Interstitial cells of Cajal (ICC) have been proposed as stretch receptors for vagal afferent nerves in the stomach based on immunohistochemical studies. The aim of the present study was to use electron microscopy and the anterograde degeneration technique to investigate ultrastructural features and survival dependency of ICC associated with vagal afferent innervation of the cat esophagus. This is the first report on the ultrastructural characteristics of ICC in the cat esophagus. Intramuscular ICC (ICC-IM) were identified throughout the musculature, whereas ICC in the myenteric plexus were rare. ICC-IM were particularly numerous in septa aligned with smooth muscle bundles. They were in synapse-like contact with nerve varicosities and in gap junction contact with smooth muscle cells. Smooth muscle cells also made contact with ICC through peg and socket junctions. Precision damage through small-volume injection of saline in the center of the nodose ganglion from the lateral side, known to selectively affect sensory nerves, was followed within 24 h by degeneration of a subset of nerve varicosities associated with ICC-IM, as well as degeneration of the associated ICC-IM. Smooth muscle cells were not affected. Nerves of Auerbachs plexus and associated ICC were not affected. In summary, ICC-IM aligning the esophageal muscle bundles form specialized synapse-like contacts with vagal afferent nerves as well as gap junction and peg-and-socket contacts with smooth muscle cells. This is consistent with a role of ICC-IM as stretch receptors associated with vagal afferent nerves; the ICC-vagal nerve interaction appears essential for the survival of the ICC.
- interstitial cells of Cajal
- stretch receptor
- afferent innervation
- intramuscular interstitial cells of Cajal
for normal esophageal function, the vagus nerves provide the major sensory and motor innervation to both striated and smooth muscle portions of the esophagus. Sensory pathways from the esophagus have been studied both electrophysiologically (40, 41) and by nerve tracing techniques (4, 11, 28). However, the precise location and structure of the sensory receptor(s) involved in control of esophageal motor function remain controversial. It is likely that more than one structurally and functionally distinct type of mechanoreceptor exists within the esophagus (4, 5, 8). As early as 1929, intraganglionic laminar nerve endings (IGLE) were considered candidates for esophageal stretch receptors (34). In an elegant study by Berthoud and Powley (4), intramuscular vagal afferents were shown to be positioned and structured as likely candidates for the gastric stretch receptor. Intramuscular arrays (IMA) enter circular or longitudinal muscle layers, run parallel to respective muscle fibers, and bifurcate to create an array of parallel terminal elements in stomach and sphincters (28). In the stomach, these terminal elements consist of interconnected parallel axonal telodendria that are arrayed in close proximity to one another and that form appositions with interstitial cells of Cajal (ICC) (20). Recent ultrastructural studies have identified synapse-like connections between IMA nerve varicosities and intramuscular ICC (ICC-IM) (29).
ICC have been implicated as a potential sensory receptor (19, 38, 45). In the stomach, it was proposed that IMA/ICC-IM/smooth muscle fiber units could constitute sensory receptor structures that mediate mechanical transmission of smooth muscle stretch to IMAs via ICC-IM (19). Because the esophagus depends on an interaction between vagal afferents and smooth muscle to perform neurally directed peristalsis, the present study set out to investigate morphological features of interactions between vagal afferents and ICC-IM, data that would benefit the evaluation of the hypothesis that ICC might be stretch receptors. Roman and coworkers (35) were the first to propose that a specific interstitial cell linked vagal afferents with smooth muscle cells in the cat esophagus although the interstitial cells were not identified as ICC. One previous study in the cat esophagus tried to identify vagal afferent nerve endings employing injection of horseradish peroxidase in the nodose ganglion and subsequent light microscopy. They indicated that numerous afferent fibers were running through the muscle layers, but it could not be established whether nerve endings were present within the muscle layers (10). ICC were not identified in that study. Therefore, we employed electron microscopy to study the relationships between nerve endings and smooth muscle cells and/or ICC. We also took advantage of the fact that motor nerves run exterior to the cat nodose ganglion as distinct superficial bundles at the ventral side (24, 25); hence, afferent neurons can be reached selectively in the center of the nodose ganglion when approached from the lateral side.
