GLIA 4:195-204 (1991)

Enteric Glia MICHAEL D.GERSHON AND TAUBE P. ROTHMAN Department of Anatomy and Cell Biology, Columbia University College of Physicians and Surgeons, New York, New York 10032

KEY WORDS

Gut, Enteric nervous system, Myenteric plexus, Submucosal plexus, Glial development, Differentiation

ABSTRACT The structure of the enteric nervous system (ENS) is different from that of extraenteric peripheral nerve. Collagen is excluded from the enteric plexuses and support for neuronal elements is provided by astrocyte-like enteric glial cells. Enteric glia differ from Schwann cells in that they do not form basal laminae and they ensheath axons, not individually, but in groups. Although enteric glia are rich in the S-100 and glial fibrillary acidic proteins, it has been difficult to find a single chemical marker that distinguishes enteric glia from non-myelinating Schwann cells. Nevertheless, two monoclonal antibodies have been obtained that recognize antigens that are expressed on Schwann cells (Ran-1 in rats and SMP in avians) but not enteric glia. Functional differences between enteric glia and non-myelinating Schwann cells, including responses to gliotoxins and in vitro proliferative rates, have also been observed. Developmentally, enteric glia, like Schwann cells, are derived from the neural crest. In both mammals and birds the precursors of the ENS appear to migrate to the bowel from sacral as well as vagal levels of the crest. These crest-derived emigres give rise to both enteric glia and neurons; however, analyses of the ontogeny of the enteric innervation in a mutant mouse (the Zslls), in which the original colonizing waves of crest-derived precursor cells are unable to invade the terminal colon, suggest that enteric glia can also arise from Schwann cells that enter the gut with the extrinsic innervation. When induced to leave back-transplanted segments of avian bowel, enteric crest-derived cells migrate into peripheral nerves and form Schwann cells. Enteric glia and Schwann cells thus appear to be different cell types, but ones that derive from lineages that diverge relatively late in ontogeny. UNIQUE PROPERTIES OF THE ENTERIC NERVOUS SYSTEM The enteric nervous system (ENS)differs functionally and structurally from any other region of the peripheral nervous system (PNS) (Furness and Costa, 1987; Gabella, 1987; Gershon, 1981,1987). With respect to function, the ENS is the only division of the PNS that is capable of mediating reflex activity in the absence of input from the brain and/or spinal cord (Trendelenburg, 1917). The efferent input to the bowel from the vagus nerves, for example, is remarkably sparse (Kirchgessner and Gershon, 1989). Experiments in which an anterograde tracer was injected bilaterally into the dorsal motor nuclei of the vagus have revealed that most enteric ganglia are not even contacted by vagal preganglionic fibers and, in those ganglia that do receive a vagal input, only a minority of neurons receive vagal synapses. Moreover, the submucosal plexus appears not to be directly innervated by vagal efferent fibers at all; 01991 Wiley-Liss, Inc.

therefore, the secretomotor neurons that are located in this plexus (Bornstein and Furness, 1988) must all be innervated via an intermediary synapse in the myenteric plexus. These considerations have given rise to the concept that much of the behavior of the bowel is regulated by the activity of local microcircuits located within the ENS (Wood, 1987). Central effects on gastrointestinal neural function, according to this hypothesis, are the result of activation of specialized “command neurons” that regulate these microcircuits. Also present in the ENS, and making the local mediation of reflexes possible, are intrinsic primary afferent neurons (Biilbring and Crema, 1958; Kosterlitz and Lees, 1964). These intrinsic sensory neurons have yet to be identified with certainty; however, neurons have been found

Address reprint requests to Taube P. Rothman, Department of Anatomy and CeLl Biology, Columbia University College of Physicians and Surgeons, 630 West 168th Street, New York, NY 10032.

196

GERSHON AND ROTHMAN

in the ENS that share differentiation antigens and enzymatic markers with dorsal root ganglia (Kirchgessner et al., 1988). The autonomy of the ENS has been recognized as conferring upon it “brain-like”attributes (Wood, 1987). Perhaps not surprisingly,therefore, the structure of the ENS resembles that of the brain. The usual coats of connective tissue that surround nerve cell bodies and s o n s in most peripheral nerves and ganglia (Thomas and Ochoa, 1984) are lacking in the ENS (Cook and Burnstock, 1976a, b; Gabella, 1972a, b; Gabella, 1987). In fact, both fibrillar collagen and basal laminae are entirely excluded from the interiors both of enteric ganglia and interganglionicconnectives.As a result, the ultrastructure of the ENS is more similar to that of the central nervous system (CNS) than it is to the remainder of the PNS (Komuro et al., 1982).This basic structural resemblance of the ENS to the brain and spinal cord implies that support for the neural components of the ENS is likely to be provided by cells analogous to the glia that subserve this role in the CNS. Observations made during the last two decades have indeed provided support for the hypothesis, derived initially from ultrastructural observations (Cook and Burnstock, 1976b; Gabella, 1971),that the ENS contains specialized supporting cells that are different from either Schwann cells or satellite cells and which are similar to astrocytes. Morphology of Enteric Glia Enteric glia differ markedly from Schwann cells in their morphology (Cook and Burnstock, 1976b;Gabella, 1971; Gabella, 1987).A major difference between the two is that the shape of the enteric glial cell is far more irregular (Fig. 1).In the myenteric plexus, long processes radiate out from a centrally located glial perikaryon to terminate in small swellings (“end feet”) that abut on the basal lamina that surrounds ganglia and interganglionic connectives (Erde et al., 1985). These end feet form a glial sheath, analogous to the pial-glial interface of the CNS; however, in contrast to the pial-glial boundary, the enteric glial sheath is incomplete and only partially separates the neurons of the myenteric plexus from the extraganglionic connective tissue. Both axonal varicosities and nerve cell bodies extend through holes in the glial wrapping. As a result, there are regions in the ENS where neural structures come into contact with the periganglionic basal lamina without an intervening glial element (Gabella, 1972a). In addition to their role in establishing the perimeter of the myenteric plexus, enteric glial processes also partition myenteric ganglia into compartments (see Fig. 1). Within these compartments axons are enveloped in bundles by the glial processes, not individually as in Schwann cells (Fig.2).Moreover,enteric glia appear not to synthesize a basal lamina. As a result, while all of the surfaces of Schwann cells become covered with laminin both in vitro and in vivo (Cornbrooks et al., 1983),the surfaces of enteric glia are mostly devoid of laminin and

