Spontaneous immortalisation of Schwann cells in culture: short-term cultured Schwann cells secrete growth inhibitory activity

A knowledge of the control of Schwann cell proliferation is critical for understanding how the peripheral nerve develops. After the correct number of Schwann cells has been generated in the peripheral nerve, they become quiescent and do not proliferate in the normal nerve. Cessation of division is an essential and important step in Schwann cell differentiation and proliferating cells do not differentiate to express the myelin protein, Po (Morgan el al. 1991). Schwann cells do not, however, lose the ability to divide and will do so in nerve repair and in tumour formation. In culture, when removed from axons, Schwann cells form an apparently homogeneous population with a phenotype that in several respects resembles that of non-myelin-forming Schwann cells in vivo (Jessen et al. 1990; Mirsky and Jessen, 1987). Under these conditions, unlike many other cell types, Schwann cells divide very slowly even in the presence of serum (Brockes et al. 1980a; Porter et al. 1986). It is therefore possible that, even in culture, Schwann cell growth remains under tight negative control.

When growth control is lost in the peripheral nerve in vivo, two types of tumours arise; the neurofibroma and the Schwannoma, both of which are usually benign. Mitogenic activity has been isolated from these tumours (Brockes et al. 1986; Ratner et al. 1990; Riccardi, 1986) and the Schwann cell culture system described in the present study might provide a tissue culture model for the loss of growth control encountered in neurofibromas and Schwannomas in vivo. The gene for NF-1 has now been cloned and shown to have homology with the catalytic domains of GTPase activating protein (GAP) (Xu et al. 1990). This has stimulated much interest in neurofibromas and control of Schwann cell mitosis.

Despite their low proliferation rate in serum, cultured Schwann cells divide quite rapidly when stimulated with mitogens. We report here that, when Schwann cells are maintained in culture for several months without the addition of exogenous mitogens, a change occurs in their growth control and they spontaneously begin to proliferate. We have previously demonstrated that these immortalised cultured Schwann cells secrete platelet-derived growth factor (PDGF) which is at least partially responsible for the proliferation of these cells. Quiescent short-term cultured Schwann cells, however, also secrete similar levels of PDGF. Thus, PDGF secretion did not account for the elevated division of immortalised long-term cultured Schwann cells (Eccleston et al. 1990). Here, we show that short-term cultured Schwann cells secrete a heat- and acid-resistant, trypsin-sensitive growth inhibitor which was not detectable in conditioned medium from long-term cultures. The secretion of a growth inhibitor could in part explain the low proliferation rate of short-term Schwann cell cultures and its down regulation may at least partially explain the elevated DNA synthesis of long-term cultures. It may also have a role in the post-natal cessation of division that occurs in development of the peripheral nerve.

Cultures were grown in Dulbecco’s modified Eagle’s medium (DME) supplemented with fetal calf serum or calf serum (10%), glucose (2mgml⁻¹), insulin (5μgml⁻¹), penicillin (100i.u ml⁻¹) and streptomycin (100μg ml¹) Some experiments were done in serum-free medium using a modification of the medium described by Bottenstem and Sato (1979) which has been described previously (Richardson et al. 1988). Cells were seeded in serum-containing medium and, for serum-free culture in some experiments, cultures were first washed three times to remove traces of serum Potential growth inhibitors were added to long-term cultures or shortterm cultured Schwann cells stimulated to divide with 8-bromo cAMP or GGF in the presence of calf serum. Human recombinant IL6 (Immunex Corporation) was used at 0.5 – 500 ng ml⁻¹; activin from a partially purified preparation from the XTC cell-hne was provided by J.C. Smith (Smith, 1987) and was used at a dilution of 1 300–1:3000, atrial natriuretic peptide (ANP) was from Cambridge Research Biochemicals and was used at 10⁻⁶−10⁻⁹3; epidermal growth factor (EGF) was from Collaborative Research and was used at 0.2–200 ng ml⁻¹.

Schwann cells were purified in 25 cm² culture flasks from the sciatic nerves of 3- to 4-day-old Sprague Dawley rats using a modification of the method of Brockes et al. 1979. Fibroblasts were removed from the cultures by including the mitotic inhibitor cytosine arabinoside (araC) (10 ^{− 5}3) in the medium for the first 2-3 days to kill these rapidly dividing cells. Short-term Schwann cell cultures were those used for experiments in the first two weeks of culture. For use, cells were passaged onto 13 mm polylysine-coated coverslips at a density of 5 – 10×10³ cells per coverslip. The monoclonal antibody DB1 was added to some cultures at dilutions of 1:10 to 1:40 (Van der Meide et al. 1986).

