Control of peripheral glial cell proliferation: enteric neurons exert an inhibitory influence on Schwann cell and enteric glial cell DNA synthesis in culture

Cessation of division is an important step in glial cell differentiation. In the peripheral nervous system (PNS), this differentiation gives rise to a cellular arrangement in the enteric nervous system (ENS) which is distinctly different from that in peripheral nerves. In peripheral nerves, Schwann cells ensheath and myelinate individual axons or ensheath groups of small axons. Enteric glial cells of the enteric nervous system support enteric neurons and axons in an organisation similar to that of astrocytes and neurons in the central nervous system (CNS) (Gabella, 1971, 1981).

In peripheral nerves, the axon has been implicated in control of glial cell numbers since the axonal membrane and axolemma fragments are potent mitogenic signals for Schwann cells (Salzer et al. 1980; De Vries, 1981; Pleasure et al. 1985). Thus, axonally associated signals, possibly together with soluble mitogens (Brockes et al. 1980; Pleasure et al. 1985; Eccleston et al. 1987), may be important for expanding Schwann cell populations during development and nerve regeneration. It is less clear, however, why Schwann cells do not divide in mature nerves. It seems likely, by analogy with other systems, that negative growth signals may be important (Braun et al. 1988; Sporn et al. 1987; Wang and Hsu, 1986).

Our preliminary results suggested that enteric neurons did not induce proliferation of enteric glial cells in explant cultures (Bannerman et al. 1987). Here we show that enteric neurons inhibit the DNA synthesis of both Schwann cells and enteric glia. Nevertheless we find that, like Schwann cells, enteric glia increase their DNA synthesis in response to axolemma isolated from the CNS. Enteric neurons, under the culture conditions used, therefore differ from sensory neurons in their influence on glial cell proliferation. Thus the DNA synthesis of enteric glia and Schwann cells can be either stimulated or inhibited by neuron-associated factors. It is possible that, during development of the PNS, both types of signals may play a role in regulating glial cell numbers.

Cultures were grown in Dulbecco’s Modified Eagle’s Medium (DME) supplemented with fetal calf serum (10%), glucose (2mg ml⁻¹), insulin (5μg ml⁻¹’), penicillin (100IUml⁻¹) and streptomycin (100 μg ml⁻¹). Nerve growth factor (10 ng ml⁻¹) was added to cultures of guinea pig dorsal root ganglia. Axolemma was prepared as described previously (Cullen et al. 1981; De Vries, 1981) and added to the culture medium at 35 or 70 μg per culture.

Purified populations of enteric glial cells were prepared from explants of myenteric plexus as described previously (Jessen et al. 1983; Bannerman et al. 19886). Essentially, segments of the myenteric plexus were dissected from the gut wall following enzyme treatment, explanted onto polylysinecoated glass coverslips and treated with cytosine arabinoside (AraC) for 4 –5 days to suppress fibroblast growth. After a further 2 –3 days without AraC, the central neuronal areas were removed and cultures treated for 1 h with anti-Thy1.1and complement to kill any remaining neurons and fibroblasts. In most cultures, >95 % of all cells were enteric glia.

Purified populations of enteric neurons were prepared from explants of myenteric plexus as described previously (Bannerman et al. 19886). Briefly, explant cultures of the myenteric plexus were treated with AraC for a prolonged period to suppress the growth of fibroblasts and enteric glial cells. Any remaining glial cells were killed using the monoclonal antibody LB1 and complement. Cultures were 85 –95% enteric neurons.

Rat Schwann cells were purified from sciatic nerves by a modification of the method of Brockes et al. 1979 (Jessen et al. 1987a). They initially had a low proliferation rate and spindleshaped morphology. With time in culture, and in the absence of any added mitogen, they gradually became more flattened in morphology and increased their proliferation rate. A similar increase in Schwann cell division rates after stimulation with mitogens has been described previously and is thought to be mediated via an autocrine mechanism (Porter et al. 1986, 1987). The change in morphology allowed us to choose cells with a higher division rate to study growth inhibition.

Guinea pig Schwann cells were purified from the sciatic nerves of 1-to 5-day-old Dunkin-Hartley guinea pigs as described previously (Eccleston et al. 1987). Once purified, Schwann cells were removed from the dishes using 0 ·25 % trypsin diluted 1:25 in calcium- and magnesium-free DME.

Dorsal root ganglia were removed from newborn Dunkin-Hartley guinea pigs and desheathed. They were dissociated using 0 ·125% trypsin and 0 ·2% collagenase for 1h as for guinea pig Schwann cells. AraC was included in the culture medium to remove dividing Schwann cells and fibroblasts.

