ACh and ATP play a role as fast neurotransmitters in peripheral synapses (Nishi & North, 1973; Buckley & Caulfield, 1992; Evans et al. 1992; Silinsky & Gerzanich, 1993; Zhou & Galligan, 1996; Burnstock, 1997). These neurotransmitters activate specific cation channels in the postsynaptic membrane by acting at pharmacologically distinct receptors, named nicotinic and P2X receptors, respectively. Such receptors, like other ligand-gated ion channels, are thought to be formed by multisubunit complexes, and several of their different subunits have already been cloned. Each subunit of the nicotinic receptor is believed to have four transmembrane domains, whereas those of P2X receptors appear to cross the membrane only twice (Brake et al. 1994; Surprenant et al. 1995; Lindström, 1996; Lindström et al. 1996; North, 1996; Burnstock, 1997).
How these subunits organize themselves to form ion channels in native membrane is not completely understood (see Bean, 1992). Recently, Nakazawa (1994a) suggested that there is an overlap of P2X and nicotinic channels, and that ATP could open a subpopulation of nicotinic channels by binding to P2X receptors in rat sympathetic neurons. This ‘channel overlap’ hypothesis was based on the findings that channel activation by ATP and ACh is not independent. Indeed, the cationic current activated by ACh (IACh; which was selectively blocked by hexamethonium) occluded the smaller cationic current induced by ATP (IATP; which was selectively blocked by suramin), in rat sympathetic neurons. Similar observations have been reported in rat phaeochromocytoma PC12 cells by Nakazawa et al. (1991).
In the present study our aim was to functionally characterize a putative inhibitory interaction between P2X and nicotinic receptors in submucosal neurons. Our findings do not support the channel overlap hypothesis but rather indicate that activation of nicotinic and P2X receptors open two different channel populations. These two channels, however, negatively modulate each other when they are simultaneously activated. This inhibitory interaction occurs as soon as whole-cell currents are detected (within a few milliseconds), are triggered by ion influx through the channels, and do not require Ca2+, Na+, Mg2+, G-proteins or protein phosphorylation, suggesting that they might be mediated by direct interaction between these receptors.
- Top of page
Young guinea-pigs (150–300 g), either male or female, were stunned and immediately exsanguinated by severing major neck blood vessels. A segment of small intestine (jejunum; about 5 cm in length) was removed, placed in modified Krebs solution and opened longitudinally. The mucosa was removed and the submucosal layer (submucosal preparation) was dissected from the underlying layers of smooth muscle. Modified Krebs solution was composed of (mm): 126 NaCl, 1.2 NaH2PO4, 1.2 MgCl2, 2.5 CaCl2, 5 KCl, 25 NaHCO3, 11 glucose; gassed with 95 % O2 and 5 % CO2.
Methods for dissociating and culturing enteric neurons have been described elsewhere (Barajas-López et al. 1996b). Briefly, the submucosal preparation was dissociated using sequential enzymatic treatments, first with a papain solution (0.01 ml ml−1; activated with 0.4 mg ml−1 of L-cysteine) and later with a collagenase (1 mg ml−1) plus dispase (4 mg ml−1) solution. After washing out these enzymes, submucosal neurons were plated on rounded coverslips coated with sterile rat tail collagen. Culture medium was minimum essential medium (97.5 %), containing 2.5 % guinea-pig serum, 2 mm L-glutamine, 10 U ml−1 penicillin, 10 μg ml−1 streptomycin and 15 mm glucose.