Fig. 1. An enteric glial cell impaled with a microelectrode and injected with horseradish peroxidase. The cell was found to be electncally inexcitable.Notice that there is a central soma that ’ves rise to a complex astrocyte-like arbor of processes. When in the $an, of focus these processes can be seen to terminate in feathery expansions or “end feet”(-+). In electron micro aphs (not illustrated) these processescan be seen to abut on the basafiamina that surrounds the ganglion (dark region) in which the injected cell is located. Axon bundles occupy the compartments (*) between the glial processes. The marker = 50 pm.

make contact with that protein only where their end feet underlie the periganglionic basal lamina (Bannerman et al., 1986). The internal structure of enteric glia, as well as their shape, is different from that of Schwann cells. The most striking difference is the high concentration and dense packing of 10 nm intermediate filaments (“gliofilaments”)in most enteric glial cells (Cook and Burnstock, 1976b; Gabella, 1987; Komuro et al., 1982).As a consequence of this concentration of gliofilaments enteric glial cells are very rich in glial fibrillary acidic protein (GFAP) (Jessen and Mirsky, 1980),the protein associated with the intermediate filaments of glia (Anderton, 1981). GFAP, however, is not an absolutely specific marker for enteric glia in the PNS. The protein is also expressed by the Schwann cells of unmyelinated, but not myelinated extra-enteric nerves (Jessen et al., 1984; Yen and Fields, 1981). In fact, it has proved to be very difficult to find a single chemical marker that distinguishes (qualitatively rather than quantitatively) enteric glial cells from nonmyelinating Schwann cells. Glial Markers Enteric glia resemble astrocytes in the abundance of their immunoreactive GFAF’, as well as in their expres-

ENTERIC GLIA

Fig. 2. An electron micrograph showinga portion of a anglion of the myentericplexus ofa mouse.Aglial soma (GI and its gliafprocesses(+I can be identified by their dense packing with intermediate filaments (“gliofilaments,” 10 nm in diameter). Notice that the glia processes divide the an lion into compartments in which s o n s are grouped, but not indivi8ua8y ensheathed. Glial end feet abut on the basement

197

membrane (BM) that se arates the ganglion from the adjoininglayer of connectivetissue cells ($1 and smooth muscle (SM). The glial end feet do not prevent a varicose axon terminal (A) or even the perikaryon of a neuron (N) from directly contacting the periganglionic connective tissue space. The marker = 1 pm.

198

GERSHON AND ROTHMAN

sion of a variety of additional markers. These include glutamine synthetase, Ran-2 (Jessen and Mirsky, 1983), the lipid transport protein, apolipoprotein E (Boyles et al., 19851, and the Ca2+ binding protein, S-100 (Bishop et al., 1985; Ferri et al., 1982;Kobayashi et al., 1986; Scheuermann et al., 1989). Nevertheless, each of these antigens may also be expressed on nonmyelinating Schwann cells as well as by the glia. Probably the most common marker used in studies of enteric glia has been S-100.This protein does not distinguishenteric glia from extra-enteric non-myelinating Schwann cells any better than any other marker, but it has been very useful in distinguishingglial from neural elements within the enteric plexuses. The S-100 protein has also been employed as a marker to recognize glial cells in cultures derived from the ENS (Bannerman et al., 1988a).Enteric neurons and glia can be separated from one another and each survive well in relatively pure culture (Bannerman et al., 1988b).These cultures have been studied as a means of gaining insight into the special properties of enteric glial cells; however, many characteristics of enteric glia (such as expression of Ran-1)have been found t o change radically from the in situ condition when these cells are grown in vitro (Jessen and Mirsky, 1983). The relevance of observations made on enteric glia in culture to the in situ properties of these cells, therefore, must remain in doubt until verified by in vivo experiments. For example, all of the surfaces of cultured enteric glia (S-100immunoreactive cells) become covered with laminin immunoreactivity when these cells are grown in vitro (Bannerman et al., 1988a); nevertheless, as noted above, glial cells in situ are only covered with laminin where they contact the periganglionic basal lamina (Bannermanet al., 1986). Distinction Between Enteric Glia and NonMyelinating Schwann Cells The similarities between enteric glia and non-myelinating Schwann cells, in chemical properties if not morphology, raises the issue as to whether they are different cell types or a single non-myelinating Schwann cell type that displays a different appearance in different locations. A number of recent observations indicate that enteric glia are fundamentally different from extra-entericSchwann cells. Functional, chemical, and developmental differences have been found. Functional properties in which enteric glia differ from Schwann cells include their reaction to the glio-toxin, 6-aminonicotinamide (6-AN) (Aikawa and Suzuki, 1985,1986). This compound is an antagonist of niacin that causes degeneration of glial cells in the CNS. When administered to newborn or suckling mice, 6-AN causes lesions to appear in enteric glia, but not in other peripheral nerves. Another functional difference between enteric glia and extra-enteric non-myelinating Schwann cells is that when examined in vitro, the glia proliferate at a rate that is significantly higher than that of the

