flanked by 10 to 15 bases of wild-type sequence. During consecutive rounds of temperature cycling, the mutant oligonucleotides prime extension of each of the strands of the parental clone by Pfu DNA polymerase. After amplification, parental (methylated) DNAs were digested byDpnI and the remaining DNAs were transfected into competent bacteria. Mutant cDNAs were later confirmed by DNA sequencing.

Sequence alignments were made using ClustalW 1.7 on the BCM server (http://dot.imgen.bcm.tmc.edu:9331/multi-align/multi-align.html) and shaded for presentation using Boxshade 3.21 (http://www.ch.embnet.org/software/BOX_form.html).

Preparation of RNA and Oocyte Injection.

After linearization and purification of cloned cDNAs, RNA transcripts were generated using the mMessage mMachine in vitro RNA transcription kit (Ambion, Austin, TX). Resultant RNA transcripts were evaluated by UV spectroscopy and denaturing agarose gel electrophoresis (visualized with ethidium bromide). RNAs were diluted to a concentration of 600 ng/μl and stored frozen in ribonuclease-free water at −80°C.

Ovarian lobes were surgically removed from anesthetized adult femaleXenopus laevis frogs and then cut open to expose the oocytes. The ovarian tissue was then treated with collagenase for about 2 h at room temperature (1 mg/ml in oocyte saline solution: 82.5 mM NaCl, 2.5 mM KCl, 1 mM NaH₂PO₄, 15 mM HEPES, 1 mM MgCl₂, pH 7.4). After harvest, healthy stage-5 oocytes were isolated and injected with 50 nl each of a mixture of the appropriate subunit RNAs. Sterile oocyte storage medium (oocyte saline solution supplemented with 1.8 mM CaCl₂, 5 U/ml penicillin, 5 μg/ml streptomycin, and 5% horse serum) was changed daily. Recordings were made 2 to 7 days after injection depending on the RNAs being tested.

Electrophysiology.

Two-electrode voltage-clamp recordings were made at room temperature in oocyte saline solution supplemented with 1.8 mM CaCl₂ and 1 μM atropine. All recordings were made using a Turbo Tec 01C amplifier (Adams & List, Westbury, NY) at a holding potential of −50 mV unless otherwise noted. Recording electrodes were filled with 3 M KCl and typically had resistances in the range of 0.5 to 3 MΩ.

Oocyte recording solution was perfused at a rate of 5 ml/min through a Lucite recording chamber via a large bore pipette (1.5 mm diameter) placed about 0.5 mm above the oocyte. Agonist and antagonist solutions were applied by loading a loop near the terminus of one arm of the perfusion line. Constant perfusion was maintained by switching to the other arm of the perfusion line during loading of the drug loop. Antagonist solution can be pre-equilibrated by switching to a second reservoir before loading of the drug loop. Perfusion of oocyte saline solution from an independent reservoir at a rate of 2 ml/min maintained bulk flow through the recording chamber at all times. Based on the rise time of current responses, solution exchange times in the range of 500 to 800 ms are achieved under these conditions. Data were collected at a sampling rate of 100 Hz on a Gateway (N. Sioux City, SD) personal computer using Clampex 7 (Axon Instruments; Foster City, CA). From the time of each drug application, 2 min of data were acquired.

Each experimental response was normalized to an initial control response to agonist alone. A second control application of agonist alone subsequent to the experimental application permitted assessment of inhibition time course and receptor rundown. Each drug application was separated by a wash period of approximately 4 min. Values for EC₅₀, the Hill coefficient, and IC₅₀ were estimated from curve fits to normalized data using Kaleidagraph 3.08 (Abelbeck/Synergy Software; Reading, PA). Data for receptor activation by acetylcholine were plotted using a nonlinear least-squares fit to the Hill equation:For voltage-dependence experiments, e-folding voltages were calculated from exponential fits to data of the form Λ-1, where Λ represents the ratio of control response to response in the presence of inhibitor (Ascher et al., 1979). The IC₅₀ value was calculated with a nonlinear, least-squares fit to the equation:Because the inhibition is noncompetitive, IC₅₀ was taken to beK_I and the ΔG° was calculated as RTlnK_I, where R is 1.987 kcal/mol and T is the absolute temperature in Kelvin. The standard deviation for ΔG° (ς_ΔG°) was calculated from the expression:and the error in ΔΔG° from:where ΔΔG° is ΔG°₁ − ΔG°₂.

