|Home | About | Journals | Submit | Contact Us | Français|
Conceived and designed the experiments: PK MM LGS. Performed the experiments: PK MM SB GH SK MB PJGK. Analyzed the data: PK MM SB GH SK MB PJGK LGS. Wrote the paper: PK MM LGS.
The neuronal nicotinic receptors that mediate excitatory transmission in autonomic ganglia are thought to be formed mainly by the α3 and β4 subunits. Expressing this composition in oocytes fails to reproduce the properties of ganglionic receptors, which may also incorporate the α5 and/or β2 subunits. We compared the properties of human α3β4 neuronal nicotinic receptors expressed in Human embryonic kidney cells (HEK293) and in Xenopus oocytes, to examine the effect of the expression system and αβ subunit ratio.
Two distinct channel forms were observed: these are likely to correspond to different stoichiometries of the receptor, with two or three copies of the α subunit, as reported for α4β2 channels. This interpretation is supported by the pattern of change in acetylcholine (ACh) sensitivity observed when a hydrophilic Leu to Thr mutation was inserted in position 9′ of the second transmembrane domain, as the effect of mutating the more abundant subunit is greater. Unlike α4β2 channels, for α3β4 receptors the putative two-α form is the predominant one in oocytes (at 11 αβ cRNA ratio). This two-α form has a slightly higher ACh sensitivity (about 3-fold in oocytes), and displays potentiation by zinc. The putative three-α form is the predominant one in HEK cells transfected with a 11 αβ DNA ratio or in oocytes at 91 αβ RNA ratio, and is more sensitive to dimethylphenylpiperazinium (DMPP) than to ACh. In outside-out single-channel recordings, the putative two-α form opened to distinctive long bursts (100 ms or more) with low conductance (26 pS), whereas the three-α form gave rise to short bursts (14 ms) of high conductance (39 pS).
Like other neuronal nicotinic receptors, the α3β4 receptor can exist in two different stoichiometries, depending on whether it is expressed in oocytes or in mammalian cell lines and on the ratio of subunits transfected.
Heterologous expression of muscle nicotinic receptors produces recombinant channels whose properties are similar to those of native channels from the neuromuscular junction. These properties are broadly similar for channels expressed in the two main heterologous expression systems, Xenopus oocytes and mammalian cell lines. The situation is different for the other main form of nicotinic receptor at peripheral synapses, the ganglion receptor. The main subunits contributing to this channel are α3 and β4 (probably together with α5 and β2 –). Receptors expressed in oocytes from these subunits resemble those from native autonomic ganglia in macroscopic pharmacology , but are very different at single-channel level. Most oocyte-expressed α3β4 channels open to a smaller conductance and produce much longer bursts than native receptors . When the α3 and β4 subunits were stably transfected in mouse fibroblasts at a 11 ratio, our outside-out recordings showed both oocyte-like channels and an additional, ganglion-like class of openings . At the time, we did not have an explanation for this phenomenon, as obtaining transient expression of neuronal nicotinic subunits at sufficient levels was problematic.
The central-type α4β2 neuronal nicotinic combination can exist in two stoichiometries, which contain either two or three copies of the α4 subunit in the channel pentamer , as first suggested by Papke and co-workers for α2β2 . The two forms of the α4β2 receptor can be distinguished in that the two-α form is activated at much lower ACh concentrations, has a different sensitivity to other agonists and does not display enhancement of submaximal ACh responses in the presence of low Zn2+ concentrations , . Given that in the Cys-loop superfamily binding sites are at the interface of subunits, different subunit stoichiometries result in pharmacological differences in binding sites (see for GABAA receptors ).
Earlier work on oocytes injected with α3 and β4 subunit cRNA has indicated the presence of a heterogeneous channel population . Here we present evidence that α3β4 receptors, like the α4β2 receptor, can also exist in two different stoichiometries, whose different properties explain our old results , . This is shown by data from receptors containing 9′ reporter mutations in either the α or β subunit in both oocytes and transiently-transfected HEK293 cells. We also provide a characterisation of the two forms of the α3β4 receptor, exploiting the fact that the stoichiometry of oocyte-expressed receptors can be manipulated by changing αβ cRNA ratios. When injections are carried out with equal amounts of α3 and β4 cRNA, the putative two-α form of the channel predominates in oocytes , whereas the three-α form is more abundant in HEK cells. The most distinctive macroscopic property of the two-α form is that it displays both potentiation and inhibition in response to Zn2+ (only inhibition is observed in the three-α form). In addition to that, single-channel recordings from HEK cells indicate that the “oocyte-like” long-burst, low-conductance channel corresponds to the two-α form, whereas the short-burst, high conductance class (“ganglion-like”, see , ) reflects the activity of pentamers containing three α subunits.
