Abstract
Human α4β2 nicotinic acetylcholine receptors (AChRs) expressed in Xenopus laevis oocytes or transfected cell lines are present as a mixture of two stoichiometries, (α4)2(β2)3 and (α4)3(β2)2, which differ depending on whether a β2 or α4 subunit occupies the accessory subunit position corresponding to β1 subunits of muscle AChRs. Pure populations of each stoichiometry can be expressed in oocytes by combining a linked pair of α4 and β2 with free β2 to produce the (α4)2(β2)3 stoichiometry or with free α4 to produce the (α4)3(β2)2 stoichiometry. We show that the (α4)3(β2)2 stoichiometry and the (α4)2(β2)2β3 and (α4)2(β2)2α5 subtypes in which β3 or α5occupy the accessory positions have much higher permeability to Ca2+ than does (α4)2(β2)3 and suggest that this could be physiologically significant in triggering signaling cascades if this stoichiometry or these subtypes were found in vivo. We show that Ca2+ permeability is determined by charged amino acids at the extracellular end of the M2 transmembrane domain, which could form a ring of amino acids at the outer end of the cation channel. α4, α5, and β3 subunits all have a homologous glutamate in M2 that contributes to high Ca2+ permeability, whereas β2 has a lysine at this position. Subunit combinations or single amino acids changes at this ring that have all negative charges or a mixture of positive and negative charged amino acids are permeable to Ca2+. All positive charges in the ring prevent Ca2+ permeability. Increasing the proportion of negative charges is associated with increasing permeability to Ca2+.
Nicotinic acetylcholine receptors (AChRs) belong to the cysteine loop superfamily of ligand-gated ion channels that includes muscle and neuronal AChRs, GABA types A and C receptors, 5-hydroxytryptamine type 3 receptor, and glycine receptors (Sine and Engel, 2006). These receptors are formed by five homologous subunits arranged around a central ion pore. Neuronal AChRs are formed from a variety of pentameric combinations of subunits α2–α10 and β2–β4 (Lindstrom, 2000; Gotti et al., 2006; Sine and Engel, 2006).
α4β2* AChRs are the predominant subtypes with high affinity for nicotine in the mammalian brain (Flores et al., 1992). The asterisk indicates that additional subunits such as α5 are present in some of these AChRs (Gerzanich et al., 1998; Gotti et al., 2006). Most α4β2* AChRs are believed to be located presynaptically and modulate the release of several neurotransmitters (Dani, 2001; Dajas-Bailador and Wonnacott, 2004). α4β2* AChRs are sufficient for nicotine reward, tolerance, and sensitization (Tapper et al., 2004; Maskos et al., 2005). Mutations in either the α4 or β2 subunits can cause autosomal dominant nocturnal frontal lobe epilepsy (Combi et al., 2004). In addition, the α4β2 AChR subtype is believed to be involved in neurodegenerative diseases such as Alzheimer's and Parkinson's diseases and Lewy body dementia because in postmortem brain tissues from patients of these diseases, α4β2* AChRs are significantly reduced with respect to age-matched normal subjects (Zanardi et al., 2002).
The stoichiometry of chicken α4β2 AChRs heterologously expressed in oocytes using a 1:1 (α/β) ratio of cRNAs was initially shown to be (α4)2(β2)3 by two different methods (Anand et al., 1991; Cooper et al., 1991). However, later studies with human AChRs showed that the stoichiometry (α4)3(β2)2 could also be formed (Zwart and Vijverberg, 1998; Nelson et al., 2003; Zhou et al., 2003; Moroni et al., 2006). The (α4)3(β2)2 stoichiometry has much lower sensitivity to activation and up-regulation by agonists and desensitizes more rapidly. It is unknown whether the (α4)3(β2)2 stoichiometry is expressed in brain. Its properties seem most appropriate for a postsynaptic AChR (Nelson et al., 2003). Long-term exposure to nicotine increased the proportion of the more nicotine-sensitive (α4)2(β2)3 stoichiometry in human embryonic kidney cells permanently transfected with human α4β2 AChRs (Nelson et al., 2003; Kuryatov et al., 2005). The proportion of the (α4)2(β2)3 stoichiometry can be increased by increasing expression of β2 (Nelson et al., 2003). The proportion of the (α4)3(β2)2 stoichiometry can be increased by increasing the amount of α4 through protein kinase A or the 14-3-3 chaperone (Exley et al., 2006). In addition, the presence of untranslated regions in α4 and β2 subunits can influence the proportion of stoichiometries expressed (Briggs et al., 2006), presumably by influencing the efficiency of translation. In X. laevis oocytes, greatly increasing the proportion of β2 or α4 mRNA injected to 1:10 or 10:1 forces the expression of one stoichiometry or the other, but the lopsided subunit ratio reduces total AChR expression (Moroni et al., 2006), presumably by producing nonproductive assembly intermediates. α3β2 AChRs, at least, are also subject to expression in two similar stoichiometries (Briggs et al., 2006).