The objectives of the present study were to provide the first ultrastructural characterization of ICC in the cat esophagus, to study the possibility of survival dependency of ICC on vagal afferent nerves, and to provide further insight into ultrastructural features of varicosities of vagal afferent nerves in association with ICC, possibly comprising together a mechanoreceptor structure.
MATERIALS AND METHODS
In five cats, under isofluorane anesthesia and aseptic conditions, 5 μl of saline were injected in both nodose ganglia. We took advantage of the fact that motor nerves run exterior to the nodose ganglion as distinct superficial bundles at the ventral side (24, 25); hence, afferent fibers were reached selectively in the center of the nodose ganglion by approaching the ganglion from the lateral side. Cats were killed after 1, 3, 5, or 28 days. Two additional cats were studied 3 days after bilateral vagotomy below the nodose ganglia. Four cats were studied for control data. The studies were performed in accordance with ethical guidelines and the approval of The Toronto Hospital Animal Care Committee.
When killed, cats were anesthetized with pentobarbital sodium (30 mg/kg iv). Just before perfusion fixation, the cat was killed by injecting pentobarbital sodium (100 mg/kg iv). Each cat was fixed by intracardiac perfusion through the left ventricle with 2.5% glutaraldehyde in 0.075 M sodium cacodylate buffer, pH 7.4, containing 4.5% sucrose and 1 mM CaCl2. The esophagus with attached cuff of fundus was removed after initial perfusion fixation, opened along the lesser curvature, and pinned flat in a petri dish to a silicon-rubber mat mucosa side up. The perfused fixed muscle strips (1 cm in width) were cut circumferentially from different esophageal regions. The esophageal body of control cats was cut at distances of 0.5, 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 cm proximal to the lower esophageal sphincter (LES). The esophageal body of experimental cats was cut at distances of 0.5, 1, 2, and 3 cm proximal to the LES. After dissection, the strips were immersed in the fresh fixative for an additional fixation overnight at 4°C. Following fixation, all tissues were washed overnight in 0.1 M cacodylate buffer containing 6% sucrose and 1.24 mM CaCl2 (pH 7.4) at 4°C. After washing, smaller (1.5 × 8 mm) circular and longitudinal strips were prepared from the cut segments and postfixed with 2% OsO4 in 0.05 M cacodylate buffer (pH 7.4) at room temperature for 90 min, stained with saturated uranyl acetate for 60 min at room temperature, dehydrated in graded ethanol and propylene oxide, and embedded in Spurr.
To locate suitable areas, 0.5-mm-thick sections were cut and stained with 1% toluidine blue. Following the examination of toluidine blue-stained sections, ultrathin sections were cut and double stained with uranyl acetate and lead citrate. The grids were examined in a JEOL-1200 EX Biosystem electron microscope at 80 kV.
General Characteristics of the Smooth Muscle Layers
The muscularis externa of the body of the esophagus was composed of the circular and longitudinal muscle layers separated by space containing the myenteric plexus and capillaries. The circular muscle layer was separated into lamellae by connective tissue septa in continuity with the myenteric plexus region (Fig. 1A). Inside the muscle lamellae, smaller septa were present between muscle bundles. Many large and small nerve fibers were always present in these smaller septa (Fig. 1, A and B), emanating presumably from the myenteric plexus region. Smooth muscle cells in the circular muscle layer were interconnected by gap junctions (Fig. 1A), but no gap junctions were found between smooth muscle cells in the longitudinal muscle layer.
Both muscle layers were comprised of smooth and/or skeletal muscle fibers, with the ratio depending on the location within the esophagus (Figs. 1A and 2, A and B). The smooth muscle content of both muscle layers was highest at the LES. In the circular muscle layer up to 3 cm above the LES, all muscular lamellae were composed exclusively of smooth muscle cells. At 3 cm above the LES, small groups of skeletal muscle fibers were dispersed among smooth muscle cells in the same lamellae, running in the same direction as smooth muscle cells. At 5 cm proximal to the LES, skeletal muscle constituted approximately one-half of the circular muscle layer. At 8–10 cm proximal to the LES, randomly scattered small smooth muscle groups were seen only in a few muscular lamellae. In the longitudinal muscle layer, the most caudal 1-cm segment always contained a few skeletal muscle fibers. At 3 cm proximal to the LES, approximately one-half of longitudinal muscle was composed of skeletal muscle (Fig. 2B). No smooth muscle cells were found in the longitudinal muscle layer at 8 cm above the LES.
