04 and A007-sulfatide antibodies bind to embryonic Schwann cells prior to the appearance of galactocerebroside; regulation of the antigen by axon-Schwann cell signals and cyclic AMP

During the development of a peripheral nerve, Schwann cells with flattened sheet-like extensions located mainly at the periphery of the primitive nerve are transformed, through proliferation and a series of morphological rearrangements with respect to axons and the extracellular matrix, into the mitotically quiescent non-myelin-forming and myelin Schwann cells found in the adult nerve (Peters and Muir, 1959; Diner, 1965; Gamble, 1966; Friede and Samorajski, 1968; Webster et al. 1973; Webster and Favilla, 1984; Ziskind-Conhaim, 1988). These two forms of mature Schwann cell are thought to originate from a common precursor cell. In the rat sciatic nerve, the first indication of divergent maturation along one of the two adult pathways is the axonally regulated appearance of galactocerebroside on the first Schwann cells induced along the myelin pathway, seen from embryo day 18 onwards in the rat sciatic nerve (Mirsky et al. 1980; Jessen et al. 1985; 1987b).

Prior to this time, during the period from embryo day 15 to day 18, early Schwann cells in the rat sciatic nerve are known to express the cell surface proteins Ran-1(217c) and A5E3, the cell adhesion molecules Ll/Ng-CAM and N-CAM, laminin and vimentin (Jessen et al. 1989; Jessen et al. 1990 and unpublished observations) and NGF receptors (Taniuchi et al. 1986; 1988; Yan and Johnson, 1988). None of these molecules, however, show significant or well-defined changes in expression during this early time period that could be used to analyse the underlying developmental process and the regulatory factors controlling it.

A developmentally regulated lipid antigen defined by the monoclonal antibody 04 was first described by Sommer and Schachner (1981) and was shown to bind to cells of the oligodendrocyte lineage, appearing on these cells before galactocerebroside (Sommer and Schachner, 1981; Schwab and Caroni, 1988; Wolswijk and Noble, 1989). It was also present on about 80 % of spindle-shaped Schwann cells in short-term neonatal mouse dorsal root ganglion cultures (Schachner et al. 1981) and on 70 %-80 % of all non-neuronal cells in 24 h cultures from chick sympathetic and dorsal root sensory ganglia throughout the period of development from embryo day 7 to 16. These 04+ cells were almost certainly Schwann or satellite cells associated with either axons or neuronal cell bodies within the ganglia (Rohrer, 1985; Rohrer et al. 1985; Rohrer and Thoenen, 1987). The developmental properties of the 04 antigen and their distribution on rat Schwann cells in situ or in culture has not previously been described. It has been reported, however, that 67 % of cells from a transformed rat Schwann cell line, D6P2T, show surface labelling with 04 antibodies on growth in serumcontaining medium. In these cells, the 04 antibody reacts with sulfatide in lipid extracts separated on TLC plates, and the antibody reacts with purified sulfatide in dot blot experiments (Singh and Pfeiffer, 1985; Bansal and Pfeiffer, 1987). Evidence from studies with polyclonal antibodies to sulfatide has indicated, however, that sulfatide can be detected on both Schwann cells and oligodendrocytes only after the first appearance of galactocerebroside (Ranscht et al. 1982) and thus it has remained an open question whether, in some cells, the 04 antibody detects another antigen in addition to sulfatide (Bansal et al. 1989).

In this paper, we have shown that the lipid antigen 04 is an early Schwann cell differentiation antigen in the rat sciatic nerve. We report on its developmental appearance, its regulation by axonal contact or cAMP and on studies of its chemical identity. In conjunction with these studies, we show that another sulfatide antibody, the monoclonal antibody A007, gives similar results to those obtained with 04 antibodies.

Cell suspensions of sciatic nerves were prepared from Sprague-Dawley rats. Embryos, 15-19 days old, newborn, 5 day and 30 day or older rats were used. The sciatic nerves or cervical sympathetic trunks were excised and where possible the epineurial sheath was removed. The tissue was placed in an enzyme mixture containing 2 mg ml-¹ collagenase (Worthington), 1.25 mg m1-¹ trypsin (Gibco) in Dulbecco’s Modified Eagle’s Medium (DMEM) without calcium or magnesium, and was then chopped finely. The tissue was incubated at 37°C and 5 % CO₂ for 90 min (30 day or older rats), 40 min (5 day old rats), or 20 min (newborn and embryo rats). An equal volume of Minimal Eagle’s Medium with 0.02 M HEPES buffer (MEM-H) containing 10 % calf serum was added and the tissue gently dissociated through a plastic pipette tip. The cells were centrifuged for 10 min at 500 g and resuspended either in MEM-H with 10 % calf serum and antibody, if they were to be labelled in suspension, or in DMEM with 10 % calf serum or supplemented DMEM (Jessen et al. 1990) with 0.1 % calf serum, if they were to be placed in culture.