Pure populations of adult rat dorsal root ganglion neurons were prepared as described previously (Wood et al. 1988). Briefly, dorsal root ganglia from rats over two months old were dissociated, enriched for neurons by a differential adhesion step and plated on laminin-coated coverslips. Further purification was achieved by including AraC in the culture medium.

Rabbit anti-Thyl.l was a gift from Dr A. F. Williams and was used at a dilution of 1:100. Rabbit antiserum to S100 protein (DAKO immunoglobulins A/S) (dilution 1:400 –1:800) was used as a specific glial cell marker in the PNS (Brockes et al. 1979; Mirsky et al. 1985; Bannerman et al. 1988a). The monoclonal antibody LB1, a gift from Dr J. Cohen, recognises a ganglioside on the surface of several cell types in the nervous system. In the ENS, it has been shown to bind specifically to enteric glial cells (Bannerman et al. 1988a). The monoclonal antibody, A5E3, is rat specific, recognising a cell surface glycoprotein on Schwann cells and fibroblasts (Mirsky et al. 1985). Tetramethyl rhodamine conjugated to goat anti-rabbit Ig (G anti-R Ig Rd) (Cooper Biomedical) was adsorbed with mouse Ig to remove cross-reacting antibodies and was used at a dilution of 1:100. Fluorescein conjugated to goat anti-mouse Ig (G anti-M Ig Fl) (Nordic Laboratories Ltd) was adsorbed with rabbit Ig and used at a dilution of 1:100.

Schwann cells and enteric glia were identified using anti-S100. Rat Schwann cells in mixed cultures with enteric neurons were identified with anti-S100 alone or the rat-specific monoclonal antibody A5E3 in combination with anti-S100 in order to distinguish the rat Schwann cells (S100⁺, A5E3⁺ ) from any enteric glial cells present in the neuron cultures (S100⁺A5E3). Guinea pig dorsal root ganglion neurons were identified using anti-Thy1.1. For demonstration of S100, cultures were fixed in 4% paraformaldehyde in phosphate-buffered saline for 20 min followed by treatment with meth- anol at —20°C for 10min. LB1, A5E3 and anti-Thy1.1were applied to living cultures. The procedure used has been described previously (Jessen et al. 1987b).

Cultures were incubated for 24h with 0 ·1 μCi [³H]thymidine (5 Ci mmol⁻¹) (Amersham International, pic). They were then washed in several changes of Eagle’s Minimal Essential Medium supplemented with Hepes buffer (0 ·015 M) and immunolabelled as described above. Coverslips were mounted cell side uppermost on gelatin-coated glass slides and processed for autoradiography as described previously (Eccleston et al. 1987). The percentage of Schwann cells or enteric glia with more than 5 silver grains over their nuclei was determined using a Zeiss microscope equipped with epi-fluorescence and phase-contrast optics. 200 –500 cells per coverslip were scored.

In myenteric explant cultures, the neurons and a population of glial cells remain in the central explant area (Fig. 1). Other enteric glial cells migrate out and proliferate forming a glial cell carpet around the central area, over which fine neuntes extend (Fig. IB). The neurites remain on the surface of the glial cells and do not extend onto the glass coverslip (Jessen et al. 1983). Myenteric explant cultures and purified populations of glial cells initiated from the same animal were exposed to [³H]thymidine for 24 h after 10 days in culture and labelled with anti-S100 to identify glial cells. [³H]thymi-dine incorporation levels were assessed in three separate areas of the expiant cultures. (1) In the outgrowth zone, excluding a narrow band of approximately 0 · 3 mm immediately surrounding the central neuronal area, the glial cells had a labelling index of 55 · 6 ± 1 · 0 (mean ± S.E.M.; n = 6). In parallel experiments, pure populations of enteric glial cells divided at a similar rate of 56 · 2 ±2 · 1 (mean ± S.E.M.; n = 8). (2) In the area immediately surrounding the central neuronal area (an approximately 0 · 3 mm zone surrounding the explant area) S100-positive glial cells incorporated [³H]thymidine at a greatly reduced rate of 13 · 7 ±1 · 9 (mean ± S.E.M.; n = 5) compared with the outgrowth zone. (3) In the central explant area, almost no tritiated thymi-dine-labelled cells could be detected although many S100-positive enteric glial cells were present. The dense arrangement of glia in close apposition to neurons made cell identification and counting of individual cells impossible.

Thus enteric glia in these experiments do not show a mitogenic response when they are exposed to neuronal membranes. Moreover, the results suggest that under some circumstances enteric neurons can exert an inhibitory influence on the division rate of enteric glial cells. This is in contrast to results obtained with Schwann cells in contact with dorsal root ganglion neurons or axolemmal fragments.