Drugs that are at least somewhat selective for each α4β2 AChR stoichiometry have been identified recently (Briggs et al., 2006; Moroni et al., 2006; Zwart et al., 2006). Pharmacological evidence suggests that the (α4)2(β2)3 stoichiometry is expressed in the brain but does not eliminate the possibility that the (α4)3(β2)2 stoichiometry is also expressed. The observation that the (α4)3(β2)2 stoichiometry is more sensitive to the channel-blocking drugs mecamylamine and chlorisondamine (Briggs et al., 2006) indicates that its channel differs from that of the (α4)2(β2)3 stoichiometry.
We take advantage of the use of α4β2 AChRs formed from linked subunits to obtain exclusively either of the stoichiometries in X. laevis oocytes (Zhou et al., 2003). Here, we show that the (α4)3(β2)2 stoichiometry is much more permeable to Ca2+ than is the (α4)2(β2)3 stoichiometry. If the (α4)3 stoichiometry were expressed in brain, this would allow it to be much more potent in triggering signaling cascades triggered by the influx of Ca2+.
Materials and Methods
cDNAs and cRNAs. Synthesis of cDNAs for human α4 (Kuryatov et al., 1997), human β2 (Anand and Lindstrom, 1990), and tandem constructs of human α4 and β2 subunits (Zhou et al., 2003) were described previously. The cDNA for rat β3 was a gift from Steve Heinemann and Jim Boulter. The cDNA for human α5 was kindly provided by Dr. Francesco Clementi (University of Milan, Milan, Italy). It was subcloned in the pSP64poly(A) vector (Kuryatov et al., 1997). cRNAs from linearized cDNA templates were synthesized in vitro using the SP6 or T7 mMessage mMachine kit (Ambion, Austin, TX). The tandem construct used in this study was the β-6-α type (according to nomenclature in Zhou et al., 2003), in which the C terminus of the β2 subunit is linked via an AGS sequence repeated six times to the N terminus of the α4 subunit. Preparation of cDNAs and cRNAs for human α1 and δ muscle AChR subunits and human α7 subunit was described previously (Luther et al., 1989; Peng et al., 1994a). β1 and ϵ human cDNAs subunits were kindly provided by Dr. A. G. Engel. Mutations E261K into human α4 subunit and K258E into human β2 subunit were introduced using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and sequenced to verify that only the desired mutation was present.
Oocyte Removal and Injection. Oocytes were removed surgically from X. laevis and placed in an OR-2 solution containing 82.5 mM NaCl, 2 mM KCl, 1 mM MgCl2, 5 mM HEPES, and pH 7.5. They were defolliculated in this buffer containing 2 mg/ml collagenase type IA (Sigma, St. Louis, MO) for 1.5 h. After defolliculation, oocytes were incubated at 18°C in semisterile L-15 medium (Invitrogen, Carlsbad, CA) diluted by half in 10 mM HEPES buffer with 10 U/ml penicillin and 10 μg/ml streptomycin, pH 7.5. Oocytes were injected cytosolically with combinations of α7, α4+β2, β-6-α+α4, β-6-α+β2, β-6-α+α5, or β-6-α+β3 subunit cRNAs subunits (50 ng of α7; 5 ng of either α4, α4EK, β2, β2KE, β3, or α5 subunit; and 10 ng of the concatamer β-6-α in a total volume of 46 nl). To express muscle AChR, 10 ng of α1 and 5 ng of each of the other subunit cRNAs (β1, δ, and ϵ) were injected in a total volume of 46 nl. Five nanograms of the mutant K258E β2 subunit (β2KE) cRNA were injected with 5 ng of either the wild-type α4 subunit or the mutant E261K α4(α4EK) subunit cRNAs. Twenty nanograms of each cRNA were injected for the α4EK and wild-type β2 combination.
