Nicotinic Acetylcholine Receptors Assembled from the α7 and β3 Subunits*

Intracellular recordings were performed in voltage-clamped Xenopus oocytes upon injection with a mixture of cDNAs encoding the β3 and mutant α7 (L247Tα7) neuronal nicotinic acetylcholine receptor (nAChR) subunits. The expressed receptors maintained sensitivity to methyllycaconitine and to α-bungarotoxin but exhibited a functional profile strikingly different from that of the homomericL247Tα7 receptor. The heteromericL247Tα7β3 nAChR had a lower apparent affinity and a faster rate of desensitization than L247Tα7 nAChR, exhibited nonlinearity in the I-V relationship, and was inhibited by 5-hydroxytryptamine, much like wild type α7 (WTα7) nAChR. Single channel recordings in cell-attached mode revealed unitary events with a slope conductance of 19 picosiemens and a lifetime of 5 ms, both values being much smaller than those of the homomeric receptor channel. Upon injection with a mixture of WTα7 and β3 cDNAs, clear evidence was obtained for the plasma membrane assembly of heteromeric nAChRs, although ACh could not activate these receptors. It is concluded that β3, long believed to be an orphan subunit, readily co-assembles with other subunits to form heteromeric receptors, some of which may be negative regulators of cholinergic function.

Intracellular recordings were performed in voltageclamped Xenopus oocytes upon injection with a mixture of cDNAs encoding the ␤3 and mutant ␣7 ( L247T ␣7) neuronal nicotinic acetylcholine receptor (nAChR) subunits. The expressed receptors maintained sensitivity to methyllycaconitine and to ␣-bungarotoxin but exhibited a functional profile strikingly different from that of the homomeric L247T ␣7 receptor. The heteromeric L247T ␣7␤3 nAChR had a lower apparent affinity and a faster rate of desensitization than L247T ␣7 nAChR, exhibited nonlinearity in the I-V relationship, and was inhibited by 5-hydroxytryptamine, much like wild type ␣7 ( WT ␣7) nAChR. Single channel recordings in cell-attached mode revealed unitary events with a slope conductance of 19 picosiemens and a lifetime of 5 ms, both values being much smaller than those of the homomeric receptor channel. Upon injection with a mixture of WT ␣7 and ␤3 cDNAs, clear evidence was obtained for the plasma membrane assembly of heteromeric nAChRs, although ACh could not activate these receptors. It is concluded that ␤3, long believed to be an orphan subunit, readily co-assembles with other subunits to form heteromeric receptors, some of which may be negative regulators of cholinergic function.
The 11 neuronal nicotinic acetylcholine receptor (nAChR) 1 subunits cloned to date fall into two classes depending on whether an ␣/␤ pair is required for assembly (obligatory heteromers) or whether they can assemble into functional homomeric receptors. Combinations of the ␤2 and ␤4 subunits with one or more of the ␣2 to ␣6 subunits form functional nAChRs in reconstituted systems (1)(2)(3)(4)(5)(6), although the particular combinations of these subunits that co-assemble in vivo are largely unknown. In contrast, the ␣7 to ␣9 subunits form homomeric nAChRs in cellular expression systems, even though it is not yet clear whether these subunits can also form heteromeric nAChRs in nerve cells (1,3,5,(7)(8)(9)(10)(11)(12). ␤3, the remaining sub-unit, was long referred to as an orphan because it could not be ascertained whether it assembled into functional nAChRs either in neurons or in cellular reconstitution systems (13)(14)(15)(16). Recent biochemical (4) and physiological evidence (17) does strongly suggest, however, that ␤3 participates in the assembly of nicotinic receptors, together with the ␣3, ␣4, ␤2, and ␤4 subunits.
Using the idiosyncratic properties of the L247T ␣7 receptor as a tool to investigate whether ␤3 can participate in functional receptor formation, we injected ␤3 and L247T ␣7 cDNAs into the nuclei of Xenopus oocytes and studied the functional properties of the expressed nAChRs. We report here that coinjection of ␤3 and L247T ␣7 cDNAs causes the formation of an heteromeric nAChR with functional properties clearly different from those of the homomeric L247T ␣7 receptor. We also show that ␤3 coassembles with wild type ␣7 ( WT ␣7), but that the resulting heteromeric nAChR is insensitive to ACh.