HEK293 cells were transfected with equal amounts of the α3 and β4 subunit cDNA to express α3β4 receptors. We characterised the change in ACh sensitivity caused by introducing a reporter mutation in either the α or the β subunit (see Figure 1). This mutation (9′ LT) replaces the conserved leucine in the middle of the pore-lining second transmembrane domain (M2) with a hydrophilic threonine and increases agonist potency in a manner proportional to the number of mutations in the channel , . The dose-response curves in Figure 1B (obtained from the peak inward currents elicited by ACh, see the traces in panel A) show that the greater shift in ACh potency is seen when the mutation is in the α subunit. Thus, α3LTβ4 receptors are almost 20 times more sensitive to ACh than wild-type channels, with an EC50 of 5.5±0.9 vs. 91.1±10.7 µM, respectively (n=7 and 8, respectively, see Table 1). Mutating the β subunit had a significantly smaller effect, reducing EC50 by about 5-fold, to 18.6±1.9 µM for α3β4LT receptors (n=6; significance from the confidence intervals for the potency ratios, Table 1). This strongly suggests that HEK-expressed α3β4 receptors contain more α than β subunits, in contrast with our observations in oocyte-expressed α3β4 receptors, where the same approach indicated a 2α3β stoichiometry .
In our previous work in oocytes, cells were injected with equal amounts of α3 and β4 subunit cRNA , , . More recent studies proved that changing the expressed ratio of αβ changes the subunit stoichiometry of α4β2 nicotinic receptors in oocytes ,  and we hypothesised that this may occur also for α3β4 receptors.
Figure 2 shows inward currents elicited by ACh in oocytes expressing α3LTβ4 or α3β4LT receptors at αβ cRNA injection ratios of 19 and 91 and the respective dose-response curves. When a subunit ratio of 1α9β was used (left side of Figure 2), mean EC50 values were very similar to those observed when equal amounts of α and β cRNA were injected (Table 2 and ). In particular, the greater increase in agonist sensitivity, approximately 150-fold, occurred when the reporter mutation was inserted in the β subunit (top half of Table 2). The effect of the mutation on the ACh dose-response curve was very different when the α subunit cRNA was injected in a 9-fold excess to the β subunit (right side of Figure 2). This time the pattern was similar to that seen for receptors expressed in mammalian cells, as the greater increase in ACh sensitivity was seen for α3LTβ4 receptors, with a leftward shift of 47-fold (bottom of Table 2), significantly greater than the 10-fold shift produced by inserting the mutation in the β subunit (cf. the confidence intervals shown in Table 2). This is what would be expected if excess α cRNA forces the oocyte to produce receptors with a three-α and two-β stoichiometry, and overrides the oocyte tendency to express receptors with two α and three β in the pentamer.
In the simplest case of a symmetrical channel made of equivalent subunits, each mutation copy would produce an equal “unitary” change in the free energy of gating, and hence a r-fold “unitary” change in EC50 values. Two or three copies of the mutation would shift EC50 by r2 or r3-fold, respectively. This gives rise to a linear relation between the logarithm of the EC50 values and the number of mutant subunits incorporated in the pentamer. Figure 3 shows that this relation holds for HEK- and oocyte-expressed receptors with up to three copies of the mutation. These data are consistent with our interpretation that α3β4 receptors from HEK and from oocytes injected with the 91 αβ ratio contain three copies of the α subunit, whereas receptors from oocytes injected with a 11 or a 19 ratio contain only two copies of the α subunit. Note also that our estimates of the elementary change in EC50, r (from the square or cube root of the potency ratios), are similar for putative three-α receptors whether they are expressed in HEK or in oocytes (see Tables 1 and and22).
Our results on α3β4 receptors, therefore, agree with those on α4β2 receptors , : depending on the α3 to β4 cRNA ratio used, oocytes can express receptors that contain either two or three copies of the α subunit in the pentamer.
The EC50 values in Table 2 also show that wild-type α3β4 receptors expressed with extreme ratios have different ACh sensitivity, 137.8±13.7 vs. 309.5±27.8 µM (n=4 for both, p=0.0015, two-tailed Student's t test) for the 19 and 91 ratios, respectively. In α4β2 receptors this difference is much greater (80-fold or more) but in the same direction , . It is possible that the two-α and the three-α form of α3β4 differ in other properties, some of which may be useful as tools in characterising native receptors. We therefore characterised the sensitivity of the two forms to a range of nicotinic agonists.