Schwann cells (Eccleston et al., 1987).This difference is striking because perturbations of culture conditions that affect the rate of cell proliferation affect enteric glia and extra-enteric non-myelinating Schwann cells similarly. For example, the rate of DNA synthesis in each of these cells is inhibited by collagen type I and stimulated by laminin, fibronectin, type IV collagen, and matrix secreted by bovine corneal endothelial cells (Eccleston et al., 1989b). Proliferation of both enteric glia and extra-enteric non-myelinating Schwann cells is also negatively affected by co-culturing either of them with enteric neurons (Eccleston et al., 1989a). In contrast, DNA synthesis by each of these cell types is stimulated when they are exposed to bovine axolemmal fractions or neurites from dorsal root ganglion cells. The fact that enteric glia and extra-enteric non-myelinating Schwann cells respond in the same way to a variety of stimuli suggests that the major difference between them in their basal rate of proliferation when grown under similar conditions reflects an intrinsic difference between the two cell types. Chemical differences have recently been reported between the enteric glia and extra-enteric Schwann cells (myelinatingand non-myelinating)of both avians and mammals. The avian Schwann cells express a surface protein (a doublet of M, 75-80 kDa) that can be detected with a monoclonal antibody called SMP (Schwann cell myelin protein) (Dulac et al., 1988).This antigen is not expressed by enteric glia and thus clearly distinguishes these cells from extra-enteric non-myelinating Schwann cells, which are SMP-immunoreactive. In the avian CNS SMP is expressed by oligodendrocytes (Cameron-Curryet al., 1989),whereas enteric glia resemble astrocytes.A similar situation has been found to exist in rats. The marker, Ran-1, is expressed on extraenteric Schwann cells and is not seen on the surfaces of enteric glia in situ (Jessen and Mirsky, 1983). Ran-1, however, is expressed by enteric glia when these cells are grown in vitro. Thus, SMP in avians and Ran-1 in rats represent markers expressed in situ by Schwann cells, but not by enteric glia. Whether avian enteric glia in long term tissue culture would acquire the ability to express the SMP antigen, as rat enteric glia do Ran-1, remains to be determined. Although two markers have now been identified, which are expressed on extraenteric non-myelinating Schwann cells but not enteric glia, no markers have yet been found that are uniquely expressed on enteric glia, which would enable the glial cells to be positively identified. One possible marker, yet to be adequately tested, is plasmalemmal 5'-nucleotidase activity (Andersson-Forsman and Gustafsson, 1985). This enzyme is found on the surfaces of smooth muscle and enteric glia cells. In glia 5'-nucleotidase activity is concentrated at specialized neural glial junctions. It has been speculated that the presence of the enzyme reflects the release of adenine nucleotides as a result of purinergic neural activity within enteric ganglia. Expression of 5'-nucleotidase activity by extraenteric non-myelinating Schwann cells remains to be studied.

ENTERIC GLIA

Development of the Enteric Nervous System The ENS has been demonstrated in avians to be entirely derived from the neural crest (Le Douarin, 1982). Deletions of the vagal crest cause a failure of ganglion formation in the bowel (Yntema and Hammond, 1954) and work with quail-chick interspecies chimeras has confirmed that crest-derived cells migrate to the gut from the vagal and sacral regions of the neuraxis (Le Douarin and Teillet, 1973). Experiments with chimeras have revealed that both neurons and supporting cells in the avian enteric plexuses are crest derived. No direct evidence of the type that exists in birds has been obtained in mammals. Since no means have been available to identify individual crest-derived cells migrating to the bowel in mammalian embryos, research on the colonization of the mammalian gut has had to rely on surrogate markers, presumptions drawn from analogies to avian systems, and deduction. For example, although crest-derived cells cannot be recognized as such when they first reach the mammalian gut, their arrival can nevertheless be accurately timed if one makes the assumption that the only cells in the bowel that can form neurons or glia are those that are crest derived. This assumption underlies the use of an indirect explant culture assay to detect the presence in the gut of crest-derived cells (Rothman and Gershon, 1982; Rothman et al., 1984). This assay was employed in the investigation of the development of enteric neurons. When segments of mouse foregut were explanted prior to day E9, the explants gave rise t o aneuronal cultures ofbowel (Jacobs-Cohen et al., 1987). On the other hand, ganglia were found to develop in vitro if the gut was explanted at day E9 or later (Rothman and Gershon, 1982; Jacobs-Cohen et al., 1987). This technique thus establishes that the murine foregut has been colonized by crest-derived cells (neuronal precursors) as early as day E9 of development. Similar experiments on the developing rat have established that crest-derived cells are present in the fetal rat foregut as early as day El0 (Baetge et al., 1990b).These types of observations have also been made on the development of enteric glia. Glial cells cannot be recognized until rather late in the development of the bowel in situ and are not seen as distinct entities until after enteric neurons are morphologically recognizable. This occurs on day E l 2 in the mouse intestine (Rothman and Gershon, 1982,1984) and day E25 in the guinea pig bowel (Gershon et al., 1981). In contrast, in situ expression of GFAP cannot be detected until day El6 in the mouse (Rothman et al., 1986).As in the case of neurons, however, glia develop in cultures of murine gut explanted as early as day E l 0 (Rothman et al., 1986). The glial cells that develop in these cultures moreover appear to be the same type of glial cell as is seen in the bowel in situ. The cells developing in vitro are rich in GFAF' and contain bundles of 10 nm gliofilaments. Interestingly, the glia also form an incomplete sheath around the developing neurons in cultures of bowel and, as occurs in situ, collagen (both basal laminae and fibrillar collagen) are excluded from the extra-

199

cellular space internal to the glia sheath. These experiments thus suggest that there is a single date on which glial and neural precursors become present in the mammalian bowel. The fact that such a date exists is compatible with the idea that in mammals, as in avians, enteric glia and neurons are each derived from a single wave of migrants from the neural crest. Recent observations have suggested that some of the crest-derived cells that migrate to the gut in mammals (although not in birds) transiently express a catecholaminergic phenotype (TC cells) while in the process of migrating to and colonizing the bowel (Baetge and Gershon, 1989; Baetge et al., 1990a, 1990b). If so, the catecholaminergic phenotype would provide a marker that would permit the pathway followed by at least a subset of the crest-derived cells that colonize the mammalian gut to be traced. TC cells appear in the foregut of mice (E9.5) and rats (E10)just after crest cells enter the bowel (see above) and disappear in each species (in mice by day E l 3 and in rats by day El51 at about the time the extrinsic sympathetic nerves grow into the gut (Baetge and Gershon, 1989; Cochard et al., 1978; Gershon et al., 1984; Jonakait et al., 1979; Teitelman et al., 1981; Teitelman et al., 1978). In both mice and rats TC cells express neural markers, including neurofilaments, peripherin, growth-associated protein (GAP)-43,microtubule-associated protein (MAP)-2,and MAP-5, and they are neurogenic (Baetge et al., 1990a, b). Because TC cells are neurogenic, they can be presumed to be derived from neurectoderm and almost certainly, therefore, to be of crest origin. TC cells are also proliferating (Baetge et al., 1990a; Teitelman et al., 1981);consequently, they should not be considered to be mature neurons, which are post-mitotic (Fujita, 19641, but neural precursors. TC cells seem to disappear because they lose their catecholaminergic phenotype; however, it is now clear that the cells do not die. The progeny of TC cells have been followed in rats where in situ hybridization has revealed that TC cells retain the catecholamine biosynthetic enzyme, tyrosine hydroxylase (TH), for a considerable period of time after TH mRNA is no longer produced (Baetge et al., 1990a; Jonakait et al., 1989). This property led to the discovery that several of the properties of TC cells, including dopamine-p-hydroxylase, are retained by mature enteric neurons, thereby revealing that many of these are descended from fetal TC cells. It has thus been proposed that TC cells are the precursors of neurons of the adult ENS that lose their catecholaminergic phenotype when they follow their definitive program of differentiation. TC cells, moreover, are found not only in the intestine but also along a pathway that eventually is also followed by the descending fibers of the vagus nerves (Baetge and Gershon, 1989). At first, the TC cells are located distal to the advancing front of vagal fibers, but at later ages they appear to be overtaken by these nerves and come to lie among the growing vagus nerve fibers. This pathway, in which TC cells can be detected, corresponds to that shown in avians to the route followed by crest-derived cells migrating from the vagal crest to the bowel (Pay-

