Intracellular recordings were made from isolated bundles of the circular muscle layer of mouse gastric antrum and the responses evoked by stimulating intrinsic nerve fibres were examined. Transmural nerve stimulation evoked a fast inhibitory junction potential (fast-IJP) which was followed initially by a smaller amplitude long lasting inhibitory junction potential (slow-IJP) and a period of excitation. The excitatory component of the response was abolished by atropine, suggesting that it resulted from the release of acetylcholine and activation of muscarinic receptors. Fast-IJPs were selectively reduced in amplitude by apamin and slow-IJPs were abolished by Nω-nitro-l-arginine. Slow-IJPs were associated with a drop in membrane noise, suggesting that inhibition resulted from a reduced discharge of unitary potentials by intramuscular interstitial cells of Cajal (ICCIM). The chloride channel blocker, anthracene-9-carboxylic acid, reduced the discharge of membrane noise in a manner similar to that detected during the slow-IJP. When recordings were made from the antrum of W/WV mice, which lack ICCIM, the cholinergic and nitrergic components were absent, with only fast-IJPs being detected. The observations suggest that neurally released nitric oxide selectively targets ICCIM causing a hyperpolarization by suppressing the discharge of unitary potentials.
The activity of gastric muscle can be modified by neuronal activity. Cholinergic nerve stimulation results in an increased force of contraction associated with each slow wave and an increase in their frequency of occurrence (Vogalis & Sanders, 1990; Hirst et al. 2002c). Conversely, inhibitory nerve stimulation results in a fall in the force of contraction associated with each slow wave without any marked change in their rate of generation (Dickens et al. 2000). In the guinea-pig gastric antrum, where inhibitory nerve stimulation involves apamin-sensitive and nitrergic components, the nitrergic component is dominant (Desai et al. 1994; Dickens et al. 2000). In other regions of the gut, inhibitory nerve stimulation has been shown to evoke a biphasic inhibitory junction potential (IJP), consisting of an apamin-sensitive fast-IJP which is followed by an apamin-insensitive nitrergic slow-IJP (Niel et al. 1983; Lyster et al. 1992; He & Goyal, 1993; Zhang & Paterson, 2002). Although it has been traditionally held that both excitatory and inhibitory transmitters produce their responses by acting on smooth muscle cells within the gastrointestinal tract, this has recently been questioned. As well as generating the secondary component of the slow wave (Dickens et al. 2001), it has been suggested that ICCIM are intermediaries in the pathway by which neuronal information modifies contractile activity (Burns et al. 1996; Ward et al. 2000). Thus the responses to both inhibitory and excitatory nerve stimulation are greatly attenuated in tissues devoid of ICCIM (Burns et al. 1996; Ward et al. 2000; Beckett et al. 2002). Similarly, the responses to cholinergic nerve stimulation in the circular layer of the guinea-pig gastric antrum have been largely attributed to activation of a chloride conductance in ICCIM (Hirst et al. 2002b,c) rather than to activation of cation-selective channels linked to muscarinic receptors on smooth muscle cells (Benham et al. 1985).
The experiments described here have examined the process of inhibitory neurotransmission in bundles of circular muscle isolated from the gastric antrum of mice, in preparations with and without ICCIM. Inhibitory nerve stimulation evoked a biphasic IJP: a fast-IJP followed by a slow-IJP when ICCIM were present but only a fast-IJP when ICCIM were absent. This suggests that ICCIM, rather than smooth muscle cells, are the major target for nitric oxide (NO), the inhibitory transmitter responsible for the slow-IJP, but not the transmitter responsible for the fast-IJP. The results are discussed in relation to the idea that during inhibitory nerve stimulation neurally released NO causes a hyperpolarization by suppressing an ongoing chloride conductance (gCl) (see Crist et al. 1991a,b; Zhang & Paterson, 2002) present in ICCIM (Hirst et al. 2002b).
