Neuroeffector apparatus in gastrointestinal smooth muscle organs


Corresponding author K. M. Sanders: University of Nevada, Physiology and Cell Biology, Department of Physiology and Cell Biology University of Nevada School of Medicine, Reno, NV 89557, USA.  Email:


Control of gastrointestinal (GI) movements by enteric motoneurons is critical for orderly processing of food, absorption of nutrients and elimination of wastes. Work over the past several years has suggested that motor neurotransmission is more complicated than simple release of transmitter from nerve terminals and binding of receptors on smooth muscle cells. In fact the ‘neuro-effector’ junction in the tunica muscularis might consist of synaptic-like connectivity with specialized cells, and contributions from multiple cell types in integrated post-junctional responses. Interstitial cells of Cajal (ICC) were proposed as potential mediators in motor neurotransmission based on reduced post-junctional responses observed in W mutants that have reduced populations of ICC. More recent studies on W mutants have contradicted the original findings, and suggested that ICC may not be significant players in motor neurotransmission. This review examines the evidence for and against the role of ICC in motor neurotransmission and outlines areas for additional investigation that would help further resolve this controversy.

[ Professor Sanders received his PhD from the Department of Physiology in the UCLA School of Medicine in 1976. He further trained as a post-doctoral fellow at UCLA and the Mayo Foundation from 1976 to 1979. He became an Associate Professor at the University of Nevada in 1982 and head of Physiology and Cell Biology department in 1988. He has studied excitability mechanisms in smooth muscles and neural regulation of smooth muscles throughout his career. In 2004 he was a Carnegie Centenary Professor in Scotland and recently served on the National Commission for Digestive Diseases at the National Institutes of Health in Washington, DC. Professor Ward received his DPhil from the Department of Biology, University of Ulster in 1987. He spent a short time as a post-doctoral fellow in the Wellcome Laboratory, University College London before taking up a fellowship in Physiology and Cell Biology at the University of Nevada School of Medicine, where he is currently a Professor of Physiology. He has studied the excitability and structure of visceral organs, particularly the gastrointestinal tract and oviducts. In 2000 he was awarded the Sixth Annual American Gastroenterological Association in Gastroenterology for ‘Basic and Clinical Research’. In 2006 he was awarded a University of Nevada Foundation Professor, and in 2007 the Eighth International Award for Studies on Neurogastroenterology and Motility.]


Ca2+-activated Cl channel


fluorescence-activated cell sorting


fibroblast-like cell




internal anal sphincter


interstitial cells of Cajal


ICC of the deep muscular plexus


intramuscular ICC


ICC within the plane of the myenteric plexus


inhibitory junction potential


lower oesophageal sphincter


non-selective cation channel


platelet-derived growth factor receptor α


NO-sensitive guanylyl cyclase


Interstitial cells of Cajal (ICC) are non-muscular cells of mesenchymal origin within the tunica muscularis of the gastrointestinal (GI) tract. Morphologists observed these cells for many decades and suggested they might have regulatory functions due to their anatomical locations, close proximity to neurons, organization into networks, and gap junction connectivity to smooth muscle cells (e.g. Cajal, 1911; Imaizumi & Hama, 1969; Faussone-Pellegrini, 1977; Thuneburg, 1982). Physiological studies during the past 20 years have provided evidence that ICC: (i) serve as pacemaker cells with unique ionic currents that generate spontaneous electrical activity known as electrical slow waves in GI muscles; (ii) provide a pathway for active slow wave propagation in GI organs, by forming electrically coupled networks and expressing voltage-dependent mechanisms capable of slow wave regeneration; (iii) express receptors, transduction mechanisms and ionic conductances allowing them to mediate post-junctional responses to enteric motor neurotransmission; (iv) regulate smooth muscle excitability by contributing to resting potential and affecting syncytial conductance; and (v) manifest stretch-receptor functions regulating excitability and regulating slow wave frequency.

The concept that ICC mediate a component of the post-junctional response to enteric motor neurotransmission grew from morphological observations showing close contacts between varicosities of enteric nerves and ICC. Imaizumi & Hama (1969), for example, suggested that “…the interstitial cell may play a role in transmitting stimuli received from the axon to surrounding smooth muscle cells by an electrotonic response.” Such observations prompted further investigation of ICC in several regions of the GI tract. A morphometric study of the lower oesophageal sphincter quantified close contacts between nerve varicosities and ICC vs. contacts between varicosities and smooth muscle cells. ICC and nerve terminals were frequently in close contact, but similar contacts between nerve terminals and smooth muscle cells were more rare (Daniel & Posey-Daniel, 1984). These authors suggested that if the ‘intercalation’ of ICC between nerve terminals and smooth muscle cells is a significant factor then “… attempts to define the characteristics of nonadrenergic, noncholinergic junction potentials in smooth muscle and compare them with effects of putative mediators may be misleading. If the smooth muscle junction potential is produced after electrotonic transmission from interstitial cells and its characteristics are determined by events in the interstitial cells, then there is no reason to expect coincidence of effects of nerve-released and exogenously added mediators. The first may act directly on interstitial cells and the second on smooth muscle cells as well as on interstitial cells, with the action on smooth muscle cells likely to be predominant.”

The hypothesis posed by Daniel & Posey-Daniel (1984) was tested using gastric fundus of W/WV mice in which intramuscular ICC (ICC-IM) are largely absent. Post-junctional electrical responses to both nitrergic (Burns et al. 1996) and cholinergic (Ward et al. 2000) nerve stimulation were greatly reduced in these muscles, suggesting that ICC have an important role in mediating enteric neurotransmission (Fig. 1). Similar findings were made on lower oesophageal and pyloric sphincter muscles (Ward et al. 1998) and small intestine (Ward et al. 2006). However, in the latter case more extreme measures were needed to remove ICC-DMP (specialized ICC-IM in the small intestine) because these cells are not disrupted in W/WV mice. In each of these experiments, post-junctional responses to transmural nerve stimulation decreased when ICC-IM were reduced. Another study, on mice (Sl/Sld) with mutations in Kit ligand (stem cell factor) and loss of ICC-IM in the gastric fundus (Beckett et al. 2002), confirmed the importance of ICC in post-junctional electrical and mechanical responses to cholinergic and nitrergic neural responses. Taken together, these studies suggested an important role for ICC in mediation of enteric motor neurotransmission.

Figure 1.

Comparison of responses in W/WV (A and B) and Ws/Ws (C and D) fundus muscles
A shows control responses of a wildtype circular fundus muscle strip to electrical field stimulation (EFS) (0.3 ms duration pulses): 1 pulse (at arrowheads, left traces all panels) and 20 pulses in 1 s (beginning at arrowheads, right traces in all panels). In the wildtype mouse, EFS elicited a small excitatory junction potential (EJP) followed by an inhibitory junction potential (IJP). 20 pulses increased these responses and a long-duration 2nd component is apparent. l-NNA (2nd line of traces in A enhanced the EJP and greatly reduced the IJP, including the long duration 2nd component. B, responses to single stimuli were minimal in W/WV muscles, and both the EJP and IJP to 20 pulses were greatly reduced. l-NNA had little effect on the single stimuli response but slightly increased the IJP and unmasked a more pronounced ‘rebound’ response to 20 stimuli in the W/WV muscle. C, nitrergic responses were not very pronounced in the rat. In wildtype animals, a small EJP and IJP were evoked by a single stimulus and a large amplitude hyperpolarization followed by a sustained 2nd component of hyperpolarization were evoked by 20 pulses. l-NNA had little effect on responses to single stimuli and reduced the duration of the response to 20 stimuli (see dotted lines in traces at right in C). D, little or no response was elicited by 1 stimulus in the Ws/Ws rat fundus. Small inhibitory junction potentials were elicited by 20 stimuli. Note the absence of the sustained 2nd component in the Ws/Ws muscle. l-NNA had very little effect, suggesting that nitrergic responses were reduced in these muscles.