When cultures as decribed above were not used in the first two weeks they were maintained by changing half of the medium once a week This was continued for approximately 3 months until populations of proliferating flattened cells appeared. Cultures were then passaged approximately once a month. Long-term Schwann cell cultures were passaged onto cover-slips for immunofluorescence or for studies of cell division at a density of 2.5 – 5 × 10³ cells per 13mm diameter polysine-coated glass covershps.

The epmeunum was removed from 4-day-old Sprague Dawley rats and dissociated using the same method of enzyme digestion as for Schwann cells Cells were grown to confluence in 25 cm² culture flasks in serum-containing medium, and conditioned medium collected or cells were passaged onto coverslips.

Cultures enriched for type 1 astrocytes were prepared as described previously (Richardson et al 1988). Conditioned medium was collected from confluent cultures.

This rat Schwannoma cell line was produced by ethylnitro-sourea treatment of rats (Fields et al. 1975). Conditioned medium was collected from confluent cultures or cells were passaged onto coverslips.

Swiss 3T3 cells (passage number not known) were grown in 25 cm flasks in serum-containing medium Conditioned medium was collected from confluent flasks or cells were passaged onto 13 mm diameter polylysine-coated glass cover-slips at a density of 1 – 2000 cells per coverslip.

Fluorescein conjugated to goat anti-mouse Ig was absorbed with rabbit Ig to remove cross-reacting antibodies (G anti-M Ig fl) and was used at a dilution of 1:100. Tetramethyl rhodamine conjugated to goat anti-rabbit Ig was absorbed with mouse Ig to remove cross-reacting antibodies (G anti-R Ig rd) and was used at a dilution of 1.100.

Rabbit anti-bovine S100 protein, used at a dilution of 1:800, was purchased from Dakopatts a/s. Rabbit anti-GFAP was purchased form Dakopatts a/s and used at 1:200. Rabbit anti-fibronectin was a gift from R. Hynes and was used at 1:100. Antibodies to rat Thy 1:1 were a gift from A. F. Williams and were used at 1:100 Monoclonal anti-galactocerebroside (Gal C) was provided by B. Ranscht (Ranscht et al. 1982). Rabbit anti-Ll and N-CAM were provided by M. Schachner and both used at 1.200. Rabbit anti-P_o was provided by J.P. Brockes and used at 1 · 200. The monoclonal antibody, 04 reacts with sulphatide (Mirsky et al. 1990) and was a gift from I. Sommer. A monoclonal anti-nerve growth factor receptor (NGFR) was a gift from E. Johnson (Yan and Johnson, 1988) and was used at 1:5000. 217c(Ranl) is now known to recognise the NGFR and was used at 1:500 (Peng et al. 1982; Fields and Dammerman, 1985).

Immunofluorescence labelling was done as described previously (Eccleston et al. 1987). Unfixed cultures were labelled when using the monoclonal antibodies A5E3, GalC, 04, 217c and NGFR and rabbit antisera against Thy 1.1, N-CAM, LI and fibronectin For intracellular demonstration of fibronectin, P_o or GFAP cultures were fixed first in methanol at —20°C. For intracellular demonstration of S100, cultures were fixed in 4% paraformaldehyde for 20 min at room temperature followed by a 10 min treatment with methanol at —20°C.

BrdU (10⁻⁵ M) was included in the medium for the final 24 h of culture to label cells replicating DNA. Nuclei that had incorporated BrdU were visualised using a mouse monoclonal antibody produced against BrdU which was kindly provided by D. Y. Mason (Gratzner, 1982). Cultures were fixed for 10min m 2% paraformaldehyde followed by methanol at —20°C for 10 min. They were treated with 2N HC1 for 10 min and the acid was neutralised by a further 10 min treatment with 0 1 M sodium borate (pH 8 5) Coverslips were incubated with anti-BrdU (1:5 – 1.20) and anti-S100 (1.800) together in PBS contaimng 0 1 % Triton X-100 for 40 min. After washing, they were incubated with G anti-R Ig rd and G anti-M Ig fl for further 30min, washed and mounted in Citifluor anti-fade medium (City University, London). The percentage of S100-positive Schwann cells with fluorescein-labelled nuclei was determined by scoring 200 – 500 glial cells per coverslip.