The animal experimentation ethics committee at the University of Nevada in Reno approved the procedures described. Experiments were carried out on BALB/c mice bred at the University of Nevada and on C57BL/6 wild-type and W/WV mutant mice obtained from Jackson Laboratory (Bar Harbor, ME, USA). Mice of either sex were killed by cervical dislocation and exsanguination. The stomach was exposed and transferred to a dissecting chamber filled with oxygenated (97 % O2-3 % CO2) physiological saline (composition (mm): NaCl, 120.7; NaHCO3, 15.5; NaH2PO4, 1.2; KCl, 5.9; MgCl2, 1.2; CaCl2, 2.5; and dextrose, 11.5). The stomach was cut along the lower curvature and the mucosa dissected away. Subsequently the preparation was re-pinned serosal surface uppermost and the longitudinal muscle layer was dissected away. Single bundles of circular muscle (diameter 50-100 µm, length 400-800 µm) were dissected free and pinned in a recording chamber (see Suzuki & Hirst, 1999). A pair of platinum stimulating electrodes was positioned, one on either side of the preparation, to allow intramuscular nerve terminals to be stimulated (Hirst et al. 2002c). Intracellular recordings were made using sharp microelectrodes (90-150 MΩ) filled with 3 m KCl. In some experiments the preparations were impaled with two independently mounted electrodes; one was used to record membrane potential changes and the other to inject current so changing the membrane potential of the preparations. Signals were amplified with an Axoprobe amplifier, low-pass filtered (cut-off frequency 1 kHz) digitized and stored on a computer for later analysis. Preparations were constantly perfused with physiological saline solution warmed to 37 °C. In many experiments, nifedipine (1-10 µm; Sigma Chemical Co., St Louis, MO, USA), which has been shown not to significantly affect the waveform of regenerative potentials (Suzuki & Hirst, 1999), was added to the physiological saline to suppress muscle movements. In some experiments, the average of moving standard deviation time courses of individual membrane potential recordings was calculated using a 0.5 s window (see Hirst & Edwards, 2001). In other experiments, spectral density curves were constructed from baseline regions before, during and after the slow-IJP (see Edwards et al. 1999).
Atropine sulphate, apamin, Nω-nitro-l-arginine (l-NA) anthracene-9-carboxylic acid (9-AC) and nifedipine (obtained from Sigma) were used in these experiments.
The distribution of cells expressing Kit in bundles of circular muscle was determined as described previously (Hirst et al. 2002a). Briefly, preparations were fixed in acetone and washed in phosphate-buffered saline before being incubated with a rat anti-Kit monoclonal antibody (ACK-2; 5 µg ml−1; Gibco BRL, Gaithersburg, MD, USA) for 24-48 h at 4 °C. After washing, preparations were incubated for 1 h at room temperature with Alexa Fluor 488 goat anti-rat IgG (1:200; Molecular Probes, Eugene, OR, USA), which was used as the secondary antibody to detect Kit labelling. Tissues were examined with a Biorad MRC 600 confocal microscope (Hercules, CA, USA) with an excitation wavelength appropriate for Alexa Fluor 488. The distributions of ICCIM were determined by optically sectioning through the entire muscle bundle, followed by construction of z-stacks. Confocal micrographs shown in this paper are digital composites of z-series scans of 5-15 optical sections through a depth of 20 µm starting at the myenteric border. Final images were constructed with Bio-Rad ‘Comos’ software.
All data are expressed as means ± standard error of the mean (s.e.m.). Student's t tests were used to determine if data sets differed; P values of less than 0.05 were taken to indicate significant differences between sets of observations.
