DEVELOPMENTAL DYNAMICS 213:271?282 (1998) Development of Pacemaker Activity and Interstitial Cells of Cajal in the Neonatal Mouse Small Intestine LOUIS W.C. LIU,1,2 LARS THUNEBERG,3 AND JAN D. HUIZINGA1,2* 1Intestinal Disease Research Program, Department of Biomedical Sciences, McMaster University, Hamilton, Ontario, Canada 2Department of Electrical and Computer Engineering, McMaster University, Hamilton, Ontario, Canada 3Institute of Medical Anatomy, Section C, The Panum Institute, University of Copenhagen, Copenhagen, Denmark ABSTRACT Intestinal motor patterns are not well developed in premature infants. Similarly, in neonatal mice, irregular motor patterns were observed. Pacemaker cells, identified in the small intestine as interstitial cells of Cajal (ICCs) associated with Auerbach?s plexus (ICC-APs), contribute to the generation of peristaltic movements. The objective of the present study was to assess the hypothesis that abnormal gut motor activity in (preterm) newborns can be associated with underdeveloped ICCs. Specifically, the aim was to identify at which point the electrical pacemaker activity is fully developed and whether or not the development of pacemaker activity has a structural correlation with the developmental stage of ICCs. Pacemaker activity was identified as that component of the slow wave that is insensitive to L-type calcium (Ca2?) channel blockers and displays a characteristic reduction in frequency in the presence of cyclopiazonic acid (CPA), a specific inhibitor of the endoplasmic reticulum Ca2? pump. In newborn, unfed neonates, action potentials occurred that were irregular in frequency and amplitude and sensitive to verapamil. CPA (5 然) abolished all action potentials. Quiescent spots were observed in approximately 50% of impalements. Six hours after birth, slow-wave activity appeared at a regular frequency and amplitude, and a welldefined plateau phase was observed. Verapamil did not affect the frequency, 5 然 CPA decreased it. The effect of CPA on the pacemaker frequency 2 days after birth was identical to that observed in adult mice. In 2-hr-old neonates, ICCs could be identified through selective uptake of methylene blue, but ultrastructural features were not fully developed. At 48 hr, a complete ICC network covering Auerbach?s plexus was formed, confirmed by electron microscopy. In summary, the pacemaker component of the slow waves can be identified in neonates as early as 6 hr after birth. The pacemaker component was fully developed 2 days after birth. These electrophysiological observations correlated with the development of full network characteristics of ICC-APs and the development of fully differentiated ICC-APs from r 1998 WILEY-LISS, INC. ??blast-like?? cells. Dev. Dyn. 1998;213:271?282. r 1998 Wiley-Liss, Inc. Key words: interstitial cells of Cajal; neonatal mice; pacemaker development; smooth muscle; intestinal motility INTRODUCTION In preterm human infants, rhythmic contractile activity can often be observed (Morris, 1991) at a dominant frequency of ?3 cycles per minute (cpm) in the stomach and ?12 cpm in the small intestine starting from 28 weeks of gestation (Tomomasa et al., 1985). These characteristic frequencies, which are similar to those in adult tissue, suggest the presence of electrical pacemaking activity imposing its characteristic rhythm onto the musculature. However, despite its rhythmic character, such contractions appear to be predominantly nonpropagating (Tomomasa et al., 1985; Berseth, 1989). A lack of normal responses to feeding indicates that regulatory mechanisms for peristalsis are underdeveloped. Such preterm infants, thus, are said not to ??tolerate?? oral feed as a consequence of impaired gastrointestinal (GI) motility (Newell et al., 1993). The characteristic rhythmic motor activity is driven by omnipresent electrical activity generated within the musculature, the so-called ??slow waves,?? which perform a pacemaker function. Electrophysiological and structural evidence acquired from different isolated and intact muscle strip preparations suggest that pacemaker activity originates from a network of interstitial cells of Cajal (ICCs; Thuneberg, 1982; FaussonePellegrini, 1992; Sanders, 1996; Huizinga et al., 1997) associated with the myenteric plexus of the stomach (Bauer et al., 1985) and small intestine (Hara et al., 1986; Ward et al., 1994; Huizinga et al., 1995) and with the submuscular plexus of the colon (Smith et al., 1987; Du and Conklin, 1989; Liu and Huizinga, 1993). ICCs have the Kit tyrosine kinase receptor embedded in their membranes. Antibodies against the Kit receptor have greatly aided in the identification of Kit-positive cells. Correlations with electron microscopy have made it *Correspondence to: Jan D. Huizinga, Intestinal Disease Research Program, HSC-3N5C, McMaster University, 1200 Main Street West, Hamilton, Ontario L8N 3Z5, Canada. E-mail: email@example.com Received 12 January 1998; Accepted 28 July 1998 272 LIU ET AL. likely that at least most Kit-positive cells within the gut musculature are ICCs (Komuro and Zhou, 1996). Kitpositive cells are markedly diminished in number in the afflicted section of the human colon in Hirschsprung?s disease (Vanderwinden et al., 1996c) and in the circular muscle layer of the pylorus in infantile pyloric stenosis (Vanderwinden et al., 1996a). The paucity of ICCs in these pathological conditions possibly contributes to the abnormal motility associated with these diseases. In the mouse, Kit-positive cells are recognized beginning at day 12 of gestation. These cells later differentiate into Kit-positive cells that are associated with Auerbach?s plexus and Kit-negative longitudinal muscle cells (Torihashi et al., 1997). Kit-positive ICCs and smooth muscle myosin heavy chain (SMMHC)-positive smooth muscle cells both develop from mesodermally derived mesenchymal pluripotential precursor cells that express both Kit and SMMHC mRNA markers (Klu?ppel et al., 1998). Although the location of Kitpositive cells just before birth strongly suggests that they will develop into ICCs, they cannot be identified as ICC by using established electron microscopic criteria (Thuneberg, 1982). Electron microscopic studies have shown that the cellular differentiation associated with Auerbach?s plexus is incomplete at birth, and