Whole-cell currents were recorded from short-term (2–40 h) primary cultures of submucosal neurons. Membrane currents were recorded using an Axopatch 1D amplifier. Patch pipettes were made as previously described (Barajas-López et al. 1996b) and had resistances between 1 and 3 MΩ. Sixty to seventy percent of the series resistance was compensated in about one-third of the experiments reported here. Series resistance compensation, however, did not affect the lack of additivity of IATP and IACh (see Results), so in most cases no compensation was made for this factor. Except when otherwise mentioned the holding potential was −60 mV and currents were induced by a maximal concentration (1 mm) of ATP or ACh. The standard solutions used, unless otherwise mentioned, have the following composition, inside the pipette (mm): 160 caesium glutamate, 10 EGTA, 5 Hepes, 10 NaCl, 3 ATPMg, and 0.1 GTP; external solution (mm): 160 NaCl, 2 CaCl, 11 glucose, 5 Hepes and 3 CsCl. The pH of all solutions was adjusted to 7.3–7.4 with either CsOH (pipette solutions) or NaOH (external solutions). With these standard solutions the usual input resistance of the neurons ranged from 1 to 10 GΩ. Whole-cell currents were recorded on a PC using AxoTape software (Axon Instruments). Analysis of these currents was performed on a Macintosh computer using AxoGraph software (Axon Instruments). Membrane potentials were corrected for the liquid junction potentials (pipette 11 mV negative). The recording chamber was continuously superfused with external solution at approximately 2 ml min−1. Rapid changes in the external solution were made by using an eight-barrelled device (Barajas-López et al. 1994). The external application of experimental substances was achieved by abruptly changing the tube delivering the external solution in front of the cell being recorded for a tube delivering the same solution plus the drug(s). Substances were washed out by switching back to the delivery tube containing only the external solution and by flushing of the bath. External solutions were delivered by gravity. Except where otherwise mentioned, experiments were performed at room temperature (∼23°C).
Staurosporine and K-252a were supplied by Kamiya (Thousand Oaks, CA, USA). Genistein, ACh and pyridoxalphosphate-6-azophenyl-2′,4′-disulfonic acid (PPADS) were purchased from Research Biomedical Inc. (Natick, MA, USA). All other substances were purchase from Sigma (St Louis, MO, USA). The pH of the external solution containing ATP, used to induce the IATP, was always readjusted with NaOH. The addition of the other substances to the external solution did not alter its pH.
Results were expressed as means ±s.e.m. and the number of cells used as n. Student'spaired t test was used to evaluate differences between mean values obtained from the same cells and Student's unpaired t test was used for data obtained from different groups of cells. Two-tailed P values of 0.05 or less were considered statistically significant.
- Top of page
Our results indicate that in guinea-pig submucosal neurons, activation of nicotinic and P2X receptors with ACh and ATP, respectively, opens two different channel populations. Activation of these channels is, however, not independent and currents carried through them are occluded when they are simultaneously activated. This would imply that there are inhibitory interactions between these two populations of channels. Such interactions occur as soon as ionic currents are detected (less than 5 ms), are triggered by ion influx through the channels, and do not specifically require Ca2+, Na+, Mg2+, G-protein activation or protein phosphorylation, suggesting that they may be mediated by direct interaction between nicotinic and P2X channels.
The currents induced by either ACh (1 mm) or ATP (1 mm) were specifically blocked by hexamethonium and PPADS, respectively, confirming that these transmitters, despite the high agonist concentrations used here, can specifically activate pharmacologically distinct receptors in submucosal neurons. The fact that the amplitudes of these two currents are independent of each other in the recorded neurons implies that these ligand-gated channels can be expressed separately in these neurons.
Inward currents carried through nicotinic and P2X receptors are not additive when these channels are simultaneously and maximally activated, indicating that activation of these channels is not independent. A similar inhibitory interaction between nicotinic and P2X receptors in rat sympathetic neurons was reported by Nakazawa (1994a), who explained these observations by proposing a ‘channel overlap’ hypothesis (see Introduction). According to this hypothesis, a subpopulation of nicotinic receptors share the same channels with P2X receptors. Our results obtained at a holding potential of +40 mV do not support this hypothesis. Thus, at this potential the currents induced by simultaneous application of ATP + ACh are similar to the sum of the currents induced by individual applications of ATP or ACh. Therefore, IATP and IACh must be carried through different populations of channels, otherwise they would not add at any potential. These results are in agreement with the fact that nicotinic and P2X channels have different structures (Brake et al. 1994; Surprenant et al. 1995; Lindström, 1996; Lindström et al. 1996; North, 1996; Burnstock, 1997).
The following observations rule out the possibility that the current occlusion reported here could be due to non-specific interactions between ATP and ACh molecules. First, in cells with no initial response to ATP or in which IATP had been previously blocked with PPADS, IACh+ATP has exactly the same amplitude and kinetics as the current induced by ACh alone. Second, nicotinic receptor desensitization also prevents any ACh effect on IATP and similarly, desensitization of P2X receptors prevents any ATP effect on IACh. Third, current occlusion was not observed for a low agonist-receptor occupancy. Fourth, in cells with no initial response to ACh or in which IACh had been blocked with hexamethonium, IACh+ATP has exactly the same amplitude and kinetics as IATP. It is also likely that this inhibitory interaction not only requires activation of receptors but might actually require channels to be open, since hexamethonium, a substance that is believed to be a nicotinic channel blocker (Ascher et al. 1979; Bertrand et al. 1990), prevents ACh effects on IATP.
