Interstitial cells: involvement in rhythmicity and neural control of gut smooth muscle

Abstract

Many smooth muscles display spontaneous electrical and mechanical activity, which persists in the absence of any stimulation. In the past this has been attributed largely to the properties of the smooth muscle cells. Now it appears that in several organs, particularly in the gastrointestinal tract, activity in smooth muscles arises from a separate group of cells, known as interstitial cells of Cajal (ICC), which are distributed amongst the smooth muscle cells. Thus in the gastrointestinal tract, a network of interstitial cells, usually located near the myenteric plexus, generates pacemaker potentials that are conducted passively into the adjacent muscle layers where they produce rhythmical membrane potential changes. The mechanical activity of most smooth muscle cells, can be altered by autonomic, or enteric, nerves innervating them. Previously it was thought that neuroeffector transmission occurred simply because neurally released transmitters acted on smooth muscle cells. However, in several, but not all, regions of the gastrointestinal tract, it appears that nerve terminals, rather than communicating directly with smooth muscle cells, preferentially form synapses with ICC and these relay information to neighbouring smooth muscle cells. Thus a set of ICC, which are distributed amongst the smooth muscle cells of the gut, are the targets of transmitters released by intrinsic enteric excitatory and inhibitory nerve terminals: in some regions of the gastrointestinal tract, the same set of ICC also augment the waves of depolarisation generated by pacemaker ICC. Similarly in the urethra, ICC, distributed amongst the smooth muscle cells, generate rhythmic activity and also appear to be the targets of autonomic nerve terminals.

Interstitial cells of Cajal (ICC) may be considered to be a specialised population of smooth muscle cells. Both arise from common mesenchymal cells (Lecoin et al. 1996; Young et al. 1996; Young, 1999). However, whereas smooth muscle cells develop an extensive array of contractile elements, ICC have few contractile elements but contain large numbers of mitochondria, an abundance of endoplasmic reticulum and distinct sets of channels in their membranes (Sanders, 1996). Many ICC express Kit, a tyrosine kinase receptor; this allows them to be recognised by their ability to bind antibodies to Kit (Sanders, 1996). Similarly ICC readily react with antibodies to vimentin whereas nearby smooth muscle cells do not (Rumessen & Thuneberg, 1996).

Many tissues, isolated from different regions of the gastrointestinal tract, contract rhythmically in the absence of neuronal or hormonal stimulation. When contractions and membrane potential are recorded simultaneously each contraction is seen to be triggered by a long lasting wave of depolarisation: because of their low frequency of occurrence and long duration, the waves of depolarisation have been termed slow waves (Tomita, 1981; Szurszewski, 1981). The origin and basis of the generation of slow waves have been debated for many years. It was initially thought that the generation of slow waves reflected some properties of gastrointestinal smooth muscle cells (Connor et al. 1974; El-Sharkaway & Daniel, 1975; Tomita, 1981) but studies on isolated smooth muscle cells have consistently failed to demonstrate a capability to generate slow wave activity (Farrugia, 1999). It has also long been recognised that the generation of slow waves does not rely on the sequential activation of voltage-dependent ion channels as do cardiac pacemaker cells. Rather, many early studies raised the possibility that rhythmical activity relied on the cycling of one or more metabolic processes within cells of the gut wall. Thus Prosser and his colleagues proposed that the generation of slow waves involved changes in the activity of the sodium pump (Connor et al. 1974). Subsequently Tomita and his colleagues suggested an involvement of glycolytic pathways, again assuming that pacemaking activity originated in smooth muscle cells (see Nakayama et al. 1997). Although ICC were first described in the intestine a century ago by Cajal, they were long viewed as an oddity. Their role in the generation of pacemaker activity in the gastrointestinal tract was suggested on the basis of histological studies (Fausonne-Pelligrini et al. 1977; Thuneberg, 1982). More recently, studies on mutants that lack subpopulations of ICC revealed their role in the generation of rhythmicity (Fig. 1A; Ward et al. 1994; Huizinga et al. 1995). Critically, whereas isolated smooth muscle cells rarely generate spontaneous electrical activity (Farrugia, 1999), isolated ICC invariably do (Langton et al. 1989; Tokutomi et al. 1995). Even more recently the key role played by ICC in communication between autonomic and enteric nerve terminals and many gut smooth muscle cells has become apparent (Burns et al. 1996; Ward et al. 2000a).