Micrographs of isolated bundles of circular muscle, dissected from BALB/c mice and stained with Kit antibody, showed that they were devoid of ICCMY but that ICCIM were distributed throughout the bundles (Fig. 1A). When intracellular recordings were made from these bundles they were found to have resting potentials in the range −51 to −60 mV (-55.9 ± 0.8 mV, n= 14). In each preparation the membrane potential recordings displayed an on-going discharge of membrane noise which in six of the preparations led to an irregular discharge of regenerative potentials (see Suzuki & Hirst, 1999). A single brief stimulating pulse (0.05-0.2 ms) delivered to the pair of stimulating electrodes, evoked a fast-IJP followed by a slow-IJP. When brief trains of impulses (five impulses delivered at 5 Hz) were applied, the inhibitory response was often followed by a wave of depolarization which, on occasion, masked the slow-IJP (Fig. 2A). Excitatory responses were abolished by atropine (1 µm), invariably revealing the slow-IJP (Fig. 2B). In some preparations neither a late phase of excitation nor a slow-IJP was detected in control solution following the fast-IJP; in these preparations the addition of atropine invariably revealed a slow-IJP suggesting that a small excitatory response had masked the slow-IJP. In four experiments, using circular bundles isolated from BALB/c mice, preparations were impaled with two electrodes, one to pass current and the other to record membrane potential changes. Depolarizing current pulses evoked electrotonic potentials which triggered regenerative potentials (Fig. 2C); similarly the break of a hyperpolarizing current was followed by a regenerative potential (Fig. 2D). Regenerative potentials had a peak amplitude in the range 7-25 mV (16.3 ± 3.9 mV, n= 4), with the signals being notably less robust than those detected in bundles of antral muscle isolated from guinea-pigs (Suzuki & Hirst, 1999). The time course of the onset, or offset, of each electrotonic potential was adequately described by a single exponential; these had time constants in the range 55-140 ms (100 ± 20 ms, n= 4). Preparations had input resistances in the range 1.2-3.5 MΩ (2.2 ± 0.5 MΩ, n= 4).
Properties of inhibitory junction potentials recorded from isolated circular bundles of mouse antrum
The properties of inhibitory junction potentials were characterized in physiological saline containing atropine (1 µm). Since individual traces were invariably dominated by discharges of membrane noise, the amplitudes of the responses were determined from the mean responses produced by 10 successive trains of stimuli (Fig. 3). Single stimuli evoked a fast-IJP with a mean peak amplitude of 4.8 ± 1.0 mV and a slow-IJP with a mean amplitude, determined 2.5 s after the stimulus, of 0.6 ± 0.1 mV (Fig. 3A; n= 5). Three impulses, delivered at 5 Hz, evoked a fast-IJP with a mean peak amplitude of 12.6 ± 1.7 mV and a slow-IJP with a mean amplitude, determined 2.5 s after the first stimulus, of 1.6 ± 0.3 mV (n= 5). Five impulses, delivered at 5 Hz, evoked a fast-IJP with a mean peak amplitude of 14.0 ± 1.8 mV and a slow-IJP with a mean amplitude of 2.1 ± 0.3 mV (n= 5). Apamin (0.5 µm) reduced the amplitude of the fast-IJP from 14.0 ± 1.8 to 6.6 ± 0.9 mV (n= 5), but not that of the slow-IJP (Fig. 4A and B). The subsequent addition of l-NA (10 µm) further reduced the amplitude of the fast-IJP to 2.5 ± 0.9 mV (n= 5) and abolished the slow-IJP (Fig. 4C). Conversely, in three preparations, the addition of l-NA (10 µm) reduced the amplitude of the fast-IJP, from 11.3 ± 1.2 to 8.2 ± 0.8 mV, and abolished the slow-IJP (see as example Fig. 9, recorded from wild-type antrum). The subsequent application of apamin further reduced the amplitude of the fast-IJP to 1.7 ± 0.4 mV.
