Abstract
The β3 neuronal nicotinic subunit is localized in dopaminergic areas of the central nervous system, in which many other neuronal nicotinic subunits are expressed. So far, β3 has only been shown to form functional receptors when expressed together with the α3 and β4 subunits. We have systematically tested in Xenopus laevis oocytes the effects of coexpressing human β3 with every pairwise functional combination of neuronal nicotinic subunits likely to be relevant to the central nervous system. Expression of α7 homomers or α/β pairs (α2, α3, α4, or α6 together with β2 or β4) produced robust nicotinic currents for all combinations, save α6β2 and α6β4. Coexpression of wild-type β3 led to a nearly complete loss of function (measured as maximum current response to acetylcholine) for α7 and for all functional α/β pairs except for α3β4. This effect was also seen in hippocampal neurons in culture, which lost their robust α7-like responses when transfected with β3. The level of surface expression of nicotinic binding sites (α3β4, α4β2, and α7) in tsA201 cells was only marginally affected by β3 expression. Furthermore, the dominant-negative effect of β3 was abolished by a valine-serine mutation in the 9′ position of the second transmembrane domain of β3, a mutation believed to facilitate channel gating. Our results show that incorporation of β3 into neuronal nicotinic receptors other than α3β4 has a powerful dominant-negative effect, probably due to impairment in gating. This raises the possibility of a novel regulatory role for the β3 subunit on neuronal nicotinic signaling in the central nervous system.
The function of the β3 subunit has been a puzzle since it was cloned more than 15 years ago (Deneris et al., 1989). β3 is present in substantia nigra, ventral tegmentum, and medial habenula (Deneris et al., 1989), and it is incorporated into nicotinic ACh receptors (nAChRs) in the cerebellum (Forsayeth and Kobrin, 1997), retina (Vailati et al., 2000), and striatum (Zoli et al., 2002), as shown by immunoprecipitation. In the striatum, β3 is restricted to nAChRs on dopaminergic terminals (Zoli et al., 2002). Data from β3-null mutant mice suggest that this is a distinct receptor population that binds α-conotoxin MII with high affinity (Cui et al., 2003).
Despite these indications that β3-containing receptors have a specific role, possibly in locomotor control, we know little of the effects of β3 incorporation into functional nAChRs. β3 does not form functional nAChRs when expressed with a “typical” α (α2-α4) or with a β-type subunit (β2 or β4) (Deneris et al., 1989; Boorman et al., 2000; Colquhoun et al., 2003), but it only incorporates into nAChRs as the third subunit of a “triplet” (i.e., in receptors that also contain a “typical” α and a “typical” β subunit) (Groot-Kormelink et al., 1998). In Xenopus laevis oocytes, β3 does assemble together with α3 and β4 into a functional receptor that contains two copies each of α3 and β4 and one of β3 (Boorman et al., 2000). The presence of β3 is hard to detect, because only single-channel conductance and kinetics are affected (Boorman et al., 2003). Because of the relatively restricted expression of the α3 and β4 subunits in the central nervous system, this combination (α3β4 + β3) is likely to be relevant only to habenular nAChRs (Sheffield et al., 2000), and other subunit combinations may predominate in most areas that express β3, such as substantia nigra and the ventral tegmentum area. Here, a wide subset of nicotinic subunits is present in a manner typical of many CNS areas (Le Novère et al., 1996; Azam et al., 2002).
So far, there are few detailed studies in the literature testing whether the β3 subunit can form functional receptors with combinations other than α3β4 and, if so, how it changes the properties of the receptor. One exception is that of α7: it has been reported that β3 forms “silent” heteromeric receptors with α7 subunits (Palma et al., 1999). There is also a brief report that β3 may change the macroscopic time course of α3β2 currents (McIntosh et al., 2000), but a detailed study of α3β2β3 receptors is not available in the literature.
We systematically examined β3 incorporation by coexpression in oocytes and found that β3 profoundly reduces responses produced by all pairwise functional nAChR subunit combinations known to date, with the exception of α3β4. This dominant-negative effect is not due to a change in receptor numbers and disappears when a gain-of-function mutant β3 is expressed rather than wild-type β3. This, together with the evidence for β3 incorporation, suggests that β3 reduces the maximum receptor open probability (Popen). This effect of β3 was also observed in the native α7-like nAChRs of hippocampal neurons in culture. Expression of β3 can exert a novel effect on neuronal nAChRs, which results in the functional down-regulation of responses from receptors that incorporate this subunit.