EXPERIMENTAL PROCEDURES
cDNAs and Expression Vectors-The cDNAs encoding chick WT ␣ 7 and L247T ␣ 7 neuronal nAChR subunits were cloned into the Flip vector, where transcription is under the control of the SV40 early promoter (7). To maximize expression of the chick ␤3 subunit, its cDNA was subcloned in the pMT3 expression vector (21) under the control of the efficient adenovirus major late promoter. The same vector was used to express the V250T ␤3 and the S243C ␤3 mutant subunits.
Site-directed Mutagenesis-Mutagenesis of ␤3 was carried out using the QuickChange site-directed mutagenesis system (Stratagene). We induced the point mutations using high performance liquid chromatography-purified mutant oligonucleotides with 15-16 nucleotides flanking the mutated bases. All mutations were confirmed by sequence analysis based on the Sanger dideoxy termination method (manually or with an ABI 377 sequenator). To verify the absence of undesired nucleotide changes, the entire coding region was sequenced.
Oocyte Injections-Full-length cDNAs encoding the WT ␣ 7 and L247T ␣ 7 subunits were expressed as described previously (18). The ␣7 plasmids * This work was supported by grants from Ministero Università Ricerca Scientifica e Tecnologica (to F. E.), by grants from the Swiss National Science Foundation (to M. B.), and by a Telethon fellowship (222/bi (to E. P.)). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Voltage-clamp Recordings-Membrane currents were recorded in the voltage-clamp mode 1-4 days after injection using 2 microelectrodes filled with 3 M KCl. The oocytes were placed in a recording chamber (0.1 ml) perfused continuously with oocyte Ringer's medium (82.5 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl 2 , 1 mM MgCl 2 , 5 mM Hepes, adjusted to pH 7.4 with NaOH) at controlled room temperature (20 -21°C). During most experiments, the oocytes were held at Ϫ60 mV to reduce the possible contribution of endogenous Ca ϩ2 -activated Cl Ϫ currents. To construct dose-response relationships, oocytes were held at Ϫ60 mV, and the drugs were applied at 3-min intervals. Current-voltage relationships were determined using repetitive exposures to ACh at various potentials, stepping the holding potential from Ϫ60 mV to the desired voltage 5 to 10 s before transmitter application. Drugs were dissolved in oocyte Ringer's solution and applied by superfusing the oocyte at a flow rate of 12 ml/min. In particular, MLA was applied for 30 min before testing ACh responses (25). Solution exchange was achieved by using electromagnetic valves (Type III, General Valve). Drugs and chemicals were purchased from Sigma, except MLA (Research Biochemicals International) and the MTS derivatives (Toronto Research Chemicals, Canada).
Experimental Analysis-The current records were digitized at 50 -200 Hz using an analog-to-digital converter (Digidata 1200 Interface, Axon) and stored on a computer for subsequent analysis using pClamp 6.0.2 routines (Axon). For more details, see Ref. 18. To determine the half-dissociation constant (EC 50 ) of ACh, data were fitted using nonlinear fitting routines (included in Sigma Plot, Jandel), to the Hill equation: where [ACh] is the transmitter concentration, n H is the Hill coefficient, and I max is the maximum response. To assess the functional behavior of nAChRs, we defined some parameters as listed below. Receptor sensitivity to the transmitter was estimated by calculating the ratio (in percentage) of current elicited by 0.2 M ACh (I 0.2 ) to that elicited at 100 M ACh (I 100 ), which concentrations, respectively, correspond to the EC 50 and to the I max of L247T ␣7 (12, 18 -20). The time to half-decay (T 0.5 ) of the inward current activated by ACh (I ACh ), defined as the time taken for the current to decay from peak to half-peak value and the time to 10% decay (T 0.1 ), were used to estimate the rate of receptor desensitization. The deviation of the I-V curve from linearity was estimated by the ratio of slope conductances at ϩ30 mV and Ϫ60 mV. This rectification coefficient (n ␥ ) (12) ranges between 100% (i.e. linear I-V relationship) and 0% (i.e. full rectification).