Electrophysiological Recordings and Data Analysis. Three to six days after injection, whole-cell membrane currents evoked by ACh (IACh) were recorded in oocytes at room temperature with a standard two-electrode voltage-clamp amplifier (Oocyte Clamp OC-725; Warner Instrument, Hamden, CT). Recordings were performed at a holding potential of –50 mV unless otherwise stated. All perfusion solutions contained 0.5 μM atropine to block responses of endogenous muscarinic AChRs that might be present in oocytes. Agonists were applied by means of a set of 2-mm glass tubes directed to the animal pole of the oocytes. Application was achieved by manual unclamping/clamping of a flexible tube connected to the glass tubes and to reservoirs with the test solutions. The recording chamber was perfused at a flow rate of 15 to 20 ml/min.
In a set of experiments, the perfusion solution was ND-96 with either normal or high Ca2+. These solutions consisted of 96 mM NaCl, 2 mM KCl, 1 mM MgCl2, 1.8 (normal) or 18 mM (high) CaCl2 and 5 mM HEPES, pH 7.5. Intracellular electrodes were filled with 3 M KCl. In another set of experiments, to prevent activation of endogenous Ca2+-dependent Cl– channels of oocytes, Cl–-free solutions were used instead for oocyte preincubation (6–16 h) and during recordings (Francis and Papke, 1996). In this case, normal or high Ca2+ solutions included 90 mM NaOH, 2.5 mM KOH, 1.8 or 18 mM Ca(OH)2, and 10 mM HEPES, buffered with methanesulfonic acid to pH 7.3. Intracellular electrodes were filled in this case with 2.5 M potassium aspartate. In addition, 40 mM dextrose was supplemented to the normal Ca2+ solutions (with or without chloride) to maintain the same osmolarity as the high Ca2+ solutions. In a third set of experiments, to determine IACh carried only by Ca2+ ions, a perfusion solution containing only 1.8 mM Ca(OH)2 buffered to pH 7.5 with HEPES was used. Dextrose (178 mM) was added to this solution to preserve normal osmolarity. Resistance of the voltage and current electrodes were 2.5 to 6 and 0.5 to 2 MΩ, respectively, in all cases.
Reversal potentials of IACh in normal and high Ca2+ solutions were determined by applying 2-s ramps from –70 to +50 mV during agonist application after the current had reached a steady-state value. Currents obtained in response to the voltage ramp in the absence of agonist application (reflecting passive membrane currents) were subtracted from the ramp currents during AChR activation (Kuryatov et al., 1997). In all experiments, oocytes were superfused for at least 5 min with every new test solution to ensure complete exchange of the bath solutions. A 1% agarose + 3 M KCl bridge was used between the bath solution and the ground electrode bath to minimize the differences in junction potentials along the recording circuit.
ACh concentration-response curves were obtained by normalizing the responses to different ACh concentrations to the response to 300 μM ACh in each oocyte, except for the α4EKβ2 combination, in which the responses were normalized to 1 mM ACh, and 3 mM ACh was used for the β-6-α+α4 and β-6-α+β3 combinations. The concentration-response curves were fitted using a nonlinear least-squares error curve-fit method (KaleidaGraph, Abelbeck Software; Synergy Software, Reading, PA) to the double independent Hill equation: I(x) = Imax1xn/(xn + EC501n) + Imax2xm/(xm + EC502m), where I(x) is the maximal current measured at the agonist concentration x, Imax1 and Imax2 are the maximal current responses at the saturating agonist concentration, EC501 and EC502 are the agonist concentrations required for the half-maximal response, and n and m are the Hill coefficients for each component of the fitted curve.
