Author's present address I. C. Wood: School of Biochemistry and Molecular Biology, University of Leeds, Leeds LS2 9JT, UK.
KCNQ1-4 potassium channels were expressed in mammalian Chinese hamster ovary (CHO) cells stably transfected with M1 muscarinic acetylcholine receptors and currents were recorded using the whole-cell perforated patch technique and cell-attached patch recording.
Stimulation of M1 receptors by 10 μm oxotremorine-M (Oxo-M) strongly reduced (to 0–10%) currents produced by KCNQ1-4 subunits expressed individually and also those produced by KCNQ2+KCNQ3 and KCNQ1+KCNE1 heteromers, which are thought to generate neuronal M-currents (IK,M) and cardiac slow delayed rectifier currents (IK,s), respectively.
The activity of KCNQ2+KCNQ3, KCNQ2 and KCNQ3 channels recorded with cell-attached pipettes was strongly and reversibly reduced by Oxo-M applied to the extra-patch membrane.
It is concluded that M1 receptors couple to all known KCNQ subunits and that inhibition of KCNQ2+KCNQ3 channels, like that of native M-channels, requires a diffusible second messenger.
Heteromeric KCNQ2+KCNQ3 channels probably constitute the M-channel in sympathetic ganglia (Wang et al. 1998). M-channels control neuronal excitability (Brown, 1988; Marrion, 1997) and can be inhibited via M1 muscarinic acetylcholine receptors (Marrion et al. 1989; Bernheim et al. 1992; Hamilton et al. 1997) to produce a slow cholinergic excitation. When expressed in Xenopus oocytes, heteromeric KCNQ2+KCNQ3 channels can also be inhibited through M1 receptors (Wang et al. 1998). In the Xenopus expression system KCNQ4 produces a potassium current similar to the M-current (Kubisch et al. 1999) and a native analogue of the KCNQ4 current has recently been identified in mouse outer hair cells (Marcotti & Kros, 1999). However, sensitivity of KCNQ4 currents to muscarinic receptor stimulation is unknown.
The present experiments were aimed at establishing whether the muscarinic inhibition of KCNQ2+KCNQ3 channels (i) involves one of these subunits or both; (ii) is membrane-delimited or requires a diffusible second messenger; and (iii) can be extended to KCNQ1 and KCNQ4 subunits which have shorter carboxy termini and, therefore, a reduced number of potential regulatory sites.
Cell culture and transfection
The procedure for Chinese hamster ovary (CHO) hm1 cell culture and transfection has been described previously (Selyanko et al. 1999). Briefly, CHO hm1 cells, expressing the human M1 receptor (Mullaney et al. 1993), were grown at 37°C and 5 % CO2 in alpha-modified Eagle's medium (MEM) supplemented with 10 % fetal calf serum, 1 % L-glutamine and 1 % penicillin-streptomycin. Cells plated in 35 mm plastic dishes were transfected 1–2 days after plating using ‘LipofectAmine Plus’ (Gibco BRL). Plasmids containing KCNQ and CD8 cDNAs, driven by CMV promoter, were co-transfected in a channel to marker ratio of 10:1. For heteromultimers equal amounts of KCNQ2 and KCNQ3 cDNAs were used. Cells expressing channels were identified by adding CD8-binding Dynabeads (Dynal A/S) before recording.
Whole-cell perforated patch recording
Recordings were made at room temperature (20–22°C) 1–4 days after transfection (longer times were required to obtain sufficiently large KCNQ2 and KCNQ3 currents). Cells were bath perfused with the following solution (mM): 144 NaCl, 2.5 KCl, 2 CaCl2, 0.5 MgCl2, 5 Hepes, 10 glucose, pH 7.4 with Tris base. Pipettes were filled with an internal solution containing (mM): 80 potassium acetate, 30 KCl, 40 Hepes, 3 MgCl2, 3 EGTA, 1 CaCl2, pH 7.4 with KOH. The liquid junction potential was small (4 mV; Selyanko et al. 1995) and so no correction was made. Amphotericin B was used to perforate the patch (Rae et al. 1991). Perforated patch recording leads to an additional undischarged Donnan potential due to the impermeability of amphotericin pores to large intracellular anions. Assuming all intrapipette anions are permeant through the pores and all intracellular anions impermeant, the Donnan potential would be ≤ 11 mV (see Horn & Marty, 1988). Pipette resistance was 2–3 MΩ and the series resistance was compensated (60–90 %).