Figure 1. Pacemaker role of ICC_MY in small intestine and gastric antrum.

A shows intracellular recordings of electrical activity in the circular muscle of the small intestine of mice; both recordings are in the presence of nifedipine. Aa shows the regular discharge of slow waves recorded from a control tissue. Ab shows that slow waves are absent in the small intestine of W/W^V mice: these tissues contain ICC_DMP and normal smooth muscle cells but lack ICC_MY. (Modified from Ward et al. 1994.) B shows simultaneous intracellular recordings from ICC_MY and the circular muscle layer of guinea-pig gastric antrum. The onset of pacemaker depolarisation is seen to precede the onset of depolarisation in the circular muscle layer. Each pacemaker potential consist of two components, an initial component, followed by a plateau component. Furthermore each slow wave consists of two components, an initial passive component, whose time course is shown by a dotted line, and a secondary regenerative component, superimposed. (Modified from Hirst et al. 2002c.)

In the gastrointestinal tract at least three separate functional groups of ICC exist. In most regions of the gastrointestinal tract, a network of ICC lies in the myenteric region (ICC_MY) between the circular and longitudinal muscle layers; ICC_MY function as pacemaker cells and trigger slow waves. A second population of ICC (intramuscular ICC or ICC_IM) are scattered amongst the smooth muscle cells. In the small intestine, ICC_IM, rather than being widely distributed, are concentrated at the inner surface of the circular layer at the region of the deep muscular plexus, ICC_DMP. In the antrum, ICC_IM contribute to slow wave activity (Dickens et al. 2001) and throughout the stomach their principal role relates to the transfer of neuronal information to smooth muscle cells (Burns et al. 1996; Ward et al. 2000a). In larger animals a further population of ICC (ICC_SEP) is distributed over the surface of muscle bundles and is said to have a septal location. ICC_SEP serve, much like Pukinje fibres in the heart, to convey pacemaker activity deep into the muscle bundles making up the wall of the antrum (Horiguchi et al. 2001).

Experimental approaches used to study ICC

The electrical properties of ICC have been studied using several different techniques. Isolated ICC have been examined using conventional patch-clamp recording techniques. This approach, which has been applied to ICC_MY, allows a description of the specific populations of ion channels present in their membranes (Langton et al. 1989; Lee & Sanders, 1993; Tokutomi et al. 1995) and an analysis of the cellular mechanisms which regulate the channels (Ward et al. 2000b; Koh et al. 2002). ICC are electrically coupled to neighbouring smooth muscle cells: any electrical activity ICC generate will produce a membrane potential change in the smooth muscle layers. With hindsight, simple intracellular recordings from smooth muscle cells in isolated segments of gastrointestinal tissues and isolated segments of urethra, after blocking smooth muscle L-type Ca²⁺ channels, record primarily the activity of the ICC in the tissues. The properties of ICC_MY can be determined in situ, using sharp electrodes, allowing one to monitor the behaviour of populations of interconnected ICC_MY and to determine how pacemaker potentials generate signals in adjacent smooth muscle layers (Dickens et al. 1999; Hirst & Edwards, 2001). A third method, used to study the properties of ICC_IM, involves recording from small isolated segments of circular muscle; if dissected appropriately the preparations are isopotential and contain up to 2000 smooth muscle cells linked to up to 200 ICC_IM. The membrane potential of both smooth muscle cells and ICC_IM, can be varied over a limited range and the effects of nerve stimulation can be analysed (Suzuki & Hirst, 1999; Edwards et al. 1999; Hirst et al. 2002c; Suzuki et al. 2003). Finally, the use of mutant mice in which specific sets of ICC are either absent or dramatically reduced in numbers has allowed an evaluation of the physiological properties of tissues, with and without different sets of ICC. Taking W/W^V mice as an example, in the small intestine, ICC_MY are dramatically reduced in numbers whereas the distribution of ICC_DMP remains little changed. Tissues containing ICC_MY generate slow waves whereas those which lack ICC_MY fail to generate slow waves (Fig. 1A; Ward et al. 1994, 1995; Huizinga et al. 1995). Conversely in the stomach of W/W^V mice, ICC_MY are retained but ICC_IM are absent; in gastric tissues lacking ICC_IM, neurotransmission is dramatically impaired (Burns et al. 1996; Ward et al. 2000a; Beckett et al. 2002; Suzuki et al. 2003).