Interstitial cells of Cajal.
ICC had bipolar or irregular contours with slender, long ramifying cellular processes (Figs. 2, A and B, and 3, B and C). No detectable structural differences were noticed between ICC in different locations within the esophagus. ICC had structural features also found in fibroblasts and smooth muscle cells. The myoid features included presence of surface caveolae and a continuous basal lamina (Figs. 1B and 3B). ICC contained mainly thin filaments although intermediate filaments and microtubules were also present (Fig. 3A). The ICC cytoplasm contained numerous mitochondria, Golgi complexes, and paranuclear centrioles. Fibroblast-like characteristics of ICC included the presence of numerous ribosomes and rough endoplasmic reticulum; the smooth endoplasmic reticulum was not prominent (Fig. 3, A–C). ICC formed many small gap junction contacts with neighboring smooth muscle cells (Figs. 1B, 3C, and 4B). ICC were interconnected by close apposition contacts between their overlapping processes (Fig. 1B).
Although ICC were located within both muscle layers, they were most prominent in the circular muscle layer. In this layer, ICC were found throughout the esophagus up to 9 cm proximal to LES. Within the longitudinal muscle layer, ICC were observed up to 2–3 cm above the LES. Some ICC were observed in association with skeletal muscle fibers. ICC were seen as networks inside septa and in communication with smooth muscle cells of adjacent lamella, and always in close connection to nerve varicosities (Figs. 1B and 3A, B, and D). These ICC were, by definition, ICC-IM (44). In both muscle layers, ICC were encountered as single cell profiles (Figs. 2 and 3, B and C) or small bundles of overlapping processes (Figs. 1B and 3, A and D). In the circular muscle layer, most ICC were dispersed throughout the septa in contact with smooth muscle cells of the lamella lining the septa (Fig. 1, A and B). In addition to septal ICC, solitary ICC were found within the smooth muscle bundles (Fig. 3, C and D). At the plane of the myenteric plexus, scattered single ICC were found adjacent to either the circular or the longitudinal muscle layer (Fig. 2A). No apparent ICC network was revealed in the myenteric plexus region; only scattered ICC were observed close to neurons of myenteric ganglia. In contrast to ICC, fibroblasts were randomly distributed in both muscle layers and were present between smooth muscle cells and between skeletal muscle fibers, without forming any close contacts with neighboring cells.
ICC formed single or multiple close contacts with smooth muscle cells. Three types were noticed: close appositions (Figs. 1B and 4, A and B), small gap junctions (Figs. 3C and 4C), and peg-and-socket-like contacts, when smooth muscle cells extended cellular projections into ICC cytoplasm forming a close apposition (Fig. 3, B and C). Often, one innervated ICC connected to multiple smooth muscle cells within the same muscle lamella (Figs. 1A and 4C). Innervated ICC processes were also seen to connect smooth muscle cells from apposing lamellae across a small septum.
ICC were seen within the striated muscle portion of the esophagus. However, ICC were seen much less frequently compared with ICC in the smooth muscle portion of the esophagus. They were seen in close apposition to striated muscle; however, no special contacts (e.g., gap junctions) were observed.
ICC were found in extended close apposition contact (<20 nm) with nerve fibers (Figs. 1B and 3, A, B, and D). Frequently, a few axons and nerve varicosities devoid of enteric glial cell covering were nearly completely enclosed by ICC cytoplasm (Fig. 3, A and D). Most nerve profiles in close contact with ICC contained predominantly small agranular vesicles (40–60 nm in diameter), mixed with large granular vesicles (80–110 nm in diameter; Figs. 1A and 3, B and D). Special synapse-like junctions were found among the close contacts between ICC and enteric nerves. Ultrastructural examination revealed that these synapse-like junctions were featured with thickened prejunctional membrane on the neurolemma side, a 20- to 30-nm space between pre- and postjunctional membranes, and synaptic vesicles toward the junctional area (Figs. 3D and 4, C and D). No complete perineurium was seen around nerves associated with ICC (Figs. 1A and 3, A and B). There was an apparent correlation between the density of ICC-nerve groupings and smooth muscle content: the numbers of both innervated ICC and smooth muscle cells declined gradually from the LES upward in both muscle layers. Innervated ICC were still observed in the circular muscle layer up to 9 cm proximal to the LES. The nerves were not observed in synapse-like contact with other cell types, including smooth muscle cells.