Cell-attached single-channel recording
Single-channel recordings were made 2–4 days after transfection, when the density of channels was enhanced. The methods of recording and analysis were similar to those previously employed for studying the M-channels (Selyanko & Brown, 1999). Intrapipette solution had the same composition as the bath solution used for whole-cell recording. In cell-attached experiments bath solutions had elevated (25 mM) K+ to pre-depolarise the cell at ∼-30 mV. Records were sampled at 4.46 kHz and filtered at 0.5-1 kHz. Pipette resistance was 2–3 MΩ.
In patches containing single channels, the effects of muscarinic stimulation on single-channel activity were deduced from changes in the open probability while the multi-channel activity was measured as the mean current through these channels.
Data acquisition and analysis
Data were acquired and analysed using pCLAMP software (version 6.0). Currents were recorded using an Axopatch 200A (or 200) patch-clamp amplifier, filtered at 1 kHz and digitised at 4–8 kHz. Activation curves were fitted by the Boltzmann equation:
where I is the tail current recorded at −70 mV following a step to membrane potential Vm, I(50)is current following a step to +50 mV, and V0.5 is the membrane potential at which I is equal to 0.5I(50). Current size and its inhibition by a muscarinic stimulant was measured from the amplitude of the deactivation tail recorded at −50 mV. The program ‘Origin’ (version 5.0, Microcal Software Inc.) was used for fitting activation curves and for creating the figures. Data are shown as means ±s.e.m. unless otherwise stated.
Drugs and chemicals
cDNAs for KCNQ1, KCNE1, KCNQ2 and KCNQ4 were human and KCNQ3 cDNA was from the rat. Oxo-M (oxotremorine methiodide) was obtained from RBI (Natick, MA, USA). All other drugs and chemicals were obtained from Sigma or BDH (Poole, UK).
Activation of KCNQ1-4 channels
Figure 1 shows examples of currents produced by different KCNQ subunits expressed in CHO cells. All currents had low thresholds (∼-60 mV), slow and sustained activations and slow deactivations. Activation curves for homomeric (KCNQ1, KCNQ2, KCNQ3, KCNQ4) and heteromeric (KCNQ2+KCNQ3) channels (Fig. 2) were fitted by the Boltzmann equation at V0.5 and k, respectively, equal to −7.8 ± 1.2 mV and 10.0 ± 1.1 mV, KCNQ1 (number ofcells, n= 6); −13.8 ± 0.4 mV and 12.1 ± 0.4 mV, KCNQ2 (n= 5); −36.8 ± 1.3 mV and 5.5 ± 1.1 mV, KCNQ3 (n= 3); −18.6 ± 0.3 mV and 9.8 ± 0.2 mV, KCNQ4 (n= 7); and −17.7 ± 0.6 mV and 11.9 ± 0.5 mV, KCNQ2+KCNQ3 (n= 6). Previously reported half-activation voltages for KCNQ1 (-24 mV, Chouabe et al. 1997), KCNQ2 (-37 mV, Biervert et al. 1998), KCNQ4 (-10 mV, Kubisch et al. 1999) and KCNQ2+KCNQ3 (-40 mV, Wang et al. 1998) channels differed somewhat from those obtained in the present experiments. The uncorrected junction potential of ∼4 mV and (uncertain) Donnan potential in our perforated-patch recording (see Methods) might contribute to some of the more positive half-activation voltages we have obtained, but this would not explain all the differences. These may therefore be associated with the different expression systems used.
The current onsets showed sigmoidal activation (in KCNQ4 only at very positive potentials) and all except KCNQ4 channels showed evidence of some inactivation at positive membrane potentials: they either slowly declined during continuous membrane depolarisation (KCNQ2 and KCNQ3) or ‘saturated’ with increased depolarization (KCNQ1 and KCNQ2+KCNQ3). Hyperpolarising voltage steps from the full activation potential of +50 mV removed inactivation and increased the currents.
M1 receptor inhibition of KCNQ1-4 channels
Figure 3A shows that stimulation of M1 receptors by oxotremorine-M (Oxo-M) strongly reduced the KCNQ2+KCNQ3 current (on average, to 11.6 ± 11.4 %, n= 3). When recorded in cell-attached patches, the activity of KCNQ2+KCNQ3 channels was reduced by adding Oxo-M to the bath (see Fig. 3B; average inhibition to 6.4 ± 2.8 %; 8 patches).