Stimulation and data acquisition were digitized at 100 Hz with a MacLab 2e interface and Scope 3.4.3 software (AD Instruments, Castle Hill, Australia). Current and voltage traces were analyzed using the Scope software. Representative traces were constructed opening data files in Scope and exporting data segments to Origin 5.0 (OriginLab Corp., Northampton, MA). Plots were also made with the Origin software.
Statistics. Data sets are expressed as means ± S.E. Statistical significance between data were determined by Student's t test. Differences were considered significant at the level of p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***).
Results
Ca2+Permeability of (α4)3(β2)2 and (α4)2(β2)3 Stoichiometries Assayed by Effects on Reversal Potential.X. laevis oocytes were injected with three different combinations of α4 and β2 AChR subunits: free α4 plus free β2 subunits (α4β2) in a 1:1 ratio to obtain the wild-type α4β2 AChR, expected to predominantly have the (α4)2(β2)3 stoichiometry (Anand et al., 1991; Cooper et al., 1991), or a mixed population of both (α4)2(β2)3 and (α4)3(β2)2 stoichiometries (Zwart and Vijverberg, 1998; Nelson et al., 2003); the β-6-α concatamer plus free α4 subunits (β-6-α+α4), which forms only (α4)3(β2)2 AChRs (Zhou et al., 2003); and β-6-α concatamer plus free β2 subunits (β-6-α+β2), which forms only (α4)2(β2)3 AChRs (Zhou et al., 2003). It is reasonable to predict that a change in subunit composition in α4β2 AChR would change the residues lining the lumen of the channel and, therefore, many of its functional properties, including ionic selectivity (Bertrand et al., 1993).
Ca2+ permeability was investigated by measuring the shift in the reversal potential of IACh when changing the extracellular Ca2+ concentration ([Ca2+]o) from 1.8 to 18 mM [10-fold increase in [Ca2+]o, based on the Goldman-Hodgkin-Katz equation (Fucile, 2004)]. The reversal potentials of IACh were estimated using ramp protocols (see Materials and Methods) applied during the stable phase of response to applications of 30 μM ACh (Fig. 1). To prevent activation of endogenous Ca2+-dependent Cl– channels, ND-96 solutions were substituted with Cl–-free media during a 6- to 16-h period of preincubation of oocytes and during the recordings. A 10-fold increase in [Ca2+]o caused the reversal potential of IACh to shift in the positive direction, as expected if Ca2+ were permeating through the channel, by 17.9 ± 3.0 mV in α4β2(n = 18), 26.9 ± 1.3 mV in β-6-α+α4(n = 32), and 7.3 ± 2.2 mV in β-6-α+β2(n = 12) (Fig. 2).
The shift in reversal potentials shows significant differences in calcium permeabilities between the three combinations tested. The β-6-α+α4 combination, which produces only the (α4)3(β2)2 stoichiometry, was the most permeable to calcium. The α4β2 combination, which produces a mixture of the two stoichiometries, had lower calcium permeability. The β-6-α+β2 combination, which produces only the (α4)2(β2)3 stoichiometry, was least permeable to calcium.
For comparison, human α7 AChR (the AChR subtype with greatest Ca2+ permeability) and adult muscle AChR [(α1)2β1γδ] (significantly less permeable to Ca2+ than other AChR subtypes) (Fucile, 2004) were also tested under the same experimental conditions. They gave shifts in their reversal potentials of 28.6 ± 2.4 mV (n = 9) for α7 AChR and –0.4 ± 3.5 mV (n = 7) for muscle AChR. In this assay, the Ca2+ permeability of the (α4)3(β2)2 stoichiometry is statistically indistinguishable from that of α7 AChRs (Fig. 2).