Cells (approx 5 × 10⁵) were grown in 3 ml of medium in 25 cm² flasks. Medium (3 ml) was conditioned for 2 days either in the presence or absence of serum, centrifuged to remove cells and the supernatant fluid was stored at 4°C. The inhibitory activity in serum-free conditioned medium from short-term Schwann cells was partially characterised. The medium was heated at 60 °C for 20 min or 100°C for 10 min or was reduced to pH 2 for 1 h using HC1 and then neutralised with NaOH. To test for enzyme sensitivity 1 ml of conditioned serum-free medium was treated with 100 μg trypsin (Gibco), 25 BEAE Units trypsin immobilised on agarose (trypsm/agarose) (Sigma) which was subsequently removed by centrifugation, or 100 μg collagenase in the presence of 200 μg of egg white trypsin inhibitor (Sigma) for 1 h at 37°C. In all cases, control serum-free medium which had not been conditioned was also treated and calf serum (10%), which inactivated the enzymes, was added for use.

Long-term Schwann cells (1 – 5 ×10²) were added to 1ml of medium containing 0 33 % agar in 35 mm culture dishes. The transformed Schwann cell-line, 33B was used as a positive control. Cultures were observed at weekly intervals for three weeks

During the first two weeks of culture, Schwann cells divided very slowly, even in the presence of 10 % serum (labelling index 0 – 5 %). They could be kept in the same flask for at least three months without overcrowding. During this time, the fibroblasts contaminating the original cultures died, presumably due to senescence. The initial observation of a change in the cell population was the appearance of groups of flattened cells growing amongst the spindle-shaped Schwann cells. These flattened cells were somewhat fibroblast-like in appearance under phase-contrast microscopy except that stress fibres could not be seen (Fig. 1). Eventually the cells became confluent, constituting of a population of flattened cells together with bipolar and/or stellate cells that exhibited contact inhibition of growth and could be kept without passaging for approximately one month before visible degeneration began. When the cells were passaged to a low density, they became flattened in morphology (Fig. 1) and divided at a considerably higher rate than primary Schwann cells (labelling index up to 99 % depending on cell density). They increased in number until at confluence a change in shape occurred. The morphology of each cell line at confluence was characteristic of that particular line. Some cultures became spindle shaped, similar to short-term Schwann cells, whilst others were stellate in morphology similar to astrocytes or oligodendrocytes. All, however, showed increased proliferation rates. Long-term cultured Schwann cells did not grow in semi-solid agar whereas the Schwannoma cell-line, 33B, formed visible colonies. These immortalised Schwann cells could be passaged for over two years and the cells could be frozen and restored several months later.

To test whether the immortalisation was caused by the particular batch of fetal calf serum in use, several other batches were tested, all of which gave similar results. All batches of fetal calf serum tested permitted the immortalisation of Schwann cell cultures, as described above. In addition to fetal calf serum, calf serum was tested and morphological changes occurred similar to those described for fetal calf serum although the process in calf serum tended to be slower than in fetal calf serum, and, on passaging, the transition to a flattened morphology did not always occur immediately. To date at least 20 independent cultures of Schwann cells have been immortalised.

The molecular phenotype of long-term cultures of Schwann cells was compared to that previously described for short-term cultures and Schwann cells in vivo using specific antisera (Jessen et al. 1990). All of the cells expressed the glial-specific calcium-binding protein, S100, indicating that they were from a glial lineage and not derived from the few S100-negative epineurial fibroblasts present in the original Schwann cell cultures. In these purified Schwann cell cultures, the fibroblasts did not proliferate and were lost on passaging. The majority of markers expressed by longterm cultured Schwann cells were the same as those expressed by short-term Schwann cells after the initial purification period. Thus, long-term Schwann cell cultures expressed GFAP, 217c/NGFR, A5E3, LI and N-CAM, although 217c/NGFR expression was at a lower level than on short-term Schwann cells. All Schwann cells examined were negative for the fibroblast marker Thy 1.1 (Table 1).

The markers of Schwann cell differentiation, GalC, 04 and P_o require contact with intact axons for expression and are thus lost from short-term cultured Schwann cells during the first few days of culture. Interestingly, some reexpression of these markers occurred in long-term cultures. Cells expressing low levels of the myelin protein P_o were seen together with a variable proportion of cells expressing the lipids Gal C and sulphatide (04).