Cell suspensions from sciatic nerves, prepared as above in DMEM with 10 % FCS or with modified DMEM containing 0.5 % calf serum were plated in 50 td droplets on to poly-L-Iysine-coated glass coverslips, 13 mm in diameter, in a 24-well multiwell,Plate and kept at 37°C in an incubator gassed with 5 % CO₂/95 % air. After 3 h, the cultures were topped up with 400 μJ DMEM containing 10 % FCS or with supplemented DMEM containing 0.5 % calf serum and maintained for various time periods. In some experiments 10-⁵ M-cytosine arabinoside was added for 48 h, 16 h after plating, to eliminate fibroblastic cells (Brockes et al. 1979). Cyclic AMP analogues, cholera toxin (150 ng ml-¹) or forskolin (2 μM) were added to some cultures from embryo day 15 or newborn sciatic nerves. The embryo day 15 cells were cultured in the presence of irradiated 3T3 cells in fully defined medium and 0.1 % calf serum. Forskolin was added at 24 h and then replaced at 72 h. Newborn Schwann cells were cultured for 5 days in DMEM plus 10 % calf serum, after which the medium was changed to DMEM plus 0.1 % calf serum. Cyclic AMP analogues, 5x10-⁴M 8-bromocyclic AMP plus 5x10-⁴M dibutyryl cAMP were added at day 5 and replaced by 5 × 10-M doses of each analogue at days 6 and 7, cholera toxin was added for 1 h at days 5 and 6, and forskolin was added at day 5 and replaced on days 6 and 7. Cells from cultures of both ages were labelled with 04 and 007 antibodies 4 days after first addition of the cAMP analogues, cholera toxin or forskolin. Dissociated cell cultures were prepared from dorsal root ganglia (DRG) of 8 day rats essentially as described previously (Mirsky et al. 1986). The culture media used were the same as those described above for sciatic nerve cultures except that NGF (lOOngml-¹) was added as a supplement to the medium.

Adult Sprague-Dawley rats, weighing 90-100 g, were anaesthetized and the left sciatic nerve cut 2-4 mm below the sciatic notch and a 2-3 mm segment of the nerve removed. The proximal stump was ligated and rerouted into an adjacent muscle. After surgery, rats were kept for up to 2 months before the nerves were removed for dissociation and analysis.

Supernatant fluid containing mouse monoclonal antibody 04 (Sommer and Schachner, 1981) was used routinely at a dilution of 1:1. It is of the IgM subclass. Rabbit antiserum to bovine S-100 protein (DAKO lmmunoglobulins) was used at a dilution of 1: 400 in dried and fixed preparations, and at l: 800 on cells in dissociated cultures. Mouse monoclonal antibody 217c, in the form of culture supernatant, was used at a dilution of 1: 500. This antibody, first described by Peng et al. (1982) has been shown by Fields and Dammerrnan (1985) to show extensive similarities with the anti-Ran-1 originally described by Fields et al. (1975). Rat monoclonal antibody A007 in the form of culture supernatant, was used at a dilution of 1: l. It is of the IgM subclass and has been well characterized using TLC and ELISA assays and is essentially specific for sulfatide (K. Stefansson, personal communication). Monoclonal antibody to galactocerebroside (Ranscht et al. 1982) in the form of ascites fluid was used at a dilution of 1: 200. Rabbit antiserum to N-CAM (Gennarini et al. 1984) was used at a dilution of 1:500. Rabbit antiserum to galactocerebroside (Hirayama et al. 1984) was used at a dilution of 1: 100. Tetramethyl rhodamine conjugated to goat anti-mouse lg (G anti-Mlg-Rd) (Cappel Labs.), absorbed with rabbit lg to remove cross-reacting antibodies, was used at a dilution of 1: 100. Fluorescein conjugated to goat anti-rabbit lg (G-anti-Rlg-Fl) absorbed with mouse lg to remove cross-reacting antibodies was used at a dilution of 1: 100. Tetramethyl rhodamine conjugated to goat anti-rat lg(G anti-Ratlg-Rd) (Nordic Ltd) was used at a dilution of 1: 100. Donkey antirabbit Jg or anti-rat Jg, biotinylated and streptavidin-fluorescein or streptavidin-Texas Red (Amersham International pie) were used at a dilution of 1: 100. Control antibodies, an irrelevant monoclonal lgM for 04 or IgG2n for 217c, and normal rabbit serum for the rabbit antibodies, were used at appropriate dilutions.