Figure 4A shows examples of inward currents elicited by low concentrations of nicotinic agonists in oocytes injected with a ratio of 19 (first row) or 91 αβ subunit cRNA (second row). Panel B shows partial dose-response curves as log-log plots (effectively the foot of the dose-response curve) to six agonists obtained in two representative experiments (corresponding to the traces in A, except for the lobeline responses that are from two different experiments, see legend). If we refer to the standard agonist, ACh (continuous lines in Fig. 4B), there are two agonists that stand out, epibatidine and carbachol. On both types of receptors, epibatidine is very much more potent than ACh (by more than 6000-fold), whereas carbachol is much less potent (by about 10-fold, see Table 3). The other agonists tested in these experiments, nicotine, DMPP and cytisine, are very much closer to ACh in their potency. Remarkably, the rank order of potency for the agonists we tested is very similar for the putative two-α and the three-α receptors, epibatidinelobeline>nicotine~cytisine>ACh>carbachol, in descending order of potency, with the single exception of DMPP (see Table 3). DMPP and ACh were equipotent (potency ratio 0.99±0.023) in 19 oocytes (n=9), but DMPP was ~10 times more potent than ACh (10±0.62, n=8) when cells were injected in a 91 ratio. The rank order of potency for the 19 receptors is almost identical to the one we reported for 11 receptors , in agreement with our conclusion from the reporter mutation experiments that in oocytes both the 11 and the 19 ratio of subunits give rise to receptors with two copies of α in the pentamer.
In the α4β2 receptor, Zn2+ potentiation of submaximal ACh responses is stoichiometry-specific and occurs only in receptors with three copies of the α subunit . Specific amino acids on the α4-α4 interface are necessary for potentiation, in particular a glutamate residue, in loop C, on the + side of the subunit-subunit interface, together with a histidine residue, in loop F, on the − side. By analogy with the α4β2 combination, we hypothesised that the potentiating effect of Zn2+ might be stoichiometry-specific also in the α3β4 receptor. We verified this hypothesis by testing the effects of Zn2+ on the two forms of α3β4 expressed in oocytes with 91 or 19 αβ cRNA ratios.
The traces in Figure 5A show the effect of increasing concentrations of Zn2+ on the inward currents elicited by applying ACh at EC20 to receptors with different stoichiometries. Normalised average responses are also plotted as dose-response curves in panel B of the Figure. As we hypothesized, no Zn2+ enhancement was seen for the putative three-α form of the α3β4 receptor (α3β4 91, filled squares in Fig. 5B): Zn2+ only inhibited ACh currents with an IC50 of 400±50 µM and an (absolute) Hill slope value of 1.4±0.1 (n=4). In contrast to that, Zn2+ had a complex effect on the putative two-α form of the α3β4 receptor (α3β4 19, filled circles in Fig. 5B). Lower Zn2+ concentrations (0.5 to 2 mM) enhanced ACh responses by up to 70±13% with EC50 and Hill slope values of 168.2±14 µM and 1.8±0.2, respectively (n=5). Concentrations higher than 2 mM inhibited ACh currents with IC50 and (absolute) Hill slope value for the inhibition of 3.2±0.3 mM and 2.1±0.2, respectively.
Our observation that ACh responses from the putative two-α form of α3β4 receptors can be either enhanced or inhibited, depending on the Zn2+ concentration, agree with similar results in oocytes injected with α3 and β4 cRNA in a 11 ratio . This is consistent with our finding that oocytes injected with equal amount of α3 and β4 cRNA express preferentially the two-α form of the receptor .
Oocyte-expressed α3β4 channels open mostly in long bursts to a low (22 pS) conductance : the results shown above suggest that these openings reflect the activity of a two-α form of the receptor (see also ref. ). This sort of channel is common also in stably-transfected mammalian fibroblasts, where we observed an additional type of channel activity, with shorter bursts and a higher conductance of 36 pS . Our new data on HEK293 cells show that they preferentially express another form of the receptor, and raise the possibility that the short-burst, high conductance channels correspond to a three-α stoichiometry. This hypothesis is confirmed by the data in Figure 6.
The traces in Figure 6A are representative single-channel currents evoked by the application of 5 µM ACh to outside-out patches at −100 mV. HEK cells were transfected with α3 and β4 cDNA at a ratio of 19 (left) or 91 (right). Openings in 19 patches are smaller in amplitude and occur in longer bursts, resembling the channels we characterised in oocytes, i.e. the putative two-α form of the channel. On the other hand, for 91 transfected cells, two different classes of openings occur: one is similar in amplitude and burst length to the putative two-α form observed in the 19 cells, whereas a second class opens in shorter bursts to a bigger conductance. These two classes of openings are never linked to each other by direct transitions (data not shown), and therefore they are likely to be produced by different receptor molecules.