Recently, several studies using W mutant mice and rats have reported that ICC are not necessary for nitrergic (e.g. Sivarao et al. 2001, 2008; Alberti et al. 2007; Huizinga et al. 2008; Zhang et al. 2009) and/or cholinergic (Zhang et al. 2009) responses. Some of these studies are difficult to interpret because portions of the GI tract with incomplete lesions in ICC populations were used (cf. Alberti et al. 2007; Zhang et al. 2010). In studies with incomplete loss in ICC, neural responses were reduced, altered or absent in some areas of muscle or in some tissues, but were described as normal in other areas or tissues. The role of ICC in neurotransmission is difficult to determine in tissues with incomplete lesions, because results are mixed and therefore equivocal. The gastric fundus of W/WV mice was chosen for the original studies of the role of ICC in neurotransmission because the lesion in ICC-IM was so extensive (Burns et al. 1996; Ward et al. 2000).

One study reported significant nitrergic responses in fundus of W mutants (Huizinga et al. 2008). It is not easy to understand differences in results from experiments using the same mutant models, but there are differences in techniques and methods of normalizing data. For example, basal contractile activity (e.g. tone or spontaneous contractions) is usually different in mutants with reduced ICC. Therefore, it is difficult to compare contractile responses in a quantitative manner between wildtype and mutant muscles. It should be noted that retention of neural responses when ICC are absent and no longer making close connections with nerve terminals is not an argument against the involvement of ICC in neurotransmission in wildtype animals. W mutants have congenital defects in important regulatory functions in the GI tract. Therefore, developmental compensations may enhance more direct neuromuscular communication in muscles of some animals with low numbers of ICC. Variability in the results of some studies (e.g. Zhang et al. 2010) might suggest that compensation could be inconsistent from animal to animal.

This review seeks to better understand the controversy surrounding the role of ICC in neurotransmission by summarizing and discussing the evidence for and against involvement of ICC in enteric motor neurotransmission. If we are ever to understand motor neurotransmission (a basic goal in neurogastroenterology for the past 60 years), it will be necessary to understand which cells are innervated by motoneurons (or at least which cells are exposed to enteric motor neurotransmitters at effective concentrations to elicit post-junctional responses) so we can determine the cell-specific pathways responsible for motor responses. Post-junctional responses are likely to be integrated by the smooth muscle–ICC–fibroblast-like cell syncytium in GI muscles. Delineation of these mechanisms and determining what happens to the neuroeffector apparatus in disease may provide better rationales for new therapeutics. It is clear from reviewing this subject that a genuine controversy exists regarding the relative roles of ICC and smooth muscle cells in receiving and transducing signals from enteric motoneurons, and little or nothing is known about the potential involvement of fibroblast-like cells, another population of cells coupled to smooth muscle cells by gap junctions (Horiguchi & Komuro, 2000). More rigorous studies will be required to settle the debate. It should also be noted that most experimental evidence comes from studies of rodents, and therefore it will also be necessary to extend our investigations to studies of human GI muscles.

Close contacts between enteric motor nerve varicosities and post-junctional cells

Morphologists noted very close contacts between ICC and the varicosities of enteric motoneurons (e.g. Imaizumi & Hama, 1969; Daniel & Posey-Daniel, 1984; Zhou & Komuro, 1992; e.g. Fig 2A). These junctions are common in several species, including humans, and pre- and post-junctional specializations at points of close contact and expression of synaptic proteins have also been described (Beckett et al. 2005; Horiguchi et al. 2003; Wang et al. 1999; Wang et al. 2003). While close contacts are common between varicosities and ICC, similar junctions between nerve varicosities and smooth muscle cells are more rare in many species (Daniel & Posey-Daniel, 1984; Horiguchi et al. 2003). Daniel & Posey-Daniel (1984), via morphometric analysis, reported that very close (<20 nm) junctions are more common with ICC. However, there are also reports of frequent close contacts between nerve varicosities and smooth muscle cells in the gastric antrum of the rat (Mitsui & Komuro, 2003). In the latter study, however, morphometric analysis was not performed, so the issue of close junctions with smooth muscle cells must be considered anecdotal. At present, we do not know whether there are species- or region-specific differences in close contacts between nerve terminals and post-junctional cells.

Figure 2.

Relationship of motor nerve varicosities to interstitial cells and smooth muscle cells
ICC of the circular muscle layer (ICC-IM) and fibroblast-like cells (PDGFRα+ cells) are commonly found closely associated with nerve bundles (NB), as in this section from rat stomach (A; scale bar is 0.5 μm; reproduced with kind permission from Springer Science+Business Media: Mitsui & Komuro, 2002, Cell & Tissue Res309, 219–227). Smooth muscle cells (SMC) surround these structures and form gap junctions with ICC-IM and PDGFRα+ cells (not shown in this image). PDGFRα+ cells have electron-lucent cytoplasm and dilated cisternae of rough endoplasmic reticulum. ICC-IM are frequently in very close contact with nerve varicosities containing many synaptic vesicles (*). Inset shows higher magnification of the varicosity (scale bar is 0.2 μm) denoted by the asterisk in A. The gap between membranes of the nerve varicosity and the ICC-IM (arrow) measures about 20 nm. Similar close associations between nerve varicosities and smooth muscle cells can also be found in the rat antrum (not shown in this image), but these appear to be more rare than junctions with ICC-IM in other species. B and C show double labelling of PDGFRα+ cells (green) and ICC (red) in murine stomach (images are reproduced with kind permission from Springer Science+Business Media: Iino et al. 2009a, Histochem Cell Biol131, 691–702). PDGFRα+-IM cells (green, arrowheads) of the gastric fundus are intermingled with ICC-IM (red). In the corpus multipolar PDGFRα+-MY cells (green; arrows) are closely associated with ICC-MY (red) in the plane of the myenteric plexus. These images indicate that ICC and PDGFRα+ cells have similar anatomical distributions in the tunica muscularis, but represent discrete populations of cells. Yellow pixels in these merged images are due to overlay of cells in stacks of optical sections not co-expression of antigens in single cells. Scale bar in B is 40 μm and applicable to C. D shows a cartoon of ICC-IM (red) and PDGFRα+-IM cells (green) within a muscle bundle. As demonstrated by ultrastructural and immunohistochemical studies, both of these cell types are found in close apposition to enteric motoneurons and make gap junctions with neighbouring smooth muscle cells. A section through the region denoted by the dotted line might generate an image similar to the electron micrograph in A. As discussed in the text, neuromuscular junctions in GI smooth muscles may reflect innervation of and post-junctional responses in all 3 classes of post-junctional cells. Transduction of neurotransmitter signals by ICC-IM and/or PDGFRα+-IM cells and activation of ionic conductances would be conducted electrotonically via gap junctions to surrounding smooth muscle cells and influence the excitability of the tunica muscularis and possibly the frequency of phasic activity (e.g. segmentation and/or peristalsis).

Another cell type, referred to as ‘fibroblast-like cells’ (FLCs; Ishikawa et al. 1997; Horiguchi & Komuro, 2000; Mitsui & Komuro, 2003), are also closely associated with nerve varicosities (Fig. 2A). An important report showed recently that FLCs are labelled robustly with antibodies for platelet-derived growth factor receptor α (PDGFRα; Iino et al. 2009a; Fig. 2B and C), and expression of this receptor may be a powerful new means of isolating and evaluating the function of FLC and the possible contribution of these cells in disease. Actually, as pointed out by Giorgio Gabella at a recent conference on ICC, FLC is a vague and poorly justified term for this class of interstitial cell in GI muscles. Thus, we propose that FLC in the tunica muscularis should be referred to as PDGFRα+ cells, just as ICC are frequently referred to as Kit+ cells. It should be emphasized that double labelling with PDGFRα and Kit antibodies shows these cells to be distinct populations, and the distribution of PDGFRα+ cells are unaffected in W mutants (Iino et al. 2009a). The fact that PDGFRα+ cells share similar anatomical distributions (e.g. within muscle bundles and between the circular and longitudinal muscle layers) suggests that terminology similar to that used for different classes of ICC (i.e. ICC-IM for intramuscular ICC and ICC-MY for ICC within the plane of the myenteric plexus) would be useful to distinguish PDGFRα+ cells (e.g. PDGFRα+-IM (Fig. 2B) and PDGFRα+-MY (Fig. 2C)).