200

GERSHON AND ROTHMAN

ette et al., 1984; Tucker et al., 1986). If the vagus nerves are explanted at a time when they contain TC cells, neurogenic and gliogeniccells leave the nerves and form ganglia in vitro (Baetge et al., 1990b). The distribution and fate of TC cells thus supports the concept that the mammalian gut, like the avian, is colonized by cells that migrate to the bowel from the vagal region of the neural crest. At the time in the development of the rat gut when TC cells are present in large numbers, are still proliferating, and have not yet begun to terminally differentiate (day E12), many cells in the enteric mesenchyme express nerve growth factor (NGF) receptors (Baetge et al., 1990a).These receptors can be detected immunocytochemically using a monoclonal antibody (192-IgG) (Chandler et al., 1984) to the rat NGF receptor. NGF receptor mRNA can also be detected in these cells by in situ hybridization (Baetge et al., 1990a). In contrast to the catechoIaminergic neurons of developing sympathetic and dorsal root ganglia in the same embryos, the enteric cells that express NGF receptors do not bind detectable amounts of 1251-NGFwhen incubated with a low concentration of that growth factor in vitro. The NGF receptors on the enteric cells, therefore, are all low-affinity NGF receptors. Simultaneous demonstration of catecholamine-related markers reveals that all of the TC cells present in the gut at day El2 express NGF receptors. These experiments also reveal, however, that additional cells are also present that contain NGF receptors but which do not express either catecholaminerelated markers or neurofilaments. Low-affinity NGF receptors (both the immunocytochemically detectable protein and mRNA detectable by in situ hybridization) persist in the bowel and, in the adult rat ENS, are expressed both by enteric neurons and glia. Low-affinity NGF receptors are known t o be present on nonneuronal supporting cells from a variety of peripheral nerves and ganglia (Carbonetto and Stach, 1982; Rohrer, 1985; Zimmermann and Sutter, 1983). It thus seems likely that TC cells represent only a subset of the vagal crest-derived cell population that migrates to the gut. Because the TC cells co-express neuronal markers, they probably represent a fraction of the total crestderived population that is committed to developing along a neuronal lineage. The additional non-neuronal cells, which express low-affinity NGF receptors, might represent another subset of crest-derived cells committed to developingas glia. The alternative possibility that they are uncommitted precursors capable of giving rise either to neurons or glia has not been excluded. The late appearance of GFAP in the developing ENS makes it difficult to distinguish between these possibilities (Rothman et al., 1986).An early appearing glia marker is needed. Neural markers can be detected in the post-umbilical gut of the rat before the descending front of apparently vagally derived TC cells (or any cells that express a neural marker) reaches the umbilicus (Baetge et al., 1990a). These post-umbilical cells are especially well demonstrated by their intense expression of GAP-43

immunoreactivity. Since the GAP-43 antigen has recently been shown to be expressed by glia as well as by neurons (Deloulme et al., 1990), it cannot be concluded that these ascending cells in the post-umbilical bowel are committed to developing as neurons; nevertheless, the observation indicates that the post-umbilical gut is colonized by crest-derived cells before cells from the vagal region of the neuraxis enter this portion of the bowel. This observation thus supports the hypothesis that the post-umbilical bowel of mammals, like that of avians, is colonized by cells from the sacral as well from the vagal crest. The observation also explains why indirect culture assays reveal that the mammalian hindgut is colonized by crest-derived neural and glial precursors almost as early in development as is the foregut (Jacobs-Cohen et al., 1987; Rothman and Gershon, 1982).

Origins of Enteric GIia Two possible origins can be envisioned for enteric glia. Each of these assumes a derivation from the neural crest. Enteric glia might develop from precursor cells that migrate to the bowel within the original populations of vagal and sacral crest-derived cells that also give rise to enteric neurons. Alternatively, enteric glial cell precursors could enter the gut later in development and reach the bowel with the extrinsic nerves. The two possibilities are not mutually exclusive. A supplementary supply of glial precursors could be added by the extrinsic innervation to an initial population of crestderived emigres that arrives with the neural precursors. Insight into the origin of enteric glia has been obtained from an investigation of their development in the bowel of lethal spotted (Zslls) mutant mice (Rothman et al., 1986). In these animals the terminal colon is aganglionic because it is not colonized by the crest-derived precursors of enteric neurons (Rothman and Gershon, 1984).Nevertheless, the aganglionic region of the gut is not denervated. In adult Zslls animals extrinsic nerves grow into the aganglionic segments of bowel as do nerve fibers from ganglia in the more proximal ganglionated portion of the gut (Payette et al., 1987).The nerve fibers in the aganglionic zone exhibit striking GFAP immunoreactivity, although the ultrastuctural appearance of the cells supporting nerves in this tissue is that of extra-enteric peripheral nerve and not enteric glia (Rothman et al., 1986). Axons of nerves in the aganglionic zone are individually enveloped by supporting cells (one axodmesaxon) and each supporting cell is completely encircled by a basal lamina; moreover, collagen is present around the nerve fibers in well-defined endoneurial coats. The lslls defect has been demonstrated to be due not to an abnormality of the crest-derived neural precursor cells but to an inability of the presumptive aganglionic tissue to support their inward migration (Jacobs-Cohen et al., 1987).If the defect in the presumptive aganglionic zone were to be neuron-specific, crestderived cells committed to a glial lineage might be able to colonize the abnormal segments of colon, while those