The data from the patches in Figure 6A are plotted as fitted amplitude histograms in Figure 6B, where each opening is counted as one data point. The continuous curves are the fits with one (19) or two (91) Gaussian components. All patches from the 19 transfection had one component with average amplitude of 2.6±0.1 pA (corresponding to a chord conductance of 26 pS; n=7). Low conductance openings were also present in three out of five 91 patches, where their average amplitude was 2.6±0.1 pA (area 24±6%), but even when present were less common than high conductance openings, which had an amplitude of 3.9±0.04 pA (39 pS; four out of five patches) and an area of 78±8%.
Because the amplitude histograms in panel B are based on openings, they emphasize the contribution of channels that open in prolonged bursts of many openings, such as the low conductance, putative two-α receptors. It is informative to re-plot these data counting the number of bursts for each conductance (of course, neither method allows us to estimate the number of channels in each subtype). Figure 6C shows that in the three 91 patches that did have low conductance openings, only a small proportion of bursts, 7.8±1.8% (n=3), belong to low conductance channels. Figure 6D quantifies another property of these two classes of channels, namely that the smaller openings always occur in distinctive long bursts (176±45 ms and 98±16 ms for the 19 and 91 transfections, n=7 and 3 patches, respectively). The higher conductance of 39 pS, which predominates in 91 cells, has a shorter burst length of 14±2 ms.
Depending on expression system and αβ subunit ratios, recombinant α3β4 neuronal nicotinic receptors contain either two or three copies of the α subunit. This is similar to α4β2 receptors, except that α3β4 channels favour the three-α form in HEK cells and the two-α form in oocytes. The two forms are similar in agonist sensitivity and agonist potency, but only the putative two-α form is susceptible to enhancement by low zinc concentrations. Furthermore, the two stoichiometries are markedly different in both single-channel conductance and kinetics.
Early suggestions that neuronal nicotinic receptors exist in different stoichiometries ,  were proven for α4β2 receptors , where the two forms differ by up to 100-fold in ACh sensitivity , , .
Heterogeneity in the properties of α3β4 receptors across expression systems may also be explained by different stoichiometries . This is supported by several lines of evidence. Firstly, the effects of the 9′ LT mutation are different in the two expression systems and in oocytes they depend on αβ cRNA ratios. In the past we showed by this approach that a single copy of β3 is incorporated in α3β4β3 nicotinic receptors , a finding recently confirmed by detecting FRET between subunits bearing appropriate fluorescent probes . In the present experiments, EC50 shifts were consistent, with a roughly linear relation between log EC50 and number of mutation copies (Figure 3). A second line of evidence comes from the effects of zinc, which enhances only the putative two-α form. In α4β2 receptors, Zn2+ potentiation requires an α4(+)/α4(−) interface, present only in the three-α form. The key two residues on the opposite sides of the interface are α4 His165 (Loop F, − side) and α4 Glu194 (Loop C, + side) . The question is which subunit interface mediates zinc enhancement in the α3β4 receptor. It cannot be the β4(+)/α3(−) interface (present in both stoichiometries), but it cannot be the α3(+)/α3(−) interface either (present only in the three-α form), as it lacks the key residues identified in α4. By exclusion, it would seem that the β4(+)/β4(−) interface could mediate Zn2+ potentiation, as it is present only in the two-α stoichiometry. This hypothesis could be tested by site-directed mutagenesis.
α4β2 receptors from 11 αβ transfections are a mixture of stoichiometries, but three-α, low-sensitivity channels predominate , , . Contrary to that, for α3β4 channels from 11 αβ ratios, the putative two-α form is the predominant one in oocytes and the three-α in HEK cells. The different behaviour could be due to differences in the binding of α3 and α4 to chaperone proteins (such as 14-3-3, known to affect the proportion of the two forms of α4β2 , given that the α-subunit intracellular domain is not conserved.