Patterns of rhythmical electrical activity generated by ICC in the gastrointestinal tract

Slow waves, recorded from many regions of the gastrointestinal tract, persist when L-type Ca²⁺ channels are blocked (Ozaki et al. 1991; Suzuki & Hirst, 1999). Since gastrointestinal smooth muscle cells largely rely on L-type Ca²⁺ channels to generate inward currents (Ward & Sanders, 1992; Farrugia, 1999), the contribution made by smooth muscle cells themselves to slow wave activity is likely to be small.

In the antrum slow waves occur at 3-4 waves per minute (Dickens et al. 1999). Antral ICC_MY, either in situ or isolated, generate large amplitude, long lasting pacemaker potentials (Dickens et al. 1999; Hirst et al. 2002a, Kim et al. 2002). Antral pacemaker potentials have two components, an initial rapid depolarising component which lasts for about 2 s, followed by a plateau component which lasts for about 8 s (Fig. 1B; Dickens et al. 1999; Hirst & Edwards, 2001; Hirst et al. 2002a; Kito et al. 2002a). Each pacemaker potential, generated by antral ICC_MY, gives rise to an attenuated wave of depolarisation in the adjacent circular and longitudinal muscle layers (Dickens et al. 1999; Hirst & Edwards 2001). As the onset of depolarisation occurs in ICC_MY before it occurs in either the circular or longitudinal muscle layers (Dickens et al. 1999; Hirst & Edwards, 2001), ICC_MY clearly initiate slow waves in the antrum (Fig.1B). In the circular muscle layer each passive wave of depolarisation, originating in ICC_MY, is augmented by a secondary regenerative component (Ohba et al. 1975; Dickens et al. 1999). In antral preparations, which contain ICC_MY but lack ICC_IM, the secondary component of the slow wave is absent (Fig. 2Ab; Dickens et al. 2001). Thus each antral slow wave reflects the sum of separate contributions from both ICC_MY and ICC_IM (Fig. 2Ab).

Figure 2. Generation of secondary regenerative component of antral slow waves by ICC_IM located in the circular smooth muscle layer.

A schematic diagram showing a transverse section of the antral wall with an outer longitudinal layer (open circles at top), separated from the circular muscle layer (ellipsoids at bottom) by ICC_MY. The circular layer consists of smooth muscle cells (dark shading) and ICC_IM (light shading). Aa and b show intracellular recordings of electrical activity in the circular layer of the gastric antrum, from control and W/W^V mice, respectively; both recordings are in the presence of nifedipine. Aa shows the regular discharge of slow waves recorded from a control tissue; each slow wave consists of a passive component, generated by ICC_MY, and a secondary regenerative component (see Fig. 1). Ab shows that slow waves in the antrum of W/W^V mice, which contain ICC_MY but lack ICC_IM, consist only of the passive component generated by ICC_MY. (Modified from Dickens et al. 2001.) B shows traces which compare the time course of a regenerative potential (Ba), initiated by injecting depolarising current into a bundle of circular muscle (Bc), with the time course of the associated change in [Ca²⁺]_i obtained by averaging 30 successive responses (Bb). Note that the onset of the increase in [Ca²⁺]_i lags the onset of applied depolarisation by some 900 ms but corresponds with the start of the regenerative response. (Modified from Hirst et al. 2002b.)

In the mouse antrum the contributions from ICC_MY and ICC_IM show regional variation. Near the greater curvature, the density of ICC_MY is very high and the depolarisation arising from ICC_MY is large. The density of ICC_MY falls as one moves from the greater to lesser curvature, and there is a corresponding fall in the depolarisation generated in the circular layer by ICC_MY. However the regenerative contribution made by ICC_IM is such that the amplitudes of slow waves remain unchanged over the region (Hirst et al. 2002a). Experimentally the regenerative component of the slow wave is initiated by depolarising single bundles of the circular muscle layer: these bundles contain ICC_IM, but not ICC_MY, which have been dissected away (Fig. 2B; Suzuki & Hirst, 1999). For convenience, these responses will be termed regenerative potentials. Regenerative potentials have a minimum latency of about 1 s after the onset of the applied depolarisation (Fig. 2Ba); each regenerative potential is associated with an increase in [Ca²⁺]_i, whilst the preceding conditioning depolarisation itself fails to produce a detectable change in [Ca²⁺]_i (Fig. 2Bb; Hirst et al. 2002b). Regenerative potentials are only initiated in bundles of circular muscle which contain ICC_IM; they cannot be initiated in bundles taken from tissues which lack ICC_IM (Suzuki et al. 2003).