Currents when Calcium Was the Only Extracellular Ion Available to (α4)3(β2)2 and (α4)2(β2)3 AChR Stoichiometries in Solutions without Cl–. Differences in Ca2+ permeability between α4β2 AChR stoichiometries were also tested in solutions without chloride by an alternative method. When all cations but Ca2+ in the extracellular buffer were replaced by an equiosmotic concentration of dextrose, all combinations tested still conducted detectable inward currents (Fig. 3), indicating that they were all permeable to Ca2+. The amplitude of the currents, however, normalized to the currents induced by the same concentration of agonist in the normal extracellular buffer, varied significantly between the stoichiometries: 20.2 ± 1.0% for β-6-α+α4(n = 9), 8.4 ± 1.3% for β-6-α+β2(n = 12), and 6.4 ± 1.6% for α4β2(n = 10). This value for human α7 AChR (used as a reference value for an AChR with high Ca2+ permeability) under the same experimental conditions was 33.8 ± 5.5% (n = 12), 49.8 ± 9.2% for β-6-α+α5(n = 9), and 22.7 ± 8.4% for β-6-α+β3(n = 9) (Fig. 4). In this assay, the Ca2+ permeability of the (α4)3(β2)2 stoichiometry is significantly lower than that of α7 (p < 0.001) or β-6-α+α5(p < 0.01) AChRs. These results further confirm greater calcium permeability for the (α4)3(β2)2 stoichiometry compared with the (α4)2(β2)3 stoichiometry.
Effect of a Ring of Charged Residues at the Mouth of the Channel on Ca2+Permeability of α4β2 AChR. The charged residues at the outer edge of the M2 transmembrane domain are known to be determinants for the cationic selectivity of the nicotinic channel (Imoto et al., 1986; Bertrand et al., 1993). Thus, the anionic residue glutamate Glu261 in the α4 subunit was changed to cationic lysine (referred as α4EK) and the homologous cationic residue Lys258 in β2 was changed to the anionic E (β2KE) (Fig. 5). Table 1 and Fig. 4 show the effects on the ring of M2 amino acids indicated in Fig. 5 of the various subunit combinations used. Their contributions to Ca2+ permeability in α4β2 AChRs were investigated in the same conditions as before for concatamers (Fig. 4). Oocytes were injected with equimolar amounts of cRNAs for wild-type α4+β2, α4EK+β2, α4+β2KE, α4EK+β2KE, β-6-α+α4EK, or β-6-α+β2KE.
Wild-type subunits produced nearly equal amounts of both stoichiometries, each of which contained either a 3:2 or 2:3 ratio of – to + charged amino acids in the putative ring of amino acids formed at the extracellular end of M2, resulting in a net average charge of approximately +0.11 averaged over the two stoichiometries (Table 1). This nearly equal mix of + and – charged residues was associated with Ca2+ permeability as indicated by a current when only Ca2+ was present equal to 6.4% of the current under control conditions. β-6-α+α4 produced only the (α4)3(β2)2 stoichiometry and a larger net negative charge of –1, resulting in higher Ca2+ permeability (20% of control current). Using β-6-α+β2 to produce only the (α4)2(β2)3 stoichiometry and a net charge of + 1 reduced the Ca2+ permeability to 8.7% of control. To further test the importance of changes in the putative M2 ring, we reversed the charges in α4 and β2 subunits at this position and tested β-6-α concatamers with α4EK and β2KE subunits. The Ca2+ permeabilities in these AChRs were virtually opposite to wild-type combinations: 12.2% for β-6-α+α4EK and 19.9% for β-6-α+β2KE (Table 1). The extreme case of α4EKβ2 resulted in a much larger net positive charge of +5, independent of stoichiometry, and no Ca2+ permeability. The opposite extreme case of α4β2KE resulted in a net charge of –5 independent of stoichiometry and the Ca2+ permeability of 20%. The double mutant α4EKβ2KE, in which the charges on both free subunits were reversed, resulted in a net negative charge of –0.42 and a Ca2+ permeability of 16.3% of control. Overall, all negative charges or a mix of charges in the ring permitted some Ca2+ permeability, with Ca2+ permeability increasing as the net charge became more negative, or decreasing to 0 if all five residues were positively charged.