Materials and Methods
Human Neuronal Nicotinic Subunit cDNAs. Sequences for nicotinic subunits are as deposited in GenBank, with accession numbers Y16281 (α2), Y08418 (α3), Y08421 (α4), Y08419 (α5), Y16282 (α6), Y08420 (α7), Y08415 (β2), Y08417 (β3), and Y08416 (β4). Subunits contained only coding sequences and an added Kozak consensus sequence (GCCACC) immediately upstream of the start codon and were subcloned into pcDNA3.1 (Invitrogen, Breda, The Netherlands) or pSP64GL (for oocyte expression). Mutations were introduced using the QuikChange Kit (Stratagene, Amsterdam, The Netherlands), and full-length sequence was verified. All pSP64GL plasmids were linearized immediately downstream of the 3′-untranslated β-globin sequence. Capped cRNA was transcribed using the SP6 mMessage mMachine Kit (Ambion, Cambridge, UK) and checked by RNA electrophoresis.
Two-Electrode Voltage-Clamp Recording of X. laevis Oocytes. Female X. laevis frogs were anesthetized by immersion in neutralized ethyl m-aminobenzoate solution (tricaine, methanesulfonate salt; 0.2% solution, w/v) and were killed by decapitation and destruction of the brain and spinal cord (in accordance with Home Office guidelines) before removal of ovarian lobes to sterile modified Barth's solution consisting of 88 mM NaCl, 1 mM KCl, 0.82 mM MgCl2, 0.77 mM CaCl2, 2.4 mM NaHCO3, and 15 mM Tris-HCl, with 50 U/ml penicillin and 50 μg/ml streptomycin (Invitrogen, Paisley, UK); pH 7.4 was adjusted with NaOH. Mature oocytes were manually defolliculated after collagenase IA treatment (Boorman et al., 2000) before cRNA was injected at a ratio of 1:1 for pair receptors and 1:1:20 for triplet receptors. The total amount of cRNA for each combination was determined empirically, with the aim of achieving a maximum ACh-evoked current of 1 to 2 μA and was 0.25 to 10 ng per oocyte, depending on the combination.
Oocytes, held in a 0.2-ml bath, were perfused at 4.5 ml/min with modified, nominally Ca2+-free Ringer solution (150 mM NaCl, 2.8 mM KCl, 10 mM HEPES, 2 mM MgCl2, and 0.5 μM atropine sulfate, pH 7.2, adjusted with NaOH; 18-20°C) and voltage-clamped at -70 mV using the two-electrode clamp mode of an Axoclamp-2B amplifier (Molecular Devices, Union City, CA), and electrodes were filled with 3 M KCl (resistance, 0.5-1 MΩ on the current-passing side). Agonist solution (acetylcholine chloride, freshly prepared from frozen stock aliquots) was applied via the bath perfusion at 5-min intervals. ACh responses from α7-expressing oocytes were recorded in the presence of 5 mM 5-hydroxyindole to enhance the response amplitude and reduce interoocyte variability (Zwart et al., 2002). Chemicals were from Sigma-Aldrich (Gillingham, UK) unless otherwise stated.
A descending-dose protocol was used for dose-response curves. All data shown are compensated for response rundown (Boorman et al., 2000). To reassure ourselves that the lack of functional expression observed for some subunit combinations was true and not a false-negative due to oocyte health or expression problems contingent to a given batch, oocyte data were obtained from a minimum of two separate oocyte batches for each combination. In every experimental batch, at least one “control” highly expressing subunit combination was injected to check for expression efficiency.
Concentration-response curves were fitted with the following Hill equation:
where I is the response, measured at its peak, [A] is the agonist concentration, Imax is the maximum response, EC50 is the agonist concentration for 50% maximum response, and nH is the Hill coefficient; least-squares fitting was performed with the use of the program CVFIT, courtesy of D. Colquhoun and I. Vais, available at http://www.ucl.ac.uk/Pharmacology/dc.html. Each curve was fitted separately, with individual responses being equally weighted, to obtain estimates for Imax, EC50, and nH. For display purposes, data points were normalized to the fitted maximum and pooled before fitting.