Thus in summary, slow waves in the antrum of mice and guinea-pigs result from the combined activity of two separate populations of ICC. Pacemaker potentials are initiated in ICC_MY. The wave of depolarisation conducts passively to the ICC_IM and ICC_IM respond to the pacemaker depolarisation by generating the secondary component of the slow wave which conducts to nearby circular smooth muscle cells.

In the small intestine slow waves occur at 10 to 40 waves per minute depending on species and location (Fig. 1A). In the small intestine of W/W^V mice, where ICC_MY are greatly reduced in number, slow waves are not detected (Fig. 1A; Ward et al. 1994). These observations suggest that in the small intestine, pacemaker activity is initiated by ICC_MY and indicate that neither ICC_DMP, which are retained in the small intestines of W/W^V mice (Ward et al. 1994), nor smooth muscle cells are able to generate rhythmical activity. Furthermore if ICC_DMP augment the depolarisation in the circular layer, as ICC_IM do in the antrum, they must have different properties from those of antral ICC_IM with antral regenerative potentials having refractory periods of some 5–10 s (Suzuki & Hirst, 1999). Recordings have been made from clusters of murine intestinal ICC_MY, maintained in short term primary culture. Under these conditions intestinal ICC_MY generate pacemaker currents with single monophasic components which last for some 1–2 s (Koh et al. 1998) and resemble the initial component of the antral pacemaker potential. Their waveforms and amplitudes are similar to those of slow waves recorded from intact tissue (Koh et al. 1998). This implies that, in situ, intestinal ICC_MY generate sufficient current to passively depolarise the muscle layers without substantial change in waveform. These preliminary observations suggest that in the small intestine, unlike the antrum, slow waves may be generated by a single population of ICC, namely ICC_MY. Unfortunately this idea has not been tested directly in the small intestine since recordings have not been made from intestinal ICC_MYin situ. Indeed when changes in the internal Ca²⁺ levels ([Ca²⁺]_i) have been monitored in murine ICC_MY their time courses are often poorly correlated with those occurring in the longitudinal muscle layer (Yamazawa & Iino, 2002).

Cellular basis of electrical activity generated by ICC

At the outset, it should be stated that the cellular basis for the generation of electrical activity in ICC is controversial. This disagreement probably arises because ICC in different regions of the gut utilise different ion channels to generate electrical activity. In the mouse small intestine, pacemaker potentials, generated by cultured ICC_MY, are monophasic (Koh et al. 1998). Intestinal pacemaker potentials are blocked by inhibitors of the mitochondrial Ca²⁺ transporter, by agents which block Ca²⁺ release from IP₃-dependent Ca²⁺ stores (Ward et al. 2000b) and are partly blocked by low concentrations of agents which block Cl⁻ channels (Koh et al. 1998). Intestinal ICC_MY contain sets of cation-selective channels which are activated when [Ca²⁺]_i falls to very low levels. These channels are blocked at normal [Ca²⁺]_i levels and by Cl⁻ channel antagonists, indicating that Cl⁻ channel blockers cannot be used to distinguish between Ca²⁺-inhibited cation-selective channels and Cl⁻ channels (Koh et al. 2002). It has been suggested that during each intestinal pacemaker potential, Ca²⁺ is released from an IP₃-dependent store located at some distance from the ICC membrane. The local increase in [Ca²⁺]_i near the inner mitochondrial membrane is thought to activate a generalised uptake of Ca²⁺ by the mitochondria so lowering the local concentration of Ca²⁺ near the ICC membrane. This local decrease in [Ca²⁺]_i then activates cation-selective channels and a pacemaker potential ensues. As Ca²⁺ enters through the cation-selective channels, the local concentration of Ca²⁺ rises and the channels are inactivated, so terminating the pacemaker potential (Ward et al. 2000b).