Single-channel Recordings-Single-channel currents were recorded from the animal pole of the oocytes using the patch-clamp technique in the cell-attached mode, as reported (26,27). Unless otherwise indicated, the ACh concentrations used were in the range 3-5 M. At these concentrations, the open-channel frequency was variable in the range 2-20 Hz. Recordings were performed using an Axopatch 200B amplifier (Axon). The patch was discarded if no events were detected within 60 s after seal formation at 0 to 40 mV pipette potential or if opening frequency was below 0.5 Hz. Typically, a successful patch was stable for 5-15 min and had Ͼ200 opening transitions. No channel openings were observed in uninjected oocytes or in injected oocytes examined with a patch pipette filled with an ACh-free solution. Current recordings were filtered at 2 kHz, sampled at 10 kHz, and analyzed by pClamp 6.0.2 routines (Axon) using a threshold-crossing criterion. Events briefer than 0.2 ms were incompletely resolved and were excluded from the open-time histograms, which thus represent apparent open times. Histograms of amplitudes, apparent open times, and log (shut times) were fitted with a single Gaussian function or with the sum of exponentials as appropriate. Each histogram included from 500 to 2000 events. Burst duration was studied by grouping openings separated by a specific critical time, which was calculated for each patch from the fitted parameters of the shut-time distribution. For each patch, slope conductances were obtained by least squares linear fitting of current-voltage relationships constructed by polarizing the patch potential in the range Ϫ80 mV to 100 mV. For more details, see Ref. 27.
Binding Assay on Oocytes-Oocytes that had been injected with WT ␣7 cDNA or with mixtures of WT ␣7 and ␤3 cDNAs were electrophysiologically tested with 2 mM ACh (28) and assayed for plasma-membrane ␣-Bgt binding. To measure binding, control and injected oocytes were incubated singly for 2 h in 70 l of oocyte Ringer's medium containing 20 nM [ 125 I]␣-Bgt (Amersham Pharmacia Biotech). After incubation, the labeled medium was removed, and the oocytes were washed five times with Ringer's medium (250 l each time) and counted individually on a Beckman gamma counter.
The ACh responses in mixture-injected oocytes were sensitive to MLA, as is the case for oocytes injected with L247T ␣7. For instance, the I 0.2 peak amplitude decreased by 34% at 100 pM MLA in L247T ␣7 oocytes and by 85% in mixture-injected oocytes. Furthermore, I ACh peak amplitude in mixture-injected oocytes decreased by 30% at 3 M ACh and 100 pM MLA, equal to the block on L247T ␣7 nAChR when activated near its EC 50 by 0.2 M ACh (Fig. 3). Taken together these data indicate that MLA acts on both putative heteromeric and homomeric nAChRs in a competitive manner.
It is well established that the clamp-holding current is far greater for cells injected with L247T ␣7 than with WT ␣7. This observation, together with the fact that MLA reduces the holding current, has been explained by the spontaneous, AChindependent activation of the mutant ␣7 receptor (29). In our experiments, oocytes injected with the cDNA mixture showed a spontaneous, MLA-sensitive inward current (2.4 Ϯ 1.3% of the I 100 peak amplitude, [5/2]), quite similar to that estimated in oocytes injected with L247T ␣7 (4.5 Ϯ 1.6%, [9/2]; not shown).
To investigate whether the ratio of cDNAs in the mixture could influence the functional properties of the resulting nAChRs, L247T ␣7 and ␤3 cDNAs were injected into oocytes at ␣7/␤3 ratios ranging from 1:1 to 1:6. No obvious differences in current decay, ACh sensitivity, or I-V relationships were observed. In contrast, the amplitude of the elicited I ACh was significantly impaired as the ␣7/␤3 ratio was increased (not shown).