Antibodies were diluted as described previously (Jessen et al. 1987b) and all incubations carried out for 30 min at room temperature.

These were immunolabelled with 04 or 217c antibodies by adding the antibody to cells suspended in MEM-H with 10 % calf serum. After 30 min cells were washed and incubated with G-anti-Mlg-Rd for 30 min, then dried on to gelatin-coated slides. In experiments where cells were double labelled with 04 antibodies and rabbit antiserum to galactocerebroside, cells were labelled sequentially with 04 antibodies, G-anti-Mlg-Rd, rabbit anti-galactocerebroside and G-anti-Rlg-Fl prior to drying on to microscope slides. Cells that were to be double labelled with 04 and S-100 antibodies were first labelled in suspension with 04 antibodies and G-anti-Mlg-Rd, then dried on to slides, rehydrated on the slide for 20 min in 4 % para formaldehyde in phosphatebuffered saline (PBS), washed, then permeabilized for 10 min in 95 % ethanol/5 % acetic acid. After thorough washing, cells were sequentially labelled with S-100 antibodies, biotinylated donkey anti-rabbit lg followed by streptavidin-fluorescein.

These were incubated with 04 antibodies followed by G-anti-Mlg-Rd, fixed in 4 % paraformaldehyde for 20 min, followed by 95 % ethanol/5 % acetic acid for 10 min. A similar procedure was followed with the rat monoclonal A007 except that G-anti-Ratlg-Rd was used as a second layer or alternatively biotinylated donkey anti-rat lg followed by streptavidin-Texas Red. The cells were then incubated in S-100 antibody followed by G-anti-RJg-Fl. In some experiments, 04 labelling was followed by sequential labelling with rabbit anti-galactocerebroside antibodies followed by G-anti-Rig-Fl. In this case, cells were postfixed for 10 min in 4 % paraformaldehyde in PBS. In some experiments, embryonic cells were cultured for 2 h prior to labelling with 04 or A007 antibodies. In this case, cells were fixed for 10 min in 4 % paraformaldehyde prior to immunolabelling with 04 or A007 antibodies, subsequent steps in the procedure being identical to those described above.

Sciatic nerves from adult rats were desheathed, fixed in 4 % paraformaldehyde in PBS for 20 min, rinsed in PBS and gently teased out on to a polylysine-coated microscope slide in a drop of PBS, using 23-gauge needles as described previously (Jessen and Mirsky, 1984). They were allowed to dry for several hours before application of 04 antibodies, followed by G-anti-Mlg-Rd or rat A007 antibodies, followed by G-anti-Ratlg-Rd.

Fibres were then labelled with antibodies to N-CAM followed by G-anti-Rlg-FI. Preparations were post-fixed for 10 min in 4 % paraformaldehyde in PBS.

All preparations were mounted in Citifluor anti-fade mounting medium and viewed for immunofluorescence with a Zeiss microscope using x63 oil-immersion or x40 dry phase contrast lenses, epi-illumination and rhodamine and fluorescein optics.

Sciatic nerves were excised from adult and newborn rats and immediately frozen on dry ice. Lipids were extracted from the nerves using chloroform: methanol: H₂O (5: 10: 3 by vol) and separated into upper and lower phases using Folch partition (partition 1) as described previously (Dubois et al. 1986; Rougon et al. 1986).

Lipids from the upper phase were desalted on a Sephadex G-50 column (20×l.2cm) in H2O then lyophilized. Lipids from the lower phase were submitted to alkali treatment to hydrolyse both neutral lipids and phospholipids using 0.3 M KOH in methanol: H₂O (95: 5 by vol) for 2 h at 37°C. After desalting on a Sephadex G25 (Pharmacia) column in chloroform: methanol: H₂O (60: 30: 4.5 by vol), lipids were passed through a Unisil (Clarkson Chemical Co. Inc) 100-200 mesh column (10x0.6cm) in chloroform. Hydrophobic lipids were first eluted with 10 column volumes of chloroform. Then, more polar lipids were eluted with 6 column volumes of two different chloroform-methanol mixtures (4:1 and 1: l by vol) successively. Lipids from both upper and lower phases were then separated into neutral and acidic lipids, respectively, by running on a column of DEAE-Sephadex in the acetate form (6x0.6cm). The neutral glycolipids were eluted with chloroform: methanol: H₂0 (30: 60: 8 by vol) and then the acidic lipids were eluted successively in batches with ten column volumes of 0.05M, 0.lM and 0.5M ammonium acetate in methanol. The fractions were desalted using a C18 sep-Pack Cartridge.