Because previous studies showed that mutations in the channel domain might alter not only ion selectivity but also sensitivity to agonists or antagonists and desensitization properties of the channel (Bertrand et al., 1993), ACh concentration-response curves were obtained for these mutants (Fig. 6, A and B) for comparison with the α4β2 wild type. Fitting the wild-type dose-response curve for the properties of each stoichiometry indicates that a 1:1 mixture of subunit mRNAs produces a 5:4 ratio of the (α4)2 to (α4)3 stoichiometries. All concentration-response curves of the mutants can be similarly resolved into two components. The proportions of the stoichiometries presumably reflect the relative affinities of the β2 or α4 subunits for assembly in the β1 position. The Ca2+ permeability of each stoichiometry reflects the number of positively charged lysine or negatively charged glutamate residues in the ring at the extracellular end of M2. The α4β2KE subunit combination presumed (α4)2(β2KE)3 component was most sensitive to ACh. With this subunit combination, in all stoichiometries the M2 ring was occupied only by glutamate. This ring of negative charge would be expected to not only select for Ca2+ permeability but also to increase cation concentration near the channel entrance and increase currents. Note that currents through α4β2KE with five negative charges are 47-fold greater than currents through α4EKβ2 with five plus charges (Fig. 3). The α4EKβ2 subunit combination presumed (α4EK)2(β2)3 component was least sensitive to ACh.
Concentration-response curves using linked subunits to fix the stoichiometries (Fig. 6C) revealed the expected high ACh sensitivity of (α4)2(β2)3 stoichiometry with the β-6-α+β2 combination and the expected much lower sensitivity of (α4)3(β2)2 with the β-6-α+α4 combination. The β-6-α+α4 combination showed a trace (6%) of high-sensitivity AChRs. These probably resulted because at the 1:1 M ratio of 5 ng of α4 mRNA to 10 ng of β-6-α mRNA used, there was not sufficient excess of free α4 subunit to completely prevent formation of the linked dimers one of each stoichiometry, which are formed when β-6-α is expressed alone (Zhou et al., 2003).
Both α5 and β3 subunits, like α4, have glutamate residues at the M2 ring position that was shown to be critical for regulating Ca2+ permeability; therefore, it would be expected that (α4)2(β2)2α5 and (α4)2(β2)2β3 AChRs would exhibit high Ca2+ permeability similar to (α4)3(β2)2 AChRs. These stoichiometries were achieved by expressing free α5 and β3 with linked β-6-α subunits. As expected, both α5 and β3 conferred much higher permeability to Ca2+ than did β2 (Fig. 4).
The β-6-α+α5 combination exhibited the same high sensitivity to activation by ACh as the β-6-α+β2 combination (Fig. 6C). This indicates that the (α4)2(β2)3 stoichiometry and (α4)2(β2)2α5 subtype are equally highly sensitive to ACh but are dramatically different in Ca2+ permeability, with the (α4)2(β2)2α5 subtype having 6-fold greater permeability to Ca2+ (Fig. 4) than the (α4)2(β2)3 stoichiometry. This will probably to be an important component of the biological significance of α4β2α5 AChRs.
The β-6-α+β3 combination exhibited even lower sensitivity to ACh than the β-6-α+α4 combination (Fig. 6C) and showed a similarly high permeability to Ca2+. The similar high Ca2+ permeabilities in the presence of either β3 or α5 accessory subunits further confirms the vital importance of the ring of charge in determining Ca2+ permeability, which was critically demonstrated using single amino acid mutations of α4 and β2 subunits. The greatly differing ACh sensitivities in the presence of β3 or α5 demonstrate that, although accessory subunits do not participate in forming ACh binding sites, they can greatly influence the sensitivity to the global conformation changes in AChRs that are required to open their cation channels.