A striking feature of PDGFRα+-IM and and PDGFRα+-MY is the expression of small-conductance Ca2+-activated K+ (SK3) channels (Klemm & Lang, 2002; Vanderwinden et al. 2002; Fujita et al. 2003; Iino et al. 2009a; Iino & Nojyo, 2009), a potential mediator of purinergic enteric inhibition. PDGFRα+ cells also form gap junctions with smooth muscle cells (Horiguchi & Komuro, 2000; Fujita et al. 2003). Thus, if PDGFRα+ cells express appropriate receptors, these cells might also be participants in motor neurotransmission (Fig. 2D).

The literature is replete with descriptions of cells associated with varicose terminals in the gut, but quantitative, morphometric analyses documenting the physical relationships between nerve terminals and post-junctional cells are needed, particularly in human muscles. With such studies, we could better understand the relative nature and extent of contacts between nerve terminals and the three major cell types that might participate in post-junctional responses to motor neurotransmission.

Neurotransmitter released into the tiny volumes formed by close contacts might reach very high concentrations. Concentration is a factor in receptor binding and in the metabolism and/or uptake of neurotransmitters to restore pre-stimulus conditions. At present, little is known about the release of neurotransmitter from enteric motoneurons and the profile of neurotransmitter concentrations within the interstitium. The debate about ‘synaptic’ transmission was fought long ago with regard to synapses in the brain and somatic neuromuscular junctions, but it has taken much longer to convince investigators about the role of synapse-like junctions in the autonomic and enteric nervous systems. Some have claimed that ‘volume transmission’ is prevalent in spite of morphological evidence documenting close, synaptic-like structures in visceral smooth muscles (Sarna, 2008). There are many important questions remaining with respect to transmitter release and the role of synaptic-like connectivity in these muscles. For example: (i) What are the spatial and temporal profiles of neurotransmitter release? (ii) Are sites of close contact the predominant sites of neurotransmission, or does transmitter escape these regions of close apposition and reach concentrations effective for eliciting post-junctional responses throughout the interstitium? (iii) Do the small volumes formed by regions of close apposition play a role in the kinetics of neurotransmitter metabolism or uptake? (iv) In many muscles innervated by autonomic nerves, release of transmitter from a given varicosity is highly intermittent, even though every impulse elicits Ca2+ transients in each varicosity along a nerve fibre. Quanta of transmitter are estimated to be released from a given varicosity with a probability of less than 0.25 (Lavidis & Bennett, 1993). Studies measuring post-junctional Ca2+ transients elicited by neurotransmitter release estimate even lower release probabilities of only about 0.02 (Brain et al. 2002). If the low coupling between varicosity Ca2+ transients and transmitter release is true for enteric motoneurons, then moment-to-moment transmitter profiles may be highly focalized, and profiles that might accomplish ‘volume-transmission’ or even widespread ‘pharmaco-mechanical’ coupling in GI muscles might be difficult to achieve. As Professor Daniels suggested (Daniel & Posey-Daniel, 1984), exogenous neurotransmitters may act via different mechanisms than neurotransmitters released from nerve varicosities. There are many instances of differences between neural responses and responses to exogenous neurotransmitters. Thus, one must consider the hypothesis that neurotransmission in the gut is accomplished via selective innervation of post-junctional cells or selective binding of receptors in specialized regions of post-junctional cells.

Are appropriate receptors, signalling pathways and effectors present in ICC, PDGFRα+ cells and smooth muscle cells?

Regardless of the concentration of neurotransmitter, responses of individual cells depend upon expression of appropriate receptors, signalling pathways and effectors. Because gap junctions connect ICC and PDGFRα+ cells to smooth muscle cells, the post-junctional response of GI muscles might be a summation of responses in any or all of these cells and might result from stimulation of electrical and/or biochemical pathways. For example, signals directed at enhancing Ca2+ entry into smooth muscle cells tend to be excitatory, but Ca2+ sensitization mechanisms may also contribute to integrated excitatory responses. Electrical and biochemical responses may require different modes of neurotransmission because activation of a biochemical apparatus distributed throughout smooth muscle cells (e.g. Rho kinase pathway regulation of myosin light chain phosphatase or Ca2+ sensitization) would be difficult to accomplish if specialized neuro-effector junctions dominate, while activation of ion channels near specialized points of innervation could affect global excitability and regulate Ca2+ entry throughout a syncytium of cells.

Many studies have investigated the direct effects of GI neurotransmitters on smooth muscle cells, and these cells clearly express receptors and show responsiveness to most putative neurotransmitters. Whether these cells generate responses consistent with the effects of neurotransmission (i.e. involving the same receptors, signalling pathways and effectors) is still under investigation. More discussion will be given later regarding the direct responsiveness of smooth muscle cells to nitric oxide (NO). ICC, isolated by fluorescence-activated cell sorting (FACS), also express receptors for all the major neurotransmitters (Chen et al. 2007). Few physiological studies have been performed on ICC-IM, so neurotransmitter responses in these cells are undetermined. Virtually nothing is known about the expression of neurotransmitter receptors and/or responsiveness of PDGFRα+ cells, so it is currently not known whether these cells respond to enteric motor neurotransmitters. There has been speculation that these cells might participate in post-junctional purinergic responses (Klemm & Lang, 2002), because they express apamin-sensitive, small-conductance Ca2+-activated K+ (SK3) channels. This is an important area for investigation because signalling pathways or ion channels activated in ICC and PDGFRα+ cells might be different than in smooth muscle cells, and this knowledge might provide a means of differentiating between responses caused by neurotransmitter directed toward specific cells and effectors vs. transmitter released ‘in volume’ and affecting all cells with receptors.

Receptors and signalling pathways in post-junctional cells activated by neurotransmitter released from motoneurons

Post-junctional responses are elicited in ICC in response to nerve stimulation in intact muscles; however, similar evidence for direct activation of responses in smooth muscle cells or PDGFRα+ cells is scant. Nitrergic responses in GI smooth muscles are largely mediated by cGMP-dependent mechanisms, as demonstrated by pharmacological blockers of guanylyl cyclase and by genetic deletion experiments (e.g. Ward et al. 1992; Franck et al. 1997; Pfeifer et al. 1998; Ny et al. 2000; Friebe & Koesling, 2009), but direct stimulation of cGMP production and activation of cGMP-dependent effectors in smooth muscle cells in response to NO released from nerves is unproven. In fact, having studied this mechanism for several years, we were struck by the relative difficulty in eliciting responses to NO and donors in GI smooth muscle cells. The receptor for NO in GI muscle tissues (NO-sensitive guanylyl cyclase; NSCG) consists of α and β subunits that form heterodimers required for catalytic generation of cGMP (Friebe & Koesling, 2003; Friebe & Koesling, 2009). Both α and β subunits are expressed robustly in ICC-IM, and these cells are closely associated with nerve processes expressing nNOS (Iino et al. 2008; Iino et al. 2009b). α and β subunits could not be resolved in smooth muscle cells by immunohistochemistry. Of course, these studies do not exclude expression of NO-sensitive guanylyl cyclase subunits in smooth muscle cells; low levels of enzyme may be unresolved by immuno-histochemistry but be sufficient to transduce NO signals. However, it seems likely that NO released from neurons would be more effective in cells expressing higher concentrations of its receptor.