When two components were detected in the concentration-response curve, free fits of the individual curves were poorly defined because of the large number of parameters fitted. Good fits were obtained when all of the concentration-response curves for this combination were fitted simultaneously with EC50 and nH values for the two components constrained to be equal across oocytes, whereas the proportion of current in the first component was allowed to vary from one oocyte to the other.
Because of the heteroscedasticity of the functional data, we used Kruskal-Wallis one-way nonparametric ANOVA followed by Dunn's post hoc multiple comparisons test (Daniel, 1978) (GraphPad Prism version 4.00 for Windows; GraphPad Software, San Diego, CA). For all comparisons, we also carried out randomization tests (Colquhoun, 1971; RANTEST, available at http://www.ucl.ac.uk/Pharmacology/dc.html), which gave similar results to the ANOVA/post hoc tests.
Radioligand Binding Studies. Mammalian tsA201 cells were maintained at 37°C in 5% CO2 in Dulbecco's modified Eagle's medium containing Glutamax (Invitrogen), plus 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (all from SigmaAldrich). Subconfluent cultures were transiently transfected overnight with a total of 0.6 μg of cDNA per 10-cm dish, using the Effectene reagent kit (Qiagen, Crawley, UK). Cells were harvested 40 to 44 h later and resuspended in Hanks' buffered saline solution (HBSS) (Invitrogen) for assay. Amounts of total cellular protein were determined by a Bio-Rad protein assay (Bio-Rad, Hemel Hempstead, UK) using bovine serum albumin standards. For the α7 experiments, cells were transfected with cDNA for human RIC-3 protein (subcloned into pcDNA3 and transfected in equimolar amount with α7 cDNA) to ensure α7 surface expression (Williams et al., 2005).
[3H]Epibatidine binding samples were incubated on ice for 2 h with a single saturating dose of [3H]epibatidine (3 nM; PerkinElmer LAS, Beaconsfield, UK), and receptor-bound ligand was isolated using Whatman GF/B filters presoaked in 0.5% polyethylenimine on a Brandel cell harvester (Semat, St. Albans, UK). Radioligand binding was measured in the presence of buffer alone, along with binding in the presence of buffer containing excess nicotine (1 mM) to define nonspecific binding. Internal binding was estimated by blocking external binding sites using an excess of the nonpermeant ligand ACh (10 mM) and the number of external sites was computed by subtraction.
125I-α-Bungarotoxin Binding. Samples were incubated at room temperature for 2 h with a single saturating concentration of 125I-α-bungarotoxin (6-10 nM; GE Healthcare UK Ltd, Chalfont St. Giles, UK) in the presence of 1% bovine serum albumin to minimize nonspecific binding. Receptor-bound ligand was isolated as above. Cell-surface binding was measured in intact cells either in the presence of buffer alone or in the presence of excess nicotine and carbachol (1 mM each) to define nonspecific binding.
Patch-Clamp Recording of Primary Hippocampal Neurons. Hippocampal neurons were cultured from embryonic day-18 rat embryos (Thomas et al., 2005): the hippocampus was incubated in trypsin (0.25% w/v; Invitrogen) for 15 min before washing in Hanks' medium (Invitrogen), and dissociation was performed using the polished tip of a Pasteur pipette. Suspended cells were plated onto 22-mm coverslips coated with 0.1 mg/ml poly(l-lysine) and maintained in B-27-Supplemented Neurobasal Medium (Invitrogen) for 1 week before transfection with cDNA for EGFP-c1 (BD Biosciences, Oxford, UK) and either α7, β3, or β3V273S (1:1 ratio, 0.4 μg/μl per 35-mm dish) using Effectene (Qiagen). Whole-cell recordings were performed on the first and second day after transfection at a holding potential of -70 mV.
Plated cells were superfused (3 ml/min) with an extracellular solution containing 150 mM NaCl, 3 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM glucose, 5 mM HEPES, 1 μM atropine, and 0.3 μM tetrodotoxin, with pH adjusted to 7.3 with NaOH. The pipette solution consisted of 147 mM CsCl, 2 mM MgCl2, 1 mM CaCl2, 10 mM EGTA, and 10 mM HEPES, with pH adjusted to 7.3 with CsOH. Pipettes were pulled from borosilicate glass (GC150-TF; Harvard Apparatus, Edenbridge, UK) to a resistance of 1 to 3 MΩ. Series resistance (4-8 MΩ) was compensated between 60 and 90%.