Discussion
Expression of an equal amount of α4 and β2 subunits in X. laevis oocytes results in nearly equal amounts of AChRs with an (α4)2(β2)3 or (α4)3(β2)2 stoichiometry. By expressing a β-6-α concatamer with free β2 subunits, a pure population of the (α4)2(β2)3 was produced, and by expressing this concatamer with free α4 subunits, only the (α4)3(β2)2 stoichiometry was produced. We assayed Ca2+ permeability of these recombinant human α4β2 AChRs by two methods: 1) the shift in the reversal potential of IACh toward more positive values when native Ca2+-activated Cl– channel effects were avoided by eliminating Cl–, and 2) measurement of inward currents when Ca2+ was the only extracellular permeant ion. Both methods showed that the (α4)3(β2)2 stoichiometry had greater Ca2+ permeability than the (α4)2(β2)3 stoichiometry. Furthermore, the differences in permeability were shown to result from charged residues at the outer side of the M2 transmembrane domain, which form a pentameric ring at the extracellular end of the cation channel. α4, α5, and β3 subunits all have a homologous glutamate in M2, which contributes to high Ca2+ permeability, whereas β2 has a lysine at this position that reduces Ca2+ permeability. All positive charges in the ring prevented Ca2+ permeability, and Ca2+ permeability increased with the increasing numbers of negative charges in this ring. The presence of a ring of negative charges near the mouth of the channel would also be expected to increase the cation concentration near the channel entrance and increase currents.
Physiological Implications of the Results. There is clear evidence that coexpression of α4 and β2 subunits can result in a mixture of AChRs in (α4)2(β2)3 and (α4)3(β2)2 stoichiometries (Nelson et al., 2003; Zhou et al., 2003; Briggs et al., 2006; Moroni et al., 2006; Zwart et al., 2006). It has not been demonstrated that both stoichiometries are expressed in brain neurons, but there is evidence for heterogeneity of apparent α4β2 AChR properties in neurons that could be accounted for either by a mixture of these stoichiometries and/or subunit combinations such as α4β2α5, α4β2β3, or α4α6β2β3 (Marks et al., 1999; Shafaee et al., 1999). The (α4)3(β2)2 stoichiometry has been shown to differ from the (α4)2(β2)3 stoichiometry in that it is less sensitive to nicotine-induced up-regulation caused by increased assembly as a result of pharmacological chaperone effects of nicotine (Nelson et al., 2003; Kuryatov et al., 2005; Moroni et al., 2006). The (α4)3(β2)2 stoichiometry desensitizes more rapidly (Nelson et al., 2003). The (α4)3(β2)2 stoichiometry is less sensitive to activation or competitive inhibition by many ligands (Kuryatov et al., 2005; Moroni et al., 2006; Zwart et al., 2006). The efficacy of some agonists differs greatly between stoichiometries. For example, cytisine fails to activate the (α4)2(β2)3 stoichiometry but gives 22% of the maximum effect of ACh on the (α4)3(β2)2 stoichiometry, whereas TC-2559 has 260% of the potency of ACh on the (α4)2(β2)3 stoichiometry but only 22% efficacy on the (α4)3(β2)2 stoichiometry (Moroni et al., 2006; Zwart et al., 2006). The newly approved drug for smoking cessation, varenicline, is believed to act like cytisine as a partial agonist on α4β2 AChRs in oocytes and on dopaminergic neurons (Coe et al., 2005; Jorenby et al., 2006). If it behaved like cytisine, it would only be a partial agonist if (α4)3(β2)2 AChRs existed in neurons and would be an antagonist on the (α4)2(β2)3 stoichiometry. Nicotine is only a 28% partial agonist on the (α4)2(β2)3 stoichiometry but a 62% partial agonist on the (α4)3(β2)2 stoichiometry (Moroni et al., 2006). The channel-blocking antagonists mecamylamine and chlorisondamine more potently block the (α4)3(β2)2 stoichiometry (Briggs et al., 2006), which is consistent with our observation that the channel properties of the two stoichiometries differ.