The two forms of the α4β2 receptor differ greatly in pharmacology, cation permeability and response to chronic nicotine , , , . Remarkably, this is not the case for α3β4 receptors, at least at macroscopic level. In oocytes, overexpressing the α subunit decreases acetylcholine sensitivity by less than three-fold. The main difference between the two forms is in the relative potency of DMPP, which is equipotent with acetylcholine on the putative two-α form and 10-fold more potent on the putative three-α form. Interestingly, the rank order of potency of DMPP is also one of the main differences between recombinant and native receptors of this type. Receptors from the soma of autonomic ganglion neurons are most sensitive to cytisine, followed by DMPP and ACh , , , a sequence that is not reproduced by rat subunits expressed in oocytes, and is seen only in a fraction of stably transfected fibroblasts , . DMPP is unlikely to be a useful tool in practice, because it is more potent on the low-sensitivity form of the receptor and this reduces the difference in potency in absolute terms. Extrapolation to EC50s, assuming identical efficacy, suggests that DMPP would have an EC50 of 30 and 140 µM on the putative three-α and two-α forms, respectively. For α4β2 receptors the difference in potency between the two forms varies widely across agonists (from complete lack of selectivity to 100-fold difference ), so α3β4 agonists with better discrimination may yet be found.
The putative two-α form of the channel opens to much longer bursts (100 ms or more vs 14 ms) with smaller conductance (26 vs 39 pS) than the three-α form. Thus, it resembles the predominant type of opening that we described after 11 expression of rat α3β4 receptors in oocytes and in stably-transfected fibroblasts ,  and may account for the slow kinetics of rat α3β4 receptors in voltage-jump relaxations . Channels similar to the putative three-α form, rare in oocytes, are more common, but still a minority, in stably-transfected fibroblasts. This form resembles neuronal nicotinic receptors recorded from rat superior cervical ganglion , , . Ganglionic receptors are thought to contain α3 and β4 possibly with α5  especially in somatic receptors  but see  and mediate fast excitatory synaptic events, which have a time course similar to the burst length of the putative three-α form (cf 13.9 ms ) and certainly much shorter than that of the two-α form.
Our results differ from those of a previous study  that found that oocyte-expressed α3β4 receptors had mostly short bursts, unless expressed with α5. It is difficult to compare the expression conditions, as subunit ratios are not specified and plasmid construction is different (in the ones we employ, the original untranslated regions were removed and replaced by those of β-globin). Conductance comparisons with this and other studies (see ) are impossible when the recording solutions contain different concentration of divalent ions (especially calcium, which affects both nicotinic conductance and kinetics ).
In nicotinic channels, conductance is controlled by three rings of charges in the pore-lining M2 domain  and by residues in an M3–M4 loop amphipathic helix that lines the intracellular fenestrations of the channel . Because α3 and β4 differ at position 20′ in M2 (Glu vs. Lys), a channel containing three α subunits would have an overall negative charge (−1) in this ring and one containing two α subunits would carry an overall positive charge (+1). In the muscle receptor, mutations that produce the same difference in charge in this positions decreased conductance by 20% . These subunits differ also in the intracellular amphipathic helix. At the −4′ position, α3 has a Gln and β4 an Asp and at the 0′ position α3 has a Lys and β4 a Gln residue. These differences would be expected to oppose the effects of the M2 residues on conductance, but it is difficult to estimate by how much (only charge reversals have been tested in the equivalent positions of α4β2 ). Our results would suggest that M2 residues have the greater effect.
It is interesting to note the profound difference in kinetics between the two forms of the receptor. The difference in stoichiometry potentially means a different number of binding sites, and it may be that the α3(+)/α3(−) interface can function as an agonist binding site, a hypothesis that warrants further investigation (the Hill slope of the dose-response curve of the putative three-α form is higher, but this could stem from any number of changes in the kinetic mechanism .
Our data show that the α3β4 neuronal nicotinic receptor can exist in two forms that differ in stoichiometry. This may be a general phenomenon for heteromeric neuronal nicotinic receptors which can form from the expression of one α and one β subunit. It will be important to assess if native channels can exist in different stoichiometries (or whether this possibility is abolished by the additional subunits native channels may contain) and whether that has physiological or pharmacological consequences.
Note added in print:
During the revision of this manuscript, a report was published showing that, in substantial agreement with our findings, oocyte expression of rat α3β4 neuronal nicotinic receptors gives rise to a mixture of three-α and two-α receptor forms which have differ in their sensitivity to isomers of α-conotoxin AuIB (Grishin AA, Wang CI, Muttenthaler M, Alewood PF, Lewis RJ et al. (2010) α-conotoxin AuIB isomers exhibit distinct inhibitory mechanisms and differential sensitivity to stoichiometry of α3β4 nicotinic acetylcholine receptors. J Biol Chem 285: 22254–22263).
cDNAs for the human α3 and β4 nAChR subunits (GenBank accession numbers Y08418 and Y08416, respectively) containing the coding sequences and an added Kozak consensus sequence (GCCACC) immediately upstream of the start codon were subcloned into the pcDNA3.1 vector (Invitrogen, Breda, The Netherlands) for expression in human embryonic kidney 293 (HEK293) cells. The 9′ LT mutated subunits (α3L279T and β4L272T, where L stands for leucine and T stands for threonine; referred from here on as α3LT and α3LT), were created using the QuickChange™ Site-Directed Mutagenesis Kit (Stratagene) and their full-length sequence was verified.