Short-term cultured Schwann cells did not express detectable levels of fibronectin and only contaminating fibroblasts were positive for this molecule (Fig. 2), in agreement with previous work (Pleasure et al. 1985). In these cultures, only contaminating fibroblasts express fibronectin. Long-term cells, however, were clearly seen to express fibronectin, both on the cell surface and within the cytoplasm. This indicates that the cells both produced fibronectin and bound it through extracellular receptors. The significance of the elevation of fibronectin levels in long-term cultures is not clear but it was the major difference in molecular phenotype detected between the non-proliferating short-term and proliferating long-term Schwann cell populations.

The reason for the very low rate of DNA synthesis in short-term cultured Schwann cells, even in the presence of serum, is not understood. We have demonstrated previously that both long- and short-term cultured Schwann cells secrete PDGF and yet only the long-term cultured Schwann cells responded to this PDGF by DNA synthesis (Eccleston et al. 1990). A possible explanation for the lack of DNA synthesis in short-term cultures is that they secrete an autocrine growth inhibitor. Loss of inhibition would then contribute to the increase in the division rate of long-term Schwann cell cultures. In order to test this hypothesis, conditioned medium from short-term cultures was transferred to long-term Schwann cell cultures. The DNA synthesis of long-term cultured Schwann cells was inhibited by short-term Schwann cell conditioned medium in comparison to cells grown in non-conditioned medium (Fig. 3A). Similarly, medium conditioned by other long-term Schwann cell cultures did not inhibit DNA synthesis in long-term cultured Schwann cells.

The possibility that short-term cultured Schwann cells secrete a growth inhibitor was investigated further by examining the effect of conditoned medium on mitogen-stimulated short-term cultures. Medium conditioned by short-term Schwann cells (STCM) inhibited DNA synthesis in short-term cultured Schwann cells stimulated to divide in the presence of serum by 8-bromo cAMP (Fig. 3B) or GGF (Fig. 3C) compared to cells stimulated in freshly changed non-conditioned medium. This growth inhibition was not due to the comitogen in serum being depleted in STCM since the inhibitory effect was observed when serum-free conditioned medium was supplemented with fresh 10% calf serum.

The growth inhibitory activity was not detectable in medium conditioned by long-term cultured Schwann cells, the transformed Schwannoma cell line 33B, astrocytes, epineurial fibroblasts or the Swiss 3T3 cell-line (Table 2), indicating a possible cell-type specificity of the inhibitory molecule. In addition STCM did not inhibit serum-induced DNA synthesis in epineurial fibroblasts, 33B cells or Swiss 3T3 cells. For example, in experiments with epineurial fibroblasts, control cultures in fresh tissue culture medium had a labelling index of 19.5± 1.3 (n=3) whereas those grown in three different batches of STCM had labelling indices of 21.8±2.5 (n=3), 20.0±0 (n=2) or 20.6±1.4 (n=3), respectively. This indicates both a specificity in the action of the growth inhibitory activity and a lack of toxicity. We do not think that we are looking at a toxic phenomenon since the Schwann cells can be cultured for several months with half of their medium changed only once a week. They then do not die, but begin to proliferate rapidly.

The growth inhibitory activity in STCM was heat stable since it was not destroyed by incubation at 60 °C for 20min or by boiling for 10min. It was also stable to acid, withstanding lh at pH 2 at room temperature. The growth inhibitory activity was destroyed by trypsin, both when trypsin/agarose and soluble trypsin were used. It was not destroyed by collagenase treatment thus indicating a non-collagenous protein component to the growth inhibitory molecule (Table 3).

The growth inhibitory activity in STCM was not transforming growth factor-β (TGF-β) since this molecule is a mitogen for short-term cultured Schwann cells (Eccleston et al. 1989b; Ridley et al. 1989) The inhibitory activity in the STCM differs from gamma-interferon since it inhibits both GGF- and cAMP-stimulated division whereas gamma-interferon only inhibits cAMP stimulated DNA synthesis. In addition, when STCM was pretreated with a neutralising mono-clonal anti-gamma interferon (DB1) (Van der Meide et al. 1986), which had previously been shown to neutralise gamma-interferon activity on Schwann cells (Kingston et al. 1989), its inhibitory activity remained (results not shown). Similarly, when DB1 antibodies (1:10 – 1:40) were added to short-term cultured Schwann cells, their DNA synthesis rate was not increased (results not shown).

Other factors that were known to inhibit growth of some cell types were tested. 116, EGF and ANP did not inhibit DNA synthesis of long-term cultured Schwann cells, or short-term cells stimulated to divide by elevation of cAMP. Partially purified activin preparations had a similar effect to TGF-β on Schwann cell DNA synthesis. Activin stimulated DNA synthesis in short-term cultured Schwann cells and inhibited longterm cells (Table 4).