Here, we show that the (α4)3(β2)2 stoichiometry has much greater Ca2+ permeability than does the (α4)2(β2)3 stoichiometry. The increased Ca2+ flux through the (α4)3(β2)2 stoichiometry would increase transmitter release in response to α4β2 AChRs located presynaptically and increase Ca2+-activated signaling cascades in response to activation of postsynaptic α4β2 AChRs. We proposed (Nelson et al., 2003) that the (α4)2(β2)3 stoichiometry might be localized presynaptically where tonic volume transmission depending on ACh released at a distance would depend on its high sensitivity and slow desensitization, whereas the (α4)3(β2)2 stoichiometry might be most effective as a postsynaptic AChR where it would be exposed transiently to high concentrations of ACh in the course of short-term rapid synaptic transmission that would be sufficient to activate it. Further investigations are required to localize α4β2 AChR stoichiometries in neurons and to determine their physiological and pharmacological significance. This would be aided by identifying drugs that are highly selective for each stoichiometry. Note also that the observation that α3β2 AChRs can be expressed in two stoichiometries (Briggs et al., 2006) suggests that this mode of regulation may apply to the expression of many heteromeric AChRs.
The presence of α5 or β3 in the β1-like accessory position would determine the (α4)2 stoichiometry of the α4β2* AChR. Depending on the extent of incorporation of α5 or β3, the presence of (α4)3 AChRs would be reduced or eliminated. However, because both α5 and β3 have glutamate in the M2 ring at the same position as does α4, one can expect that the (α4)3(β2)2,(α4)2(β2)2α5, and (α4)2(β2)2β3 AChRs would all have similarly high permeability to Ca2+. Kuryatov et al. (1997) showed that, as expected, coexpressing α5 with α4 and β2 increased Ca2+ permeability. Here we show that (α4)2(β2)2α5 has the highest Ca2+ permeability of any combination tested and that (α4)2(β2)2β3 has Ca2+ permeability similar to that of (α4)3(β2)2 stoichiometry. The combination in (α4)2(β2)2α5 of the high ACh sensitivity of the (α4)2(β2)3 stoichiometry and even higher Ca2+ permeability than the (α4)3(β2)2 stoichiometry would make this subtype exceptionally potent at functional roles in which Ca2+ influx was important, such as presynaptic promotion of transmitter release or postsynaptic triggering of signaling cascades.
The (α4)2(β2)2β3 subtype was found to have even lower ACh sensitivity than the (α4)3(β2)2 stoichiometry and a similarly high Ca2+ permeability. Both β3 and α5 can assemble only in the accessory position, comparable with that of β1 in the muscle (α1)2β1γδ AChRs, in which they do not contribute to the formation of ACh binding sites. As shown here and elsewhere, accessory subunits can influence both sensitivity to the conformation changes involved in channel opening and channel conductance. For example, α5 increases Ca2+ permeability and desensitization of both α3β2 and α3β4 AChRs and greatly increases ACh sensitivity of α3β2 but not α3β4 AChRs (Gerzanich et al., 1998). β3 greatly increases assembly and sensitivity to nicotine-induced up-regulation of α6β2 and α6β4 AChRs (Tumkosit et al., 2006). It has recently been reported that β3 led to nearly complete loss of function of α4β2, α4β4, and α3β2 AChRs and proposed that β3 served a universal role as a dominant-negative regulator (Broadbent et al., 2006). However, here we show that β3 formed functional but much lower affinity AChRs with α4β2. Thus, β3 does not behave as a universal dominant-negative regulator. The greatly reduced AChR function that Broadbent et al. (2006) observed in the presence of β3 resulted from using a cRNA ratio in oocytes of α/β/β3 of 1:1:20, which probably provokes the assembly of nonproductive intermediates, thereby reducing the total amount of AChR expressed, and the low sensitivity of α4β2β3 AChRs to activation would further reduce detectable function.
Acknowledgments
We thank Barbara Campling for her comments on the manuscript.
Footnotes
-
This work was supported by National Institutes of Health grant NS11323 (to J.L.) and the Phillip Morris External Research Program.
-
L.T. and A.K. contributed equally to this work.
-
Article, publication date, and citation information can be found at http://molpharm.aspetjournals.org.
-
doi:10.1124/mol.106.030445.
-
ABBREVIATIONS: AChR, acetylcholine receptor; ACh, acetylcholine; TC-2559, (E)-N-methyl-4-[3-(5-ethoxypyridin)y1]-3-buten-1-amine.
- Received September 5, 2006.
- Accepted November 28, 2006.
- The American Society for Pharmacology and Experimental Therapeutics