The reason for the failure of enteric glia to show a mitogenic response to enteric neurons could be that these neurons do not express the mitogenic molecule found in association with other neuronal membranes. Alternatively, enteric glia might lack receptors for the neuronal mitogen. To test the latter possibility, purified populations of guinea pig enteric glial cells were exposed to 35 or 70 μ g of bovine axolemma for 24 h or 4 days. In all cases, axolemma increased the labelling indices of the enteric glial cells (Table 1). Controls were done using short-term cultures of rat Schwann cells to show that the axolemma preparations were active, and guinea pig Schwann cells to show that the bovine axonal membrane preparation was able to cross-react with cells from guinea pig. In both cases, Schwann cell DNA synthesis was increased by the axolemma (Table 1). In addition, like short-term cultures, dividing rat Schwann cells from long-term secondary cultures (see Methods) showed increased [³H]thymidine incorporation rates in response to axolemma. Control cultures had a labelling index of 12 μ4 ±1 μ0 whereas those exposed to axolemma had a labelling index of 24 μ0 ±3 μ 0 (mean ± S.E.M.; n = 6).

Since enteric glia, like Schwann cells, could clearly respond to the axolemmal mitogen by an increase in proliferation rate, the alternative possibility, i.e. that enteric neufons lacked the mitogen or even exerted an inhibitory effect on glial division, was explored further.

For this purpose, we employed purified populations of guinea pig enteric glia or rat Schwann cells and purified explant cultures of enteric neurons from which essentially all glial cells had been removed. When purified enteric glial cells were added back to the neuronal cultures a slight inhibition of glial cell proliferation was detected. Pure enteric glial cell controls had a labelling index of 47 · 2 ± 3 · 2 (mean ± S.E.M.; n = 3) and enteric glial cells in the neuronal area of the mixed cultures had a labelling index of 33 · 0 ± 2 · 6 (mean ±S.E.M.; n = 5). In similar experiments, enteric neurons inhibited the DNA synthesis of rat Schwann cells, only Schwann cells in the neuronal area of the cultures being affected. For these experiments dividing long-term secondary Schwann cell cultures were used. Cultured alone these cells had a labelling index of 31 · 9 ±3 ·5 (mean ± S.E.M.; n = 6). In co-cultures with enteric neurons, those Schwann cells that were in the neuronal area had a labelling index of 8 ·5 ± 0 ·7 (mean ± S.E.M.; n = 9) whereas those Schwann cells distant from the neuronal explant area synthesised DNA at a similar rate to Schwann cell controls, with a labelling index of 28 · 8±2 ·0 (mean±S.E.M.; n = 5) (P<0 · 001 for Schwann cells in the neuronal area compared with Schwann cells distant from neurons). These results are summarised in Table 2.

These results indicated that a mitotic inhibitory signal was associated with guinea pig enteric neurons. To test whether this was a general feature of guinea pig neurons or, alternatively, whether the inhibitory effect was specifically associated with enteric neurons, purified rat Schwann cells were added to purified cultures of guinea pig dorsal root ganglion neurons. Schwann cells in control cultures, where no neurons were present, had a labelling index of 4 ·8 ± 1 ·1 (mean ± S.E.M.; n = 3). In the neuronal area of mixed cultures, Schwann cells had a labelling index of 23 · 8 ± 2 ·5 (mean ± S.E.M.; n = 4) and in areas where no neurons were identified the Schwann cell labelling index was 5 · 2 ±1 · 1 (mean ± S.E.M.; n = 3).

The mitogenic effect seen with guinea pig dorsal root ganglion neurons indicated that an inhibitory effect on glial DNA synthesis might be a particular feature of enteric neurons. In the guinea pig, the enteric plexuses are, however, relatively well developed at birth and three to four weeks of culturing are required for depletion of glial cells. The enteric neurons have therefore also had a considerable period to mature in vitro by the time they are re-exposed to glial cells. To test the possibility that an inhibitory effect of cultured neurons on glial DNA synthesis was a feature of mature neuronal membranes, purified Schwann cells were added to cultures of adult rat dorsal root ganglion neurons that had been maintained for 2 –4 weeks in culture. Schwann cells in control cultures grown on polylysine in the absence of neurons had a labelling index of 2 · 5 ±0 ·7 (mean ± S.E.M.; n = 3). In mixed cultures, Schwann cells had a labelling index of 12 · 3 ±0 ·6 (mean ± S.E.M., n = 3). Since the neurons were grown on a laminin-coated substratum, controls were done where the neurons were gently removed from the coverslip with a jet of medium. Schwann cells grown on this surface synthesised DNA essentially at the same rate as those on polylysine, having a labelling index of 2 ·5 ±0 ·7 (mean ± S.E.M., n = 2). The results shown are from one experiment using one batch of neurons and Schwann cells. Similar results were obtained in two further experiments where the Schwann cells had higher labelling indices.