ENTERIC GLIA

able to differentiate as neurons are excluded. Under those circumstances, the GFAP-immunoreactive supporting cells in the aganglionic bowel could be descended from a glia-committed subset of the vagal or sacral crest-derived precursors that colonize the gut. Alternatively, if the effect of the lslls lesion extends to the entire population of colonizing crest-derived precursors, the GFAP-immunoreactive cells could all be Schwann cells that enter the aganglionic tissue with the extrinsic innervation. This possibility would envision a difference in migratory properties between crest derived precursor cells and Schwann cells. In order to determine whether the colonizing wave of crest-derived cells that migrates down the intestine does contain the precursors of enteric glia, segments of fetal bowel were explanted prior to the innervation of the gut by extrinsic nerves and prior to the in situ appearance of intrinsic neurons or GFAP-immunoreactive elements (Rothman et al., 1986). These explants of gut were grown in vitro. Cells that are not present in the original explants cannot enter the bowel in the cultures; therefore, the in vitro development of glia under these conditions establishes that glial precursors are already present in the gut at the gestational age at which the explants are made. GFAP-immunoreactive cells were found to develop in all explants (removed from the animals after day E10) of fetal bowel from control mice. It can thus be concluded that cells with the capacity to give rise to glia are present in the wall of the gut before extrinsic nerves reach the bowel. These observations thus establish that enteric glia can be derived from the original wave of crest-derived cells that colonizes the intestine. Extrinsic nerves cannot be the sole source of such cells, although the possibility that they contribute additional precursors to the enteric gliogenic population is not excluded. In contrast to explants derived from control mice, most explants of presumptive aganglionic bowel from lslls animals were observed to give rise to cultures that contained neither glia nor neurons. For the most part, therefore, the presumptive aganglionic segments oflslls gut are as devoid of glial precursors as they are of cells able to give rise to neurons. The supporting cells in the aganglionic bowel in mature lslls mice, therefore, must be predominantly derived from Schwann cells entering these segments of the bowel with the extrinsic nerves. The Schwann-like morphology of these cells (see above), differing as it does from that of enteric glia, is certainly consistent with this conclusion. These experiments on control and lslls mice support the view that the glial components of the ENS may be derived from separate lineages of cells. One is the wave of crest-derived precursors that colonizes the gut early in development. The other is the Schwann cell population that reaches the bowel much later in ontogeny along with the ingrowing extrinsic nerves. It is possible that the supporting cells present in the aganglionic bowel of lslls mice look like Schwann cells rather than enteric glia because Schwann cells are a distinct cell type that is unable to assume an enteric glial morphology. Still to be ruled out is the alternative

201

possibility that Schwann cells change their phenotype to that of enteric glia when exposed to the influence of a normal enteric microenvironment but do not do so when exposed to the abnormal microenvironment of the aganglionic bowel of lslls mice. In contrast to the presumptive aganglionic segments of lslls bowel, the more proximal lslls gut, in which neurons do develop, invariably provides explants containing GFAP-immunoreactive cells (Rothman et al., 1986). Glial precursors thus colonize the relatively normal regions of the lslls bowel about as early as do neural precursors. In fact, the timing of each is indistinguishable in the mutant animals from that in controls. Nevertheless, although glial precursors, like their neural counterparts, are usually excluded from the presumptive aganglionic bowel, islands of GFAP-immunoreactive cells do develop in rare cultures of presumptive aganglionic lslls bowel in the absence of neurons or extrinsic nerves. This phenomenon suggests that some glial precursors may enter the aganglionic tissue even when neural precursors do not. Neurons thus are not required for the development of enteric glia. A similar conclusion can be drawn from clonal analyses of the development of crest-derived cells in vitro. Again, both Schwann and non-Schwann glial cells have been found to arise in the absence of neurons (Dupin et al., 1990). The observation (Rothman et al., 1986) that the behavior of neural and glial precursors with respect to the aganglionicbowel of lslls mice can be dissociated (even if only rarely) is consistent with the hypothesis that the precursor population in the colonizing wave of crestderived cells moving down the bowel is already separated into glial and neuronal lineages. Potentialities of the Initial Population of Crest-Derived Cells That Colonizes the Bowel Insight into the capacity of the crest-derived cells that colonize the bowel to give rise to Schwann cells has been derived from studies of the migration and fate of such cells leaving embryonic quail gut back-transplanted into younger chick host embryos (Rothman et al., 1990). When a segment of foregut that contains crest-derived cells is removed from a quail embryo and is back-grafted into a neural crest migration pathway, the crest-derived cells will leave the donor bowel and follow that route. Destinations in the host embryo that become colonized by the donor’s crest-derived cells thus are appropriate for crest-derived cells migrating from the sites at which the grafts are placed and seem not to be influenced by the cells’ prior history of having migrated to the gut or their original derivation from the vagal level of the crest. Moreover, the terminally differentiated cells to which the enteric emigres give rise are also host site specific. For example, although catecholaminergic neurons are not normally found in the avian gut, crestderived cells leaving segments of bowel grafted into the trunk region of a host embryo migrate to sympathetic ganglia and there give rise to TH-immunoreactive neu-

202

GERSHON AND ROTHMAN

Fig. 3. The fore t of a quail embryo was removed at day E4 and was back-transplantegetween the neural tube and 19-20th somites of a 20 somite chick host embryo (day E2). The host was permitted to survive until day E15. Enteric cells have migrated from the graft and entered a peripheral nerve. The cells that do so have been shown with immunochemical markers to be derived from the neural crest (Rothman et al., 1990).Within the nerve (A),the enteric crest-derived cells display the quail nuclear marker ( 41, demonstrated by fluores-

cence microscopy on material stained with ropidium iodide. Simultaneous immunostaining (B)with the SMP n%b, generously donated by Dr. C. Dulac and Dr. N. Le Douarin (Dulac et al., 1988), reveals that enteric-glial cells are doubly labeled (+I and express this Schwann cell s ecific marker. The same field is illustrated in A (propidium iodide f&orescence) and B (fluorescein isothiocyanate fluorescence). The markers = 25 bm.