For expression in oocytes, all cDNAs were subcloned into the pSP64GL vector, which contains 5′ and 3′ untranslated regions (UTRs) for the Xenopus β-globin. The cDNA/pSP64GL plasmids were linearized immediately downstream of the 3′ UTR, and capped cRNAs were transcribed using the SP6 mMessage mMachine kit (Ambion, Cambridge, UK). The quality of the cRNA molecules was checked by RNA gel-electrophoresis and by comparison with RNA size and concentration markers.
HEK293 cells (ATCC-CRL-1573; American Type Culture Collection) were maintained in a 95% air/5% CO2 humidified incubator at 37°C in Dulbecco's modified Eagle's medium (DMEM), supplemented with 0.11 g/L sodium pyruvate, 10% (v/v) heat-inactivated foetal bovine serum, 100 U/ml penicillin G, 100 µg/ml streptomycin sulphate, and 2 mM L-glutamine (all from Invitrogen, Breda, The Netherlands). Cells were passaged every 2–3 days (up to 20 times) and were plated on polylysine-coated coverslips in 35 mm culture dishes, approximately 24 hrs before transfection with a calcium phosphate-DNA co-precipitation method . Cells were transfected with cDNA mixtures for α3β4, α3LTβ4 or α3β4LT. For whole-cell recordings the α3 to β4 cDNA ratio in the mixtures was kept equal (11), whereas for single-channel recordings we used extreme ratios of 19 or 91 of wild-type subunit cDNAs. In all cases, plasmid coding for the marker enhanced green fluorescent protein (eGFP-c1; BD Biosciences, Oxford, UK) was also co-transfected in order to visualise successfully transfected cells. Each culture dish was transfected with 2–3 µg of DNA; this amount was enough for a high rate of transfection efficacy . All recordings were performed 14–48 hrs after transfection, at room temperature.
Female Xenopus laevis frogs, anaesthetized by immersion in neutralized ethyl m-aminobenzoate solution (tricaine, methanesulphonate salt; 0.2% solution w/v, Sigma), were killed by concussion immediately followed by decapitation and destruction of the brain and spinal cord before removal of the ovarian lobes. This procedure is in accordance with United Kingdom Home Office regulations. The study protocol does not fall under the remit of the UCL Research Ethics Committee (see http://www.ucl.ac.uk/research/images/research-ethics-framework).
Clumps of stage V–VI oocytes were dissected in a sterile modified Barth's solution consisting of (in mM): 88 NaCl, 1 KCl, 0.82 MgCl2, 0.77 CaCl2, 2.4 NaHCO3, 15 Tris-HCl, with 50 U/ml of penicillin and 50 µg/ml of streptomycin (Invitrogen, Breda, The Netherlands) in HPLC water; pH 7.4, adjusted with NaOH. The dissected oocytes were treated with collagenase (type IA, Sigma; for 50–65 minutes at 18°C, 245 collagen digestion units/ml in Barth's solution, 10 to 12 oocytes per ml), then rinsed and stored at 4°C overnight. Treated cells were manually defolliculated the following day prior to cRNA injection. The cRNA mixtures of α3β4, α3LTβ4 or α3β4LT (in a α3β4 ratio of 19 and 91) were injected into oocytes using a Nanoject Automatic Oocyte Injector (Drummond Scientific, Broomall, PA). The total amount of cRNA injected (in 46 nl of RNase-free water, 0.25–10 ng/cell) was determined empirically, with the aim of achieving a maximum ACh-evoked current no greater than 2 µA, in order to contain the magnitude of series resistance errors in our recordings. After injection, the oocytes were incubated for ~48–60 hrs at 18°C in fresh Barth's solution containing 5% heat-inactivated horse serum (Invitrogen, Breda, The Netherlands) and stored at 4°C. Experiments were carried out at 18°C, 2.5 to 5 days from injection.
Oocytes, held in a 0.2 ml bath, were perfused at 4.5 ml/min with modified Ringer solution containing (in mM): 150 NaCl, 2.8 KCl, 10 HEPES, 2 MgCl2 and 0.5 µM atropine sulphate; pH 7.2, adjusted with NaOH. The solution was (nominally) calcium-free in order to minimize the activation of calcium-gated chloride channels (endogenous to the oocyte) by calcium entry through nicotinic channels. Oocytes were held at −70 mV using the two-electrode clamp mode of an Axoclamp-2B amplifier (Molecular Devices, CA, USA). Electrodes were pulled from thin-walled Clark borosilicate glass GC150TF (Harvard Instruments) and were back-filled with 3 M KCl (resistance of 0.5–1 MΩ on the current-passing side).