ACh (3 mM) was applied via a modified U-tube. Exchange time was tested by the application of an 80% diluted extracellular solution both before obtaining the seal and after the end of recording and rupture of the seal. Only neurons in which the 10 to 90% exchange time was less than 1 ms were included. In experiments with methyllycaconitine (Tocris Cookson, Bristol, UK), the antagonist was added to the bath perfusion (2 min before the application of ACh) and to the ACh solution in the U-tube. Recordings were acquired using a Digidata 1322A with Clampex software (Molecular Devices, Union City, CA), filtered at 1 kHz with an 8-pole Bessel filter (built inhouse), and digitized at 10 kHz.
Results
Expression of the β3 Subunit, but Not of β3VS, Abolishes the Functional Responses of Most Recombinant Neuronal Nicotinic Subunit Combinations. The traces in Fig. 1 (top) show that coexpression of the β3 subunit effectively suppressed responses evoked by 1 mM ACh from neuronal nicotinic receptors expressed in oocytes from α/β subunit “pair” combinations, α2β2, α2β4, α3β2, α3β4, α4β2, and α4β4. With the exception of the α3β4 combination, adding β3 to the subunits expressed completely abolished functional nicotinic current responses. We repeated the measurement in several oocyte batches, obtaining consistent results. We have shown previously in experiments with mutant β3 that expressing α3β4 and β3 in a 1:1:20 ratio is necessary to ensure that the majority of receptors contain β3 (Groot-Kormelink et al., 1998). In the present series of experiments, the decrease in ACh currents was nearly complete when this injection ratio was used (i.e., 90-100% depending on the combination). Smaller but substantial decreases in functional responses were observed for less extreme transfection ratios, such as 1:1:1 (data not shown). We checked that the effect of β3 was not due to a decrease in ACh sensitivity in β3-containing receptors by testing higher ACh concentrations (up to 20 mM, data not shown) but failed to see any increase in responses with increases in concentration.
As shown by the summary of the data in the bar charts in Fig. 1 (bottom), functional expression of the α3β4 combination was some what spared by coexpression with β3, with a reduction of approximately two thirds (from 1.9 ± 0.36 to 0.71 ± 0.13 μA, n = 10 and 12, respectively).
To test whether the striking dominant-negative effect of β3 is a specific effect that requires the incorporation of β3 in the receptor pentamer, we repeated these experiments, expressing a point mutant of β3, V237S β3 (which we will refer to as β3VS), instead of the wild-type subunit. This mutant carries in position 9′ of the second, pore-lining transmembrane domain a hydrophilic residue, serine, instead of the hydrophobic amino acid (leucine or valine) present in this position in all subunits of the nicotinic superfamily. This type of mutation (Labarca et al., 1995) is believed to facilitate channel gating by destabilizing part of the hydrophobic girdle that keeps the channel closed (Miyazawa et al., 2003). The reason for performing this experiment is that this sort of point mutation could change the effect of β3 only if this effect is mediated by the incorporation of the subunit into the pentamer rather than by a nonspecific effect of β3 on subunit production, receptor assembly, or trafficking. We found that introducing this mutation in the β3 subunit abolished the dominant-negative effect of the wild-type subunit. Thus, expression of β3VS did not suppress functional expression of any of the subunit combinations tested (Fig. 1), where functional expression was measured as the response to 1 mM ACh. This result strongly supports the conclusion that the effect of wild-type β3 is due to a specific reduction in the Popen value of receptors that incorporate this subunit. When the β3VS subunit is incorporated instead of β3, the facilitation in gating produced by the mutation is such that the decrease in Popen seen with wild-type β3 does not take place or is much reduced, and receptor function is preserved.
As a negative control, we systematically tested whether β3VS can produce functional receptors when expressed alone or together with one other subunit (i.e., α2, α3, α4, β2, or β4). None of these combinations (β3VS, α2β3VS, α3β3VS, α4β3VS, β2β3VS, and β4β3VS) gave detectable functional responses to the application of 1 mM ACh (n = 10, two batches of oocytes for each combination).