several candidates for ICC precursor cells were identified up to 2?3 weeks after birth in the mouse small intestine (Faussone-Pellegrini, 1985) and colon (Faussone-Pellegrini, 1987). The fact that the development of the ICCs can be interrupted by injection of antibody against the Kit receptor in the first 4 days after birth, but not after day 9, is also consistent with ICCs not being fully developed at birth (Maeda et al., 1992; Torihashi et al., 1995). ICCs can selectively accumulate methylene blue (Mikkelsen et al., 1988; Liu et al., 1993), and, subsequently, characteristic changes in ribosomes and chromatin occur (Malysz et al., 1996), so that methylene blue-positive cells can be recognized and identified with electron microscopy. Because precursor cells of ICCs are difficult to identify, we used methylene blue as an aid in the identification of such cells. In the mouse, intestinal pacemaker activity appears to be immature at birth (Maeda et al., 1992). Although contractile activity has been observed on day 16 of gestation (Gershon and Thompson, 1973), and electrical activity was seen to start at day 19 of gestation (Torihashi et al., 1997), a pacemaker component, if it is present, has not been identified in embryonic or neonatal mice. To determine electrophysiologically the maturity of pacemaker activity, we used specific pharmacological properties of adult pacemaking activity. First, slow waves have a component that is insensitive to L-type Ca2? channels blockers (Huizinga et al., 1991; Ward and Sanders, 1992). Second, the slow-wave activity reacts to cyclopiazonic acid (Liu et al., 1995a), which is a specific inhibitor of the endoplasmic reticulum Ca2? pump, with a characteristic and marked reduction in frequency. The objectives of this study were to investigate in the mouse the electrophysiological development of the intrinsic pacemaker activity of the intestinal musculature and the structural development of ICCs, in particular, network formation and ultrastructural characteristics. An account of these data was presented at the 14th International Symposium on Gastrointestinal Motility (Liu et al., 1995b). The expectation was that this might lead to the formulation of a hypothesis addressing the mechanism behind certain motor abnormalities in preterm infants. RESULTS Electrical Activity of Neonatal Mouse Small Intestine The maturity of pacemaker activity in neonatal mice was investigated by comparing the characteristics of the recorded activity with pacemaker activity that was identified previously in adult mice. The pacemaker activity of the small intestine is the component of the slow-wave activity that is characterized by 1) insensitivity of amplitude and frequency to L-type calcium channel blockers (Malysz et al., 1995), 2) dose-dependent reduction in frequency by cyclopiazonic acid (CPA; Liu et al., 1995a), 3) slow but eventually complete inhibition by the removal of extracellular calcium, and 4) amplitude inhibition by Ni2?. Newborn, unfed. All tissues obtained from the jejunum of newborn, unfed neonatal mice exhibited spontaneous electrical activity in Krebs solution (Table 1, Fig. 1a). Neonates were discarded if traces of milk were found in any part of the exposed GI tract. The electrical activity was irregular in both frequency and upstroke amplitude (Fig. 1a). Similarly, the isolated but intact jejunum contracted spontaneously at a low and irregular frequency. Furthermore, no propagating, ringlike contractions were observed. The plateau phase of slow waves was not observed in any of 47 stable impalements from six different muscle strips from five neonates. In all preparations, many quiescent spots (53 out of 112 impalements) were observed with resting membrane potentials ranging from ?72 mV to ?58 mV. The L-type calcium channel blocker, verapamil (1 然) decreased the frequency, the upstroke amplitude, and the rate of rise of the electrical oscillatory activity (Table 2, Fig. 1a). Within 1 min of perfusion, 5 然 CPA completely abolished all electrical activity (Fig. 2a, Table 3). The effects of CPA were completely reversible. Ni2? (1 mM), with or without verapamil, completely abolished all electrical activity, with a slight decrease in membrane potential (2?4 mV). In three additional preparations obtained from the ileum, the electrical activity was completely abolished by 1 然 verapamil without a change in the resting membrane potential; in addition, there were more frequent encounters of electrically quiescent cells. In summary, the electrical activity observed in newborn, unfed neonates does not include a 273 PACEMAKER ACTIVITY AND ICCs IN NEONATAL MOUSE INTESTINE TABLE 1. Spontaneous Electrical Activity in Neonatal Mouse Small Intestine Measure Resting membrane potential (mV) Frequency (cpm) Duration (sec) Upstroke amplitude (mV) Plateau amplitude (mV) Rate of rise (mV/sec) Newborn, unfed group (n ? 5) 6?12 Hr group (n ? 8)a 24?48 Hr group (n ? 5) 2?7 day group (n ? 13) ?63.1 ? 2.8 14.1 ? 0.4 0.9 ? 0.2 22.1 ? 3.8 ? 119.2 ? 26.3 ?61.5 ? 1.7 17.4 ? 1.1** 1.1 ? 0.1 20.3 ? 3.5 14.9 ? 3.3 146.5 ? 20.9* ?62.0 ? 2.3 19.1 ? 0.9** 1.2 ? 0.1 23.2 ? 3.5 16.7 ? 2.6 136.3 ? 21.6* ?64.8 ? 2.1 19.8 ? 0.8** 1.2 ? 0.1 24.7 ? 2.6 19.6 ? 2.4 157.1 ? 28.9* aOnly five animals showed slow wave plateaus. *Statistically significant difference of different age groups were compared with the newborn, unfed group (P ? 0.05). **P ? 