When the individual traces were examined, slow-IJPs appeared to be associated with a suppression of membrane noise (see also Zhang & Paterson, 2002), which results from the discharge of unitary potentials (see Edwards et al. 1999). Furthermore, the suppression of unitary potentials during the slow-IJP appeared to be abolished by l-NA but to be largely unaffected by apamin (Fig. 4). The possibility that slow-IJPs were associated with a change in the discharge of unitary potentials was tested in two ways. Firstly, the average of moving standard deviation time courses of individual membrane potential recordings was calculated (Hirst & Edwards, 2001). This confirmed that slow-IJPs were associated with a marked suppression of membrane noise. An experiment is illustrated in Fig. 5. The mean response to 10 successive trains of stimuli (five impulses at 5 Hz; Fig. 5A) shows that the stimuli evoked a fast-IJP with a peak amplitude of 18 mV and that this was followed by a slow-IJP which lasted a further 15 s. The three sample traces, from which the average was constructed, show that the discharge of unitary potentials was largely suppressed during the slow-IJP and that they began to reoccur in an irregular manner. The three moving standard deviation traces, calculated from the sample membrane potential traces, confirm that immediately after the fast-IJP the variability of the membrane potential recording had fallen to a low value (Fig. 5B). The lower trace (Fig. 5C) shows the mean time course of the change in membrane potential standard deviation, calculated from the entire sequence of 10 traces. It can be seen that the standard deviation fell from a mean value of 0.4 mV to 0.2 mV and then returned to its control value with a similar time course to that of the averaged response (Fig. 5A). The mean standard deviation from this and seven further experiments for the initial baseline was 0.57 ± 0.06 mV. During the slow-IJP, the mean standard deviation fell to 0.36 ± 0.05 mV (n= 8). Some 10 s after the slow-IJP, the mean standard deviation of the baseline had returned to 0.55 ± 0.07 mV (n= 8). The mean standard deviations determined during initial and late baseline periods did not differ significantly but the mean standard deviation during the slow-IJP differed significantly from both baseline measurements. In the presence of l-NA (10 µm), nerve stimulation failed to evoke a slow-IJP and a significant fall in the standard deviation of membrane potential was no longer detected (n= 8). On the other hand, a significant fall in standard deviation of the membrane potential continued to occur during the slow-IJP when recordings were made from preparations bathed in solutions containing apamin (0.5 µm) in the absence of l-NA (n= 3).
The second way in which the effects of neurally released NO on the discharge of unitary potentials were assessed involved the construction of power spectral density curves (Edwards et al. 1999). This was done from control recordings before the trains of stimuli, during the slow-IJP and during control periods after the slow-IJP (see Fig. 5C). Both sets of control power spectral density curves were very similar and had characteristic shapes similar to those determined from recordings from the circular layer of guinea-pig antrum (Edwards et al. 1999). Energy increased from low values at frequencies above 10 Hz to reach a plateau at frequencies above 1 Hz. At frequencies above 10 Hz, the power spectral density curves were dominated by electrode recording noise (Fig. 6B). During the slow-IJP, the discharge of low frequency noise was suppressed (Fig. 6B) with the peak spectral density power falling, during the slow-IJP, to 35 ± 10 % of the control peak spectral density (n= 5). Subsequently l-NA (10 µm) was added to the physiological saline to abolish the slow-IJP. When power spectral density curves were constructed from the region where the slow-IJP had previously occurred (Fig. 6C), the peak spectral density now only fell to 78 ± 10 % (n= 5) of its control peak spectral density (Fig. 6D). The changes in spectral density during the slow-IJP recorded in control solution and during the corresponding period in the presence of l-NA were significantly different.
Together these observations show that slow-IJPs are associated with a fall in the rate of occurrence of unitary potentials. Given the small amplitudes of the slow-IJPs it is unlikely that this could result from an increase in membrane conductance. Rather the observations suggest that slow-IJPs are generated by a cessation of noise discharge.
Effect of 9-AC on the discharge of membrane noise recorded from isolated circular bundles of mouse antrum
In the circular layer of the guinea-pig antrum, unitary potentials, which give rise to membrane noise, result from the activation of calcium-activated chloride channels (Hirst et al. 2002b). The effects of the chloride channel blocker 9-AC (1 mm) on circular muscle bundles obtained from mouse antrum were examined. 9-AC reduced the discharge of membrane noise and this was associated with a hyperpolarization of some 2-4 mV (Fig. 7A, B and D). When power spectral density curves were constructed from data obtained in control solution and in the presence of 9-AC, a fall in low frequency power was apparent (Fig. 7C and E). In four experiments where power spectral density curves were calculated after addition of 9-AC (1 mm) to the physiological saline, the power spectral density fell to 15 ± 7 % of control value. As has been shown recently (Zhang & Paterson, 2002), in the presence of 9-AC, nerve stimulation failed to evoke a slow-IJP (n= 3, data not shown).