Global deactivation of NSGC by gene deletion of β1 subunits caused severe GI motor dysfunction, including hypertrophy, pyloric stenosis, caecum enlargement, and greatly increased transit times (Friebe & Koesling, 2009). Animals with ubiquitous lesions in β1 subunits died at weaning, suggesting lethality was due to GI motor dysfunction. The severe motor problems in ubiquitous β1 subunit knock-out animals did not appear to be due to loss of β1 subunits (and therefore loss of NSGC function) in smooth muscle cells, because smooth muscle-specific knock-out of β1 subunits resulted in animals with normal contractile responses to electrical field stimulation and normal survival. Future experiments using cell-specific knockouts of NSGC in ICC and PDGFRα+ may help determine whether expression of NO receptors in these cells is obligatory for normal regulation of motility. Activation of enteric inhibitory neurons increases cGMP in ICC (Shuttleworth et al. 1993; Iino et al. 2009b), but evidence demonstrating that NO released from neurons also activates cGMP production in smooth muscle cells has not been presented.

Besides activation of K+ channels by NO (see Koh et al. 1995, 2001), some authors have suggested that Ca2+-activated Cl channels, which are active under basal conditions, can be suppressed as part of the post-junctional response to NO (Zhang & Paterson, 2002, 2009; Hirst et al. 2004). There are no clear examples of Ca2+-activated Cl channels in muscle cells of the tunica muscularis, and such a conductance should be readily apparent with standard voltage-clamp protocols: step depolarization (activating Ca2+ entry) followed by repolarization to negative holding potentials results in large inward tail currents with slow deactivation kinetics. Such a conductance, so striking in tracheal or portal vein smooth muscle cells (e.g. Saleh et al. 2007), has not been observed in studies of smooth muscle cells of the tunica muscularis of the GI tract. A small, inwardly rectifying current observed in opossum oesophagus smooth muscle cells was described as a Cl conductance, and there was speculation that it might have sensitivity to Ca2+ (Zhang et al. 1998), but further, more definitive experiments to characterize this conductance, its regulation by Ca2+ and its suppression by NO have not appeared. Recent studies have demonstrated robust expression of Tmem16a, which encodes Ca2+-activated Cl channels, and a prominent Ca2+-activated Cl conductance in ICC (Gomez-Pinilla et al. 2009; Hwang et al. 2009; Zhu et al. 2009). Thus, if suppression of a Ca2+-activated Cl conductance is responsible for nitrergic responses, as claimed by Zhang & coworkers (2002, 2010), then these responses are likely to be mediated by ICC and not by smooth muscle cells.

Excitatory neurotransmitters activate post-junctional responses in ICC. Cholinergic nerve stimulation induced translocation of protein kinase Cɛ (PKCɛ) in murine small intestinal ICC of the deep muscular plexus (ICC-DMP). Translocation was blocked by atropine and tetrodotoxin, so it was initiated by nerves and post-junctional muscarinic receptors (Wang et al. 2003). Neurokinin released from nerves caused internalization of NK1 receptors (Iino et al. 2004). NK1 receptors are also expressed by smooth muscle cells in the murine small intestine, but an interesting feature of smooth muscle receptors is that pre-stimulation is necessary to ‘unlock’ their immunoreactivity (Portbury et al. 1996). These experiments suggested that smooth muscle receptors do not receive ongoing stimulation from neurokinins in vivo, because pre-stimulation (i.e. pre-exposure to a neurokinin) was required for NK1 receptor immunoreactivity. In our experiments, there was no evidence that neurokinins released from neurons reached smooth muscle cells because immunoreactivity to NK1 receptor antibody was not unlocked by nerve stimulation (Iino et al. 2004). We also found no evidence of receptor internalization in smooth muscle cells upon release of transmitter from neurons. These studies do not exclude the possibility of parallel excitatory neurotransmission to ICC-DMP and smooth muscle cells; different cells may utilize different receptors and signalling molecules. These findings make the point that ICC are innervated and transmitters reach high enough concentration to activate post-junctional signalling pathways in ICC. Equivalent evidence showing direct innervation of smooth muscle cells is not available.

Functional defects in neurotransmission from losing ICC

Many GI motility disorders, due to a variety of disease factors, have been associated with loss of ICC (see Burns, 2007; Farrugia, 2008; Ordog et al. 2009). If ICC are important intermediaries in motor neurotransmission, then loss of these cells could reduce communication between the enteric nervous system and the smooth muscle syncytium, resulting in reduced neural regulation of motility. Cause and effect is difficult to establish in human motor dysfunction because patients do not typically present until long after development of symptoms. Thus, the significance of ICC loss in motor pathologies has been studied with animal models with damaged ICC networks. It is unknown whether PDGFRα+ cells may also be affected in disease; however, if this occurs, then it may be possible to find animal models with defects in these cells as well. This line of investigation has developed novel hypotheses regarding the cause of several GI motor disorders, and further investigation may provide novel treatments.

Ramón y Cajal, noting the associations between nerves and ICC, suggested that ICC are involved in transduction of neural impulses from peripheral nerves to bowel smooth muscle cells (Cajal, 1911). Later, very close associations between enteric neurons and ICC were observed with electron microscopy, and junctions of 20 nm or less were found to be common in many species (discussed above). Thus, when we found that ICC-IM were severely depleted in gastric fundus of W/WV mice, these animals were an ideal model to test the classic hypothesis of Ramón y Cajal (Burns et al. 1996). If motor neurotransmission is reduced in tissues depleted of ICC, then this would indicate the importance of these cells in neurotransmission. However, other explanations for changes in neural responses are also possible, including: (i) concomitant defects in enteric neural development, (ii) reduced neurotransmitter release, and/or (iii) reduced responsiveness of cells in tissues lacking ICC. If no defect is seen in tissues with depleted ICC, i) then these cells may have no significance in neurotransmission or remodelling of junctions in mutants without ICC might obscure the normal physiological role of ICC. In the latter case, it is possible that neurotransmitter abundance may be affected by the presence of ICC, either due to changes in the amounts released or metabolism, (ii) released transmitter might become more available to additional targets when ICC are not interposed between varicosities and other post-junctional cells, or (iii) neuromuscular transmission may change and become reliant on other transmitters.

We found that post-junctional responses to both nitrergic and cholinergic nerves were reduced in gastric fundus and antrum, and lower oesophageal and pyloric sphincters of W/WV mice (Burns et al. 1996; Ward et al. 1998, 2000; Suzuki et al. 2003). Responses were reduced, but quantification showed that small responses were retained in W/WV muscles (Burns et al. 1996). The same question was not originally addressed by experiments on colonic or small intestinal muscles of W/WV mice because there is an incomplete lesion of ICC-IM (colon) (Fig. 3) and ICC-DMP (small intestine). Later, we found that ICC-DMP could also be reduced by incubation of muscles from birth with neutralizing Kit antibodies. With this technique, post-junctional cholinergic and nitrergic responses were also reduced in the small bowel (Ward et al. 2006). Cholinergic and nitrergic responses were also reduced in diabetic animals (Ordog et al. 2000). ICC-IM were not missing in these animals, but the close associations between nerve varicosities and ICC-IM were rare, supporting the idea that neuro-ICC connectivity is an important feature of motor neurotransmission.

Figure 3.

Incomplete lesions in ICC networks in colons of W/WV mice
A shows a digital reconstruction of a confocal Z-stack of ICC through the thickness of the muscularis in a wildtype mouse proximal colon. ICC form dense networks within the plane of the myenteric plexus region (ICC-MY; arrowheads) and within the circular and longitudinal muscle layers (ICC-IM; arrows). The density of ICC is greatly reduced and irregular in colons of W/WV mice, but loss of ICC is incomplete. B and C show ICC imaged at the same laser intensity, pixel time, detector gain and pinhole size as the image in A. The images in B and C are from a region where there were relatively few cells (B) and from another region where the ICC-MY network was considerably more intact (C). In some studies the extent of ICC lesions in W mutants may be overstated because the overall immunogenicity of ICC is reduced, most probably because of reduced Kit protein in these mutants. D demonstrates that increasing the detector gain from 643 to 750 to scan the same field as in C allows more complete resolution of ICC networks. Scale bar in B represents 100 μm and is the same in all panels.