0.01. Fig. 1. Effects of verapamil on the electrical activity in neonatal mouse small intestine of different age groups. a: In Krebs solution, the spontaneous electrical oscillations in the small intestine of newborn, unfed neonatal mice were various in amplitude and frequency (top traces). The plateau phase had not developed. One micromolar of verapamil, an L-type calcium (Ca2?) channel blocker, decreased both the amplitude and the frequency without affecting other parameters (bottom). Right traces show recordings at a faster chart speed. b: In the 6?12 hr group, both the frequency and the amplitude became steady. The plateau phase of the slow-wave activity was well defined, which is illustrated better at a faster chart speed (right top trace). Recordings were made from a an 8-hr-hold neonate. Verapamil (1 然) decreased the upstroke amplitude and rate of rise (bottom traces). c: Recordings were made from a 30-hr-old neonate. In Krebs solution, the electrical activity of this age group was not significantly different from the 6?12 hr group. Consistent with other age groups, verapamil (1 然) decreased the upstroke and rate of rise of the slow waves. d: Recordings were made from a 7-day-old mouse. Spikes superimposed on the plateau phase of slow waves started to be observed in 2-day-old mice. To better illustrate the spike activity, traces are shown in different time scales (note different calibrations). Verapamil (1 然) abolished all spiking activity and decreased the amplitude and rate of rise of the slow waves. pacemaker component of the slow-wave activity corresponding to the one described in adult tissue or neonatal tissue at 48 hr (see below). Six to twelve hours. Unlike tissues from newborn, unfed neonates, at 6?12 hr, tissues showed slow wavelike activity with a well-defined upstroke and a plateau 274 LIU ET AL. TABLE 2. Effects of Verapamil on Electrical Activitya Measure Newborn, unfed group (n ? 4) Krebs Verapamil 6?12 Hr group (n ? 4) Krebs Verapamil 24?48 Hr group (n ? 4) Krebs Verapamil 2?7 Days group (n ? 6) Krebs Verapamil Resting membrane potential (mV) ?59.9 ? 1.4 ?58.8 ? 1.3 ?61.3 ? 1.0 ?61.3 ? 1.0 ?59.5 ? 2.1 ?58.5 ? 1.5 ?59.3 ? 0.5 ?58.8 ? 0.6 Frequency (cpm) 13.9 ? 0.5 9.9 ? 1.2** 18.4 ? 1.3 18.0 ? 1.1 18.9 ? 1.2 18.4 ? 1.4 20.7 ? 1.7 19.3 ? 1.6 Duration (sec) 1.1 ? 0.2 0.9 ? 0.1 1.2 ? 0.3 1.1 ? 0.2 1.2 ? 0.2 1.1 ? 0.1 1.2 ? 0.1 1.1 ? 0.1 Upstroke amplitude (mV) 21.4 ? 5.4 16.3 ? 5.3* 18.5 ? 2.8 16.8 ? 1.8* 15.8 ? 1.1 12.7 ? 1.4* 16.4 ? 1.4 14.3 ? 1.1* Plateau amplitude (mV) ? ? 11.2 ? 0.6 11.3 ? 0.8 11.1 ? 0.6 9.8 ? 0.8 12.8 ? 0.8 11.2 ? 1.1 Rate of rise (mV/sec) 103.2 ? 29.1 50.3 ? 16.8** 146.4 ? 34.8 121.9 ? 34.0* 129.3 ? 20.7 98.3 ? 23.3* 134.7 ? 15.7 89.2 ? 13.9** aActivity in the presence of verapamil (1 然) was compared with the corresponding activity in Krebs solution in the same age group. Oscillations with amplitude less than 2 mV were neglected. *P ? 0.05. **P ? 0.01. Fig. 2. Effects of cyclopiazonic acid (CPA) on the electrical activity in neonatal mouse small intestine of different age groups. The beginning of all tracings show electrical activities in the presence of 1 然 verapamil for at least 20 min. CPA (5 然), a specific inhibitor of the endoplasmic reticulum Ca2?-pump, was added to the perfusion solution at arrows. Wash-out segments show electrical activities after removing CPA from the perfusion solution for 30 min. Experiments shown were obtained from impalements of the same cell. The effects of CPA were reversible after wash out for 30 min. a: In the presence of verapamil, similar to Figure 1a, action potentials were irregular in amplitude and frequency. CPA first reduced the frequency and then decreased the amplitude gradually. Activity was completely abolished within 5 min of perfusion with CPA. b: Recordings were made from a 10-hr-old neonate. Ten minutes of the recording were omitted between the first and the second traces. CPA significantly reduced the slow-wave frequency. Unlike in the newborn, unfed group, the upstroke amplitude reduced only slightly. c: The neonate used in this experiment was 42 hr old. There was a gap of 17 min between the first and second traces. CPA slightly depolarized the cells and subsequently decreased the upstroke amplitude. The frequency was reduced significantly. d: Experiment performed on a 5-day-old neonate. Twenty minutes of the recording were omitted between the first and second traces. Similar to other age groups, CPA significantly decreased the frequency and slightly depolarized the tissue. phase. Both the frequency and the amplitude were regular (Fig. 1b). Consistently, regular rhythmic contractions were observed in the whole intestine of these neonates. The rate of rise of the slow wave-like activity in this group was significantly higher than that of newborn, unfed animals (Table 1). Verapamil (1 然) did not affect the frequency and plateau amplitude of the slow wave-like oscillations (Table 2, Fig. 1b) but still decreased the upstroke amplitude and the rate of rise (Table 2). Addition of CPA (5 然) in the presence of verapamil significantly decreased the frequency of the slow wave- 275 PACEMAKER ACTIVITY AND ICCs IN NEONATAL MOUSE INTESTINE TABLE 3. Effects of Cyclopiazonic Acid on Electrical Activity Measure Newborn, unfed group (n ? 4) Verapamil CPA 6?12 Hr group (n ? 4)a Verapamil CPA 24?48 Hr group (n ? 4)b Verapamil CPA 2?7 Day group (n ? 5)b Verapamil CPA Resting membrane potential (mV) ?59.8 ? 1.3 ?57.2 ? 0.9* ?62.7 ? 1.3 ?60.0 ? 1.2* ?60.0 ? 0.8 ?59.0 ? 1.0 ?59.5 ? 0.6 ?57.8 ? 0.7* Frequency (cpm) 9.9 ? 1.2 ? 18.6 ? 1.2 1.9 ? 0.1** 20.7 ? 0.5 4.8 ? 0.7c** 19.6 ? 1.9 7.8 ? 1.5d** Duration (sec) 0.9 ? 0.1 ? 1.2 ? 0.2 1.0 ? 0.3 1.0 ? 0.1 1.3 ? 0.3 1.2 ? 0.2 1.1 ? 0.1 Upstroke amplitude (mV) 16.3 ? 5.3 ? 21.5 ? 3.3 19.0 ? 3.5 15.2 ? 0.4 13.3 ? 0.8* 15.8 ? 1.2 12.2 ? 1.4* Plateau amplitude (mV) ? ? 12.7 ? 1.5 ? 10.1 ? 0.5 10.4 ? 0.3 11.5 ? 1.5 11.2 ? 1.1 Rate of rise (mV/ sec) 50.3 ? 16.8 ? 169.2 ? 35.0 102.5 ? 7.5** 110.0 ? 16.7 67.5 ? 27.5** 103.6 ? 21.7 71.2 ? 11.9** aOnly one animal exhibited the plateau phase; thus, plateau amplitude is not included in the table. plateau phase was observed in three animals from each experimental group. Activity in the presence of 5 然 cyclopiazonic acid (CPA) was compared with the corresponding activity in Krebs solution containing 1 然 verapamil in the same age group. cSignificantly larger than in the bathing solution in the 6?12 hr group. dSignificantly larger than in the same bathing solution in the 6?12 hr group and the 24?48 hr group. The effects of CPA were completely reversible in all experiments. *P ? 0.05. **P ? 0.01. bThe like activity to 10 % of that in verapamil (Table 3, Fig. 2b). Both the amplitude and the rate of rise were also decreased. The effects of CPA were completely reversible. Ni2? (1 mM) abolished all electrical oscillations accompanied by a depolarization of 2?3 mV. Twenty-four to forty-eight hours. At 24?48 hr, electrically quiescent areas were no longer identified. Verapamil (1 然) decreased only the upstroke amplitude and the rate of rise (Table 2, Fig. 1c) of slow wave-like oscillations. At 24 hr, regular and forceful contractions were observed. At 48 hr, spatially coordinated, regular, and forceful contractions had been developed in the intact intestine, creating propagating ring contractions. Addition of 5 然 CPA decreased the slow-wave frequency to 23 % (Table 3, Fig. 2c). In the presence of verapamil, 1 mM Ni2? abolished all of the oscillations in two of five preparations; in the remaining three preparations, periodic oscillations of 2?4 mV in amplitude were observed at a frequency of 12.1 ? 