At least four different observations indicate that IACh+ATP are carried through both nicotinic and P2X channels and not only through one population of these channels. (1) IACh+ATP desensitizes with different kinetics than IACh or IATP. Thus, IACh+ATP desensitizes faster than IATP but more slowly than IACh. (2) IACh+ATP desensitization kinetics are changed to resemble IACh kinetics when P2X receptors are blocked with PPADS, and can be made to resemble IATP kinetics by blocking nicotinic channels with hexamethonium. (3) When ATP and ACh are applied simultaneously, both nicotinic- and P2X-mediated currents are desensitized, whereas no cross-desensitization is observed when nicotinic and P2X receptors are desensitized individually. (4) In experiments in which one of the individual currents is significantly larger than the current induced by the other transmitter, IACh+ATP amplitude is the same as that of the current induced by the most effective neurotransmitter, and is significantly larger than the weaker transmitter. Altogether, these observations imply that IACh+ATP is carried through both populations of channels and that inhibition between these channels is reciprocal.
The short latency for this current occlusion suggests that no second messenger is required for these inhibitory interactions. In agreement with such an interpretation, inhibition of G-proteins (with N-ethylmaleimide), functional modification of G-proteins (by replacing GTP with GTP-γ-S in the pipette solution) and inhibition of protein phosphorylation (with staurosporine, K-252 or genistein) do not modify the occlusion observed between IACh and IATP. Consistently, this occlusion is still present during experiments carried out at 9°C. In rat sympathetic neurons, Nakazawa (1994b) reported that occlusion between IACh and IATP was prevented by K-252a and by using a non-hydrolysable analogue in place of ATP in the pipette solution. In apparent contradiction, these procedures failed to prevent current occlusion in submucosal neurons of the guinea-pig (present study). Such inconsistency might be the result of a real difference between the two types of neurons used. However, these two studies also differ in that we used maximal concentrations of both agonists (1 mm), whereas Nakazawa (1994b) used submaximal concentrations (100 μm; see Nakazawa, 1994a). With submaximal agonist concentrations, currents might appear to be additive if a potentiatory effect was unmasked by a particular experimental procedure.
Nicotinic and P2X channels of neurons are permeable to Na+ and Ca2+, suggesting that their intracellular accumulation could mediate the inhibitory interaction between these channels. This hypothesis is supported by the fact that nicotinic and P2X receptors of enteric neurons are known to be permeable to these cations (Trouslard et al. 1993; Barajas-López et al. 1994, 1996a; Christofi et al. 1997). However, we have ruled out such a possibility, by showing that the interaction is not affected by the absence of Ca2+ or Na+ in the intracellular and extracellular solutions. Mg2+ are also not required because occlusion is still observed in the absence of these ions.
In conclusion, our results indicate that there is a very fast inhibitory interaction between nicotinic and P2X channels, which occurs only for inward currents. These interactions occur within just a few milliseconds, supporting the hypothesis that nicotinic and P2X channels are located very close to each other in the neuronal membrane, perhaps forming functional units constituted by at least one channel of each type. This hypothesis is an extension of the well-known phenomenon that transmitter-gated channels, like other ion channels, do not diffuse freely in the membrane but are clearly localized underneath the appropriate presynaptic terminal. Several proteins have been proposed to be involved in the clustering of ion channels at specific membrane sites, e.g. 43/rapsyn is believed to be responsible for the clustering of nicotinic channels (Sheng & Kim, 1996). The fact that such inhibitory interactions appear to occur only when nicotinic channels are permeable and only for inward currents suggests the existence of a sensory mechanism specific for inward ion movements. Therefore, we propose a model in which the detection of ion influx in one channel (e.g. nicotinic) would induce allosteric changes in the neighbouring channel (e.g. P2X) that would block it, so that two channels of a particular functional unit would never be open at the same time. The inhibitory effect on the second channel would disappear once the ion influx through the first channel was stopped. A functional significance of this inhibitory interaction between these ligand-gated channels would be the saving of cellular energy by limiting ion movements through the cellular membrane.