rons. Crest-derived cells also leave back-transplanted segments of foregut and migrate into peripheral nerves and spinal roots. In these locations the enteric crestderived cells assume the morphology of Schwann cells and have recently been observed to express the SMP antigen (Fig. 3), which, as noted above (Dulac et al., 1988),is expressed by Schwann cells and not by enteric glia. These experiments demonstrate that the vagal crest-derived cell population that colonizes the avian bowel contains precursor cells that are able to give rise to Schwann cells as well as to enteric glia. On the other hand, since none of the cells leaving the back-grafts of gut have been demonstrated to be terminally differentiated as glia, it cannot be concluded from these studies that enteric glia themselves change their phenotype to give rise to Schwann cells or that there is a relationship between the two cell types. On the contrary, in view of the evidence that crest-derived cells are multipotent (Baroffio et al., 1988; Bronner-Fraser, 1988; Dupin et al., 1990; Sieber-Blum and Cohen, 19801, it is more

likely that the cells that leave the bowel are uncommitted and ultimately follow programs of development that are influenced by and thus appropriate for the sites in which terminal differentiation takes place.

CONCLUSIONS Recent investigations have provided support for the concept that enteric glial cells are a distinct cell type, fundamentally different from the Schwann cells of the extra-enteric PNS. As yet, morphological and functional differences are more reliable than chemical markers in distinguishing between enteric glia and non-myelinating Schwann cells, although SMP in birds and Ran-1 in rats are expressed in situ on Schwann cells,but not enteric glia. The enteric glial cell, like the enteric neuron, is crest derived. Most of the enteric glia, moreover, are likely to originate from the initial waves of crest-derivedcells that colonize the gut. These waves in

203

ENTERIC GLIA

both mammals and birds are probably derived from both vagal and sacral regions of the neural crest. An additional complement of glial precursors may enter the bowel later in development along with the ingrowing extrinsic nerve fibers. The initial colonizingpopulations of crest-derived cells appear to contain uncommitted cells, which when induced by back-transplantation to leave the enteric microenvironment, are capable of entering peripheral nerves and developing as Schwann cells. Nevertheless, there is some evidence, as yet inconclusive, that some crest-derived emigres arriving in the bowel are separately committed to neural or glial lineages.