Lipids were chromatographed on aluminium-backed highperformance thin-layer chromatography plates (HPTLC) (silica gel 60, E Merck) either in chloroform: methanol: 0.25 % KCl in H2O (5: 3:1 by vol), solvent A, or in chloroform: methanol: H2O (65:25:4 by vol), solvent B.

Chromatographed glycolipids were visualized with 0.5 % orcinol in 2 M H₂SO₄ or with Azur A reagent (Green and Robinson, 1960; Kean, 1968) which is specific for the sulfate groups on carbohydrates.

Lipid antigens were then analysed by immunostaining of chromatograms with the monoclonal antibody 04 as described previously (Magnani et al. 1982). Hybridoma supernatant containing the antibody 04 was diluted 1: 4 with buffer A (0.05 M Tris, 0.15 M NaCl pH 7.8 containing 1 % bovine serum albumin (BSA) and 0.01 % sodium azide) and the HPTLC plate was incubated in the mixture for 3-5 h at room temperature or overnight at 4°C. After washing in buffer A, the plate was incubated with 2×1a6ctsmin-¹ m1-¹ of mI-labelled goat anti-mouse IgM antibodies (Jodogen method). After 1 h at room temperature, the plate was washed in cold phosphatebuffered saline (PBS), dried and exposed to Kodak XAR 5 X-Ray film. After exposure to X-ray film glycolipids on HPTLC plates were visualized with Orcinol reagent.

Glycolipid antigens recognized by the 04 antibody were desulfated using 0.05 M HCl in dry methanol at room temperature for 16 h.

In some experiments, antigens contained in the non-alkalitreated upper phase were alkali treated as described above. After neutralisation, lipids were desalted on Sephadex GSO in H₂O. Treated glycolipid antigens were then analysed by immunostaining of chromatograms with the monoclonal antibody 04 as described above.

This was carried out as described previously (Jessen et al. 1984) on samples of newborn and adult sciatic nerve and adult brain, using 04 monoclonal supernatant diluted 1: 10 with 3 % hemoglobin in PBS. The second layer was ¹²⁵1-labelled rabbit anti-mouse lg (Amersham International pie).

To determine when 04 binding first appears on Schwann cells during the development of the rat sciatic nerve, sciatic nerves were dissected from rats at various ages from embryo day 15 to postnatal day 1 and dissociated. Cells were plated on to coverslips and immunolabelled within 2 h using 04 antibodies. The results are shown in Fig. 1. Briefly, very few (0<0.5 %) 04⁺ cells were present at embryo day 15 to 16, whereas over 95 % of all the cells in the nerve were 04⁺ on the first postnatal day. To demonstrate that the 04⁺ cells are restricted to the Schwann cell lineage, cells from embryo day 20 and postnatal day 1 sciatic nerves were double labelled with S-100 and 04 antibodies. At embryo day 20, 99 % of S-100⁺ cells were also labelled with 04 antibody and at postnatal day 1, this figure was 100% of S-100⁺ cells. No 04⁺, S-100^- cells were seen at these time points showing that 04⁺ cells are restricted to the Schwann cell lineage.

To test whether another antibody that is highly specific for sulfatide gives similar results to those obtained with 04 antibody, these experiments were repeated using rat monoclonal A007-sulfatide antibody.

The results were essentially similar, with few positive cells present at embryo day 16, 35.3 % labelled at El 7 and over 95 % of cells labelled on the first postnatal day, all of which were also labelled with S-100 antibodies.

The appearance of the first 04+ cells, or cells binding the A007 sulfatide antibody at embryo day 16, precedes the appearance of the first galactocerebroside+ Schwann cells by two days. Several experiments were carried out to confirm that 04 and A007 binding appears significantly before galactocerebroside is detectable. In one experiment in which cells from embryo day 18 were double labelled with 04 and galactocerebroside antibodies, 55 % of the cells present were 04⁺ while none were galactocerebroside . In another experiment, 93 % of cells from embryo day 20 were 04⁺ and of the 04⁺ cells only 71 % were also galactocerebroside⁺. Furthermore, at postnatal day 0 all Schwann cells bind 04 and A007 sulfatide antibodies (Fig. 2), whereas previous work has shown that approximately 60 % of Schwann cells are galactocerebroside⁺ at this time (Mirsky et al. 1980; Jessen et al. 1985). We also found that in the cervical sympathetic trunk 93 % of Schwann cells were 04⁺ at embryo day 20, and of these only 8 % were galactocerebroside+ in agreement with a previous study of galactocerebroside appearance in this nerve (Jessen et al. 1985).