Agonist solutions (freshly prepared from frozen stock aliquots) were applied via the bath perfusion at 5 min intervals for a period sufficient to obtain a stable plateau response (at low concentrations) or the beginning of a sag after a peak (at the higher concentrations). The 5 min intervals were sufficient to ensure reproducible responses. Agonist responses were recorded on a flat bed chart recorder (Kipp & Zonen, Lincoln, UK) for later analysis. A standard concentration of ACh (approximately EC20) was applied every third response to monitor agonist sensitivity throughout the experiment and allow correction of the data for rundown, as described previously . Experiments were started only after checking that this standard concentration gave reproducible responses. For ACh dose-response curves, a descending dose protocol was used.
For the potency-ratios experiments, low agonist concentrations were tested in each cell in order to obtain partial concentration-response curves (2–3 points for each agonist). The purpose of this was to reduce the impact of desensitization and agonist self-block and to ensure that the slope of all agonist curves was similar when dose-response curves were plotted on double-logarithmic coordinates. Care was also taken to match the size of the responses to different agonists. In all cells, ACh was used as the standard agonist. The maximum agonist concentrations for the 19 and the 91 ratio of α3 to β4were, respectively: 10 nM and 20 nM for epibatidine; 10 µM and 2 µM for lobeline; 25 µM and 5 µM for 1,1-dimethyl-4-phenylpiperazinium iodide (DMPP); 20 µM for nicotine and cytisine; 50 µM for ACh; 0.5 mM and 1 mM for carbachol.
The effect of Zn2+ was assessed by co-applying increasing concentrations of Zn2+ together with ACh (at its EC20 concentration, 120 and 50 µM ACh for the 91 and 19 αβ ratios, respectively). Before each application, oocytes were preincubated for 30 s with Zn2+ alone. ACh responses in the presence of Zn2+ were normalised to the response from ACh alone, at the same concentration.
Whole-cell recordings from transfected HEK293 cells were performed with an Axopatch 200B amplifier (Molecular Devices, CA, USA) at a holding potential of −30 mV. Cells were superfused with extracellular solution containing (in mM): 120 NaCl, 3 KCl, 3 HEPES, 2 CaCl2, 2 MgCl2, 25 D-glucose, pH 7.4 (adjusted with Tris base). Electrodes (2–4 MΩ final resistance) were pulled from thin-walled borosilicate glass GC150TF (Harvard Instruments) and backfilled with a solution containing (mM): 110 Tris-phosphate dibasic, 28 Tris-base, 11 EGTA, 2 MgCl2, 0.1 CaCl2, 4 Na-ATP, pH 7.3 (adjusted with Tris base). This K+-free intracellular solution was adopted in order to minimize the receptor rundown during recording . Series resistance (between 4–10 MΩ) was routinely compensated to 80–95%. Agonist solution (freshly prepared from frozen ACh stock aliquots), was applied through a modified U-tube system with solution exchange times of the order of 2–6 ms, measured by recording the open tip potential change to the application of a 50% diluted extracellular solution before the experiment. Different ACh concentrations were applied at 2 min intervals in a random order and a standard concentration of ACh (~EC20) was applied every third response in order to allow us to monitor for rundown.
Whole-cell records were filtered at 0.5 kHz and digitized at 2 kHz to a computer with a Digidata 1322A interface and Clampex software (both from Molecular Devices, CA, USA).
Recordings from HEK293 cells were performed in the outside-out configuration at −100 mV. The extracellular solution contained (in mM): 150 NaCl, 1.1 KCl, 2 MgCl2, 1 CaCl2, 10 HEPES, pH 7.2 (adjusted with NaOH). Electrodes were pulled from thick-walled Clark borosilicate glass GC150F (Harvard Instruments), coated near the tip with Sylgard 184® (Dow Corning, UK) to improve their dielectric properties and fire-polished prior to use, to a final resistance of 8–12 MΩ. Electrodes were filled with high potassium intracellular solution to minimise the contribution of potassium channels at the holding potential used for recording (−100 mV). The solution contained (in mM): 107.1 KCl, 1 CaCl2, 1 MgCl2, 10 HEPES, 11 EGTA, 20 TEACl, 2 MgATP (pH was adjusted to 7.3 with KOH and osmolarity was corrected to 320 mOsm with sucrose when needed).