monoclonal antibod modulates the interaction of nerve growth factor with PC12 celL. J . Biol. Chem., 259:6882-6889. Cochard. P.. Goldstein. M.. and Black. I.B. (1978)Ontoeeneticamearance and disappearance oft osine hydroblase and &echolamines. Proc. Nutl. Acad. Sci. U . S x 75:2986-2990. Cook, R.D. and Burnstock, G. (1976a)The ultrastructure ofAuerbach's plexus in the guinea-pig. I. Neuronal elements. J. Neurocytol., 5:171-194. Cook, R.D. and Burnstock, G. (197613)The ultrastructure ofAuerbachs plexus in the guinea-pig. 11. Non-neuronal elements. J. Neurocytol., 5~195-206. Cornbrooks, C.J., Carey, D.J., McDonald, J.A., Timpl, R., and Bunge, R.P. (1983) In vivo and in vitro observations on laminin production by Schwann cells. Proc. Natl. AcadSci. U S A . , 80:3850. Deloulme,J.-C., Janet, T., Au, D., Storm, D.R., Sensenbrenner,M., and Baudier, J. (1990) Neuromodulin (GAP-43): A neuronal rotein kinase C substrate is also present in 0-2A lial cell lineage: {haracterization of neuromodulinin secondary cuiures of oligodendrocytes and comparison with the neuronal antigen. J . Cell. Biol., 111: 1559-1569. REFERENCES Dulac, C., Cameron-Curry,P., Ziller, C., and Le Douarin,N.M. (1988)A surface protein expressed by avian myelinating and non m elinating Schwann cells but not by satellite or enteric glial cells. &euron, Aikawa, H. and Suzuki, K. (1985) Enteric gliopathy in niacin-defi1:211-220. ciency induced by CNS glio-toxin.Brain Res., 334:354-356. Aikawa, H . and Suzuki, K. (1986) Lesions in the skin, intestine, and Dupin, E., Baroffio, A., Dulac, C., Cameron-Curry,P., and Le Douarin, N.M. (1990) Schwann-cell differentiation in clonal cultures of the central nervous system induced by an antimetabolite of niacin. Am. neural crest, as evidenced by the anti-Schwann cell myelin protein J. Pathol., 122:335-342. monoclonal antibody. Proc. Nutl. Acad. Sci. U S A . ,87:1119-1123. Andersson-Forsman, C. and Gustafsson, L:E. (1985) Cytochemical localization of 5'-nucleotidase in the enteric ganglia and in smooth Eccleston, P.A., Bannerman, P.G.C., Pleasure, D.E., Winter, J., Mirsky, R., and Jessen, K.R. (1989a) Control of eripheral glial cell muscle cells of the guinea-pi . J Neurocytol., 14:551-562. pliferation: Enteric neurons exert an inhilitory influence on Anderton, B.H. (1981)Intermefiate filaments: A family of homologous chwann cell and enteric glial cell DNA synthesis in culture. Deuelstructures. J. Mus. Res. Cell Motzl. 2:141-166. opment, 107:107-112. Baet e G. and Gershon,M.D. (1989)Transient catecholaminergic(TC) celfs in the vagus nerves and bowel of fetal mice: Relationship to the Eccleston, P.A., Jessen, K.R., and Mirsky, R. (1987) Control of peripheral glial cell roliferation: A comparison of the division rates of develo ment of enteric neurons. Deu. Biol., 132:189-211. enteric glia a n 1 Schwann cells and their response to mitogens. Dev. Pintar, J.E., and Gershon, M.D. (1990a)Transiently cateBaetge, BioZ., 1%:409-417. cholaminergic (TC)cells in the bowel of fetal rats and mice: Precursors of non-catecholaminergic enteric neurons. Dev. Biol., 141: Eccleston, P.A., Mirsky, R., and Jessen, K.R. (1989b) Type I collagen preparations inhibit DNA synthesis in glial cells of the peripheral 353-380. nervous system. Exp. Cell Res., 182373-185. Baetge, G., Schneider, K.A., and Gershon, M.D. (1990b)Development and ersistence of catecholaminergic neurons in cultured explants of Erde, S.M., Sherman, D., and Gershon, M.D. (1985) Morphology and serotonergic innervation of ph siologically identified cells of the fetafmurine vagus nerves and bowel. Develo ment, 110:689-701. guinea pig's myenteric plexus. Neurosci., k617-633. Bannerman, P.G., Mirsky, R., and Jessen, #.R. (1988a) Antigenic markers and laminin expression in cultured enteric neural cells. Ferri, G.L., Probert, L., Cocchia, D., and Michetti, F. (1982)Evidence for the presence of S-100protein in the dial component of the human Bruin Res., 440:89-98. entericnervous system. Nature, 297:4@410. Bannerman, P,.G., Mirsky, R., and Jessen, K.R. (1988b)Establishment and roperties of separate cultures of entenc neurons and enteric Fuiita. S. (1964) Analvsis of neuron differentiation in the central i e d o u s system by tiitiated thymidine autoradiography. J. Comp. glia.%rain Res., 440:99-108. Neurol., 122:311328. Bannerman, P.G., Mirsky, R., Jessen, K.R., Timpl, R., and Duance, V.C. (1986) Light microscopic immunolocalization of laminin, type Furness, J.B. and Costa, M. (1987)The enteric nervous system. In: The Enteric __ Nervous System. Churchill Livingstone, New York, pp. IV collagen, nidogen, heparan sulfate proteoglycan and fibronectin ci5-fjY. in the enteric nervous system of the rat and guinea-pig.J. NeurocyGabella, G. (1971)Glial cells in the myenteric plexus. 2. Naturforsch., tol., 15:733-743. 26B:244-245. Baroffio, A,, Du in, E., and Le Douarin, N.M. (1988) Clone-forming ability and difeerentiation otential of migratory neural crest cells. Gabella, G. (1972a) Fine structure of the myenteric plexus of the guinea- ig ileum. J. Anat. (Land.), 111:69-97. Proc. Natl. Acad. Sci. U.S.i!, 85:5325-5329. (197213) Innervation of the intestinal muscular coat. J. Bishop, A.E., Carlei, F., Lee, V., Trojanowski,J., Marangos,P.J., Dahl, Gabella, Neurocytol., 1:341-362. D., and Polak, J.M. (1985) Combined immunostaining of neurofilaments, neuron specific enolase, GFAP, and S-100:A possible means Gabella, G. (1987)Structure of muscles and nerves of the gastrointestinal tract. In: Physiology ofthe Gastrointestinal Tract. Raven Press, for assessing the morphological and functional status of the enteric New York, pp. 335-382. nervous s stem Histochemistry, 82:93-97. Bornstein, f D . and Furness, J.B. (1988)Correlated electrophysiologi- Gershon, M.D. (1981)The enteric nervous system.Ann. Reu. Neurosci., 4:227-272. cal and histochemical studies of submucous neurons and their contribution to understanding- enteric neural circuits. J. Auton. Gershon, M.D. (1987)The enteric nervous system. In: Encyclopedia of Neuroscience. G. Adelman, eds. Birkhauser Boston, Boston, pp. Nerv. Syst., 25-13. 398-399. Boyles, J.K., Pitas, R.E., Wilson, E., Mahley, R.W. and Ta lor, J.M. (1985)ApolipoproteinE associated with a s t r o c F glia of t l e central Gershon, M.D., Rothman, T.P.,Joh, T.H., andTeitelman, G.N. (1984) Transient and differential ex ression of aspects of the catecholaminnervous system and with nonmyelinating g ia of the peripheral ergic phenotype during deveyopment of the fetal bowel of rats and nervous system. J . Clin. Invest., 76:1501-1513. mice. J . Neurosci., 4:2269-2280. Bronner-Fraser, M. and Fraser, S.E. (1988) Cell lineage analysis reveals multipotency of some avian neural crest cells. Nature, Gershon, M.D., Sherman, D., and Gintzler, A. (1981) An ultrastructural analysis of the developingenteric nervous system of the guinea 335:161-164. pig small intestine. J . Neurocytol., 10:271-296. Bulbring, E. and Crema, A. (1958)Observations concerning the action of 5-hydroxytryptamineon the peristaltic reflex. Br. J. Pharmacol., Jacobs-Cohen, R.J., Payette, R.F., Gershon, M.D., and Rothman, T.P. (1987) Inability of neural crest cells to colonize the presumptive 13:444-457. aganglionicbowel of I d s mutant mice: Requirement for a permissive Cameron-Curry,P., Dulac, C., and Le Douarin, N.M. (1989)Expression microenvironment. J. Comp. Neurol., 255:425438. of the SMP antigen by oligodendrocytes in the developing avian Jessen, K.R. and Mirsky, R. (1980) Glial cells in the enteric nervous central nervous system. Development, 107:825-833. system contain glial fibrillary acidic protein. Nature, 286:73&737. Carbonetto, S. and Stach, R.W. (1982) Localization of nerve growth factor bound to neurons growing nerve fibers in culture. Deu. Brain Jessen, K.R. and Mirsky, R. (1983)AstrocJlte-like lia in the perpheral nervous system: An immunohistochemical stuiy of enteric glia. J. Res., 3:463-473. Neurosci., 3:2206-2218. Chandler, C.E., Parsons, L.M., Hosang, M., and Shooter, E.M. (1984)A

b.,

~

^ ^

8.