Nodose ganglia injection.
In animals killed 1, 3, 5, and 28 days after bilateral saline injection in the nodose ganglia, injury to and disappearance of many ICC were observed in conjunction with degenerating nerve structures (Figs. 4A and 5C). Postinjection (1 day), mild injury to ICC was recognized by slight shrinkage of their cytoplasm, resulting in an increase of their electron density (Fig. 4A) or swollen subcellular organelles and lamella bodies within the cytoplasm (Fig. 4, B–D); 3–5 days postinjection, injury to ICC involved all of the cell body (Fig. 5C), and partial depletion of cytoplasmic organelles was more frequently observed (Fig. 5B). In animals 28 days postinjection, only small terminal ICC processes connected by gap junctions or close apposition contacts to normal-looking smooth muscle cells were revealed (Fig. 5A). Severe ICC injury was most often characterized by mitochondrial damage, with swelling of cellular processes that had lost most or all of their usual organelle content, including the filament system (Fig. 5, A and C). Only the presence of plasma membrane caveolae and close contacts with an adjacent smooth muscle cell identified empty cellular profiles as ICC processes (Fig. 5A). Injury to ICC in the striated portion of the lower esophagus was also observed after bilateral saline injection. Overall, at day 28 after injection, approximately one-third of ICC was affected.
Nerve degeneration usually started before the injury of ICC: nerve damage started 1 day postinjection when some of their associated ICC were still intact (Fig. 4D). All of the degenerating nerve terminals were featured with multiple mitochondria, predominant small agranular vesicles, and a few large granular vesicles within the axoplasm. Damaged axons were characterized by axonal swelling, rupture of mitochondria, the appearance of large/swollen agranular vesicles, and partial depletion of axon content (Figs. 4, A–D, and 5, A and B). Some slightly damaged ICC still kept synapse-like junctions with nerve varicosities and gap-junctional contacts with surrounding smooth muscle cells (Fig. 4, A–D). Injection into nodose ganglia did not result in complete denervation. The majority of myenteric plexus neural structures, as well as some free running intramuscular nerves, was unaffected (Fig. 6, A and B). Similar structural changes in the ICC-nerve groupings were observed in the distal esophagus when animals were killed 5 days after bilateral injection of the physiological solution into nodose ganglia. No structural changes of injury were seen in smooth muscle cells adjoining severely damaged ICC-nerve groupings (Figs. 1–7).
Vagotomy caudal to the nodose ganglion.
The distal esophagus of cats killed 3 days after vagotomy displayed similar structural changes in ICC as seen following nodose ganglia injection, indicating pronounced cellular injury. These changes included swelling of cellular processes, partial displacement of the filament system, and accumulation of large lipid droplets in their cytoplasm (Fig. 7B). Many ICC processes were still in close contact with adjoining smooth muscle cells of normal ultrastructure (Fig. 7A). ICC were closely associated with nerve varicosities in various states of degeneration (Fig. 7, A and B). As with injection in the nodose ganglia, vagotomy did not result in complete denervation of the distal esophagus. Bundles of unaffected nerves were always noticed in both the muscle layers and the myenteric plexus region (data not shown). Similar to saline injection of the nodose ganglia, vagotomy resulted in injury to ICC in striated muscle.
The cat esophagus has been used extensively as a model for the human esophagus (6, 30, 32, 33) since it is one of the few animal models with a large smooth muscle component in the esophageal musculature (14). The present study is the first electron microscopic description of feline esophageal ICC. The dominant ICC subtype in the cat esophagus was ICC-IM, which was found dispersed throughout the muscle layers, similar to human esophagus (17). ICC-IM were almost always encountered sandwiched between a nerve varicosity and several smooth muscle cells. There were gap junction contacts with smooth muscle cells and specialized synapse-like structures with nerve varicosities. Synapse-like junctions were not observed between nerve varicosities and smooth muscle cells, with the distance between nerve and muscle usually 100 nm or greater. Hence the ICC-IM were preferentially innervated.