Initial experiments in which cocaine was coapplied with acetylcholine exhibited variability in terms of degree of inhibition by cocaine. Presumably, this variability was related to nonuniformities in perfusion of the entire surface area of the oocyte. Therefore, in later experiments, including all those described in this article, cocaine was pre-equilibrated with the oocyte for a period of about 30 s before application of agonist. This protocol allowed more reliable quantification of the degree of inhibition by cocaine. Moreover, the IC₅₀ value for cocaine inhibition of the α3β4 subunit combination was approximately 17 ± 2 μM with coapplication, compared with 6 ± 1 μM using a 30-s preincubation.

Results

Cocaine Inhibition of nAChR Subtypes.

Because the oocyte expression system permits regulated expression of nAChRs of known subunit composition, it is possible to associate pharmacological effects with specific subunit types. The effect of application of 10 μM cocaine with 30 μM ACh on a number of nAChR subunit combinations is shown in Fig. 1A, whereas inhibition curves for the various subunit combinations are shown in Fig. 1B (IC₅₀ values are summarized in Table1). The relative efficacy of the 30 μM agonist concentration in the absence of cocaine ranged from 41 to 58% of the maximal current (I_max) for all of the wild-type receptor combinations tested with the exception of α2β2 and α2β4 (12 and 18%, respectively; Table2). Because of the lower potency for activation of α2β2 and α2β4 receptors, the effects of cocaine were also tested in combination with an agonist concentration near the EC₅₀ value for these receptor subtypes (130 μM ACh). No significant difference in cocaine IC₅₀value was detected for either subunit combination (data not shown).

The effects of cocaine were evaluated initially on nAChRs formed from the α3 subunit in combination with either the β2 or β4 subunit. Nicotinic receptors, including the α3 subunit, predominate in the peripheral nervous system, where they function to mediate transmission in the autonomic ganglia (Xu et al., 1999). Recent reports have described functional roles for brain α3-containing nAChRs as well (Luo et al., 1998; Quick et al., 1999). Cocaine inhibits α3β4 receptors with an IC₅₀ value of approximately 6 μM. The role of the β subunit in mediating the inhibitory effects of cocaine was evaluated by comparison with expression of the β2 subunit in combination with α3. The IC₅₀ value for cocaine inhibition of α3β2 nAChRs is about 10-fold higher (60 μM) than observed for α3β4 receptors, indicating lower apparent affinity for this subunit combination. Based on the results for α3-containing receptors, it seems likely that a high-affinity site for cocaine binding is determined by the presence of the β4 subunit, whereas any site(s) associated with the β2 and α3 subunits are relatively low-affinity.

Because the most widespread nAChR subunit combination in the brain with a high affinity for nicotine contains the α4 and β2 subunits (Flores et al., 1992), potential effects of cocaine on this subunit combination are also of interest. Expression of the α4 subunit with the β2 subunit increases apparent cocaine affinity about 4-fold compared with α3β2 nAChRs (Fig. 1B), indicating that in the case of the α4β2 subunit combination, the presence of the α4 subunit increases cocaine binding affinity. In contrast, α2β2 receptors exhibit a lower apparent cocaine affinity, similar to that of α3β2 receptors. Consistent with the results described above for the β4 subunit, α2β4 receptors exhibit higher apparent cocaine affinity than α2β2 receptors. Pairing of the β4 subunit with the α4 subunit further increases apparent cocaine affinity. Cocaine inhibits α4β4 receptors with an IC₅₀ value of about 2 μM, suggesting that sites on each of the α4 and β4 subunits determine apparent affinity for this subunit combination.