204

GERSHON AND ROTHMAN

Jessen, K.R., Thor e, R , and Mirsky, R. (1984) Molecular identity, distribution and ieterogeneity of glial fibrillary acidic protein: An immunoblotting and immunohistochemical study of Schwann cells, satellite cells, enteric glia and astrocytes. J . Neurocytol., 13:187-200. Jonakait, G.M., Rosenthal, M., and Morrell, J.I. (1989) Regulation of tyrosine hydroxylase mRNA in the catecholaminergic cells of embryonic rat: analysis by in situ hybridization. J . Histochem. Cytochem., 37:l-5. Jonakait, G.M., Wolf, J., Cochard, P., Goldstein, M., and Black, LB. (1979) Selective loss of noradrenergic henotypic characters in neuroblasts of the rat embryo. Proc. 8atl. Acad. Sci. U.S.A., 76: 4683-4686. Kirchgessner, A.L., Dodd, J., and Gershon, M.D. (1988) Markers shared between dorsal root and enteric ganglia. J . Comp. Neurol., 276:607-621. Kirchgessner, A. and Gershon, M.D. (1989) Identification of vagal efferent fibers and putative target neurons in the enteric nervous system of the rat. J. Comp. Neurol., 285:3%53. Kobayashi, S., Suzuki, M., Endo, T., Tsuji, S., and Daniel, E.E. (1986) Framework of the enteric nerve plexuses: An immunocytochemical study in the guinea pigjejunum using an antiserum to S-100 protein. Arch. Histol. Jpn., 49:15%188. Komuro, T., Bauk, P., and Burnstock, G. (1982) An ultrastructural study of neurons and non-neuronal cells in the myenteric plexus of the rabbit colon. Neuroscience, 7:1797-1806. Kosterlitz, H.W. and Lees, G.M. (1964) Pharmacological analysis of intrinsic intestinal reflexes. Pharmacol. Rev., 16:301-339. Le Douarin, N.M. (1982) The Neural Crest. Cambridge University Press, Cambridge. Le Douarin. N.M. and Teillet. M.A. (1973)Themieration ofneural crest cells to the wall of the digestive tract in avianembryo. J. Embryol. Exp. Mor hol ,30:31-48. Payette, R.$., Bennett, G.S., and Gershon, M.D. (1984) Neurofilament expression in vagal neural crest-derived precursors of enteric neurons. Deu. Biol., 105:273-287. Payette, R.F., Tennyson, V.M., Pham, T.D., Mawe, G.M., Pomeranz, H.D., Rothman, T.P., and Gershon, M.D. (1987) Ori 'n and morphology of nerve fibers in the aganglionic colon of the letfa1 spotted (Ids) mutant mouse. J. Comp. Neurol., 257:237-252. Rohrer, H. (1985) Non-neuronal cells from chick sym athetic and dorsal root sensory gan lia express catecholamine uptafe and rece tors for nerve growtg factor during development. Deu. Biofl 111:95-107. Rothman, T.P. and Gershon, M.D. (1982) Phenoty ic expression in the developing murine enteric nervous system. J. &urosci., 2:381-393. Rothman, T.P. and Gershon, M.D. (1984) Regionally defective colonization of the terminal bowel by the precursors of enteric neurons in lethal spotted mutant mice. Neuroscience, 12:1293-1311. Rothman, T.P., Le Douarin, N.M., Fontaine-Perus, J.C., and Gershon,

M.D. (1990) Developmental potential of neural crest-derived cells migrating from segments of develo ing quail bowel back-grafted into younger chick host embryos. Deveipment, 109:411-423. Rothman, T.P., Nilaver, G., and Gershon, M.D. (1984) Colonization of the developing murine enteric nervous system and subsequent phenot ic expression by the precursors of peptidergic neurons. J. Comp. Kurol., 22513-23. Rothman, T.P., Tennyson, V.M., and Gershon, M.D. (1986) Colonization of the bowel by the precursors of enteric glia: Studies of normal and congenitally aganglionic mutant mice. J. Comp. Neurol., 252:493-506. Scheuermann, D.W., Stach, W., Timmermans, J.-P., Adriaensen, D., and De Groodt-Lasseel, M.H.A. (1989) Neuron-specific enolase and S-100protein immunohistochemistry for defining the structure and topogra hical relationship of the different enteric nerve plexuses in the smajl intestine of the pig. Cell Tiss. Res., 256:65-75. Sieber-Blum, M. and Cohen, A.M. (1980) Clonal analysis of quail neural crest cells: They are pluripotent and differentiate in vitro in the absence of non-crest cells. Dew. Biol., 80:96-106. Teitelman, G., Gershon, M.D., Rothman, T.P., Joh, T.H., and Reis, D.J. (1981)Proliferation and distribution of cells that transiently express a catecholaminergic phenotype during development in mice and-rats. Deu. Biol., 86:34%355. Teitelman, G., Joh, T.H., and Reis, D.J. (1978) Transient expression of a nonadrenergic phenotype in cells of the rat embryonic gut. Brain Res., 158:22%234. Thomas, P.K. and Ochoa, J. (1984) Microscopic anatomy of eripheral nerve fibers. In: Peripheral Neuro athy P J Dyck, P . 2 Thomas, Saunders Co., Philadelphia, E.H. Lambert, and R. Bunge, eds. PA, pp. 39-96. Trendelenburg, P. (1917) Physiologische und pharmakologische Versuche uber die Dunndarm Peristaltick. Naunyn-Schmiedebergs Arch. Exp. Puthol. Pharmakol., 81:55-129. Tucker, G.C., Ciment, G., and Thiery, J.-P. (1986) Pathways of avian neural crest cell migration in the developing gut. Deu. Biol., 116:439-450. Wood, J.D. (1987) Physiology of enteric neurons. In: Physiology ofthe Gastrointestinal Tract. L.R. Johnson, eds. Raven Press, New York, pp. 1-41. Yen, S. and Fields, K.L. (1981) Antibodies t o neurofilament, glial filament and fibroblast intermediate filament proteins bind to different cell types in the nervous system. J . Cell Biol., 88:115-126. Yntema, C.L. and Hammond, W.S. (1954) The origin of intrinsic ganglia of trunk viscera from vagal neural crest in the chick embryo. J. Comp. Neurol., 101:515-542. Zimmermann, A. and Sutter, A. (1983) P-Nerve owth factor (P-NGF) receptors on glial cells: Cell-cell interaction !$tween neurons and Schwann cells in cultures of chick sensory ganglia. EMBO J., 21879-885.

6.B.