Schwann cells normally grow very slowly in tissue culture, even in the presence of serum. They divide in response to GGF, acidic and basic FGF, TGF-βs and PDGF (Davis and Stroobant, 1990; Eccleston et al. 19896, 1990; Ratner et al. 1988; Ridley et al. 1989; Weinmaster and Lemke, 1990) all of which synergise with forskolin to mount a more significant response. In the present study, we show that Schwann cells, in tissue culture, secrete a growth inhibitory activity that could in part explain the lack of response to serum. In addition, our results are consistent with the idea that a reduction or lack of response to the growth inhibitor might contribute to the increase in DNA synthesis that occurs on Schwann cell immortalisation.

Whilst cell immortalisation has not been defined at the molecular level, it is thought to represent an early change in progression towards transformation. The present study shows that if Schwann cells are cultured for several months they undergo a spontaneous change in growth control and they can be passaged indefinitely. This change occurs in the presence of serum but without the addition of specific mitogens and without passaging. This is in contrast with a similar immortafisation of Schwann cells described by Porter et al. (1986, 1987) who passaged Schwann cells in the presence of mitogens for several months before they gained the ability to proliferate without added mitogens. Their results suggested that both immortalised and short-term Schwann cells secrete an autocrine growth factor. We have shown that PDGF is one of the factors responsible for the proliferation of long-term cultured Schwann cells and that short-term cultured Schwann cells also secrete a PDGF-like molecule (Eccleston et al. 1990).

In the rat Schwann cell culture system described here, we suggest that negative growth control is gradually lost or reduced, giving rise to a dividing population of cells that responds to serum and autocrine growth factors. Mitogenic activities have been extracted from neurofibromas (Brockes et al. 1986; Ratner et al. 1990; Riccardi, 1986) but it is not clear whether such mitogens can be extracted from normal, mature peripheral nerves. However, Raff et al. (1978) reported that mature bovine peripheral nerves, in contrast to extracts of other tissues tested, contain an inhibitory activity for the growth of cultured rat Schwann cells. It is possible that a change during tumour formation might involve increased responsiveness of the Schwann cells to mitogens. This could involve loss of or reduction in response to a growth inhibitor which could alter the balance between mitogens and growth inhibitors leading to a net increase in growth without increasing the amount of available growth factors. The best characterised tumour suppressor genes (p53, retino-blastoma) are nuclear phosphoproteins (for reviews see Levine, 1990; Seizinger and Breakfield, 1990). Loss of a growth inhibitory factor gene is another genetic alteration that could lead to loss of growth regulation. The results presented here indicate that the intracellular growth inhibitory pathway of the long-term cultured Schwann cells is intact and introduce the possibility that an autocrine growth inhibitory molecule could be lost from these cells.

Fibronectin is a high molecular weight glycoprotein of the extracellular matrix, which has a role in cell migration, adhesion and differentiation (Alitalo and Vaheri, 1982; Hynes, 1985; Yamada, 1983). Its ex-pression is clearly elevated in long-term Schwann cells while it is undetectable in association with short-term Schwann cell cultures. Fibronectin stimulates proliferation of Schwann cells (Baron-van Evercooren et al. 1982). Agents that elevate cAMP in the Schwann cell have been shown to induce fibronectin expression in Schwann cells, as well as stimulating proliferation (Baron-Van Evercooren et al. 1986). It is uncertain whether this fibronectin elevation is necessary for cAMP to stimulate cell division. Fibronectin is expressed not only by long-term Schwann cells, but also by three different Schwannoma cell lines (P. A. Eccleston, unpublished observations) and by tumour cell lines from the central nervous system (Kennedy et al. 1987; Rajaraman et al. 1978). In vivo, fibronectin expression is elevated in many types of tumours including those in the central and peripheral nervous systems (Kennedy et al. 1987; Peltonen et al. 1988). Thus, fibronectin expression can be correlated with glial cell proliferation, both in vivo and in some culture systems. Future studies are necessary to determine whether Schwann cells express fibronectin in early development when they are proliferating rapidly and if there is a role for fibronectin in Schwann cell profiferation.