Properties of IJPs recorded from bundles of circular muscle taken from antral regions of wild-type and W/WVmice
The previous observations suggest that slow-IJPs result from a reduction in the rate of discharge of unitary potentials, which are thought to be generated by ICCIM (Dickens et al. 2001). To test this idea, neural responses were recorded from bundles of circular muscle isolated from antral regions of strain-matched C57BL/6 and W/WV mice. Micrographs of bundles of circular muscle from C57BL/6 mice showed that these preparations also contained ICCIM but lacked ICCMY (Fig. 1B) and that the density and distribution of ICCIM was the same as in BALB/c mice. When recordings were made from such preparations, they were found to have resting membrane potentials of −56.6 ± 1.0 mV (n= 8). Again membrane potential recordings were dominated by membrane noise, which, in preparations taken from C57BL/6 mice, was reduced but not abolished by 9-AC (1 mm, n= 2). We have not examined the properties of the residual membrane noise further other than to note that it was abolished by apamin (0.5 µm) and that it was also present in preparations obtained from W/WV mice. When impaled with two intracellular electrodes, one to pass current and one to record membrane potential changes, preparations were found to have input resistances of 2.0 ± 0.3 MΩ and membrane time constants of 90 ± 20 ms (n= 8). Depolarizing, or the break of hyperpolarizing, current pulses evoked either an increase in the rate of discharge of unitary potentials or a regenerative potential. Transmural nerve stimulation (five stimuli at 5 Hz) evoked a fast-IJP which was followed by a slow-IJP. In four of the preparations, the inhibitory responses were followed by late phases of excitation (Fig. 8A) which were abolished by atropine (1 µm; Fig. 8B). In two of the other preparations, an atropine-sensitive phase of excitation could be evoked by increasing the number of stimuli to 10, delivered at 1, 5 or 10 Hz. In the presence of atropine, five stimuli delivered at 5 Hz evoked a fast-IJP with a peak amplitude of 10.1 ± 1.1 mV and a slow-IJP with an amplitude of 2.0 ± 0.3 mV (n= 8).
When recordings were made from the corresponding bundles of muscle obtained from W/WV mice, they had resting membrane potentials of −56.0 ± 1.8 mV, input resistances of 2.1 ± 0.4 MΩ and membrane time constants of 80 ± 10 ms (n= 5). These values were not significantly different from those determined from preparations obtained from strain-matched C57BL/6 mice. Depolarizing current pulses failed to evoke regenerative potentials but, if of sufficient amplitude, readily evoked a discharge of action potentials. As pointed out, membrane potential recordings obtained from antral preparations of W/WV mice also frequently displayed a discharge of membrane noise (Fig. 8C); this was unaffected by 9-AC (1 mm, n= 2) but was abolished by apamin (0.5 µm, n= 5). Transmural nerve stimulation (five impulses at 5 Hz) invariably evoked a fast-IJP but neither a slow-IJP nor an atropine-sensitive component was detected in any of the preparations (Fig. 8C), even when the number of stimuli was increased. Bundles of circular muscle taken from W/WV mice were found to lack both ICCIM and ICCMY (Fig. 1C).
As with antral preparations taken from BALB/c mice, slow-IJPs recorded from antral bundles of C57BL/6 mice were abolished by l-NA (10 µm; Fig. 9B). In five preparations where l-NA was applied first, it reduced the amplitude of the fast-IJP, evoked by five impulses at 5 Hz, from 10.1 ± 0.8 to 8.5 ± 0.3 mV (Fig. 9B). The subsequent addition of apamin (0.5 µm) further reduced the amplitude of the fast-IJP to 2.4 ± 0.6 mV (Fig. 9C). In contrast, l-NA had no effect on the inhibitory responses recorded from each of the five antral preparations obtained from W/WV mice (Fig. 9E), but again apamin (0.5 µm) reduced the amplitude of the fast-IJP from 8.4 ± 0.5 to 2.6 ± 0.4 mV (Fig. 9F).