2.2 cpm (18.2 ? 1.0 cpm in verapamil). To completely abolish the remaining oscillations, 2 mM Ni2? was required . Two to Seven Days. Similar to other age groups, verapamil consistently decreased the upstroke amplitude and the rate of rise of the slow-wave activity (Table 2) but there was no effect on frequency and slowwave plateau. The effects of CPA in the presence of verapamil were similar to the 24?48 hr group, except for the effects on frequency (Table 3). The frequency was reduced to 40% of that in verapamil alone (Fig. 2d). Electrical activity was abolished by 2 mM Ni2? in three out of four preparations; in the remaining preparation, 5 mM Ni2? were needed to completely abolish the activity, accompanied by a 4-mV depolarization. The sensitivity of the slow waves to the presence of extracellular calcium was investigated by removing Ca2? salts from the Krebs solution, leaving only nominal calcium ([Ca2?]nom ? 10?7 M). [Ca2?]nom first decreased the frequency and then decreased the amplitude to 2?6 mV (n ? 4; Fig. 3a); in three different preparations all oscillations were abolished. The resting membrane potential changed from ?61.8 ? 0.5 to ?59.5 ? 0.6 mV (P ? 0.05). There was a time delay of 10?26 min for calcium removal to take effect. Addition of 1 mM of the calcium chelator EGTA to nominal-Ca Krebs solution reduced the oscillation amplitude to less than 2 mV in 8?15 min (n ? 4; Fig. 3b). The effects of Ca2? removal with or without EGTA were completely reversible, although the wash-out period was longer if the tissue had been challenged previously with EGTA. Action potentials superimposed on the plateau phase of slow waves occurred in 2-day-old mice (11 of 13 mice). The appearance of action potentials became progressively more frequent with age. The frequency and amplitude of these action potentials were 148.5 ? 21.9 cpm and 3.5 ? 0.5 mV, respectively (n ? 11). These action potentials were abolished by 1 mM verapamil (Fig. 1d). Distribution and Identification of Methylene Blue-Positive Cells as ICCs Associated With Auerbach?s Plexus Light microscopy. Small intestinal wholemounts from 2-hr-old mice, after methylene blue staining, revealed scattered ICCs associated with Auerbach?s plexus (ICC-APs; Fig. 4a). A full network structure had not yet developed. At 12 hr, the methylene blue-positive cells formed a partial, incomplete network of increased density. At 48 hr (Fig. 4b), the staining pattern was indistinguishable from that in the adult mouse. Electron microscopy. In newborn mice, a firm identification of ICC-APs based on electron microscopic criteria alone was not possible, as noted previously (Faussone-Pellegrini, 1985). Most cell types were undergoing differentiation and frequent divisions and, consequently, had a less differentiated cytoplasm with numer- 276 LIU ET AL. Fig. 3. Effects of removal of extracellular Ca2? on the slow-wave activity of small intestine of two 3-day-old neonates. Both experiments were performed in the presence of 1 然 verapamil. Verapamil had been perfused for at least 20 min before the perfusion solution was switched to nominal calcium ([Ca2?]nom). Experiments shown were continuous recordings made from the same cell. a: Removal of extracellular Ca2? from Krebs solution containing 1 然 verapamil resulted in slow reduction of the upstroke frequency and amplitude. Small oscillations of approximately 3?4 mV were observed at a lower frequency after an 18-min perfusion with [Ca2?]nom-Krebs solution containing 1 然 verapamil. The effects were reversible after a 30-min wash out with Krebs solution containing 1 然 verapamil. b: Addition of 1 mM EGTA to the [Ca2?]nom-Krebs solution containing 1 然 verapamil accelerated the effects of removal of extracellular Ca2?. Reduction in the frequency was observed after a 4-min perfusion. In 10 min, oscillations were abolished completely. ous free ribosomes. In the Auerbach?s plexus region, interstitial cells were present. These could be identified tentatively as fibroblast-like or macrophage-like cells. In addition, we observed occasional, scattered interstitial cells with a higher than usual mitochondrial content (Fig. 5). These cells were always identified at the interface between the longitudinal and the circular muscle. Each of these cells was characterized by a large nucleus, a small amount of perinuclear cytoplasm, extensive smooth endoplasmic reticulum, and numerous free ribosomes, caveolae, and processes. At the same stage of development, ICCs at the deep muscular plexus (ICC-DMPs) could be identified clearly (Fig. 6), they presented with a pattern and relative density similar to adult tissue. After 48 hr postpartum, ICC-APs were easily recognized by cellular characteristics similar to those of adult ICC (Fig. 7). We then investigated these tissues after they were prestained with methylene blue. Methylene blue caused characteristic changes in the ultrastructural features of the ICC cytoplasm, namely, fine granulation of ribosomal material and a nucleus that showed a pattern of patches of heavily stained heterochromatin (Fig. 8). No other cell type exhibited such a staining pattern. These data allowed us to investigate tissues at 12 hr postpartum that were prestained with methylene blue. At 12 hr postpartum, scattered interstitial cells in the Auerbach?s plexus region were positive for methylene blue (Fig. 9). All of these cells had a higher than usual mitochondrial content, an abundance of free ribosomes, and occasional caveolae. These mitochondria-rich cells and processes were therefore identified as premature ICC-APs. PACEMAKER ACTIVITY AND ICCs IN NEONATAL MOUSE INTESTINE 277 Fig. 4. Network characteristics of methylene blue-positive cells. a: Jejunum of a 2-hr-old neonate. Interstitial cells of Cajal (ICC) associated with Auerbach?s plexus (ICC-APs) are stained with methylene blue. The network structure has not been developed (compared with