Loss of ICC-IM also reduced post-junctional responses in Sl/Sld mice, in which Kit signalling was compromised by mutations in Kit ligand. These animals also had reduced post-junctional cholinergic and nitrergic electrical and mechanical responses (Beckett et al. 2002). It should be noted that purinergic inhibitory neurotransmission increased in W/WV muscles (Burns et al. 1996), and a ‘super-sensitivity’ phenomenon seemed to occur when ICC-IM were reduced because expression of P2Y receptors (post-junctional mediators of purinergic responses) increased in W/WV muscles (Sergeant et al. 2002).

Several control studies tested other interpretations of these findings. For example, it is possible that smooth muscle development might be negatively affected in W/WV mutants. This does not appear to occur, however: smooth muscle cells in W/WV mice displayed normal ultrastructure (Burns et al. 1996) and contractile responses to exogenous nitrergic and cholinergic agonists were intact (Burns et al. 1996; Ward et al. 2000; Beckett et al. 2002). There are important differences in the responses of normal and W/WV muscles, however, as discussed below in the section comparing responses to nerve-released and exogenous agonists. Another defect in W/WV muscles could be failure of enteric motoneurons to innervate the muscle layers, but the density of nerve processes and 3H–choline overflow in response to nerve stimulation was not changed significantly in W/WV muscles (Ward et al. 2000; Beckett et al. 2002).

In contrast to these findings, others have reported that responses to nitrergic and cholinergic transmitters released from enteric neurons are normal in W/WV mouse and Ws/Ws rat fundus and lower oesophageal sphincter (LES) (Huizinga et al. 2008; Zhang et al. 2010). In the first of these studies, contractile and electrical studies were performed and there were some clear and interesting differences between fundus muscles of wildtype and W mutants: (i) Ws/Ws rat fundus had significantly less basal tone than wildtype muscles. The possibility that this was due to reduced loss of cholinergic input was untested, but our data would predict such a defect in these animals. (ii) Nerve stimulation in the presence of atropine consistently relaxed wildtype muscles, but the response was variable (i.e. contraction or relaxation) in W mutants. One wonders if this is a result of differential compensation or remodelling of post-junctional cells due to congenital loss of ICC. (iii) Non-cholinergic excitation was revealed in wildtype muscles after blockade of nitrergic effects, but non-cholinergic excitation was already evident without block of inhibitory nerves in W/WV fundus. It is unclear from these studies whether this was due to reduced inhibitory input or up-regulation of the non-cholinergic excitatory pathway. The authors believed the latter, and this may suggest remodelling or compensation in muscles of W mutants. (iv) Wildtype fundus muscles displayed tone, but the W mutant muscles often displayed phasic contractions. This is possibly another example of remodelling in W mutant muscles, and it is not clear that inhibition of the emergent phasic contractions in response to nerve stimulation is due to the same mechanisms as inhibition of tone in wildtype animals. With these changes in the intrinsic behaviour of W mutant muscles, it seems that the conclusion stated in these studies, that nitrergic responses were normal in W mutant muscles, is speculative. A more conservative conclusion, when the reasons for the phenotypic changes apparent in the W mutants are not understood, may be to suggest that mechanical responses to nitrergic nerve stimulation were retained in W mutants; however, the many changes in neuromuscular behaviour make it difficult to determine whether the remaining responses were normal or a product of tissue remodelling. In spite of differences in interpretation, these studies demonstrate that nitrergic contractile responses survive to some extent in W mutants.

Another study investigated neuromuscular responses in the W/WV LES (Zhang et al. 2010). These authors found evidence for post-junctional nitrergic responses in some animals, but others lacked nitrergic responses. Many animals retained ICC-IM in some areas of the muscles, but the variability in responses did not appear to correlate with the presence of ICC-IM. Variability in responses in mutant animals of the same genotype suggests variable compensation in response to reduced ICC-IM. Although nitrergic inhibitory junction potentials (IJPs) were reported in some muscles, most of the study was performed on muscles (see Figs 3–6 of Zhang et al. 2010) with electrical responses as previously described in LES of W/WV mice (Ward et al. 1998) i.e. with greatly reduced nitrergic IJPs and loss of noisy fluctuations in membrane potential (unitary potentials) that are attributable to ICC-IM. Loss of unitary potentials in tissues without nitrergic responses is revealing because unitary potentials are generated in ICC-IM.

Figure 4.

Neural responses in a region of muscle with incomplete ICC lesion in W/WV mice
A and B, basal electrical activity is similar in wildtype (A) and W/WV (B) colonic muscles and consists of spike complexes with intermittent periods of quiescence. The frequency of spike complexes was greater in W/WV muscles (2.45 ± 0.26 cycles min−1 in control muscles (n= 11) vs. 4.9 ± 0.24 cycles min−1 for W/WV muscles (n= 12; P < 0.00001). We noted defects in post-junctional inhibitory responses to electrical field stimulation in many W/WV muscles (compare C and D). C is the wildtype control; D is the mutant muscle. C, in wildtype muscles EFS (stimulus applied at arrowhead; 1 pulse, 0.5 ms) evoked a single phase EJP (*) followed by an IJP that consisted of a fast and slow component (open circle). D, in W/WV muscles, EJPs were not evoked by EFS, and only the fast initial components of IJPs were resolved in 13 of 20 mice. E, the slow component is the nitrergic portion of the response and was blocked by l-NNA. The nitrergic component averaged 11 ± 0.7 mV (n= 18; P < 0.00001). l-NNA caused significant depolarization of wildtype muscles (see leftmost numbers on scale bars (F), but little depolarization was noted in W/WV muscles in response to l-NNA (numbers in brackets in F apply to the W/WV muscle after l-NNA). In the post-l-NNA W/WV muscle, a series of action potentials followed the IJP, possibly indicating phase advancement of the next action potential complex. G and H, in 7 of 20 W/WV animals, we observed small secondary components (filled circle) that were significantly smaller in amplitude than in wildtype colons and averaged 4 ± 0.6 mV (n= 7/20; P < 0.000001; 2 representative traces are shown in G and H).

Figure 5.

Reduced tonic inhibition in W mutant animals
Another approach to evaluating nitrergic regulation in W mutants is to characterize the magnitude of tonic inhibitory drive that is normally imposed upon colonic muscles. l-NNA treatment of wildtype muscles caused dramatic depolarization (18.6 ± 2.3 mV; n= 6) and nearly continuous firing of action potentials (A), whereas l-NNA caused only a small degree of depolarization (3.0 ± 0.75 mV; n= 9; P < 0.00001) in W/WV colonic muscles and did not disrupt the pattern of action potential complexes (B). These data demonstrate many of the points elaborated upon in the text. Post-junctional neural responses are abnormal in W/WV muscles, but a portion of nitrergic regulation is retained. There are reduced numbers and heterogeneity in the distribution of ICC in the colons of W/WV muscles, which may explain the reduced and heterogenous post-junctional responses. The nitrergic component of the tonic inhibition normally imposed upon the murine colon is significantly depressed. We are currently testing how this partial loss of nitrergic responses affects colonic transit.

Figure 6.