All axons so far tested from the central and peripheral nervous systems have had a stimulatory effect on Schwann cell DNA synthesis (De Vries et al. 1982; Mason et al. 1989; Salzer et al. 1980; Pleasure et al. 1985). The present work shows that the enteric neuron has an inhibitory influence on the DNA synthesis rates of both Schwann cells and enteric glia in culture. This is in contrast to the axons of sensory neurons and axolemmal fragments, which stimulate division of peripheral glia. A heparan sulfate proteoglycan is required for the proliferation of Schwann cells along the neunte membrane (Ratner et al. 1985,1988). Several growth factors, including fibroblast growth factors, bind to heparin with high affinity (Lobb et al. 1986; Rybak et al. 1988), and it seems probable that the axonal mitogen is a proteoglycan-growth factor complex (Ratner et al. 1988). Enteric neurites might lack the heparan sulphate proteoglycan associated with their plasma membrane. If so they would not bind the mitogenic growth factor and thus not stimulate glial cell division. Furthermore, it is possible that the mitogenic effect of dorsal root ganglion neurites is due to binding of a factor or factors that also can act as soluble mitogens, e.g. glial growth factor. The presence of receptors for the neuriteassociated mitogen on enteric glia would then be consistent with our previous observations that guinea pig enteric glia and Schwann cells respond to the same range of soluble and extracellular-matrix-derived mitogens (Eccleston et al. 1987). They both respond to fibroblast growth factor and glial growth factor but not to epidermal growth factor. In addition, laminin, fibronectin and a basement membrane-like extracellular matrix stimulate the proliferation of both enteric glial cells and Schwann cells and both divide in response to cAMP and cholera toxin (Eccleston et al. 1987).

In development, it is conceivable that the same axon might sequentially act, first as a mitogen to expand cell numbers and then in a more mature state, as an inhibitor of cell division when the correct number of glial cells has been attained. In culture, a bound growth factor might mask any inhibitory effect that the axon would otherwise exert on peripheral glial cells. Thus, the results presented here might not simply indicate that enteric and sensory neurons differ in their control of glial cell proliferation, but could be revealing an inhibitory signal common to all neurons under the correct conditions. Indeed, a precedent for this exists in the CNS where both axolemmal fragments and granule neurons have been shown to inhibit the proliferation of astrocytes in culture (Sobue and Pleasure, 1984; Hatten, 1987).

Cerebellar granule neurons inhibit astrocyte division in the CNS via a non-diffusible neuronal membrane molecule (Hatten, 1985, 1987; Edmondson et al. 1988). We were unable to use the method of Hatten (1987), where fixed neurons were added to glial cells to determine whether the inhibitory molecule was membrane bound, since sufficient numbers of enteric neurons could not be purified by present methods. Attempts to use neurons, fixed directly on coverslips, using the same procedure as described for experiments reported here using unfixed neurons have so far failed since the neurons fixed using a mild treatment with paraformaldehyde began to disintegrate before the experimental procedure was complete. However, only glial cells close to the neuronal cell area showed a reduced DNA synthesis rate. Those in the periphery of the cultures had labelling indices that were the same as controls where no neurons were present. This indicates that a membrane-bound growth inhibitor is likely to be involved in the inhibition of glial cell DNA synthesis by enteric neurons.

When cultures of Schwann cells or enteric glia were added to pure populations of enteric neurons glial cell division was inhibited. In explant cultures of myenteric plexus that contained both enteric neurons and enteric glia, the results were somewhat different. The inhibition of glial cell DNA synthesis was only detectable close to the neuronal cell bodies i.e. in the central neuronal area of the cultures and in a zone approximately 1mm wide surrounding the neuronal area. If a membrane-associated molecule is involved in mitotic inhibition, these results indicate that it may be present at a higher concentration over the cell bodies than in the neurites. Alternatively, the very fine network of neurites growing over the glial carpet may not provide a sufficient area of contact for inhibition to be initiated.

Since the enteric neuron and its neurites do not stimulate enteric glial cell proliferation the mechanism of generation of glial cells in development of the ENS may be different from that for Schwann cells, where in vitro evidence suggests that the axon plays an important role in expanding cell populations (Salzer and Bunge, 1980; Salzer et al; 1980; Wood, 1976). The present work suggests that the generation of adequate numbers of enteric glia depends more on soluble mitogenic growth factors than on growth signals associated with neuronal membranes.

This work was supported by the Medical Research Council of Great Britain, Action Research for the Crippled Child and the Multiple Sclerosis Society of Great Britain. We thank Drs A. F. Williams and J. Cohen for their gifts of anti-Thyl.l and LB1 respectively and M. Hardy for preparation of axolemma.