b). Magnification: ?210. b: Jejunum of a 48-hr-old neonate. Note the change from scattered, not visibly connected ICC-APs at 2 hr to a fully developed network structure of ICC-APs. Magnification: ?210. DISCUSSION The main findings of the present study are that ICC-APs are not fully differentiated at birth, although their precursor cells can be identified by specific structural alterations due to selective uptake of methylene blue. The ICC-AP precursor cells are mitochondrionrich interstitial cells with occasional caveolae. These precursor cells are scattered and do not form a complete network. Forty-eight hours after birth, ICC-APs can be identified without the aid of methylene blue, and their network structure is complete. Electrophysiologically, at birth, the adult-type pacemaker activity is not present. Forty-eight hours after birth, the pacemaker activity is stable and has attained adult characteristics. Fig. 5. Identification in an unfed, 2-hr-old neonate of a mitochondria rich cell in the Auerbach?s plexus region of the proximal jejunum. a: In the Auerbach?s plexus region, the typical ICCs of the adult, differentiated type were absent. Furthermore, it was not possible to identify with certainty a precursor cell, because many cells contained a high number of mitochondria and free ribosomes. Nevertheless, scattered throughout the region, ICC-like cells were present (ICC) that had several structural characteristics of adult ICC-AP cells, in particular, a higher than average number of mitochondria in the nuclear region. The ICC-like cells, similar to the smooth muscle cells, contained more free ribosomes than the adult type. Magnification ?14,300. CM, circular muscle; LM, longitudinal muscle. b: Enlargement of part of a. Note the mitochondria (m) in both ICC-like cells and smooth muscle cells, free ribosomes (R), and caveolae (arrows). Magnification ?20,000. Scale bars ? 1 痠 in a, 0.5 痠 in b. Ontogenesis of the Pacemaker Activity in Neonatal Mouse Small Intestine Electrophysiology. In newborn mice, the mature pacemaker activity is not present, but action potentials are generated, as noted previously (Torihashi et al., 1997). To characterize the action potentials, reference can be made to adult tissue. In the mouse small intestine, action potentials are identified as superimposed on the slow-wave activity and are very sensitive to L-type calcium channel blockers (Malysz et al., 1995). Action potentials are further reported in W mutant mice, which do not have ICC-APs (Malysz et al., 1996). In these mice, the action potentials 1) occur at variable frequencies of 0?50 cpm, 2) consist of a slow component with superimposed spikes, and 3) are completely abolished by L-type calcium channel blockers. 278 LIU ET AL. Fig. 7. Identification of ICC-APs at 48 hr. Jejunum was first preincubated with lysolecithin, which selectively destroys the mesothelium (M). Thereafter, incubation with methylene blue was followed by uptake into ICC-APs (ICC); note the changed nuclear morphology, seen as an increased contrast of condensed chromatin in these two cells. LM, longitudinal muscle layer. Magnification ?3,000. Scale bar ? 2 痠. Fig. 6. Identification in an unfed, 2-hr-old neonate of ICCs at the deep muscular plexus (ICC-DMPs). a: A long process of ICC-DMPs (asterisk) occupies its characteristic position between the main outer layer of the circular muscle (oCM) and the inner layer of circular muscle (iCM) opposite the deep muscular plexus. The ICC-DMP process is closely contacted by a nerve fascicle of the DMP (N). ICC-DMP cells and their cellular processes were present at a cellular density that was comparable to the adult pattern. SUB, submucosa. Magnification ?14,300. b: Enlargement of part of a. Note the numerous caveolae (small arrows), close contact with the outer circular muscle layer (arrowhead) but not with the inner circular muscle layer, and close contact with bundles of varicose axons of the DMP (N; large arrow). Magnification ?20,000. Scale bars ? 1 痠 in a, 0.5 痠 in b. Similar action potentials have been described in the canine colonic circular smooth muscle, when devoid of the submuscular pacemaker ICC network (Liu and Huizinga, 1994). The electrical activity in newborn mice is highly sensitive to L-type calcium channel blockers and is abolished by Ni2?. This is in contrast to the adult slow waves, which possess a pacemaker component that is insensitive to L-type calcium channel blockers and which always have an Ni2?-insensitive component (up to 2 mM). The lack of pacemaker activity in newborn mice is illustrated further by the abolition of the electrical activity by CPA (1 然). In adult mice, CPA (1 然) causes a reduction in frequency but not abolition. These observations are all consistent with the hypothesis that the electrical activity exhibited in the newborn mouse small intestine is generated primarily by smooth muscle cells with little influence from ICCs. During the first 2 days after birth, the pacemaker component in the electrical activity becomes stronger; by 48 hr, the pharmacology of the slow-wave activity is similar to that in adult tissue. Morphology. Faussone-Pellegrini (1985) identified two possible cell types in neonates as candidates for precursors of ICC-APs. ??Blast-like cells?? were elongated in shape with short, lateral branches and possessed a large, ovoid, electron-lucent nucleus; numerous free ribosomes; abundant, large mitochondria; poorly differentiated cytoplasm; and a close association with nerve fibers and nerve endings. The second candidate was described as a fibroblast-like cell, which presented with bundles of filaments in their peripheral cytoplasm. In the present study, the methylene blue-positive cells that developed into ICC-APs, as identified by electron microscopy, were very similar to the blast-like cells. The absence of caveolae in early stages of development was also noted in smooth muscle cells (Gabella, 1989). Therefore, we propose that the mitochondrion-rich interstitial cells identified in the present study are the PACEMAKER ACTIVITY AND ICCs IN NEONATAL MOUSE INTESTINE 279 Fig. 9. Neonate at 12 hr. Only mitochondria-rich, ICC-like cells, as identified in Figure 7, were methylene blue positive. The region shown is between the longitudinal muscle layer (LM) and the circular muscle layer (CM). Only the mitochondria-rich processes show the cytoplasmic change associated with methylene blue uptake: The common ribosomal appearance (as in the muscle cell marked LM) changes to a finely granular or flocculent appearance. R, ribosomes; m, mitochondria. Magnification ?20,000. Scale bar ? 