Calculation of differences in free energy change (ΔΔG°) associated with cocaine binding across subunit combinations permitted comparison of the effects of individual subunits (Fig. 1C). In general, this analysis was consistent with the hypothesis that differences in apparent affinity were caused by independent contributions of the α and β subunits. However, cocaine binding to the α2β4 subunit combination has a lower ΔG° value than would be predicted from the other measurements, suggesting a more complicated interaction with this subunit combination. The ΔΔG° value associated with the presence of α4 is consistent for comparisons across all subunit combinations tested with the exception of α2β4 (mean of −0.67 ± 0.27 kcal/mol for α4β2 versus α3β2, α4β2 versus α2β2, and α4β4 versus α3β4 compared with −1.17 ± 0.17 for α4β4 versus α2β4). Likewise, the ΔΔG° associated with the presence of β4 is consistent across subunit combinations with the exception of α2β4 (average of −1.24 kcal/mol for α3β4 versus α3β2 and α4β4 versus α4β2 compared with −0.54 ± 0.20 for α2β4 versus α2β2). Moreover, the ΔΔG° value between the subunit combination with the lowest apparent affinity (α3β2) and the receptor type with the highest apparent affinity (α4β4) is about −1.9 ± 0.2 kcal/mol, consistent with additive rather than synergistic effects of the two subunits.

Contribution of β Subunit to Inhibition by Cocaine.

Individual nicotinic receptor subunits share a conserved membrane topology composed of an N-terminal extracellular segment followed by four hydrophobic transmembrane domains. Comparison of the β2 and β4 subunit amino acid sequences reveals a number of differences in the N-terminal putative extracellular domains of the two subunits (Fig.2A). In contrast, only a single amino acid differs between the pore-lining second transmembrane (M2) domains of the two subunits. To distinguish between these two potential regions of cocaine interaction (N terminus versus membrane-spanning domains), the effects of cocaine were tested on receptors incorporating chimeric β subunits in which all sequence N-terminal to M1 is exchanged between β2 and β4 (Fig. 2B). Functional effects of these exchanges on the kinetics of inhibition by neuronal bungarotoxin have been evaluated previously (Papke et al., 1993).

Receptors incorporating the α3 subunit with β subunits that exchange the N-terminal domain between β2 and β4 are not appreciably different from wild-type receptors in terms of cocaine sensitivity (Fig. 2, C and D; Table 1), indicating that the site of cocaine interaction with the β subunit is located C-terminal to M1. As measured from the application of 30 μM ACh with a range of cocaine concentrations, the IC₅₀ values for cocaine inhibition of wild-type α3β4 receptors and chimeric α3β2₂₂₈/β4 receptors are nearly identical (6 μM for each), whereas wild-type α3β2 receptors and chimeric α3β4₂₃₀/β2 receptors show comparable IC₅₀ values (60 and 35 μM, respectively). The mean response of the chimeric receptors to 30 μM acetylcholine alone was 32% of I_max for α3β2₂₂₈/β4 receptors but only 3% ofI_max for α3β4₂₃₀/β2 receptors. Therefore, the effects of cocaine were also evaluated at an agonist concentration closer to the EC₅₀ value for α3β4₂₃₀/β2 receptors (250 μM). No significant difference in the cocaine IC₅₀ value was detected.