Antral ICC_MY and ICC_IM generate spontaneous depolarisations in the absence of stimulation (Edwards et al. 1999; Hirst & Edwards 2001). In bundles of circular muscle containing ICC_IM, the discharge of transient depolarisations is reflected as a ‘noisy’ resting membrane potential (Fig. 3Ba); when [Ca²⁺]_i is buffered to low levels, individual depolarising potentials become readily detectable (Fig. 3Bb). These transient depolarising potentials have been termed unitary potentials (Edwards et al. 1999) or spontaneous transient depolarisations (van Helden et al. 2000) and are the membrane potential analogues in the antrum of the spontaneous transient inward currents detected in isolated urethral ICC (Sergeant et al. 2001b). For clarity the term unitary potential will be used to describe the subthreshold monophasic depolarisations generated spontaneously by antral ICC. It will be argued that much of the electrical activity generated by antral ICC_MY or ICC_IM results from changes in the frequency of occurrence of unitary potentials. Antral pacemaker potentials consist of two components, an initial component, which resembles the pacemaker potential generated by intestinal ICC_MY, and a long lasting plateau component (Fig. 1A; Hirst & Edwards, 2001; Kim et al. 2002). The initial component of the pacemaker potential triggers the plateau component; the latter consists of a high frequency discharge of unitary potentials (Hirst & Edwards, 2001). Immediately after each pacemaker potential, the spontaneous discharge of unitary potentials falls to a very low rate and then gradually increases until it triggers the next pacemaker potential. The rate of generation of pacemaker potentials is slowed by chloride channel antagonists' at the same time the plateau component is abolished (Kito et al. 2002a). Similarly a decrease in [Cl⁻]_o transiently increases both the rate of generation of pacemaker potentials and the amplitude of the plateau component (Hirst & Edwards, 2001). These observations suggest that unitary potentials, which both trigger the initial component of the pacemaker potential and give rise to the plateau component, result from the activation of Cl⁻ channels (see Large & Wang, 1996). The membrane events underlying the initial component of antral pacemaker potentials are not understood. Antral ICC_MY can be paced electrically and the rate of rise of the initial component is sensitive to small changes in [Ca²⁺]_o (Hirst & Edwards, 2001; Hirst et al. 2002c). These observations suggest that the initial component is generated by voltage-dependent Ca²⁺ channels. Presumably they normally are activated when the discharge of unitary potentials causes the membrane potential ICC_MY to reach threshold for the channels (Hirst & Edwards, 2001). The scheme does not explain the role of mitochondria, which clearly play a role in the maintenance of pacemaker activity in the antrum, with antral slow waves being abolished by inhibitors of mitochondrial Ca²⁺ uptake (Ward et al. 2000b).

Figure 3. Generation of electrical activity by antral ICC depends upon IP₃ receptors and discharge of unitary potentials.

Aa and b show intracellular recordings of electrical activity in the circular layer of the gastric antrum, from control and mutant mice which lack type-1 IP₃ receptors; both recordings were made in the presence of nifedipine. Aa shows the regular discharge of slow waves recorded from a control tissue. In contrast when recordings were made from tissues lacking type-1 IP₃ receptors, slow waves were not detected (Ab). (Modified from Suzuki et al. 2001.) B shows membrane potential recordings from an isolated bundle of antral circular muscle from the guinea-pig. In control solution the resting potential displays an ongoing discharge of membrane noise (Ba); after buffering [Ca²⁺]_i to low levels, individual unitary potentials (•) are detected (Bb). Ca shows a regenerative potential initiated in control solution. After buffering [Ca²⁺]_i to low levels, depolarising currents (Cc) cause an increase in the frequency of occurrence of unitary potentials (•; Cb). (Modified from Edwards et al. 1999.)

Bundles of antral circular muscle, separated from ICC_MY, generate regenerative responses spontaneously (Suzuki & Hirst, 1999; van Helden et al. 2000; Fukuta et al. 2002). The pattern is irregular and usually occurs at a lower frequency than that of ICC_MY, indicating that ICC_IM do not normally generate pacemaker activity in the intact tissue but are paced by ICC_MY. A similar pattern of rhythmical activity is detected in the urethra (Hashitani & Edwards, 1999). It is not known whether the urethra has a dense anastomosing network of ICC, analogous to the ICC_MY network found in the gastrointestinal tract. However isolated urethral ICC, but not smooth muscle cells, spontaneously generate unitary inward currents (Sergeant et al. 2000, 2001b). In the intact urethra each unitary potential increases the probability that another potential will occur. Consequently several depolarising potentials tend to occur in bursts and initiate a slow wave. Unlike the slow waves in most regions of the gastrointestinal tract, the slow wave in the urethra depends heavily upon the activation of L-type Ca²⁺ channels. Thus nifedipine abolishes the plateau component of the urethral slow wave and only the spontaneous potentials generated by ICC remain (Hashitani & Edwards, 1999).