0.5 痠. Fig. 8. Changes in ultrastructure due to accumulation of methylene blue. a: Increased nuclear contrast due to condensed chromatin in ICC (at right). Fibroblast nuclei (at left) retain the normal contrast. Cells were identified during scanning of the whole cell. b: Normal ribosomal particles in the fibroblast (at left) in contrast to the finely granular or flocculent ribosomal material in the ICC (at right). precursor cells of ICC-APs, consistent with the tentative conclusion of Faussone-Pellegrini. A close spatial relationship was observed between ICC-AP precursor cells and nerves in newborn neo- nates. However, neural interaction should not be interpreted as essential for ICC development. ICCs in the mouse (Young et al., 1996) and chicken (Lecoin et al., 1996) have been shown to develop in the absence of the enteric nervous system. The growth factor, ??steel factor,?? is essential for maturation of the ICC-AP network in the mouse small intestine (Ward et al., 1994; Klu?ppel et al., 1998). Steel factor in the mouse may be produced by enteric nerves (Torihashi et al., 1996), although this apparently is not the case in humans (Vanderwinden et al., 1996c). In the present study, the identification of ICC-DMPs was unequivocal when using electron microscopic characteristics, even in neonates. Faussone-Pellegrini (1984) was more cautious, in that she also referred to these cells as ??ICC blast-like cells,?? which could not be identified with certainty. Torihashi and coworkers (Willenbucher et al., 1992; Torihashi et al., 1997) reported that such ICC blast cells were not observed at day 18 of gestation. Thus, in combination with our observation, it is suggested that a significant development of this ICC network is undertaken in the last 3 days before birth. It is noteworthy that pacemaker activity was not present in newborn neonates, even with well-developed ICCDMPs. This is consistent with the existing data, suggesting that the intestinal pacemaker activity is associated with ICC-APs rather than with ICC-DMPs. Functional Implications of an Underdeveloped ICC-AP Network In preterm infants, dominant, nonpropagating, rhythmic, contractile activity can be associated with food 280 LIU ET AL. intolerance. We showed in the normal mouse that the slow-wave activity plays a major role in normal peristaltic activity of the proximal small intestine (DerSilaphet et al., 1998). In W mutant mice, in which ICC-APs as well as slow-wave activity were absent, nonpropagating, rhythmic, contractile activity was also dominant. Hence, the presence of underdeveloped ICCAPs, as observed in newborn mice, may lead to abnormalities in peristaltic, propagating, contractile activity. It would lead to inadequate signal transmission among ICCs, nerves, and smooth muscle cells: hence the hypothesis is that, in preterm infants with a dominant pattern of nonpropagating motor activity, the network of ICCs is not fully developed. Consistent with this hypothesis is a recent finding that a premature infant without peristaltic activity in the colon was found to have no ICCs. Muscle cells as well as distribution of neuronal tissue were normal (Kenny et al., 1998). It is noteworthy that ICC networks can mature normally after such developmental delays (Vanderwinden et al., 1996b; Kenny et al., 1998). It is also important to note that, in the colon, the ICC network is responsible for coordinating motor activities along the long axis of the colon, across circular muscle lamellae. These data encourage further studies in preterm infants into the development of ICCs, both in relation to ultrastructural maturation and the forming of a network structure. Pacemaker Ion Channel in Neonatal Mouse Small Intestine The slow wave in the adult mouse small intestine is likely initiated by a Ca2?-dependent, nonspecific cation channel (Malysz et al., 1995; Thomsen et al., 1998). In the canine colon, a similar pacemaker channel has also been proposed (Huizinga et al., 1991; Ward and Sanders, 1992). More specifically, it has been demonstrated recently that the frequency of activation of the pacemaker channel is entrained with the calcium-refilling cycle in the endoplasmic reticulum associated with the plasma membrane (Liu et al., 1995a). The observations with CPA in the present study show that a similar coupling mechanism is likely operating in the mouse small intestine. In summary, this is the first documentation of the development of the pacemaker component of the electrical activity in the small intestine. 1) Methylene blue was employed successfully to identify ICC precursor cells. 2) At birth, the absence of slow-wave activity was correlated with ICCs that were not fully differentiated and were not organized yet as a network These data provide the hypothesis that, in preterm infants, dominant patterns of nonpropagating motor activity may be due to an underdeveloped network of ICCs. EXPERIMENTAL PROCEDURES Tissue Acquisition and Preparation Neonatal mice (birth to 7 days) were decapitated. Pregnant mice (CD1) were purchased from Charles River Laboratories (Wilmington, MA) at 15?16 days of gestation and monitored for delivery (time zero) after 19?20 days of gestation. The GI tract, starting from the lower esophagus to the colon, was removed with the intact mesenteric vascular bed to minimize stretch when the gut was transferred to a dissecting dish, which was filled with prewarmed Krebs solution. After releasing the gut from the mesenteric vascular bed, the gut was mounted without stretching onto a Sylgard (184 silicone elastomer; Dow Corning Corporation, Midland, MI) surface with insect pins (Fine Science Tools Inc.; 0.1 mm in diameter) at the stomach and the ileocaecal junction. The musculature of the proximal small intestine (20?50% of the entire length, as measured between the pylorus and the ileocaecal junction) was carefully dissected from the submucosa without opening the gut under a dissection microscope (Zeiss, Thornwood, NY) at a magnification of ?40. Electron microscopic examination of the dissected tissue revealed that the musculature cleaved along the deep muscular plexus, leaving the outer circular muscle layer, the myenteric plexus, and the longitudinal muscle layer intact. The isolated musculature was mounted with the serosal surface facing up in between two pieces of Sylgard that were anchored with insect pins. The top piece of Sylgard had a circular opening approximately 1 mm in diameter, through which microelectrodes were able to access the tissue. Before experimentation began, all preparations were equilibrated for at least 2 hr at 37.0 ? 