Together these observations indicate that nitrergic and cholinergic responses could only be detected in preparations which contained ICCIM but that apamin-sensitive IJPs persisted when ICCIM were absent.
These experiments show that inhibitory nerve stimulation evoked a biphasic IJP in the circular muscle layer of the gastric antrum of mice. The initial component was selectively reduced in amplitude by apamin, suggesting that it resulted largely from the opening of small conductance calcium-activated potassium channels (Niel et al. 1983; Komori & Suzuki, 1986). The second component, the slow-IJP, was abolished by l-NA, indicating that it resulted from the release of NO (Li & Rand, 1990; Dalziel et al. 1991; Lyster et al. 1992; He & Goyal, 1993). The slow-IJP was associated with a fall in the rate of discharge of unitary potentials and this response was absent in tissues which lacked ICCIM. Chloride channel-blocking agents produced a similar hyperpolarization and fall in power spectral density to that detected during a slow-IJP. These observations suggest that neurally released NO selectively targets ICCIM where it prevents the discharge of unitary potentials. This gives rise to a hyperpolarization which therefore results from a suppression of gCl (Crist et al. 1991a,b; Zhang & Paterson, 2002), predominantly located in ICCIM (see Hirst et al. 2002b).
The experimental observations support the view that, in many regions of the gastrointestinal tract, intrinsic nerve terminals selectively innervate ICCIM (Wang et al. 2000). Thus the responses to inhibitory and excitatory nerve stimulation are much reduced in the fundus when ICCIM are absent in W/WV mutant mice and in Steel mutants (Burns et al. 1996; Ward et al. 2000; Beckett et al. 2002). In the mouse antrum, cholinergic responses were detected in antral preparations from BALB/c mice and from C57BL/6 mice. They resembled regenerative potentials in that they resulted from a summed discharge of unitary potentials (Fig. 2; see also Hirst et al. 2002c). Both these responses were absent in antral preparations from W/WV mice supporting the view that both involve ICCIM (Dickens et al. 2001; Hirst et al. 2002a,c).
The observations suggest that neurally released NO exerts its inhibitory effects by suppressing the discharge of unitary potentials (Fig. 6). This idea is consistent with the observation that inhibitory nerve stimulation does not evoke a nitrergic component in mutant mice devoid of type 1 IP3 receptors (Suzuki et al. 2000). The discharge of unitary potentials by ICCIM relies upon the release of calcium from IP3-dependent stores (Hirst et al. 2002b). As this will not occur when IP3 receptors are absent, unitary potentials will not be generated and the target for nitrergic inhibition will be absent. Direct evidence for the role of ICCIM in the generation of nitrergic responses came from the finding that slow-IJPs were not detected in preparations taken from W/WV mutant mice (Fig. 8 and Fig. 9). This observation implies that NO exerts its effects by selectively inhibiting ICCIM. In many ways this is a surprising finding. Although it is generally accepted that conventional transmitters are released from specialized release points, which lie in close apposition to target cell membranes, NO is thought to be synthesized in the cytosol of pre-junctional nerve terminals and to diffuse out in a non-targeted manner (Rand & Li, 1995). The present findings might imply that NO is preferentially released from the surface of varicosities near the target cell, in this case ICCIM (Wang et al. 2000), but how this would occur is not clear. Alternatively NO may be released from all surfaces of the varicosity but only be effective when present in a high concentration near the membrane of the closest cells, presumably ICCIM (Wang et al. 2000). If this were the case it implies that there is a large degree of wasteful NO production. Alternatively, it may simply be that smooth muscle cells are less sensitive to NO than are ICCIM.