Comparison of electrical responses to exogenous transmitter and nerve evoked cholinergic excitation of circular muscles of the murine gastric antrum
A shows responses of wildtype, strain-matched control (top trace) and W/WV (bottom trace) antral muscles to exogenous carbachol (CCh, 100 nm; arrowheads). Both traces show spontaneous slow wave activity characteristic of antral muscles. CCh elicited depolarization and increased the frequency of electrical slow waves (chronotropic effect) in wildtype muscles. CCh caused depolarization in W/WV muscles, but no chronotropic effect was elicited in the absence of ICC-IM (Forrest et al. 2006). Both responses were blocked by atropine (1 μm; arrows). B, the responses to exogenous CCh are compared to responses to release of acetylcholine from nerves during EFS. EFS (5 Hz) was sustained through each trace in B. EFS produced positive chronotropic effects in wildtype muscles (2nd trace), but no response in W/WV muscles (not shown here but shown in Forrest et al. 2006, from which traces were redrawn). Chronotropic responses were blocked by 4-DAMP (3rd trace), an M3-specific antagonist, and no further effect was elicited by atropine (4th trace). Release of ACh from nerves did not depolarize cells as observed with the exogenous muscarinic agonist (A). EFS responses were due primarily to stimulation of M3 receptors, blocked by 4-DAMP (3rd trace) and not further affected by atropine (4th trace). C shows the direct response of CCh on a smooth muscle cell isolated from the murine antrum. As shown in many studies in the literature (e.g. Benham et al. 1985; Inoue & Isenberg, 1990), direct application of a muscarinic agonist to GI smooth muscles causes significant depolarization due to activation of non-selective cation channels encoded by Trpc4 and Trpc6 (Tsvilovskyy et al. 2009). Conclusions are summarized at the right of the traces. When ICC are present, nerve-evoked effects are mainly chronotropic. Muscarinic stimulation of smooth muscle cells causes depolarization, which is manifest when exogenous transmitter is applied to cells or tissues, but not elicited when transmitter is released from nerve terminals with constant EFS at 5 Hz. These data suggest that ACh released from neurons may be restricted to neuro-ICC junctions, most probably due to the action of acetylcholine esterases (see Ward et al. 2000).

Some authors have suggested that if nitrergic responses are reduced in W/WV muscles, then these animals should have similar dysfunction as nNOS knock-out mice (Sivarao et al. 2001). In fact, we would have predicted, as was observed by Sivarao & coworkers (2001, 2008), that nNOS−/− mice would have quite a different phenotype than W/WV mice, because defects in both nitrergic and cholinergic neurotransmission occur in the latter (Burns et al. 1996; Ward et al. 2000). The LES of nNOS−/− mice was hypertensive with compromised relaxation responses to swallowing, and the LES of W/WV mice was hypotensive with apparently normal relaxation responses to swallowing. Our data (Ward et al. 1998) suggest that sphincter hypotension in W/WV mice might result from loss of cholinergic tone. The conclusion that ‘normal’ relaxation occurred in the LES of W/Wv mice is problematic (Sivarao et al. 2001). Responses were recorded from different levels of tone, but data were normalized as per cent of relaxation. Different degrees of inhibition may be required to accomplish the same ‘per cent of relaxation’ from different states of contractile activation. Nevertheless, pressure changes were affected by an inhibitor of nNOS in W/WV mice, and if the responses the authors measured were not due to problems with maintenance of the position of manometry ports (e.g. with sphincter hypotension, it may be difficult to localize the point of the physiological sphincter or whether ports moved during contraction of oesophageal skeletal muscles), then the study demonstrates partial retention of nitrergic responses in W/WV animals.

The same group reported manometric studies of the pyloric sphincter that were similar to their results on LES (Sivarao et al. 2008): nNOS−/− mice had normotensive pyloric sphincters but lack nitrergic inhibition and the pylorus in W/WV mice is hypotensive with preservation of nitrergic regulation. Contractile responses were again quantified on the basis of per cent of basal tone, which was considerably different in wildtype and W/WV mice. As above, contractile responses beginning at different levels of basal activity are difficult to interpret. Inspection of the raw data chosen for the paper clearly shows less robust nitrergic responses in W/WV mice than in wildtype animals. Thus, while nitrergic responses were sustained in W/WV animals, it appears that resting tone and spontaneous activity, which might be cholinergic in nature, and responses to nitrergic nerve stimulation were attenuated in animals with reduced numbers of ICC-IM, as would be predicted from in vitro studies (Ward et al. 1998).

There may be species differences between rats and mice in terms of the role of ICC in neurotransmission. In rats, close contacts between varicosities and smooth muscle cells were readily identified (Mitsui & Komuro, 2003), whereas similar contacts are more rare in mice. Structural differences might affect the relative innervation of one type of cell vs. another. If species differences occur in the neuro-effector apparatus, then this underscores the need of more extensive studies of human muscles using morphometric techniques to quantify the nature of neuro-effector structures.

Studies of neural responses in colonic and internal anal sphincter (IAS) muscles of Ws/Ws rats and W/WV mice also seem contrary to the role of ICC in motor neurotransmission (Alberti et al. 2007; de Lorijn et al. 2005). Lesions in ICC networks in Kit mutants are heterogenous (Ward et al. 1994; Burns et al. 1996). Specific classes of ICC are largely missing in some regions of the gut, but in other regions, ICC networks are intact; however, the density of ICC is usually reduced. It is more difficult to resolve remaining ICC in muscles of W mutants with Kit immunohistochemistry because the absolute expression of Kit, and therefore immunoreactivity, is reduced in surviving ICC (Fig. 3). Thus, tissues selected for some studies of W mutants are not optimal for testing the role of ICC in neurotransmission. For example, substantial numbers of ICC were retained in internal anal sphincters and colons of Ws/Ws rats (Alberti et al. 2005; de Lorijn et al. 2005). Nitrergic responses (i.e. slow components of inhibitory junction potentials) were recorded in only about half of electrical impalements of Ws/Ws colons, but nitrergic responses were not resolved in the remainder of cells. Such observations might be expected with partial loss of ICC.

Why are results (and conclusions) of studies of W mutants so different in different laboratories? The most likely explanation for these differences is that experiments have been performed using different parameters. For example, single stimuli or brief trains of stimuli were used in most of our studies (Burns et al. 1996; Ward et al. 2000), but a contradictory study of the same anatomical region used far more intensive stimulation (e.g. 40 Hz stimulation) to elicit nitrergic responses. More intense stimulation would be expected to produce greater transmitter outflow and may elicit responses in a broader receptive field. If this is the main difference between studies, then it might be pertinent to consider whether losing motor responses to low-frequency activation of enteric neurons is compatible with normal GI motor control. Can normal neural regulation of motility be accomplished with part of the post-junctional response lost?

A common theme in several studies claiming that ICC are not important in enteric motor neurotransmission is variability in responses from tissue to tissue or cell to cell (e.g. Alberti et al. 2005; Zhang et al. 2009). As previously mentioned, differences in post-junctional neural responses between mutants sharing the same genotype raise the possibility of differential compensation. W mutants, experiencing congenital loss of an important pathway for neural regulation, may compensate for this loss by remodelling of the neuromuscular apparatus. One study also suggested that cells other than Kit+ cells, possibly immature ICC that do not fully develop into the adult ICC (i.e. Kit+) phenotype, might compensate for loss of ICC (Farre et al. 2007). Compensation might occur in a variable manner within a population of animals and might depend upon a number of factors, including strain, age, diet, housing, etc. Studies of the molecular phenotype of W mutants, displaying and not displaying cholinergic and nitrergic responses, are lacking, but the fact that purinergic responses (Sergeant et al. 2002) and non-cholinergic excitatory responses (Huizinga et al. 2008) are up-regulated in W mutants makes the issue of neuromuscular remodelling an intriguing hypothesis to investigate. Denervation supersensitivity is a well-known phenomenon in visceral smooth muscles (Auintas & Noel, 2009). The developing gut may compensate, to some extent, for loss of motor innervation in the absence of ICC.

An important question raised by experiments on tissues where the loss of ICC is incomplete is to what degree of ICC loss leads to abnormal neural regulation? In studies on Ws/Ws rat colon (Alberti et al. 2005) and W/WV LES (Zhang et al. 2010), post-junctional nitrergic electrical responses were observed during impalements of some cells, but not in a significant number of other impalements, and as argued above, this might be consistent with partial loss of ICC. We have reproduced these variable responses in intracellular studies of colonic muscles from W/WV mice, where ICC loss is incomplete (Figs 3 and 4). Nitrergic IJPs were recorded in 7 of 20 W/WV animals tested, but responses in these muscles were significantly smaller in amplitude and, therefore, abnormal (Fig. 4). Reduced and variable responses were also observed in studies of the W/WV LES (Zhang et al. 2010), but these authors concluded that Kit-positive ICC are not essential for neurotransmission in the mouse LES. Variability in responses of W/WV LES was attributed to abnormal Ca2+ handling by the sarcoplasmic reticulum in smooth muscle cells (Zhang et al. 2010), but this idea is speculative because no direct measurements of Ca2+ transients or Ca2+-dependent conductances in smooth muscle cells were provided.