Although exchange of the N-terminal domain between β2 and β4 has little effect on cocaine sensitivity, the exchange does seem to have reciprocal effects on response kinetics (Fig. 2C). Wild-type α3β2 and chimeric α3β2₂₂₈/β4 receptors show qualitatively similar rapid response profiles during the time course of the agonist application (15 s), whereas wild-type α3β4 and α3β4₂₃₀/β2 receptors show more protracted maximal responses. Certain difficulties, such as temporal averaging associated with physical limitations in achieving rapid and homogeneous solution exchange over the surface of the oocyte membrane, are intrinsic to the oocyte system and can complicate reliable quantification of receptor desensitization. Although this caveat should be kept in mind, for this relatively moderate agonist concentration (30 μM), the observed changes in response kinetics with incorporation of the chimeric β subunits are most consistent with altered receptor desensitization properties. Previous studies have described differences in steady-state dose-response relationships (Cohen et al., 1995) and single-channel kinetics (Papke and Heinemann, 1991) between α3β2 and α3β4 receptors. Our observations extend the studies of these subunit combinations by demonstrating that sequence elements contributing to desensitization kinetics are contained within the N-terminal domain of neuronal β subunits.

Owing to the high sequence homology in the M2 domains of receptor subunits in the nicotinic superfamily, a common nomenclature has been adopted that designates the approximately 20 amino acids that comprise an element of the channel pore (Charnet et al., 1990). Homologous amino acid positions in M2 are designated 1′ to 20′ from the intracellular to extracellular extent of the membrane-spanning domain. Because the β2 and β4 subunits differ by only a single residue located at the 13′ position (V in β2; F in β4) and this substitution has been implicated previously as a determinant of affinity for another noncompetitive inhibitor of neuronal nAChRs (Substance P; Stafford et al., 1998), the effects of reciprocal mutation of this residue on cocaine sensitivity were evaluated (Fig.3, A and B; Table 1). Expression of the β2_13′F and β4_13′Vsubunits with the α3 subunit results in functional nAChRs that exhibit altered sensitivities to inhibition by cocaine. Specifically, α3β4_13′V mutant receptors show a decreased sensitivity to inhibition by cocaine to a level comparable with that of α3β2 receptors (IC₅₀ values of 50 μM). Conversely, α3β2_13′F receptors show an almost 4-fold increase in sensitivity to cocaine inhibition (IC₅₀ ∼ 17 μM). However, in this case, the valine to phenylalanine exchange at the 13′ site does not fully recapitulate the cocaine affinity of wild-type α3β4 receptors. Notably, substitution of valine for phenylalanine in the β4 subunit also produces an approximately 6-fold increase in the EC₅₀ value for activation by acetylcholine (Fig.4A) such that 30 μM ACh only produces 4% of the maximal current (versus 45% for α3β4). Again, for α3β4_13′V receptors, the inhibitory effects of cocaine were evaluated at a higher agonist concentration (300 μM) without any significant effect on IC₅₀ values.

A similar effect on acetylcholine potency was noted above for α3β4₂₃₀/β2 (5.4-fold increase in EC₅₀) receptors suggesting that the 13′ position is responsible for much of the effect on agonist potency associated with incorporation of the chimeric β subunit. Reciprocal mutation of the 13′ residue of the β2 subunit produces more modest effects (1.7-fold) on agonist potency (Fig. 4B; Table 2), indicating that this residue may not serve functionally homologous roles across the two classes of β subunit. From these data, it is not possible to distinguish between effects of this substitution on agonist binding and channel gating. However, this functional distinction between the 13′ position of β2 and β4 may underlie our inability to achieve fully reciprocal effects on cocaine inhibition with mutation of the 13′ residue. Alternatively, other sequence elements, such as adjacent transmembrane domains, may also directly or indirectly contribute to the binding site for cocaine.

Contribution of α Subunit to Inhibition by Cocaine.

The N-terminal domain of nicotinic receptor α subunits includes at least three stretches of amino acids previously demonstrated to contribute to agonist binding (loops A–C; for review, see Changeux and Edelstein, 1998). Thus, it is possible that cocaine inhibits α4β2 receptors by a competitive mechanism involving amino acids in the N terminus of the α4 subunit. Alternatively, cocaine may inhibit α4β2 receptors by a noncompetitive mechanism involving binding to either extracellular or pore sites (including the N and C termini, transmembrane domains, and extracellular loop region between M2 and M3).