The β3VS Subunit Is Incorporated into Functional nAChRs. One possible alternative explanation is that when the β3VS mutant is transfected, function remains because the mutant subunit, contrary to wild-type β3, cannot participate in receptor formation. In the nicotinic superfamily, channel pore 9′ mutants have been widely used in stoichiometry studies, and no effects on subunit incorporation have been reported (Boorman et al., 2000). Nevertheless, we tested this possibility by characterizing the ACh sensitivity of receptors expressed in oocytes from α4β2 + β3VS cRNA. Figure 2 shows the dose-response curves of recombinant α4β2 receptors expressed alone (•, solid line) or together with β3VS (▴, broken line). The α4β2 dose-response curve shows a typical feature of this receptor combination, in that it has two components, one at 6.6 ± 1.8 μM and one at 160 ± 14 μM(nH values were 0.78 ± 0.09 and 2.4 ± 0.2, respectively; n = 4). The presence of these two components has been ascribed to the existence of two distinct receptors that differ in the number of α subunits contained in the pentamer, the high-sensitivity form of the receptor containing two copies of α4 (Nelson et al., 2003). When β3VS is coexpressed with α4 and β2, the ACh doseresponse curve is shifted leftward and seems to have only one component (EC50 of 1.0 ± 0.1 μM, nH of 1.1 ± 0.1, n = 4), which is clearly different from either of the two components observed for α4β2 alone. These effects of β3VS coexpression show that the mutant subunit is incorporated in the α4β2 type receptor; in addition to that, the presence of only one component suggests that this new receptor can only take one stoichiometry, which is likely to include two copies each of α4 and β2 and one of β3VS. Similar results were obtained for the α3β2 combination expressed alone and with β3VS (data not shown).
Combinations Containing the α6 Subunit. The expression of “pair” α6 subunit combinations produced only 0 or very small nicotinic responses in oocytes, even when relatively large amounts of cRNA were injected, up to 1 ng per subunit (i.e., 0.92 ng, namely 80 times more than was injected for the α3β4 combination). For α6β2 transfections, only two oocytes of eight injected with this cRNA amount responded to 1 mM ACh with an average 12-nA response. Responses were also poor but were more consistent for α6β4 (49 ± 15 nA, n = 8, two batches of oocytes). Coexpression of the β3 subunit in its wild-type form did not affect the amplitude of responses elicited from α6 combinations: only two of eight oocytes injected with α6β2β3 responded to ACh (average, 4 nA), whereas six of eight oocytes injected with α6β4β3 responded to 1 mM ACh (18 ± 5.6 nA). Somewhat unexpectedly, coexpression of the gain-of-function mutant β3VS significantly increased nicotinic responses in both combinations to 120 ± 17 and 560 ± 140 nA for α6β2β3VS and α6β4β3VS, respectively (eight of eight injected oocytes for both combinations; p < 0.001, Kruskal-Wallis one-way nonparametric ANOVA followed by Dunn's post hoc multiple comparisons test).
Transfection with β3 Suppresses Functional ACh Responses by Native Nicotinic Receptors in Hippocampal Neurons in Culture. The powerful and consistent dominant-negative effect of β3 on recombinant receptors raises the question of whether this effect is relevant for native receptors in neurons. The best way to test this is to transfect neurons in primary culture with β3. We chose hippocampal primary cultures, because hippocampal neurons have robust responses to nicotinic agonists but do not express the β3 subunit (Deneris et al., 1989; Sudweeks and Yakel, 2000; Cui et al., 2003).