Single-channel currents, elicited by the application of 5 µM ACh, were recorded with an Axopatch 200B amplifier, prefiltered at 10 kHz (using the 4-pole Bessel filter of the amplifier) and stored on a DAT tape (Bio-Logic Science Instruments, Claix, France). For off-line analysis, recordings were played back from the tape, filtered at 0.5–2 kHz with a 8-pole Bessel filter, and digitized to a computer with a sampling rate of 30 kHz, using a Digidata 1322A and Clampex.
ACh dose-response curves obtained either from HEK293 cells or oocytes, and also Zn2+ dose-response curves from oocytes were fitted with the Hill equation:
where I is the current response, measured at its peak, Imax is the maximum response, A is the agonist concentration, EC50 is the agonist concentrations for 50% maximum response and nH is the Hill coefficient. The program CVFIT, courtesy of D. Colquhoun and I. Vais, (available at http://www.ucl.ac.uk/Pharmacology/dc.html) was used for least square fitting. Data for each cell or oocyte were fitted separately to give the values for Imax, EC50, and nH in the Tables (shown as mean ± standard error of the mean).
In order to estimate the shift in the dose-response curve caused by the 9′ LT mutation, individual curves were normalised each to its fitted maximum and pooled, to give one normalised curve for each receptor combination. These sets were then fitted again simultaneously with the Hill equation, with the constraint of having equal Hill slopes (“parallel” fits). From the parallel fits, the program could estimate the distance between the wild-type and mutant curves (i.e. the dose ratios, see Table 1) and the 2.01-unit likelihood intervals for the dose ratios, which are equivalent to a confidence interval of ±2 standard deviations for a Gaussian variable .
For the potency ratios experiments, all partial dose-response curves obtained from each oocyte were fitted simultaneously with equal weighting, using CVFIT, with power functions constrained to be parallel. This is equivalent to fitting the dose-response curves with a Hill equation, with the constraint of having a Imax which is much higher than the responses analysed (see equation 1; as Figure 2 shows, the responses to the concentrations of the standard agonist used in these experiments, ACh, were less than 10% of Imax); at low concentrations, the slope of this log-log plot (Figure 4B) tends to the value of the Hill slope. For each cell, dose ratios for each agonist were estimated relative to ACh (equal to dose ratio of 1). Potency ratios for each agonist are the reciprocals of the dose ratios (i.e. equivalent to saying that each agonist is x times more potent than ACh). Errors and confidence intervals for potency ratios were calculated using Fieller's theorem . In order to confirm that the curves were indeed parallel, all data were also fitted without the constraint of parallelism to obtain the slope for each individual agonist (Table 3).
The dose-response curve to zinc was fitted to the sum of two Hill equations (parameterised as logistic equation by the fitting program Prism 5.01, GraphPad Software, San Diego, USA):
where I is the ACh response, measured at its peak and normalized to the value in the absence of Zn2+, w is a fitted coefficient between 0 and 1 that gives the weight of the first component, Max and Min are fitted parameters that express the maximum enhancing or inhibiting effect of Zn2+ (as a fraction of the control response), EC50 and IC50 are the Zn2+ concentrations for 50% maximum enhancement or inhibition, respectively, nH1 and nH2 are the Hill coefficients of the two components and x is log10 [Zn2+]. Data from each experiment were fitted separately and the resulting EC50 values were used to obtain the mean EC50 and the standard error of the mean reported in the text. For the experiments in Figure 5B, the fitted values of w, Max and Min were 0.31, 1.78 and −0.83, respectively.
The recordings chosen for single-channel analysis were idealized by time-course fitting using the program SCAN. After choosing and imposing the appropriate resolution (40–200 µs), fitted amplitude diagrams were created and fitted with the appropriate number of Gaussian curves using EKDIST (both programs are available at http://www.ucl.ac.uk/Pharmacology/dc.html). More than 100 openings were used from each experiment and these openings were included only if they were longer than twice the rise time of the filter and therefore had well-defined amplitudes. Current amplitudes were converted into chord conductances, assuming a reversal potential of 0 mV (with no correction for junction potential). For patches with both high- and low-conductance openings, burst duration was analysed separately for the two channels.
We are grateful to Ana I. Sánchez-García of Madrid University for sharing unpublished data.
Competing Interests: The authors have declared that no competing interests exist.
Funding: This work was supported by the Wellcome Trust (http://www.wellcome.ac.uk/) (project grant 064652, LGS; PhD studentship, SK); the MRC (http://www.mrc.ac.uk/index.htm) (programme grant G0400869, LGS; PhD studentship, SB); and the Royal Society (http://royalsociety.org/) (University Research Fellowship, MB; European Exchange grant, GH). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.