Initial experiments evaluated the mechanism of cocaine inhibition. For a competitive inhibitor, application of the IC₅₀concentration of cocaine would be expected to produce little inhibition in the presence of saturating agonist concentration. Coapplication of 30 μM cocaine (a concentration near the IC₅₀value for α4β2 receptors as measured in the presence of 30 μM ACh; Table 1) with a range of ACh concentrations (30 μM–1 mM; Fig.5) indicates that cocaine is a noncompetitive inhibitor of this receptor subtype. Application of cocaine in the presence of a higher ACh concentration (3 mM) also produces significant inhibition (45%), consistent with noncompetitive inhibition.

To localize structural determinants of cocaine affinity on the α4 subunit, a similar strategy to the one described above was used. Because α4β2 receptors exhibit a higher apparent affinity for cocaine than do α2β2 or α3β2 receptors, a series of chimeric subunits that exchange increasing amounts of N-terminal sequence between the α4 subunit and either the α2 or α3 subunits were created. The effects of these exchanges on cocaine inhibition were evaluated. The chimeric α3₂₆₇/α4 and α4₂₆₇/α3 subunits exchange sequence at a conserved BstXI restriction site located at the approximate midpoint of the subunits in a region of sequence coding for the extracellular loop between M2 and M3 (Fig.6, A and B, top). The chimeric α4₁₂₈/α2 and α2₁₂₈/α4 subunits exchange sequence at a conserved PstI restriction site located in a region coding for the cysteine loop of the N-terminal extracellular domain (Fig. 6, A and B, bottom). Application of 30 μM cocaine to receptors formed from the coinjection of the β2 subunit with either the α3₂₆₇/α4 or α4₂₆₇/α3 chimeric subunits indicates that a site of high-affinity cocaine binding is associated with the N-terminal half of the α4 subunit (Fig. 6, C and D; IC₅₀values summarized in Table 1). Conversely, application of the same concentration of cocaine to receptors formed from β2 with either the α2₁₂₈/α4 or α4₁₂₈/α2 chimeric subunits indicates that a high-affinity cocaine binding site is associated with sequence C-terminal to the initial cysteine residue of the cysteine loop in the mature α4 protein (Fig. 6, C and D; Table 1). For the α2₁₂₈/α4β2 and α4₂₆₇/α3β2 subunit combinations, application of 30 μM ACh is sufficient to elicit 44 and 52% of the maximum current, respectively, whereas in the case of α3₂₆₇/α4β2 and α4₁₂₈/α2β2 receptors, 30 μM ACh elicits only 23% of the maximum current. Again, evaluation of cocaine inhibition at higher agonist concentrations (170 μM for α3₂₆₇/α4β2 and 220 μM for α4₁₂₈/α2β2) yielded no significant change in estimated IC₅₀ values. Taken together, these results localize a determinant of cocaine affinity to a portion of the α4 subunit that includes approximately 90 amino acids of the N-terminal extracellular domain with M1 and M2. Because α2β2 and α3β2 receptors show similarly mild sensitivity to inhibition by cocaine and α4β2 nAChRs exhibit a higher apparent affinity (Fig.1), it seems reasonable to hypothesize that sequence determinants on the α subunits are conserved between α2 and α3, whereas they are nonconserved in α4. Using this logic in the context of the high degree of sequence identity that is preserved across the transmembrane domains of α subunits, it is possible to identify a stretch of approximately 50 amino acids immediately preceding M1 that likely determine cocaine affinity in α4 (Fig. 6A). However, reciprocal mutation between α4 and either α2 or α3 of eight different nonconserved residues considered to be the most likely candidates for mediating cocaine binding in this region produced receptors that differed only slightly in cocaine sensitivity, suggesting that cocaine binding may involve multiple amino acids. Given the complex folding that is presumably required for formation of the agonist binding pocket in this region, regulation of cocaine affinity for the α4 subunit may be a distributed phenomenon requiring contributions from multiple residues in the stretch of sequence identified by the subunit chimeras. Alternatively, the data do not rule out a contribution from the few nonconserved residues in M1 and M2.