In the circular muscle layer of the antrum, depolarisation transiently but dramatically increases the frequency of unitary potentials to give rise to a regenerative response (Fig. 3Ca). When [Ca²⁺]_i is buffered to low levels so that the background frequency of unitary potentials is reduced, then the increase in frequency of unitary potentials produced by periods of depolarisation can be readily detected (Fig. 3Cb). The resting discharge of unitary potentials and regenerative responses are abolished by agents which interfere with mitochondrial Ca²⁺ uptake (Kito et al. 2002b), by 2-aminoethoxydiphenyl borate (2-APB) and by Cl⁻ channel blockers, and by appropriate anion substitution (Hirst et al. 2002b), suggesting that Cl⁻ channels are activated by Ca²⁺ released from IP₃-dependent stores. In the urethra, unitary currents with pharmacological properties similar to unitary potentials generated by antral ICC_IM, reverse at E_Cl and are similarly blocked by agents which block Ca²⁺-activated Cl⁻ channels (Sergeant et al. 2001a,b). Together these observations suggest that ICC_IM which can generate rhythmical activity rely upon Ca²⁺-activated Cl⁻ channels to do so. It is to be stressed however that not all ICC_IM generate rhythmical activity. In the fundus, ICC_IM give rise to an ongoing discharge of unitary potentials (Beckett et al. 2002) but these unitary potentials show no tendency to co-ordinate their activity: moreover in the fundus the resting discharge of unitary potentials is not blocked by Cl⁻ channel blockers (E. A. H. Beckett, S. M. Ward & G. D. S. Hirst, unpublished observations).

From the studies on different tissues, several common principles arise. Firstly rhythmical electrical activity generated by ICC, unlike the heart, does not depend entirely upon the activation of voltage-dependent ion channels. Rather the generation of electrical activity depends upon release into the cytoplasm of Ca²⁺ from IP₃-dependent internal Ca²⁺ stores and the subsequent activation of ion-selective channels in the membranes of ICC. Thus the antrum is electrically silent in mutant mice lacking type-1 IP₃ receptors (Fig. 3A; Suzuki et al. 2000), indicating that neither ICC_MY nor ICC_IM are functioning. Similarly slow waves are abolished by 2-APB (Hirst & Edwards, 2001), an inhibitor of IP₃-induced Ca²⁺ release (Maruyama et al. 1997); by xestospongin, a membrane permeable inhibitor of IP₃-dependent stores; and, in cultured ICC_MY networks, by heparin placed in the patch pipette (Ward et al. 2000b). On the other hand, Ca²⁺ release into the cytoplasm from stores linked to ryanodine receptors appears to make little or no contribution to slow wave activity (Chowdhury et al. 1995; Ward et al. 2000b). Secondly cytosolic Ca²⁺ concentrations vary in a pulsatile manner and each local increase in [Ca²⁺]_i is thought to either directly (Hirst & Edwards, 2001; Kito et al. 2002a) or indirectly (Koh et al. 2002) trigger the opening of sets of membrane channels, so giving rise to a unitary potential. Thirdly, mitochondria play a key part in regulating cytosolic [Ca²⁺]_i in ICC. Thus during pacemaker activity in cultured networks of intestinal ICC, periodic increases in the concentration of Ca²⁺ within mitochondria are detected. Furthermore slow waves and regenerative potentials are abolished in several tissues by agents which impair Ca²⁺ handling by mitochondria (Ward et al. 2000b).

How individual ICC co-ordinate their activity with their neighbours is not understood. In the gastrointestinal tract, ICC_MY form an interconnected network of cells, with numerous gap-junctions between neighbouring ICC_MY (Sanders, 1996) that allow injected dyes to pass between cells (Dickens et al. 1999). It has been suggested that in antral ICC_MY the initial voltage-dependent component co-ordinates activity between cells (Hirst & Edwards, 2001). In the urethra, although each unitary potential increases the probability that another will occur, the rate of discharge of unitary potentials in isolated cells appears to be independent of membrane potential (Sergeant et al. 2001b). The spread of pacemaker potentials in different regions of the GI tract occurs in an anisotropic manner, conducting more rapidly circumferentially than longitudinally. However the distribution of Kit immunoreactivity does not reveal a polarisation in ICC pacemaker networks and the mechanism underlying the behaviour is not understood.