‘Normal neural responses’, as elicited by electrical field stimulation, may be more difficult to evaluate than commonly assumed, because this method of stimulating GI muscles would be expected to activate several types of neurons within the field of stimulation simultaneously: e.g. excitatory and inhibitory motoneurons, interneurons and intrinsic primary sensory afferent neurons within ganglia. Thus, post-junctional responses to field stimulation are complex and cannot be assumed to be purely the products of direct activation of motoneurons. Specific components of responses to motoneurons, though easily dissected with pharmacological blockers (i.e. nitric oxide synthase (NOS) inhibitors, muscarinic antagonists, P2Y receptor antagonists, etc.), may be more difficult to resolve than assumed. Responses to field stimulation, however, are clearly not ‘normal’ from a physiological standpoint, because the spatial and temporal organization of excitatory and inhibitory motor pathways during physiological reflexes is not present during electrical field stimulation (Spencer et al. 1999). More focused evaluation of specific pathways might be accomplished by studying specific oral and anal responses to intrinsic reflexes, stimulation of vagal efferent fibres to the stomach where there is a dominance of inhibitory input, or to evaluate tonic inhibition of GI muscles (Wood, 1972) that is attributed to nitrergic input to the gut in many laboratory animals (Gustafsson & Delbro, 1993). An example of differences in tonic inhibitory input in wildtype and W/WV colonic muscles is shown in Fig. 5. The NOS inhibitor l-NNA caused a much larger depolarization and release of control on intrinsic excitability in wildtype muscles than in W/WV muscles suggesting reduced tonic nitrergic inhibition in colons of W/WV mice. Thus, the conclusion that nitrergic responses are ‘normal’ in W mutant animals may be too simplistic.

It is also possible that electrical behaviour (i.e. depolarizing excitatory junction potentials or hyperpolarizing inhibitory junction potentials) is only partially responsible for smooth muscle responses, and a significant part of post-junctional mechanical responses are mediated by non-electrical, pharmaco-mechanical coupling mechanisms (e.g. Ca2+ release from internal stores and changes in Ca2+ sensitivity; Somlyo & Somlyo, 2003). If this is true then records of electrical responses would not necessarily be a thorough assay of motor neurotransmission, and different cells might mediate different aspects of post-junctional responses. For example, parallel innervation might result in electrical responses in ICC that are primarily directed at regulation of membrane potential, electrical rhythmicity (i.e. frequency of phasic activity) and/or general excitability of the ICC/smooth muscle syncytium, and the smooth muscle response could be mainly focused at regulation of contractile force (e.g. generation or inhibition of tone or modulation of the amplitude of phasic contractions). Application of exogenous transmitter substances to intact or α-toxin permeabilized GI muscles clearly shows that modulation of Ca2+ sensitivity is capable of substantial regulation of contractile responses (Kitazawa et al. 1991; Ozaki et al. 1992; Khromov et al. 2006), but activation of this pathway in response to enteric motor neurotransmission has never been clearly documented. It is also possible that the electrophysiological responses of tissues are a consequence of integrated responses to neurotransmitters in ICC, smooth muscle cells and PDGFRα+ cells. Such hypotheses are largely untested at the present time.

Is transmitter release and/or diffusion limited by autoreceptors, uptake or metabolism?

Maintenance of contractile responses to neurotransmission in muscles lacking ICC does not obviate the role of ICC in enteric motor neurotransmission. The close association between nerve terminals and ICC, along with having consequences for initiating responses via activation of specific post-junctional mechanisms, may also have consequences for metabolism (i.e. deactivation) of neurotransmitter, transmitter uptake and effectiveness of pre-junctional stimulation of auto-receptors. If transmitter release is focused in active zones that are in close apposition with ICC, then concentrations will reach high levels in these spaces relative to the rest of the interstitium. Concentration can affect metabolic rates of transmitters, uptake and binding to auto-receptors. Removing ICC could facilitate transmitter release and broaden the spatial distribution of transmitter by reducing negative feedback from auto-receptors or reducing the rate at which transmitter is metabolized.

Some investigators have suggested that NO released from enteric inhibitory neurons would be impossible to confine to neuro-effector junctions because NO, generated in nerves, is freely diffusible (Sarna, 2008), and therefore not likely to be focused toward a specific aspect of a varicosity (transmitter release zones). It may even be questioned whether NO release is confined to varicosities because demonstrable expression of nNOS occurs all along processes of motoneurons. Synthesis of NO by nNOS is regulated by Ca2+, and therefore voltage-dependent Ca2+ entry mechanisms may spatially limit synthesis of NO to nerve varicosities, where specialized populations of voltage-dependent Ca2+ channels (e.g. CaV2.2) appear to be concentrated (Brain & Bennet, 1997). Little is known about the expression of CaV channel isoforms or spatial and temporal dynamics of Ca2+ transients in processes of enteric motoneurons, but if these events are analogous to autonomic neurons that innervate other visceral smooth muscles, then nerve stimulation initiates Ca2+ transients in both varicosities and inter-varicosity regions of axons, but Ca2+ transients are intensified in varicosities (see Brain & Bennet, 1997). nNOS also occurs in multiple states within neurons; not all are catalytically active. For example, a natural protein inhibitor of nNOS (PIN or LC8) (Jaffrey & Snyder, 1996) is associated with cytoplasmic nNOS, but not with membrane-bound nNOS in varicosities (Chaudhury et al. 2008). Thus, active nNOS may be targeted to specific active zones in nerve terminals (Zhai & Bellen, 2004). On the basis of these arguments, inter-varicosity nNOS may not be a major source of NO in response to nerve action potentials. In fact, we know little about the ‘release sites’ and spatial and temporal distribution of NO released by enteric inhibitory neurons. Thus, the question of whether release of NO can be limited spatially within GI muscles is unresolved. The suggestion that NO can diffuse freely through the smooth muscle syncytium at concentrations effective to yield post-junctional responses is speculative at the present time, but this is an interesting question that may be possible to answer with NO-sensitive dyes and imaging techniques or by measuring post-junctional responses with functional immunohistochemistry. In wildtype GI muscles, cGMP-like immunoreactivity increases in ICC-IM in response to nitrergic nerve stimulation (Shuttleworth et al. 1993; Iino et al. 2009b), but increased cGMP in response to NO released from nerves has not been resolved in smooth muscle cells.

A high density of mitochondria relative to surrounding cells is an ultrastructural feature of ICC. Superoxide is a by-product of mitochondrial ATP synthesis (Lambert & Brand, 2009), and reaction of superoxide with NO is a means of NO degradation. This metabolism might be reduced in the absence of ICC. Reduced rates of degradation may preserve NO released from nerves and increase its spatial distribution. Little is known about negative feedback regulating NO release by pre-junctional receptor-mediated mechanisms. NO concentration (or concentration of purine co-transmitters) may be a factor in regulating the amount of NO released from inhibitory neurons. As a means of solving the controversy about the role of ICC in inhibitory neurotransmission, it will be important to determine whether pre-junctional mechanisms (auto-receptors) regulate transmitter release and whether different pathways and effectors are activated in wildtype and mutant animals lacking ICC-IM or specific components of post-junctional transduction mechanisms. Cell-specific knock-outs of signalling molecules affected by nitrergic mechanisms will be important tools for this investigation (see Friebe & Koesling, 2009).