Single Channel Recordings-To investigate if the co-expression of ␤3 could alter the L247T ␣7 channel parameters, we performed single-channel recordings on mixture-injected oocytes (ratio 1:1). Because the coinjection of ␤3 and L247T ␣7 reduces I ACh , single-channel recordings were only successful in the subset of oocytes that had a high level of expression. A Properties of nAChRs Obtained upon Injection into Oocytes of ␤3 and WT ␣7 cDNAs-A number of experiments were performed in oocytes injected with a mixture of ␤3 and WT ␣ 7 cDNAs to determine whether these subunits can co-assemble into functional nAChRs. In agreement with the experiments using a mixture of ␤3 and L247T ␣7 cDNAs, the amplitudes of the current responses to ACh were considerably reduced in oocytes injected with ␤3 and WT ␣7 cDNAs, as compared with those injected with WT ␣7 cDNA alone. For instance, at 100 M ACh the average I ACh recorded in oocytes injected with WT ␣7 cDNA was Ϫ693.4 nA [24/3], and Ϫ45.4 nA [21/3] with a mixture of WT ␣7 and ␤3 at 1:1. Despite this strong inhibition, no differences in the functional properties of the expressed nAChRs were observed. Furthermore, when estimating receptor sensitivity to nicotine by the ratio (in percentage) of the currents elicited by this agent at 10 M and at 100 M (which approximate the half-maximum and maximum current amplitudes for WT ␣7) (30), we found no significant difference between WT ␣7/␤3 and WT ␣7 oocytes [4/1]. Other experiments tested the functional and pharmacological properties of WT ␣7/␤3 nAChRs (Table I). The apparent affinity, desensitization, current rectification, and sensitivities to MLA or 5HT of this receptor species were identical to the corresponding parameters in WT ␣7 injected oocytes.
Coinjection of Mutant ␤3 and WT ␣7 cDNAs-Given that coinjections of ␤3 and WT ␣7 did not provide evidence for the participation of ␤3 in functional WT ␣7 nAChR complexes, we sought a direct proof by injecting appropriate ␤3 mutants in combination with WT ␣7 cDNAs. We reasoned that particular mutations in the M2 membrane-spanning segment of the ␤3 subunit should yield ACh responses whose parameters would clearly establish co-assembly with ␣7. We injected oocytes with WT ␣7 and ␤3 bearing the mutations V250T or S243C. The first mutant aligns with L247T in the ␣7 subunit, whereas the second aligns with S240C in the ␣5 subunit, both previously shown to be exposed in the nAChR channel (2,23). We observed similar functional and pharmacological receptor profiles in oocytes injected with V250T ␤3 and WT ␣7 cDNAs, as compared with oocytes injected with WT ␣7 cDNA alone ([9/3], not shown). We coinjected the S243C ␤3 mutant and WT ␣7 into oocytes to test whether the MTS-derived thiol reagents MTSET, MTSEA, and MTSES were able to inhibit I ACh, as reported elsewhere for muscle and neuronal nAChRs mutated to cysteine in the homologous residues (2,22,23). We found that MTSET, MTSEA, and MTSES (1, 2.5, and 10 mM, respectively) were all unable to affect I 100 ([8/3], not shown) in the same experimental conditions as described previously (2,23). On the other hand, oocytes injected with S243C ␤3 and L247T ␣7 exhibited a profile similar to that of oocytes injected with wild-type ␤3 and L247T ␣7, but MTSET irreversibly inhibited I ACh by ϳ65%, indicating that the S243C ␤3 mutant is functionally expressed in oocytes and readily assembles with L247T ␣7 subunits.
Surface Expression of nAChRs upon Injection of Mixed ␤3 and WT ␣7 cDNAs-Measurements of plasma-membrane toxin binding were carried out in oocytes coinjected with WT ␣7 and ␤3, and the results compared with those obtained upon injecting WT ␣7 alone. It was found that the level of ␣-Bgt binding in the coinjected oocytes ( WT ␣7/␤3 ratio 1:3) decreased by 20%, whereas the I ACh peak amplitude decreased by ϳ90% of control. Remarkably, a significant number of these oocytes bound large amounts of radiolabeled toxin but had null current responses to 2 mM ACh (Fig. 5). This finding indicates that ␣-Bgt receptors readily assemble in the plasma membrane of coinjected oocytes and that they cannot be activated by ACh. In oocytes coinjected with ␣7 and ␤3 at the ratio 1:6, the relative level of nAChR surface expression established by ␣-Bgt binding remained relatively high, whereas ACh responses all but disappeared (Fig. 5), confirming that the heteromeric WT ␣7␤3 receptor binds ␣-Bgt but is insensitive to ACh.