It is also unclear as to how NO exerts its inhibitory effects on antral motility. Although the present observations fully support the suggestion that the slow nitrergic component of IJPs results from a suppression of gCl (Crist et al. 1991a,b; Zhang & Paterson, 2002), it seems unlikely that inhibition of antral contractility would result simply from the small amplitude of hyperpolarization occurring during each slow-IJP. In the gastric antrum of guinea-pigs the large changes in membrane potential, evoked by the apamin-sensitive component of the response to inhibitory nerve stimulation, appear to make little or no contribution to the inhibition of mechanical activity (Dickens et al. 2000). Conversely, blocking the synthesis of NO, which provokes a much smaller membrane hyperpolarization (Fig. 4), abolishes the inhibitory effect of vagal nerve stimulation in the guinea-pig antrum (Dickens et al. 2000). Two explanations seem possible. Firstly, it may be that the electrical responses involving ICCIM have little relationship to the inhibitory mechanical responses detected in intact tissue. This seems to be unlikely since electrical and mechanical responses are both severely attenuated in tissues which lack ICCIM (Beckett et al. 2002). A second explanation might be that NO exerts its inhibitory influence on both mechanical and electrical activity by a mechanism directly linked to the suppression of unitary potential discharge. This suggestion implies that an excitatory messenger, presumably IP3, synthesized by ICCIM diffuses into nearby smooth muscle cells so participating in the regulation of their internal concentration of calcium ions. Alternatively, the suppression of unitary potentials may involve the formation of a second inhibitory messenger, or messengers, in ICCIM. Given the long time courses of slow-IJPs and the fact that they appear to involve the suppression of IP3-dependent events, this seems a reasonable suggestion. The inhibitory second messenger, produced in ICCIM, may then diffuse to neighbouring smooth muscle cells and inhibit their contractility in a manner largely independent of changes in membrane potential (Dickens et al. 2002).
Whereas the nitrergic component of the responses to inhibitory nerve stimulation required ICCIM, this was not the case for the fast-IJP with an apamin-sensitive IJP being detected in all tissues examined. Presumably this component results from an inhibitory transmitter acting directly on smooth muscle. In other tissues it has been suggested that ATP is responsible for the initiation of the fast-IJP (however, see Ohno et al. 1996) and that ATP and NO may be released from separate populations of inhibitory nerve terminals (Bridgewater et al. 1995). If this is the case, our observations might suggest that purinergic nerves selectively innervate smooth muscle cells; however, histological studies have failed to detect specialized neuroeffector junctions between nerve terminals and smooth muscle cells in the gastric antrum (Horiguchi et al. 2003). Alternatively, if ATP and NO are released from the same nerve terminals, since the amplitude of fast-IJPs differed little between preparations with and without ICCIM (Fig. 9), either ATP must fail to evoke a response in ICCIM or, in the absence of ICCIM, the sensitivity of smooth muscle cells to ATP must be increased as has been recently reported (Sergeant et al. 2002a).
In summary, these experiments have shown that, in the antrum, nitrergic and cholinergic responses, evoked by nerve stimulation, both depend upon the presence of ICCIM (Burns et al. 1996; Ward et al. 2000; Beckett et al. 2002). The extent to which ICCIM play a similar role in other tissues is not known; clearly in the longitudinal layer of the ileum where excitatory nerves form junctions with smooth muscle cells (Klemm, 1995) this cannot be the case (see Cousins et al. 1993). On the other hand, in tissues where ICCIM are present (Sergeant et al. 2000), these cells also appear to be intermediaries in the process of neuroeffector transmission (Sergeant et al. 2002b). Furthermore, the membrane potential changes appear to involve a modulation of a chloride conductance, with gCl being increased by cholinergic stimulation in the antrum (Hirst et al. 2002c) and by noradrenergic stimulation in the urethra (Sergeant et al. 2002b) and with gCl being decreased by nitrergic inhibition in the intestine (Crist et al. 1991a). The simplest explanation for the findings is that neurally released excitatory transmitters, acetylcholine in the gut and noradrenaline in the urethra, increase the rate of generation of unitary potentials by ICCIM whereas neurally released NO slows their rate of occurrence.
This project was supported by a grant from the Australian NH&MRC and by a grant from the National Institutes of Health (grant DK57236 to S.M.W.). We also wish to thank Dr N. J. Bramich for her careful reading of the manuscript. Professor H. Suzuki was supported by a grant from the Japan Society for the Promotion of Science.