Differences in responses to exogenous transmitter substances and neurotransmitter released from neurons

There is no reason to assume a priori that responses to neurotransmitters released from neurons and exogenous transmitter substances are mediated by the same cells, receptors or post-junctional (transduction) signalling pathways. Neurotransmitters released from varicosities may be spatially limited to specific populations of receptors, whereas transmitters added to organ baths may bind to receptors on a variety of cells. Studies of intact GI muscles are complicated by the fact that the multiple cell types, including enteric neurons, express receptors for all known GI motor neurotransmitters. Release of transmitter from a nerve varicosity within a muscle bundle is unlikely to have effects on nerve cell bodies, but this is certainly a plausible contaminant of responses to exogenous transmitters. Likewise, pharmacological blockade of receptor populations are not limited to post-junctional receptors mediating neuromuscular transmission. Eliminating contamination from activation of neural pathways has not been routine in most studies of exogenous transmitters. Thus, it is possible that some of the nitrergic or cholinergic responses reported to be normal in W mutants could arise from ganglionic responses in muscle strips or intact animals. Another difficulty is that some transmitters bind to auto-receptors at nerve terminals and this contamination is very difficult to control. In fact, autoreceptor regulation at terminals of enteric motoneurons is poorly understood, but this is a common phenomenon at autonomic neuromuscular junctions (e.g. Bennett et al. 1998; Bennett, 2000).

There are many examples of differences in responses to neurotransmitters and exogenous transmitter substances, and these observations indicate that different sets of post-junctional receptors are bound by exogenous drugs and transmitters released from nerves. Different signalling pathways and cellular effectors may be activated by post-junctional mechanisms in smooth muscle cells, ICC and PDGFRα+ cells. As specific cellular mechanisms are discovered, it may be possible to dissect the differences in mechanisms activated by nerve-released transmitters and bath-applied transmitters, and this information will help explain why drugs designed to work via one type of neurotransmitter receptor have multiple side-effects.

An example chosen to illustrate differences in responses to a neurotransmitter released from motoneurons and an exogenous transmitter is cholinergic stimulation of murine gastric muscle (Fig. 6). It is well documented that cholinergic agonists bind M2 and M3 receptors coupled to increasing the open probability of non-selective cation channels (NSCCs) in GI smooth muscle cells (e.g. Benham et al. 1985; Inoue & Isenberg, 1990). Different channels, possibly Ca2+-activated Cl channels (CaCCs), may be targeted by muscarinic pathways in ICC. In the former case, cholinergic agonists activate inward currents, tending to cause tonic depolarization of the smooth muscle syncytium. In the latter case, Ca2+ release from stores via a spark- or puff-like mechanism is likely to activate Ca2+ transients in near-membrane compartments and transient activation of CaCC, producing spontaneous transient inward currents (STICs). Summation of these events in a syncytium of ICC and smooth muscle cells could result in tonic depolarization, but in a phasic smooth muscle, the result may be a positive chronotropic effect. Based on this logic, we compared the effects of sustained cholinergic nerve stimulation with bath-applied ACh, and there were important differences in the responses. In wildtype muscles, muscarinic stimulation caused significant tonic depolarization and increased slow wave frequency (Forrest et al. 2006), but in muscles lacking ICC-IM, ACh depolarized muscles without chronotropic effects. Tonic activation of cholinergic nerves had little effect on resting potential (i.e. no tonic depolarization), but slow wave frequency was enhanced. Previous studies demonstrated that phase-advancing effects (i.e. positive chronotropic effects) of cholinergic nerve stimulation are mediated by ICC-IM (Beckett et al. 2003). Loss of chronotropic responses in muscles lacking ICC-IM and manifestation of mainly chronotropic effects when ACh is released from motoneurons suggests that neurotransmitter binds primarily receptors of ICC-IM. Similar investigations using specific pharmacology and genetically modified mice with cell-specific receptors and molecular effectors knocked-out may eventually sort out which cells are activated by each transmitter pathway, and this level of scrutiny should be a goal for future studies of enteric motor neurotransmission. Imaging studies may also be able to directly demonstrate cellular innervation as localized Ca2+ transients may be resolvable in ICC and smooth muscle cells.

Issues for future investigation

Motor neurotransmission in GI muscles is fundamental to normal bowel function, and understanding this process and the mechanisms utilized is critical to understanding neurogastroenterology and GI motility disorders. While some appear intransigent about ‘volume transmission’ and the view that only smooth muscle cells must mediate post-junctional effects in the tunica muscularis (Sarna, 2008), there is considerable evidence that ICC, at a minimum, are also transducers and integrators of enteric motor neurotransmitter signals. It seems likely that the PDGFRα+ cells may also contribute to the integrated motor responses of smooth muscle tissues. Thus, neuromuscular regulation of GI muscles is likely to include multiple modes of neurotransmission (volume and synaptic) and several cell types may receive and transduce inputs from motoneurons. Based upon current knowledge, it is simplistic to measure electrical, mechanical or biochemical responses elicited by exogenous, ‘bath-applied’ transmitter substances and conclude that the pathways and mechanisms activated recapitulate responses elicited during motor neurotransmission. We know that exogenous compounds applied to solutions bathing muscles have complicated effects due to binding of multiple receptors expressed by many cell types present in muscle strips, and responses generated must be contaminated by many factors (e.g. stimulation of multiple cells, stimulation of intrinsic neural pathways, release of paracrine substances, etc.). Under these circumstances contractile responses are the product of multiple complex and interlinking signalling pathways, which may or may not be activated in response to neurotransmitters released from motoneurons. Substances released by neurons are likely to impinge upon a much more restricted group of receptors expressed by a more limited population of cells or even upon receptors in areas of post-junctional specialization in cells (e.g. such as might occur at neuro-effector junctions). It will be important for future studies to unravel the post-junctional mechanisms activated by endogenous neurotransmitters rather than continue to attempt to simulate these responses with bath application of exogenous substances.

Answers to several important questions would greatly advance our knowledge of motor neurotransmission in the gut. Among topics for investigation are: (i) What are the relative volume densities of specialized contacts between nerve terminals and post-junctional cells? Morphometric and 3-dimensional reconstruction of neuro-effector junctions, particularly in human muscles, are needed. (ii) What is the specific expression of post-junctional receptors, signalling and effector molecules in each of the three cell types clustered around motor nerve varicosities? (iii) What is the profile of motor neurotransmitters released from enteric motor neurons? What is the frequency of varicosities (density in a volume)? (iv) Do varicosities of enteric motoneurons release transmitter with each action potential, and what are typical frequencies of action potential firing in motoneurons during ‘physiological’ reflexes? (v) There has been criticism that the evidence for functional innervation (and a synapse-like) relationship between nerve terminals and ICC is circumstantial (Goyal & Chaudhury, 2010), but evidence for direct innervation of smooth muscle cells is unavailable. Do neurotransmitters released from motoneurons bind receptors of and elicit direct responses in smooth muscle cells in GI tissues? (vi) Functional remodelling of post-junctional cells in mutants lacking specific cell types may alter the nature of responses and obscure the importance of ICC in responses in wildtype animals. How are post-junctional receptors, signalling pathways and effectors affected in the GI muscles when ICC are absent? (vii) What are the relative roles of ICC, smooth muscle cells, and PDGFRα+ cells in metabolism of transmitter? Does removing the close contacts between nerve terminals and ICC affect the rate in which neurotransmitter can be metabolized? (viii) What is the role of PDGFRα+ cells in enteric motor neurotransmission and development and maintenance of neuro-effector relationships?



Work on ICC has been funded by the NIH through grants P01 DK41315 to K.M.S. and S.M.W. and DK40569 to K.M.S. and DK57236 to S.M.W. The authors would like to thank Professor Terence Smith for helpful discussions during preparation of this review. We are grateful to Drs Laura Dwyer and Sang Don Koh for the trace in Fig. 6C showing the effects of carbachol on isolated murine gastric smooth muscle cells. There are no conflicts of interest to disclose.