Almost all hippocampal neurons tested responded to the U-tube application of 3 mM ACh with fast inward currents as shown in Fig. 3A (250 ± 54 pA, 25 of 27 neurons tested). As reported previously by Albuquerque et al. (1997), these responses are likely to be mediated by α7 nicotinic receptors (note their fast time course and the fast and complete sag in the response with sustained ACh application). The α7-like nature of these responses was confirmed by their sensitivity to the application of the nicotinic blocker methyllycaconitine (MLA). At a concentration (1 nM) selective for α7 receptors (Alkondon and Albuquerque, 1993), MLA completely blocked the response to 3 mM ACh in 14 of 14 neurons (Fig. 3A). We also transfected hippocampal neurons with the α7 subunit together with EGFP-c1 as a marker: responses in transfected neurons (identified because of their green fluorescence) had the same time course as in control but were much bigger (3800 ± 510 pA, 26 of 26 neurons tested; p < 0.001 against the 27 control neurons; Kruskal-Wallis one-way nonparametric ANOVA followed by Dunn's post hoc multiple comparisons test; statistical tests were carried on all the neurons tested, including nonresponders). Having established that nicotinic responses in the preparation are robust, consistent, and α7-like, we proceeded to test the effect of transfection with β3. Transfection with wild-type β3 had a striking dominant-negative effect: 34 of 38 transfected (i.e., fluorescing) neurons did not respond to 3 mM ACh at all, as shown in the middle trace of Fig. 3A. In the remaining four neurons, currents similar to those recorded in untransfected neurons were observed (490 ± 240 pA, n = 4).
As in oocyte-expressed recombinant receptors, transfection of hippocampal neurons with the mutant β3VS subunit failed to produce the dominant-negative effect observed for the wild-type transfections. As shown in Fig. 3A, almost all neurons transfected with β3VS responded to 3 mM ACh with inward currents comparable in amplitude with those measured in control neurons in the same dishes (240 ± 36 pA, 48 of 53 neurons; 5 neurons did not respond to ACh), although the time course of these responses was somewhat slower than that of the untransfected controls.
We carried out a similar experiment in oocytes expressing the α7 subunit: coexpression of the β3 wild-type subunit produced an 88% decrease in the responses to 1 mM ACh (from 0.97 ± 0.15 to 0.10 ± 0.027 μA, n = 8 and 14, respectively), as shown by the middle trace in Fig. 3B. If the mutant β3VS subunit was coexpressed, instead of wild-type β3, the decrease was by 60% (to 0.38 ± 0.057 μA, n = 13; last trace in Fig. 3B).
Changes in Receptor Surface Expression Cannot Account for the Dominant-Negative Effect of β3 on Function. The effect of expressing β3VS is already a strong indication that β3 impairs receptor function at the level of the receptor molecule and reduces its open probability rather than reducing the number of receptors in the membrane. Nevertheless, we checked for that by carrying out a binding assay to measure the number of nicotinic sites expressed on the surface of tsA201 cells transiently transfected with α3β4, α4β2, or α7 alone or together with either wild-type β3 or with β3VS. The ligand was [3H]epibatidine for the α3β4* and α4β2* sites and 125I-α-bungarotoxin for α7* receptors. In the α7 experiments, tsA201 cells were transfected with both α7 and the human RIC-3 protein to ensure reliable surface expression of the α7 receptor (Williams et al., 2005).
As shown in Fig. 4, coexpression of wild-type β3 did not abolish surface expression of nAChR. After coexpression of β3WT, there were relatively small but significant changes in the number of surface binding sites. The number of α3β4β3 sites was 65.0 ± 7.8% of the number of α3β4 sites (Fig. 4, left, see the 63% reduction in maximum ACh response shown in Fig. 1), whereas the number of α4β2β3 sites was somewhat increased with respect to the number of α4β2 sites (140 ± 16%, n = 7, Fig. 4, middle; cfr. the 97% reduction in maximum ACh response for the same combination; Fig. 1). α7 Sites were approximately halved (to 52 ± 16% of control, n = 6, Fig. 4, right; cfr. the 88% suppression in functional ACh responses for the same combination, Fig. 3B) by the coexpression of β3 wild type but not by coexpression of β3VS (86 ± 14% of control). Thus, the wild-type form of β3 halved the surface expression of α7 receptors but effectively suppressed (by 88%) their functional responses. Clearly the dominant-negative effect of β3 cannot be explained by a change in the number of receptors on the surface.
Discussion
The β3 subunit is present at high levels in CNS regions that express many other nicotinic subunits (Deneris et al., 1989). To our surprise, we found that expressing β3 together with every known type of pairwise functional recombinant neuronal nicotinic combination (except for α3β4) abolished functional nicotinic responses. Our data confirm and substantially extend those of Palma and coworkers (1999) that chick β3 and α7 subunits coassemble into nonfunctional receptors. We found that the dominant-negative effect of β3 could not be accounted for by a reduction in the number of surface nAChRs and that it was reversed by a V9′S mutation in the second transmembrane domain of β3.