Faussone Pellegrini and Cortesini (17) provided an important assessment of ICC in the human esophagus and suggested that a muscle bundle should be considered a functional unit with ICC mediating innervation to smooth muscle. Furthermore, they proposed that esophageal ICC have spontaneous activity that stimulates contraction of smooth muscle cells, based on ICC being most abundant in areas where myogenic contractile activity is prominent. Because esophageal smooth musculature does not normally show spontaneous electrical and mechanical activity, we propose that the correlation between the presence of ICC and muscle activity may be primarily related to ICC acting as mechanoreceptors and conveying sensory information on the contractile activity via the vagus to the central nervous system (CNS). The present study shows that degeneration of vagal afferent nerves by careful injection of saline in the center of the nodose ganglion concurrently causes degeneration of associated ICC. This suggests that vagal afferent nerves may provide a trophic factor for ICC and also suggests an important functional relationship. Consistent with this is the degeneration of ICC after bilateral vagotomy as shown in the cat esophagus in the present study and in the monkey esophagus by Wong and coworkers (52). Interestingly, despite gap junction contacts between degenerating ICC and smooth muscle cells, the latter remained undamaged in many of the animals, suggesting that the trophic influence is not essential for smooth muscle cells. It is likely that the ICC are associated with vagal afferent sensory fibers since injection in the nodose ganglion labeled these neural structures and not motor fibers (10). Furthermore, Fox and coworkers (20, 49) showed an intimate relationship between IMAs and ICC-IM in the fundus. In addition, in the c-Kit mutant mouse where the number of ICC-IM in the fundus is markedly reduced, IMAs were also distinctly reduced (21). This report together with our present study suggests that vagal afferents and ICC are dependent on each other for their survival and maintenance. The majority of the nerve endings in the musculature were affected by nodose ganglion injection, and, indeed, the majority of vagal fibers are afferent (1). Functional support of ICC-IM mediating vagal afferent activity came from a study by Liu and coworkers (26) who observed that vagal traffic to the CNS upon stomach distention was reduced in Ws/Ws rats that lack ICC-IM. With their location in the muscle of the esophagus, ICC-IM may act in series with the IMAs, responding to stretch or contraction within the esophagus to serve either motor or cognitive sensation functions. This is supported by the prominence of peg-and-socket junctions formed between ICC-IM and smooth muscle cells in the cat esophagus. The ICC-IM received protruding pegs from smooth muscle cells, but ICC-IM formed sockets. Strong evidence has recently emerged that the peg-and-socket junctions of the ICC deep muscular plexus of the small intestine are associated with stretch sensing (45). The marked loss of vagal afferents and ICC after nodose ganglion injection or vagotomy in the cat may be responsible for the temporary paralysis of peristalsis in the smooth muscle portion of cat esophagus, as demonstrated by many early studies (7, 27).
The dependence of ICC on neuronal survival may be restricted to sensory nerves, since pacemaker ICC Auerbach's plexus of the mouse (53) and human (22) small intestine are normal in the absence of enteric motor neurons. The mechanism may be similar to that of transneuronal degeneration proposed to be due to failure of one neuron to transport a trophic agent to another (39). The ligand of the c-Kit receptor, steel factor, can be found in nerves (47), but this cannot be confirmed in all tissues or species (48). The varicosities of the vagal afferent nerves contain small agranular vesicles and large granular vesicles and are not distinct from varicosities of motor neurons. Neurotransmitters of vagal afferent nerve terminals may be involved in the maintenance of the target cell, as demonstrated for the β-cell of the pancreas (31), and/or may have a modulatory function on mechanosensation (3).