Voltage Dependence of Cocaine Inhibition.

As a second approach to localizing determinants of cocaine affinity, we sought to evaluate the disposition of cocaine binding sites with respect to the membrane electric field. The effects of changes in holding potential on inhibition were tested for the α3β4, α3β2_13′F, and α4β2 subunit combinations (Fig. 7). These subunit combinations were chosen to ensure that only a single class of binding site was evaluated in each case. For example, in the case of α3β4 receptors, any effects of holding potential on inhibition are likely associated primarily with a site on the β4 subunit because α3β2 receptors show little inhibition by cocaine at the concentration used in these experiments (10 μM; Fig. 1). Similarly, in the case of α4β2 receptors, it is likely that any observed effects of holding potential are primarily associated with the α4 subunit because the β2 subunit shows only low cocaine affinity.

Consistent with binding to a site in the channel pore, inhibition of α3β4 receptors by cocaine shows voltage dependence. Application of 10 μM cocaine at a standard holding potential of −50 mV results in approximately 56% inhibition (Fig. 1). Application of the same concentration of cocaine at a holding potential of −100 mV results in about 80% inhibition, whereas raising the holding potential to −20 mV decreases the observed inhibition to about 30% of control. Inhibition appears linear over the range of holding potentials tested, with an e-fold change in inhibition for a 34-mV change in holding potential. Inhibition of the α3β2_13′F mutant also exhibits appreciable voltage dependence (e-folding voltage of 45 mV, data not shown).

In contrast to the voltage dependence of inhibition observed for α3β4 receptors, inhibition of α4β2 receptors by cocaine exhibits a much weaker voltage dependence across the range of potentials tested. Specifically, application of 20 μM cocaine to α4β2 receptors (a concentration approximately equipotent to the 10 μM concentration used for α3β4 receptors) results in about 70% inhibition at −100 mV, whereas 50% inhibition is observed at a holding potential of −20 mV (e-folding voltage of 90 mV).

Differences in the voltage dependence of cocaine inhibition across the two receptor types suggest that cocaine inhibition arises via mechanistically distinct processes. In the case of α3β4 receptors, the data are consistent with binding to a channel site. However, the voltage-dependence data argue against a similar mechanism for inhibition of α4β2 nAChRs. Given the high degree of sequence identity in the M2 regions of α2–α4 (97%, Fig. 6A), the voltage-dependence data are most consistent with the presence of an extracellular binding site for cocaine associated with the α4 subunit.

Discussion

Although the addictive properties of cocaine are likely mediated via interactions with neurotransmitter transporters, our data indicate that cocaine can also inhibit members of the nicotinic family of nervous system LGICs. Serum concentrations of cocaine can rise to a peak near 1 μM within 30 min after nasal insufflation (Jeffcoat et al., 1989) and brain concentrations are likely to be higher because of the lipophilic nature of the drug (Nayak et al., 1976). Our data indicate that cocaine inhibits particular nAChR subtypes with an affinity comparable with that displayed for the open/inactivated state of voltage-gated sodium channels (∼10 μM; Wright et al., 1997), suggesting the possibility of neuronal nAChR-mediated cocaine effects. Cocaine potency shows considerable variability across nAChR subtypes, with IC₅₀ values ranging between 2 μM (α4β4) and 60 μM (α3β2). Although inhibition of the α3β4 subunit combination is voltage-dependent and affected by mutation of the 13′ residue of the pore-lining M2 domain, inhibition of the α4β2 subtype shows only very weak voltage dependence and is probably associated with N-terminal sequence immediately preceding M1 (Fig. 8).