Involvement of ICC in neuroeffector transmission

Most muscular organs containing smooth muscle cells are innervated by post-ganglionic axons which arise from either sympathetic, parasympathetic or enteric ganglion cells. In the gastrointestinal tract, muscle layers are often innervated by both excitatory and inhibitory projections. The traditional view has been that sympathetic, parasympathetic and enteric nerve varicosities release transmitter and the transmitter diffuses through the extracellular fluid directly activating any receptors present on the surface of the smooth muscle cells. However when this hypothesis has been tested directly it has only rarely been found to be correct. In many tissues, autonomic varicosities form organised neuroeffector junctions with nearby smooth muscle cells and neurally released transmitters appear to activate specific sets of postsynaptic receptors located close to the sites of transmitter release (Hirst et al. 1996). This also appears to be the case in the longitudinal muscle layer of the small intestine where neurally released and applied acetylcholine (ACh) activate different sets of muscarinic receptors on the smooth muscle cells (Cousins et al. 1995; Klemm, 1995). However several lines of evidence now suggest that, in several regions of the gastrointestinal tract, the specific sets of receptors activated by neurally released transmitter are not on smooth muscle cells but are located on ICC_IM, in the stomach, and on ICC_DMP, in the small intestine.

The idea that ICC_IM were intermediaries in neural control pathways initially arose from studies which examined the relationship between enteric nerve terminals and ICC_IM (Daniel, 1985). Their key role was demonstrated when neuroffector transmission in the stomach of W/W^V mice was analysed. In the fundus of control mice, simultaneous stimulation of excitatory and inhibitory nerves evoked a complex response consisting of cholinergic excitatory and nitrergic inhibitory components. Surprisingly in the fundus of W/W^V mice, where ICC_IM were absent, both the excitatory and inhibitory components were greatly attenuated (Burns et al. 1996; Ward et al. 2000a) even though enteric varicosities were present and capable of releasing transmitters (Ward et al. 2000a). More recently a similar finding was made using tissues isolated from Steel mutant mice (Sl/Sl^d mice) where ICC_IM are also absent from the stomach (Beckett et al. 2002). In the fundus of control mice, nerve stimulation evoked a biphasic response consisting of an excitatory junction potential (EJP), followed by an inhibitory junction potential (IJP; Fig. 4Aa). The IJP was abolished by N^ω-nitro-L-arginine (L-NA), an inhibitor of NO production, suggesting that fundal IJPs largely result from the release of NO. In the presence of L-NA, nerve stimulation evoked an EJP (Fig. 4Ab) whose amplitude was only slightly increased by inhibiting cholinersterases with neostigmine (Fig. 4Ac). However in neostigmine, the cholinergic EJP was followed by a slower depolarisation (Fig. 4Ac). Both the slow response and the EJP were abolished by atropine (Fig. 4Ad). As with other tissues where ICC_IM are present, the resting membrane potential displays an ongoing discharge of membrane noise, resulting from the random occurrence of unitary potentials (Fig. 4A). In contrast when similar recordings were made from the fundus of Sl/Sl^d mutants, nerve stimulation failed to evoke either an EJP or an IJP (Fig. 4Ba and b); at the same time membrane noise was absent (Fig. 4B). Clearly the cholinergic nerve terminals were functional, since in the presence of neostigmine an atropine-sensitive slow depolarisation was detected (Fig. 4Bc and d). The simplest explanation for these findings is that both inhibitory and excitatory nerve terminals selectively target ICC_IM. When these cells are absent insufficient transmitter is released to activate ‘extra-junctional' receptors located on nearby smooth muscle cells. However, when the hydrolysis of ACh was inhibited then sufficient ACh diffused to the smooth cells where it activated muscarinic receptors, presumably with slower activation kinetics (see also Cousins et al. 1995). Qualitatively similar observations have been made on the circular layer of the mutant mouse antrum. In the control antrum, nerve stimulation evokes a complex IJP, consisting of an initial apamin-sensitive IJP and a slower nitrergic IJP; the inhibitory responses are followed by an atropine-sensitive excitatory response. In the antrum of W/W^V mice, which lack ICC_IM, although the initial apamin-sensitive component remained, both the nitrergic and cholinergic responses were absent (Suzuki et al. 2003).

Figure 4. Involvement of ICC_IM in responses to inhibitory and excitatory nerve stimulation in the stomach.