DISCUSSION
The ␣5, ␣6, and ␤3 neuronal nAChR subunits were long believed incapable of assembling with other subunits to form functional receptors. However, recent evidence demonstrates that ␣5 assembles with ␣4␤2 and that ␣6 assembles with ␤2 and ␤4 to form functional receptors in reconstituted systems (2,6,(31)(32)(33). Here we give evidence that ␤3 assembles with an ␣7 mutant to form a heteromeric nAChR exhibiting functional and pharmacological profiles drastically different from those of oocytes expressing the homomeric mutant ␣7 nAChR. As compared with oocytes expressing the L247T ␣7 receptor, oocytes injected with a mixture of the ␤3 and L247T ␣7 cDNAs exhibit the following distinctive properties. (i) The ACh-activated currents decay much faster, (ii) the I-V relationship becomes nonlinear, (iii) the channel open time, 4.7 ms on average, is much shortened (11 ms for L247T ␣7 channels) (27), (iv) the channel mean slope conductance, 19 pS, is markedly lowered (44 and 58 pS for the two L247T ␣7 conductance classes) (20,27), (v) the apparent binding affinity and the cooperativity are both reduced, and (vi) serotonin acts as a weak agonist/antagonist, being a potent agonist in oocytes expressing L247T ␣7 nAChR (18). Taken together, these findings provide convincing evi-dence that the ␤3 subunit coassembles with L247T ␣7 to form a functional heteromeric receptor.
The L247T ␣7␤3 channels share functional profiles matching those of most of the nicotinic heteromeric receptors studied in oocytes. For instance, (i) channel conductance (19 pS) is similar to that described for chick ␣4␤4, ␣4␤2, ␣3␤2, and ␣3␤4 nAChRs in Xenopus oocytes (18 -24 pS conductance) (2,34,35), (ii) a strong inward rectification develops with oocyte depolarization as for most of the heteromeric nicotinic receptors (34), (iii) the time course of nAChR desensitization compares to that described for other heteromeric receptors and is very different from that of the homomeric ␣7 receptor (7). In addition, its pharmacological behavior appears similar to the profiles described for other heteromeric receptors, with both 5HT and MLA behaving as antagonists, as they do for other heteromeric receptors (36,37).
The only striking difference in the profile of L247T ␣7␤3 nAChR versus the other heteromeric nAChRs described to date is in the sensitivity to ␣-Bgt, a potent blocker of both the L247T ␣7␤3 and the homomeric ␣7 nAChR that is essentially ineffective on all the heteromeric neuronal nAChRs that have been described (reviewed in Ref. 38). This result is in contrast to the general assumption that ␣-Bgt sensitivity is restricted to the homomeric nAChRs and suggests that the ␣-Bgt binding activity detected in native preparations does not prove that homomeric nAChRs are expressed but merely reveals the presence of nAChRs containing the ␣7 or ␣8 subunits.