Figure 4 shows that KCNQ1-4 subunits expressed individually can also be inhibited by Oxo-M, to 3.0 ± 2.9 % (n= 5), 9.8 ± 1.1 % (n= 3), 0.0 ± 0.0 % (n= 6) and 3.9 ± 3.9 % (n= 5), in subunits 1–4, respectively. When recorded in cell-attached patches, the activities of KCNQ2 and KCNQ3 channels were inhibited by Oxo-M added to the bath, to 2.2 ± 2.0 % (5 patches) and 5.3 ± 4.5 % (3 patches), respectively (not shown).
All effects of Oxo-M on KCNQ1-4 channels were sustained and slowly reversible.
The idea that KCNQ2+KCNQ3 channels form M-channels is based on the overlapping distributions of KCNQ2-, KCNQ3- and M-channels in the nervous system and the similarities of the biophysical and pharmacological properties of their macroscopic currents (Wang et al. 1998). The present experiments give additional new support to this idea. Thus, KCNQ2+KCNQ3 channels were inhibited via activation of M1 receptors (as previously reported in oocytes by Wang et al. 1998); and furthermore, cell-attached recording showed that, like native M-channels (Selyanko et al. 1992), this inhibition required a diffusible second messenger. Since in CHO cells M1 receptors couple to the G protein Gq (Mullaney et al. 1993), M1-induced inhibition of KCNQ2+KCNQ3 channels, like that of native ganglionic M-channels (Haley et al. 1998), may be mediated by the αq subunit. Interestingly, average inhibition of KCNQ2+KCNQ3 currents by 10 μM Oxo-M (88 %, present experiments) was not different from that for the M-currents in mammalian sympathetic neurones (∼90 %; Hamilton et al. 1997; Haley et al. 1998).
The present experiments showed that homomeric KCNQ2 and KCNQ3 channels can also be inhibited via activation of M1 muscarinic receptors, and also through a diffusible second messenger. This suggests that both subunits forming the M-channel couple to M1 muscarinic receptors, probably through the same mechanism.
Two other members of the subfamily, KCNQ1 and KCNQ4, have not previously been considered as contributing to M-channels largely because one of them, KCNQ1, is seen as a ‘cardiac’ channel, with low expression in the brain (Yang et al. 1997). The other, KCNQ4, has not yet been extensively studied, and its distribution in the brain remains to be determined (Kubisch et al. 1999). The results obtained in the present experiments with expressed KCNQ1 and KCNQ4 channels are interesting in three respects. First, when recorded with the standard deactivation M-current protocol, KCNQ1 and KCNQ4 currents had kinetics strikingly similar to those of the M-current, and these channels too were inhibited via stimulation of M1 muscarinic receptors. Thus, all members of the KCNQ family reported so far can generate ‘M-channels’ as defined kinetically and pharmacologically. Second, acetylcholine is a transmitter in both the heart and the inner ear, so if M1 receptors and corresponding KCNQ subunits are co-expressed in the same cells, their coupling may influence cardiac function and hearing. (Acetylcholine has been reported to reduce IK,s when enhanced by isoprenaline (Harvey & Hume, 1989; Yazawa & Kemeyama, 1990), but this was probably due to inhibition of adenylate cyclase, and so was probably mediated by M2 rather than M1 receptors. Also, Freeman & Kass (1995) reported that IK,s in guinea-pig sino-atrial node cells was inhibited by cholinergic agonists but via a direct interaction with IK,s channels.) Third, because KCNQ1 and KCNQ4 subunits have short carboxy termini (compared with those of KCNQ2+KCNQ3 subunits; see Wang et al. 1998; Kubisch et al. 1999) this allows us to eliminate substantial parts of KCNQ2+KCNQ3 C-termini as obligatory sites involved in channel inhibition.
Thus, although the mechanism of muscarinic inhibition of M- (and KCNQ-) channels is still unknown, it is clearly conserved among all four members of the KCNQ family.
This work was supported by the MRC and the Wellcome Trust. We thank Dr D. McKinnon for KCNQ2 and KCNQ3 cDNAs and Dr M. T. Keating for KCNQ1 cDNA.