Nerve stimulation evoked a biphasic response consisting of an EJP, followed by an IJP in the fundus of control mice (Aa). After abolishing the IJP with N^ω-nitro-L-arginine (L-NA), nerve stimulation evoked an EJP (4Ab) whose amplitude was little changed by inhibiting cholinesterases with neostigmine (Ac); in neostigmine, the cholinergic EJP was followed by a slower depolarisation (Ac). Both excitatory responses were abolished by atropine (Ad). When similar recordings were made from Sl/Sl^d mutants, which lack ICC_IM, nerve stimulation failed to evoke either an EJP or an IJP (Ba and b) but in the presence of neostigmine an atropine-sensitive slow depolarisation was detected (Bc and d; modified from Beckett et al. 2003). In an isolated bundle of circular muscle, dissected from guinea-pig antrum, after abolishing the effects of inhibitory nerve stimulation with apamin and L-NA, excitatory nerve stimulation evoked a regenerative potential (Ca). The regenerative potential, but not the response of smooth muscle cells to neurally released ACh, was abolished by hyperpolarisation (Ca). The excitatory response involving ICC_IM was abolished by anthracene-9-carboxylic acid (9-AC) but the response of smooth muscle cells to neurally released ACh persisted (Cb). (Modified from Hirst et al. 2002c.)

The ideas derived from studies on murine tissues deficient in ICC_IM are supported by analyses of neuroeffector transmission in guinea-pig antrum. In the circular layer of the gastric antrum, after the effects of inhibitory transmitters had been abolished pharmacologically, cholinergic nerve stimulation initiated regenerative responses identical to those initiated when ICC_IM are depolarised (Fig. 4C; Hirst et al. 2002c). The responses, both to neurally released ACh and to direct depolarisation of ICC_IM, depend upon the release of Ca²⁺ from IP₃-dependent stores and the activation of Cl⁻ channels (Fig. 4Cb). A small proportion of neurally released ACh appears to access the smooth muscle layer directly; there it triggers a rapid depolarisation that involves neither Cl⁻ channels nor Ca²⁺ release from IP₃-dependent stores (Fig. 4Cb). Furthermore the neural pathway which involved ICC_IM included a voltage-dependent step, such that a moderate hyperpolarisation blocked activation of the pathway (Fig. 4Ca); in contrast the responses to neurally released ACh on smooth muscle were little affected by membrane hyperpolarisation (Fig. 4Ca). These experiments show that most functional excitatory neuronal connections in the antrum involve ICC_IM, with neurally released ACh increasing the rate of discharge of unitary potentials so that a regenerative potential is initiated. Moreover in intact tissues excitatory neural activity increases the frequency of discharge of regenerative potentials so that ICC_IM become the dominant pacemaker cells (Hirst et al. 2002c). Similarly neurally released nitric oxide selectively targets ICC_IM (Suzuki et al. 2003). An analysis of nitrergic IJPs in the antrum indicates that neurally released NO suppresses the resting discharge of unitary potentials (Suzuki et al. 2003). Since unitary potentials result from an increase in chloride conductance (g_Cl), nitrergic hyperpolarisation manifests itself as a suppression of g_Cl (Suzuki et al. 2003; see also Crist et al. 1991a,b; Zhang & Paterson, 2002). In the urethra, ICC_IM also appear to be intermediaries in the process of neuroeffector transmission. There the contractions evoked by noradrenergic nerve stimulation also involve Cl⁻ channels (Sergeant et al. 2002) with added noradrenaline activating a different pathway presumably located in urethral smooth muscle cells (G. Sergeant personal communication).

Summary

The role of ICC in the control of gastrointestinal motility has been demonstrated using a range of different approaches. The studies suggest that although smooth muscle cells are the final mediators of contraction, their contribution to rhythmical electrical activity and their direct involvement in neuroeffector transmission is minimal. Recordings from mutant mice, in which selected sets of ICC are missing, have shown that in the small intestine ICC_MY are essential for the generation of pacemaker activity. Similarly, recordings from the stomachs of such mice, where ICC_MY are present but ICC_IM are absent, have revealed roles for ICC_IM in both excitatory and inhibitory neuroeffector transmission. Direct recordings of antral ICC_MYin situ have shown that they generate pacemaker potentials which depolarise the adjacent muscle layers. Analyses of neuroeffector transmission in the stomach have shown that neurally released transmitters produce effects that can be attributed almost entirely to the transmitters acting solely on ICC_IM.