A Mechanism for the Dominant-Negative Effect of β3 on Nicotinic Function. The changes in the α4β2 dose-response curves produced by β3VS and our previous data on α3β4β3 receptors (Boorman et al., 2000, 2003) show that β3 is incorporated into nAChRs. It follows that the suppression of nicotinic responses must result mainly from impaired function of β3-containing nAChRs. Function is not restored by increasing the ACh concentration, so the effect is not due to a shift in agonist sensitivity. The amplitude of the maximum agonist response of a ligand-gated ion channel is affected by the number of receptors, the unitary channel current, and the maximum channel Popen value for the agonist, and we shall examine these factors in turn.
Our data show that the effect of coexpressing β3 or β3VS on the number of surface receptors expressed in mammalian cells cannot account for the profound inhibition of functional responses we observe. There is a one-third reduction in the number of α3β4-type sites (cfr. a two-thirds decrease in α3β4 currents), no change in the number of α4β2 sites, and a halving of α7 sites, in contrast with the nearly complete abolition of α4β2 and α7 responses. This agrees with the finding that β3 only slightly reduced surface α7 sites (Palma et al., 1999).
Agonist responses will also be affected by the size of the single-channel current. The nearly complete suppression of functional responses by β3 makes it difficult to measure the single-channel conductance of most β3-containing receptors, but we already know that β3 increases α3β4 channel conductance (Boorman et al., 2003). In nAChRs, conductance is determined by conserved pore-lining domain residues that form three rings of charges. β3 has a negatively charged glutamate in the external 20′ ring, in which other neuronal β subunits have a positively charged lysine. Because β3 takes the place of a β subunit (Boorman et al., 2000), its incorporation increases the negative charge in 20′ and explains the observed increase in conductance (Imoto et al., 1988). In addition, it is hard to see how anything short of a complete loss of channel conductance could account for the near-total suppression of receptor function we observed. Furthermore, a conductance change would probably not be reversed by the 9′ mutation, which does not affect single-channel conductance (Filatov and White, 1995). Finally, residues that determine conductance are highly conserved in neuronal nicotinic subunits, and it is hard to see why the effect of β3 differs depending on subunit combination.
This leaves us with the possibility that β3WT reduces the maximum channel Popen value by reducing gating efficacy, E. This would explain why the dominant-negative effect is suppressed (or counterbalanced) by the V9′S mutation, which is believed to facilitate gating (Labarca et al., 1995). For many simple mechanisms, maximum Popen value is given by E/(E + 1) (where E = β/α is the gating equilibrium constant for the fully liganded channel, and β and α are the opening and closing rate constants, respectively). The effect of reducing E on the maximum Popen value will be relatively small if E is large in the first place. Hence, α3β4 function would be relatively spared by β3 if α3β4 has a higher value of E than the other combinations. Unfortunately, nothing is known of the gating efficacy of neuronal nAChRs. Efficacy values in the nicotinic superfamily go from very high (muscle nAChRs, Colquhoun and Sakmann, 1985; and glycine receptors, Burzomato et al., 2004) to relatively low values (GABAA receptors, Jones and Westbrook, 1995).
Because β3 produces such a complete suppression of nicotinic currents, obtaining direct estimates of the maximum Popen value by single-channel measurements is not feasible for most β3-containing combinations. The exception is α3β4β3, and our single-channel data from this combination are consistent with the hypothesis that β3 primarily impairs gating. Incorporation of β3 greatly shortens the duration of bursts of openings at low agonist concentration (receptor activations;Boorman et al., 2003), mainly by reducing the number of events per burst (from 38.3 ± 11.3 to 2.9 ± 0.31 gaps per burst, n = 10 and 5, respectively; M. Beato and L. Sivilotti, unpublished data), but does not affect the macroscopic EC50 value of α3β4 nAChRs (Groot-Kormelink et al., 1998). The mean number of openings, m, in an activation is
where β is the opening rate constant, and koff is the dissociation rate constant. If β3 reduced the number of openings per burst primarily by increasing the dissociation rate constant, we should also observe a change in EC50, because the EC50 value is linearly related to the dissociation equilibrium constant. In the simplest case of a receptor opened by two agonist molecules (Colquhoun, 1998),
where kon is the association rate constant. Equation 3 shows that changes in the gating constants have smaller effects on EC50 values, because EC50 is a function of the square root of the gating constants. This explanation is not unique (and does not consider the possibility that β3 changes the proportion of missed single-channel events), but it corroborates indications from the other experiments (i.e., binding assays and the effect of the valine-serine gain-of-function mutation).