The present study also identified β subunit residues important in the processes of receptor desensitization (Fig. 2) and activation (Fig. 4). Although the observed effects on receptor desensitization are associated with the N-terminal portion of the β subunit, the most significant changes in receptor activation properties are associated with mutation of the 13′ residue of the β4 subunit. The processes of desensitization and binding/gating do not seem to be determined by common sequence elements because mutation of the 13′ site does not yield appreciable differences in response time course (Fig. 3). A number of previous studies have illustrated effects of mutation of pore residues on channel gating. Most notably, the 9′ position has also been implicated in the gating/binding process (Revah et al., 1991; Filatov and White, 1995; Labarca et al., 1995; Kearney et al., 1996). The observation that mutation of the 13′ position has differential effects on receptor activation across the β2 and β4 subunits emphasizes the idea that sequence homologues in the pore region are not necessarily functional homologues.

Inhibition of α3β4 Receptors May Involve Open-Channel Block.

Receptors incorporating the α3 subunit with chimeric subunits that pair N-terminal β2 sequence with β4 sequence C-terminal to M1 show sensitivity to cocaine inhibition comparable with wild-type α3β4 receptors whereas the reciprocal chimeric receptors (N-terminal β4 sequence with C-terminal β2 sequence) are inhibited by cocaine with a potency similar to wild-type α3β2 receptors. This result localizes the site of high-affinity cocaine binding to the C-terminal portion of the β4 subunit beyond M1. Within this region, exchange of the 13′ site between β2 and β4 has reciprocal effects on the concentration dependence of cocaine inhibition, implicating the 13′ position as a determinant of cocaine binding affinity in α3β4 receptors. Consistent with binding to a site in the ion channel pore, cocaine inhibition of α3β4 receptors exhibits voltage dependence. Given this observation and the dependence on sequence in the M2 region, cocaine may act as a conventional open-channel blocker on the α3β4 subtype of nAChR. However, the observation that cocaine has a lower potency at α3β2_13′F receptors compared with wild-type α3β4 receptors or chimeric α3β2₂₂₈/β4 receptors may indicate that other sequence elements not contained within the M2 domain can contribute to the disposition of the cocaine binding site. Residues in the adjacent membrane-spanning domains or connecting loops could contribute directly to a cocaine-binding site or otherwise affect the exposure of the 13′ residue to the channel pore. Additionally, because the functional role of the 13′ position in receptor activation does not seem to be maintained across classes of β subunit (Fig. 4), exposure of this residue may differ during gating of the two receptor subtypes.

Inhibition of both the dopamine transporter and sodium channels by cocaine involves binding to phenylalanine and/or other aromatic residues (Ragsdale et al., 1994; Lin et al., 1999), suggesting a common mechanism of action across classes of protein. The importance of cation-π interactions between charged groups and aromatics in receptor-ligand interactions and protein structure has recently emerged (for review, see Dougherty, 1996). Our data may indicate that the protonated amine group of cocaine interacts with the aromatic ring of phenylalanine. In this case, voltage dependence might arise from the essentially electrostatic nature of the cation-π interaction. Alternatively, the phenyl ring of cocaine could interact with the aromatic group of the 13′ phenylalanine. In this case, the amine moiety may project deeper into the pore. The size of cocaine is estimated to be on the order of 6 × 12 Å. Given the partially helical nature of nAChR subunit M2 regions (Akabas et al., 1994; Unwin, 1995), the protonated amine group of cocaine may project as deep as the 6′ position (∼5.5Å/turn). In either scenario, selectivity for β4-containing receptors arises from the presence of phenylalanine at the 13′ site, a feature that distinguishes cocaine inhibition of β4-containing neuronal nAChRs from both quaternary local anesthetic inhibition of muscle nAChR (Charnet et al., 1990) and inhibition of α3β4 receptors by mecamylamine (Webster et al., 1999), both of which involve binding to the 6′ and 10′ positions. As the human and rat forms of the nAChR β4 subunit are the only subunits in the nicotinic superfamily that include a phenylalanine at this position, inhibition of nAChRs by this mechanism is restricted to receptors containing the β4 subunit. A third possibility that seems less likely is that the 13′ site contributes allosterically to a cocaine binding site located distal from the pore.