0.5蚓 in a tissue chamber with continuously aerated (95% O2 and 5% CO2) Krebs solution perfusing at a rate of 500 ml/hr (P-1; Pharmacia LKB, Uppsala, Sweden). Electrophysiological Measurements Intracellular recordings were made by microelectrodes (50?80 M?) filled with 3 M KCl. A microelectrode was inserted into a microelectrode holder, which was connected to an electrometer (Duo773; World Precision Instruments, New Haven, CT). Microelectrodes were driven into the tissue vertically by using a micromanipulator (MN-151; Narishige). The output of the electrometer was displayed on a Gould oscilloscope (1421; Gould, Inc., Cleveland, OH) and recorded on a Gould ink-writing recorder (2400S). Drugs and Solutions All solutions perfused into the partition chamber were prewarmed to 37.0 ? 0.5蚓 and equilibrated with 95% O2 and 5% CO2. The composition (in mM) of the Krebs solution was NaCl, 120.3; KCl, 5.9; CaCl2, 2.5; MgCl2, 1.2; NaHCO3, 20.2; NaH2PO4, 1.2; and glucose,11.5. The [Ca2?]nom Krebs solution was prepared by omitting CaCl2 in the Krebs solution formula. CPA (Sigma, St. Louis, MO) was dissolved in dimethyl sulfoxide (DMSO; Sigma) to prepare a stock solution of 10 mM. Verapamil (verapamil hydrochloride; Sigma) stock solution (1 mM) was prepared in deionized, PACEMAKER ACTIVITY AND ICCs IN NEONATAL MOUSE INTESTINE distilled water. Vehicles in the concentrations applied did not have any effect on the electrical activity. Result Presentation and Statistical Analysis All data were expressed as mean ? standard error of the mean. ?n? represents the number of mice used in each set of experiments. Statistically significant differences between data sets were determined by one-way repeated measures analysis of variance (KWIKSTAT 4; TexaSoft, Cedar Hill, TX). The slow-wave duration was measured at the half maximum of the slow-wave plateau amplitude. Because of the temporal variation of the slow-wave activity in the presence of CPA, representative slow-wave parameters were obtained from analyses over periods of at least 5 min. Light and Electron Microscopy Unfed 12-hr-old, 24-hr-old, and 48-hr-old neonatal mice (n ? 4 mice of each age) were decapitated. After exposing the intestines, half of the animals were incubated for methylene blue staining, as described below, followed by fixation, whereas the other half of the animals were directly immersed in fixative. The primary fixative was glutaraldehyde 2%, formaldehyde 2%, picric acid 0.2%, and phosphate buffer 0.1 M, pH 7.5. Methylene Blue Staining The decapitated animals, after removal of the abdominal wall, were preincubated for 30 sec in calcium-free, phosphate-buffered saline containing 0.7 mM lysolecithin (Sigma) to permeabilize the mesothelial cell layer. They were transferred to and incubated for 45 min at room temperature under subdued light in a wellaerated (95% O2 and 5% CO2) Krebs solution supplemented with 50 然 methylene blue B (Merck, Darmstadt, Germany; Mikkelsen et al., 1988). The animals were transferred to the aldehyde fixative, which contained picrate to precipitate the methylene blue in the tissue. The exposed parts of the intestines were well stained and were selected under the magnification of a stereomicroscope. Wholemounts of intestinal segments as well as the isolated external muscle were studied and photographed under a Leitz Orthoplan microscope (Wetzlar, Germany). Processing for Thin and Ultrathin Sectioning After an overnight aldehyde fixation at 4蚓, samples of duodenum, jejunum, and ileum from the methylene blue-stained and the directly fixed intestines were cut in millimeter-sized pieces, washed with 0.1 M phosphate buffer, postfixed in 2% osmic acid/0.1 M phosphate buffer for 1 hr, dehydrated in a graded series of ethanol with block-staining for 1 hr in 1% uranyl acetate in absolute ethanol, followed by propylene oxide, and Epon embedding. One-micrometer-thick sections were stained with toluidine blue and examined under a Leitz Orthoplan microscope. Ultrathin sections were poststained with alcoholic uranyl acetate and lead 281 citrate and were examined under a Philips 300 electron microscope (Eindhoven, The Netherlands). ACKNOWLEDGMENTS The Medical Research Council (MRC) of Canada provided operating grants and an MRC Scientist Award to J.D.H. and a studentship to L.W.C.L. NATO provided travel funds for L.T. REFERENCES Bauer AJ, Publicover NG, Sanders KM. Origin and spread of slow waves in canine gastric antral circular muscle. Am. J. Physiol. 1985;249:G800?G806. Berseth CL. Gestational evolution of small intestine motility in preterm and term infants. J. Pediatr. 1989;115:646?651. Der-Silaphet T, Malysz J, Arsenault AL, Hagel S, Huizinga JD. Interstitial cells of Cajal direct normal propulsive contractile activity in the small intestine. Gastroenterology 1998;114:724?736. Du CA, Conklin JL. Origin of slow waves in the isolated proximal colon of the cat. J. Auton. Nerv. Syst. 1989;28:167?177. Faussone-Pellegrini MS. Morphogenesis of the special circular muscle layer and of the interstitial cells of Cajal related to the plexus muscularis profundus of mouse intestinal muscle coat. An EM study. Anat. Embryol. 1984;169:151?158. Faussone-Pellegrini MS. Cytodifferentiation of the interstitial cells of Cajal related to the myenteric plexus of mouse intestinal muscle coat. An EM study from foetal to adult life. Anat. Embryol. 1985;171: 163?169. Faussone-Pellegrini MS. Cytodifferentiation of the interstitial cells of Cajal of mouse colonic circular muscle layer. Acta Anat. 1987;128:98? 109. Faussone-Pellegrini MS. Histogenesis, structure and relationships of interstitial cells of Cajal (ICC): From morphology to functional interpretation [review]. Eur. J. Morphol. 