Implications for Native Receptors Containing the β3 Subunit. β3-Containing nAChRs, whether formed by a subunit pair with β3 or by α7 with β3 (including native hippocampal α7-like nAChRs), have profoundly reduced function, except for α3β4-type receptors. This “sparing” of α3β4 receptor function may mean that, in a typical CNS neuron that expresses a wide range of subunits together with β3, nicotinic responses would have a predominantly α3β4-type profile. The precise functional consequences for nicotinic signaling would depend on the physiological ACh concentrations that activate these central nAChRs. Because α3β4 receptors are less sensitive to ACh than the α4β2-type (by up to 100-fold; Gerzanich et al., 1995), the predominance of α3β4 responses may mean that higher ACh levels are needed to produce nicotinic responses in neurons in which β3 is expressed. This might be important presynaptically, where transmitter levels may not reach the saturating concentrations at which peripheral fast synapses operate. Differences in the extent and rate of desensitization of the different receptors may also be important.
Caution must be exerted in extrapolating our findings to native receptors, because our results apply to relatively simple recombinant receptors (i.e., “triplet” receptors made of an α/β pair plus β3). We do not know whether native receptors can have this sort of composition or contain more than three different subunits. Furthermore, β3 expression is often associated with α6 expression (Le Novère et al., 1996), and β3 may facilitate α6* receptor trafficking, because it increases functional expression of α6/α3 β2 chimeric receptors (Kuryatov et al., 2000; McIntosh et al., 2004). Nevertheless, α6-containing receptors have proven very hard to characterize because of low functional expression. In our hands, the expression of α6β2 or α6β4 produced at best very small functional responses which were not increased by β3 coexpression. Robust responses were observed only when the gain-of-function mutant β3VS was coexpressed. A further complication is that efficient surface expression of α6β4 ligand binding sites may require both the β3 and the α5 subunit (Grinevich et al., 2005).
Results from knockout mice suggest that β3-containing nAChRs on striatal dopaminergic terminals are either α6β2β3 or α6α4β2β3 (Luetje, 2004; Salminen et al., 2004). At present, we cannot determine what the effect of β3 would be on this complex α6-containing receptor combination because of the difficulty of expressing receptors that contain four different subunits when subsets of these four are also functional. To obtain a pure population of these receptors, it may be necessary to use concatamer techniques (Zhou et al., 2003; Groot-Kormelink et al., 2004, 2006).
The disappearance of a specific striatal α6-containing receptor (α-conotoxin MII-sensitive) in β3-null mice has been taken to mean that efficient formation of this receptor requires β3. However, β3 deletion increased another type of nicotinic response (i.e., the α-conotoxin MII-resistant component of dopamine release produced by nicotine) (Cui et al., 2003). Hence, this distinct receptor population was believed not to contain β3. However, their enhanced function after β3 deletion could be explained if the receptors normally do contain β3 and if their function is reduced by the presence of β3. The precise subunit composition of the receptor could thus determine whether β3 stabilizes the receptor or reduces its function.
β3 expression profoundly reduces nAChR function in a variety of subunit combinations. The magnitude of this effect depends on the channel subunit composition and may result in the switching to a different profile of functional receptors.
Acknowledgments
We are indebted to Dr. Philip Thomas and Professor Trevor Smart for providing us with the hippocampal primary cultures.
Footnotes
-
This work was supported by the Wellcome Trust (project grants 064652 to L.G.S. and 074041 to N.S.M.) and by the Medical Research Council (PhD Studentship to S.B. and Training Fellowship to M.B.).
-
ABBREVIATIONS: nAChR, nicotinic acetylcholine receptor; ACh, acetylcholine; Popen, channel open probability; CNS, central nervous system; ANOVA, analysis of variance; MLA, methyllycaconitine.
-
↵1 Current affiliation: Novartis Horsham Research Centre, Horsham, United Kingdom.
- Received May 16, 2006.
- Accepted July 5, 2006.
- The American Society for Pharmacology and Experimental Therapeutics