Potential Allosteric Inhibition of α4β2 Receptors.

The α subunit chimeras localize a determinant of cocaine affinity to a region that includes a portion of the N-terminal extracellular segment with M1 and M2. On the basis of sequence identity alone, contributions from a large number of residues in this region, particularly residues in the membrane-spanning domains, can be eliminated (Fig. 6A). By concentrating on residues that differ between α4 and the other α subunits, a likely determinant of cocaine affinity can be further localized to a relatively nonconserved stretch of 50 amino acids immediately preceding M1. A region containing the vicinal cysteine pair thought to be a component of the acetylcholine binding site (loop C) is also contained within this stretch of amino acids, suggesting that cocaine may act in the vicinity of the agonist binding site. Consistent with a previous study (Lerner-Marmarosh et al., 1995), our results indicate that cocaine is not competitive with ACh over a 100-fold range in ACh concentration (30 μM–3 mM; Fig. 5). Therefore, cocaine probably binds to an allosteric site on α4β2 receptors and inhibits activation in a noncompetitive fashion.

Possible Functional Relevance of Cocaine Effects on nAChRs.

Nicotinic receptors containing the α4 and β2 subunits are the predominant form of high-affinity nicotine receptors in the brain and the activation and/or subsequent desensitization of this subunit combination by nicotine is believed to be responsible for the addictive properties of the drug (Peng et al., 1994; Dani and Heinemann, 1996;Picciotto et al., 1998). Thus, this receptor type could provide a site of interaction for cocaine and nicotine effects. Synergistic effects between nicotine and the noncompetitive inhibitor mecamylamine on receptor turnover have been described previously (Peng et al., 1994).

Because nicotine is a desensitizing agonist with residual inhibitory properties on nAChRs expressed in oocytes (de Fiebre et al., 1995;Fenster et al., 1997), the physiological agonist acetylcholine was used in these studies to examine the inhibitory properties of cocaine independent of those of nicotine. However, another recent study has examined cocaine antagonism of both behavioral measures of nicotine-elicited effects, such as antinociception, and nicotine activation of heterologously expressed receptor subtypes, demonstrating significant inhibition in both cases (Damaj et al., 1999).

A few recent reports have demonstrated the functional significance of α3β4-containing nAChRs in the CNS. Functional studies employing an α3β4-selective conotoxin have demonstrated the importance of α3β4-containing nAChRs in the medial habenula (Quick et al., 1999) and in mediating nicotine-elicited release of noradrenaline in the hippocampus (Luo et al., 1998). Moreover, nicotine-elicited hippocampal noradrenaline release is inhibited by cocaine with an IC₅₀ value of <1 μM (Hennings et al., 1999). Although these native subunit combinations are likely to be more complex than the simple pair-wise combinations used in our studies, these results are consistent with our data and underscore the possibility of central effects associated with inhibition of neuronal nAChRs.

In light of the growing appreciation that nicotinic receptors play a modulatory role centrally via regulation of the release of other neurotransmitters and activation of certain classes of interneurons, interactions between cocaine and nAChRs could be of considerable importance. Although the euphoric effects of cocaine are probably mediated by the dopaminergic system, given the widespread involvement of nAChRs in the peripheral nervous system (Xu et al., 1999), nAChR inhibition could also be contributory to peripherally mediated toxic effects of cocaine. Although additional studies in situ are obviously required, the present results indicate that neuronal nAChRs constitute an additional site of potential cocaine action in the nervous system that should be incorporated into conceptual models of the drug's pharmacology.