Two types of negative growth regulators have been described; those that, like the TGF-βs, act on a wide variety of cell types and those that exhibit cell type specificity (Wang and Hsu, 1986). The growth of Schwann cells in culture might be regulated by stimulatory and inhibitory factors. Purified short-term Schwann cells secrete a PDGF-like molecule (Eccleston et al. 1990) and it now seems likely that they also secrete an autocrine growth inhibitor, which holds their proliferation in check and may explain their lack of response to the PDGF that they secrete. This growth inhibitory autocrine loop may account for the lack of division of short-term cultured Schwann cells and a perturbation in its function may have a role in the elevated DNA synthesis rate of long-term cultures and of tumour Schwann cells in vivo. We have not demonstrated growth inhibitory activity in the medium of long-term cultured Schwann cells. However, the lack of inhibition by conditioned medium from these cells might be due to an increase in mitogen secretion by these cells, which could mask the effects of an inhibitor. Long-term cultured Schwann cells have an elevated intracellular cAMP level (Stewart et al. 1991), which might render the cells more responsive to the mitogens that they secrete as well as those in serum by elevating growth factor receptors (Weinmaster and Lemke, 1990). It is possible that the Schwann cell cAMP growth inhibitor could decrease Schwann cell cAMP levels thus reducing their response to mitogens.

Although the growth inhibitory activity described here remains to be characterised, it is not TGF-β1 or TGF-β2 since these molecules are mitogens for shortterm cultured Schwann cells (Eccleston et al. 19896; Ridley et al. 1989), although it does inhibit growth of long-term cultured Schwann cells (Eccleston et al. 19896). Similarly, activin synergised with GGF and 8-bromo cAMP to elevate DNA synthesis and like TGF-β1 it inhibited DNA synthesis of long-term cultured Schwann cells. A molecule that has no effect when added to short-term Schwann cell cultures, but which inhibits long-term and cAMP-stimulated short-term cultures is gamma-interferon (Eccleston et al. 19896). Thus, an intrinsic Schwann cell interferon-like molecule might have a role in growth inhibition in the peripheral nerve. However, gamma-interferon does not inhibit DNA synthesis in short-term Schwann cells stimulated by GGF, and neutralising anti-gamma interferon had no effect on Schwann cell DNA synthesis indicating that the Schwann cell secreted growth inhibitor is distinct from gamma-interferon as well as TGF-β. The Schwann cell growth inhibitor was not type 1 collagen, the only known protein that inhibits proliferation of cultured Schwann cells (Eccleston et al. 1989a), since collagenase treatment did not destroy the inhibitory activity and, unlike type 1 collagen, the activity was heat stable. Since the growth inhibitory activity was destroyed by trypsin, it seems likely that the molecule is a protein although the possibility that it is a proteoglycan has not been ruled out.

A recent study by Muir et al. (1991) reported a similar autocrine Schwann cell growth inhibitory activity associated with a protein of M_r 55K secreted by Schwann cells and purified from the RN22 Schwannoma. The activity reported here appears to differ in at least two respects from this protein. First, it is heat stable whereas heating to 90°C for 20 min greatly diminished inhibition by the 55K protein described by Muir et al. Second, the medium from our short-term Schwann cells appears to inhibit the DNA synthesis of immortalised Schwann cells to a much greater extent than the medium from RN22 cells which contains the 55K protein (Muir et al. 1990).

In conclusion, our results suggest that short-term cultured Schwann cells secrete a growth inhibitory factor(s) into the medium that has a growth regulatory function. This growth inhibitor might have an important role in the developmental cessation of division that precedes Schwann cell differentiation (Morgan et al. 1991). It is distinct from TGF-β1, activin, gamma interferon, IL-6, EGF and ANP and some cell-type specificity is apparent. When cultured for several months Schwann cells begin to proliferate and non-clonal cell fines can be established, which can be passaged for over a year. This immortalisation could be related to the loss of the growth inhibitory activity from long-term cultured Schwann cells.

This work was supported by a grant from Action Research for the Crippled Child. We are grateful to W. D Richardson and J. Gavnlovic for helpful discussion and to J. Gavnlovic for comments on the manuscript. We thank M. Rosendaal for help with culture in agar and Suzie Okyere-Debrah for typing the manuscript. We are grateful to K. L. Fields, R. Hynes, J Brockes, D. Y. Mason, E. Johnson, A. F. Williams, B. Ranscht and I. Sommer for providing antibodies and to A Goodearl, E. C. Collarini, J. C. Smith and J. Saklatvala for providing GGF, cultured astrocytes, activin (XTC-MIF) and ÌL3, respectively.