An open question raised by our data is the subunit composition of the heteromeric L247T ␣7␤3 nAChRs. Upon injection of an equimolar mixture of L247T ␣7 and ␤3 cDNAs, one would expect the assembly of a variety of pentameric receptor species made up of L247T ␣7 and ␤3 subunits, with probability n(1/2) 5 for each of the compositions likely to be functional ( 5 L247T ␣7, n ϭ 1; 4 L247T ␣7 1 ␤3, n ϭ 5; 3 L247T ␣7 2 ␤3, and 2 L247T ␣73␤3, n ϭ 10). On the assumption that the inward current elicited by 5HT is because of the activation of the homomeric ␣7 nAChR, whereas the heteromeric receptors are blocked by 5HT, the estimated homomeric population as a fraction of the I 5HT /I ACh is ϳ5% that of the total nAChR expressed in oocytes injected with the cDNA mixture, in excellent agreement with the theoretical value. The remaining 95% of the nAChR-channel responses result from the activation of L247T ␣7␤3 populations of uncertain subunit composition. Single channel recordings showing a unique channel conductance class and a unique exponential open time component, together with the reduced Hill coefficient of the whole-cell dose-response relationships all suggest that substituting one or more ␤3 subunits into the L247T ␣7 receptor reduces the number of binding sites for the transmitter and affects cooperativity, channel conductance populations, and channel kinetics. Given that data obtained under voltageclamp conditions with the cDNAs mixture are not so scattered as to suggest inhomogeneous channel populations, it could be that formation of nAChR of the standard heteromeric stoichiometry, 2 L247T ␣7 3 ␤3, is favored. As shown in another context (4), ␤3 is thought to enhance the assembly of ␣4␤2 and ␣4␤4 dimers into pentamers of 2 ␣4 1 ␤2 1 ␤3 1 ␤4 composition. We speculate that ␤3 may likewise enhance pentamer assembly by associating with an L247T ␣7␤3 dimer to make a trimer, thus facilitating pentamer formation with available dimers. This predicts that an excess of ␤3 subunits would sharply curtail the amount of assembled pentamers by sequestering all of the L247T ␣7␤3 dimers into 1 L247T ␣7 2 ␤3 trimers incapable of functional assembly. The sharp reduction of mean I ACh we observe at elevated ␤3/␣7 ratios is in qualitative agreement with this model.
A key feature of the homomeric ␣7 nAChR is its high Ca ϩ2 permeability, as compared with heteromeric receptors (39). We note a reduction in the null potential from Ϫ17 mV in homomeric L247T ␣7 to Ϫ10 mV in L247T ␣7␤3, suggesting a reduction of Ca ϩ2 entry through the heteromeric channel, with consequent impairment of the Ca ϩ2 -activated Cl Ϫ component of the ACh response. Because the L247T ␣7 receptor inactivates slowly even in the presence of an intracellular Ca ϩ2 chelator (39), the faster decay of the ACh-evoked currents in the heteromer is unlikely to result from an enhanced Cl Ϫ component and appears to be an intrinsic property of the receptor. This argument is strengthened by the observation that the heteromer maintains its inactivation kinetics at or close to the Cl Ϫ reversal potential.
We have no evidence that ␤3 can assemble with WT ␣7 to form a functional heteromeric receptor in oocytes, although we attempted the co-assembly of ␣7 not only with wild-type but also with mutant ␤3 subunits that should have detectably altered receptor properties. Our data argue that oocytes block the assembly of functional ␣7␤3 wild-type heteromers, while allowing the assembly of functional L247T ␣7␤3 and L247T ␣7 S243C ␤3 receptors. The large plasma membrane binding of ␣-Bgt coupled to the small ACh responses in oocytes injected with a mix of WT ␣7 and ␤3 cDNAs is interpreted as reflecting the expression of a major population of nonfunctional heteromeric WT ␣7␤3 nAChRs (of undetermined stoichiometry) and of a minor population of homomeric WT ␣7 nAChRs. Although it is possible that the heteromeric WT ␣7␤3 nAChRs could be activated by an unidentified ligand, their insensitivity to the natural transmitter, together with the reduced expression of functional receptors upon increasing the amount of injected ␤3, indicate that co-assembly of the ␤3 subunit negatively regulates the function of nicotinic receptors formed by WT ␣7. It is possible that WT ␣7 is prevented from functionally assembling with ␤3 by post-translational processes inherent to the Xenopus oocyte expression system and that in native systems the same constraints do not develop. There is a large variability in the nicotinic responses attributed to the expression of the ␣7 nAChRs in native systems (38,40), indicating that ␣7 may indeed assemble with other subunits capable of modifying receptor parameters.
In conclusion, our findings provide conclusive evidence that ␤3 is capable of forming functional receptors in combination with other subunits, in this case an ␣7 mutant. Remarkably, the co-assembly of ␤3 with WT ␣7 appears to form "silent" heteromeric receptors insensitive to ACh. This, together with a recent demonstration that the V273T ␤3 mutant participates in the assembly of functional receptors with the ␣3 and ␤4 subunits (17), adds further members to the repertoire of nicotinic receptors and increases the potential regulatory complexity of the neural cholinergic system.