No ICC-AP, not all ICC-IM, and not all nerve varicosities were degenerating, suggesting that only a subpopulation of ICC-IM are involved in vagal afferent innervation. Undamaged ICC associated with undamaged nerve varicosities could represent ICC associated with motor nerve endings since ICC-IM are also thought to be involved in efferent innervation. Examination of the structural relationship between ICC and nerve varicosities in the opossum esophagus (12) and mouse LES (50) suggested that ICC may be intermediary cells picking up signals from nonadrenergic noncholinergic nerves and transmitting that information to the smooth muscle cells. There is little doubt that ICC are innervated by motor neurons (43, 50, 51), but direct innervation to smooth muscle cells occurs as well (2, 13, 15, 42). Because the nodose ganglion injection does not necessarily affect all afferent nerves, it is not possible to discriminate between undamaged sensory nerves and motor nerves.
In the opossum esophagus (9, 12), no ICC were found in the longitudinal muscle, whereas ICC are frequent in the longitudinal muscle of the human and cat esophagus, making the latter a suitable model for the human esophagus. In the present study and in the human (16), ICC were found in the striated part of the esophagus. Rumessen et al. (37) proposed a role for ICC as proprioceptors, sensing movement of the striated muscle cells.
In addition to IMAs, IGLEs are mechanoreceptors associated with the myenteric ganglia of the cat esophagus (34). IGLEs are not physically associated with ICC (20, 29), and, in W/Wv mutant mice lacking ICC-IM, IGLEs were not affected (21). It is possible that IGLEs have a particularly important function in the striated esophagus (54).
The hypothesis that ICC functions as stretch receptor needs to be confirmed with functional data. This will be difficult, especially in cats where lesions to specific ICC without neural damage are impossible to achieve. Trying to confirm the results functionally with activation of reflexes some time after vagal section or after damage to afferent nerve pathways in the nodose ganglion is also very difficult for other reasons. There is remarkable redundancy of mechanisms and pathways responsible for the reflexes. This fact is particularly so for the induction of peristalsis by esophageal stimulation and for many of the esophago-upper esophageal sphincter reflexes. Both primary and esophageal distension induced peristalsis returns to the smooth muscle esophagus after vagotomy in as short a time as several hours and frequently within a day or so in a number of species, including cat, dog, and nonhuman primate (27, 33, 36, 46). This returned peristalsis, called “tertiary peristalsis” is a function of one or another peripheral neural and/or myogenic (30) mechanism that becomes operative. Peristalsis can even return to the upper striated muscle esophagus in the dog and cat after cervical vagotomy because of dual innervation of the region from both the recurrent laryngeal nerves below and a pharyngoesophageal nerve above (23). In preliminary studies, we did esophageal motility studies on two of the cats before and after bilateral vagotomy at 1 day after in one and at 1 mo after in the other. Both cats showed the presence of swallow and balloon-induced peristalsis with normal peristaltic velocity. We did not assess any of the upper esophageal sphincter reflexes. Therefore, functional studies to establish the validity of the hypothesis will need to be highly complex, probably done on several species, and use acute and chronic techniques to take into account the redundancy that allows auxiliary mechanisms to take over function. We see the present study as providing solid evidence to proceed along these lines in future experiments.
The present study establishes synapse-like junctional communication between vagal afferents and ICC in the cat esophagus. This is consistent with a recent study in the rat fundus (29). The communication between vagal afferents and ICC is likely to be bidirectional if ICC is to provide information on smooth muscle function to the nerve varicosity. It is possible that ICC release membrane-permeable mediators (e.g., nitric oxide, carbon monoxide, or lipid-derived signaling molecules) that diffuse to the vagal afferents. Bidirectional coupling involving synapses has been documented (18).
In conclusion, ICC in the cat esophagus were predominantly of the intramuscular subtype (ICC-IM). They were found to be gap junctionally coupled to smooth muscle cells and to nerve varicosities by synapse-like junctions. Damage to vagal afferent fibers caused various stages of cell injury to ICC-IM, suggesting that ICC-IM in the esophagus are dependent on afferent vagal innervation for survival. The position of ICC makes them possibly involved in transduction of mechanical stimuli to vagal afferent nerves. Future studies are required to determine the exact role of ICC-IM in sensory transduction and regulation of esophageal peristalsis.
This study was supported by operating grants from the Canadian Institutes of Health Research to J. D. Huizinga and N. E. Diamant and from the Canadian Association of Gastroenterology to L. W. C. Liu.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society