1992;30:137?148. Gabella G. Development of smooth muscle: Ultrastructural study of the chick embryo gizzard. Anat. Embryol. (Berlin) 1989;180:213? 226. Gershon MD, Thompson EB. The maturation of neuromuscular function in a multiply innervated structure: Development of the longitudinal smooth muscle of the foetal mammalian gut and its cholinergic excitatory, adrenergic inhibitory, and non-adrenergic inhibitory innervation. J. Physiol. (London) 1973;234:257?277. Hara Y, Kubota M, Szurszewski JH. Electrophysiology of smooth muscle of the small intestine of some mammals. J. Physiol. (London) 1986;372:501?520. Huizinga JD, Farraway L, Den Hertog A. Generation of slow-wavetype action potentials in canine colon smooth muscle involves a non-L-type Ca2? conductance. J. Physiol. (London) 1991;442:15?29. Huizinga JD, Thuneberg L, Kluppel M, Malysz J, Mikkelsen HB, Bernstein A. The W/kit gene required for interstitial cells of Cajal and for intestinal pacemaker activity. Nature 1995;373:347?349. Huizinga JD, Thuneberg L, Vanderwinden JM, Rumessen JJ. Interstitial cells of Cajal as pharmacological targets for gastrointestinal motility disorders. Trends Pharmacol. Sci. 1997;18:393?403. Kenny SE, Vanderwinden JM, Rintala RJ, Connell MG, Lloyd DA, Vanderhaegen JJ, De Laet MH. Delayed maturation of the interstitial cells of Cajal: A new diagnosis for transient neonatal pseudoobstruction. Report of two cases. J. Pediatr. Surg. 1998;33:94?98. Klu?ppel M, Huizinga JD, Malysz J, Bernstein A. Developmental origin and Kit-dependent development of the interstitial cells of Cajal in the mammalian small intestine. Dev. Dyn. 1998;211:60?71. Komuro T, Zhou DS. Anti c-kit protein immunoreactive cells corresponding to the interstitial cells of Cajal in the guinea-pig small intestine. J. Auton. Nerv. Syst. 1996;61:169?174. Lecoin L, Gabella G, Le Douarin N. Origin of the c-kit positive interstitial cells in the avian bowel. Development 1996;122:725? 733. Liu LWC, Huizinga JD. Electrical coupling of circular muscle to longitudinal muscle and interstitial cells of Cajal in canine colon. J. Physiol. (London) 1993;470:445?461. 282 LIU ET AL. Liu LWC, Huizinga JD. Canine colonic circular muscle generates action potentials without the pacemaker component. Can. J. Physiol. Pharmacol. 1994;72(1):70?81. Liu LWC, Thuneberg L, Daniel EE, Huizinga JD. Selective accumulation of methylene blue by interstitial cells of Cajal in canine colon. Am. J. Physiol. 1993;264:G64?G73. Liu LWC, Thuneberg L, Huizinga JD. Cyclopiazonic acid, inhibiting the endoplasmic reticulum calcium pump, reduces the canine colon pacemaker frequency. J. Pharmacol. Exp. Ther. 1995a;275:1058? 1068. Liu LWC, Thuneberg L, Huizinga JD. Simultaneous development of pacemaker activity and interstitial cells of Cajal network in neonatal mouse small intestine [abstract]. Neurogastroenterology and Motility 1995b;7:270. Maeda H, Yamagata A, Nishikawa S, Yoshinaga K, Kobayashi S, Nishi K. Requirement of c-kit for development of intestinal pacemaker system. Development 1992;116:369?375. Malysz J, Richardson D, Farraway L, Christen MO, Huizinga JD. Generation of slow wave type action potentials in the mouse small intestine involves a non-L-type calcium channel. Can. J. Physiol. Pharmacol. 1995;73:1502?1511. Malysz J, Thuneberg L, Mikkelsen HB Huizinga JD. Action potential generation in the small intestine of W mutant mice that lack interstitial cells of Cajal. Am. J. Physiol. 1996;271:G387?G399. Mikkelsen HB, Thuneberg L, Wittrup IH. Selective double staining of interstitial cells of Cajal and macrophage-like cells in small intestine by an improved supravital methylene blue technique combined with FITC-dextran uptake. Anat. Embryol. 1988;178:191?195. Morriss FH, Jr. Neonatal gastrointestinal motility and enteral feeding [review]. Semin. Perinatol. 1991;15:478?481. Newell SJ, Chapman S, Booth IW. Ultrasonic assessment of gastric emptying in the preterm infant. Arch. Dis. Child. 1993;69(Suppl.):32? 36. Sanders KM. A case for interstitial cells of Cajal as pacemakers and mediators of neurotransmission in the gastrointestinal tract [review]. Gastroenterology 1996;111:492?515. Smith TK, Reed JB, Sanders KM. Origin and propagation of electrical slow waves in circular muscle of canine proximal colon. Am. J. Physiol. 1987;252:C215?C224. Thomsen L, Robinson TL, Lee JCF, Farraway L, Hughes MJG, Andrews DW, Huizinga JD. Interstitial cells of Cajal generate a rhythmic pacemaker current. Nature Med. 1998;4:848?851. Thuneberg L. Interstitial cells of Cajal: Intestinal pacemaker cells? Adv. Anat. Embryol. Cell Biol. 1982;71:1?130. Tomomasa T, Itoh Z, Koizumi T, Kuroume T. Nonmigrating rhythmic activity in the stomach and duodenum of neonates. Biol. Neonate 1985;48:1?9. Torihashi S, Ward SM, Nishikawa S, Nishi K, Kobayashi S, Sanders KM. c-kit-Dependent development of interstitial cells and electrical activity in the murine gastrointestinal tract. Cell Tissue Res. 1995;280:97?111. Torihashi S, Hisahiro Y, Nishikawa S, Kunisada T, Sanders KM. Enteric neurons express Steel factor-lacZ transgene in the murine gastrointestinal tract. Brain Res. 1996;738:323?328. Torihashi S, Ward SM, Sanders KM. Development of c-Kit-positive cells and the onset of electrical rhythmicity in murine small intestine. Gastroenterology 1997;112:144?155. Vanderwinden JM, Rumessen JJ, Liu H, Descamps D, De Laet MH, Vanderhaeghen JJ. Interstitial cells of Cajal in human colon and in Hirschsprung?s disease. Gastroenterology 1996a;111:901?910. Vanderwinden JM, Liu H, De Laet MH, Vanderhaeghen JJ. Study of the interstitial cells of Cajal in infantile hypertrophic pyloric stenosis. Gastroenterology 1996b;111:279?288. Vanderwinden JM, Liu H, Menu R, Conreur JL, De Laet MH, Vanderhaeghen JJ. The pathology of infantile hypertrophic pyloric stenosis after healing. J. Pediatr. Surg. 1996c;31:1530?1534. Ward SM, Sanders KM. Upstroke component of electrical slow waves in canine colonic smooth muscle due to nifedipine-resistant calcium current. J. Physiol. (London) 1992;455:321?337. Ward SM, Burns AJ, Torihashi S, Sanders KM. Mutation of the proto-oncogene c-kit blocks development of interstitial cells and electrical rhythmicity in murine intestine. J. Physiol. (London) 1994;480:91?97. Young HM, Ciampoli D, Southwell BR, Newgreen DF. Origin of interstitial cells of Cajal in the mouse intestine. Dev. Biol. 1996;96: 97?107.