The distribution of 04 and A007 labelling on Schwann cells in adult rats was investigated using teased nerve preparations. The Schwann cells of unmyelinated fibres were selectively labelled with antibodies to N-CAM (Mirsky et al. 1986) and the myelinated fibres were recognized morphologically. Both non-myelin-forming and myelin Schwann cells were labelled with 04 antibody (Fig. 3). These experiments were repeated with A007 sulfatide antibody, which gave similar results with both myelinated and unmyelinated fibres being labelled.

To study the regulation of 04 antigen expression we first asked whether cells that have not yet become 04⁺in vivo would become 04⁺ on schedule in culture. Schwann cells were prepared from embryo day 15 sciatic nerves, placed in culture and examined with 04 antibodies after 2 days and 5 days. No 04⁺ cells were found in these cultures. The same results were obtained when the cells were examined with A007 sulfatide antibodies. This suggests that the triggering of 04 and A007 antigen expression is not intrinsically programmed but depends on extrinsic signals.

Further studies on the regulation of 04 and A007 binding were carried out on cells from postnatal nerves using methods similar to those used previously in (DRG) from 8 day old rats, 04+ cells also disappear over the first eight days in culture. These cultures contain regenerating neurites which clearly are unable to provide the signal provided by mature axons in the nerve, behaviour also seen in the case of the disappearance of galactocerobroside from Schwann cells in freshly dissociated DRG cultures (Mirsky et al. 1980). Evidence has suggested that cyclic AMP is an important second messenger in Schwann cells, mediating some of the effects of axons on Schwann cell differentiation. We therefore tested whether cAMP analogues and agents such as cholera toxin and forskolin, which elevate intracellular cAMP, would induce re-expression of the antigen(s) recognized by the 04 and A007 sulfatide antibodies. It was found that these treatments induce both 04 and A007 binding on a substantial proportion of cultured Schwann cells from both poststudies on the regulation of galactocerebroside and myelin associated proteins (Mirsky et al. 1980). Schwann cells dissociated from sciatic nerves of 5 day old rats were placed in culture and 04 labelling monitored over the next few days. It was found that the Schwann cells, all of which initially were 04⁺, slowly lost surface 04 binding so that by day 8 only 0.2±0.2 % of the Schwann cells were still positive (Fig. 4). When these experiments were repeated with A007 sulfatide antibody similar results were obtained, less than 1 % of S-100⁺ cells being labelled at 5 days in culture. These observations suggested that the ability of Schwann cells to bind 04 or A007 antibodies depends on signals from axons, or possibly on other factors in the endoneurium.

To show that the level of 04 antigen expression in vivo is likely to be regulated by axonal contact, rather than by other endoneurial factors, the 04 binding was examined on Schwann cells deprived of axonal contact in vivo by nerve transection. Schwann cells were removed from the distal stump of the sciatic nerve at various times after nerve cut, and 04 labelling monitored by immunohistochemical methods. It was found that 04 binding was substantially down-regulated on the surface of Schwann cells in the distal stump, which are deprived of contact with axons (Fig. 5), although traces of 04 labelling were still detectable at high magnification when examined eight weeks after nerve cut. By contrast, in nerves in which reinnervation had occurred after a crush lesion, 04 labelling was found associated with both myelinated and unmyelinated axons. Reinnervation was judged to have occurred by return of function in the leg muscles and by the morphological appearance of typical myelinated fibres and unmyelinated fibres in teased nerve preparations from the reinnervated distal stump.

In freshly dissociated cultures of dorsal root ganglia natal day 5 and embryo day 15. At postnatal day 5, Schwann cells treated for 72 h with cAMP analogues, cholera toxin or forskolin were 88±2 % 04⁺ (analogues), 78±1 % 04⁺ (CTx), and 36±11 % 04⁺ (forskolin) while at 6 days over 99 %, 89 % and 89 % respectively were 04⁺. Corresponding figures for the induction of A007 sulfa tide binding at 72 h were 73±8 %, 36±5 % and 12±6 %, respectively. At embryo day 15, the number of Schwann cells induced by forskolin was 60-70%04⁺ and 20-30%A007⁺, at 72h post-treatment. The same hierarchy of efficiencies is seen in induction of galactocerebroside, with cAMP analogues being relatively more potent than either cholera toxin or forskolin (unpublished observations).

It has been reported that the 04 antibody recognizes purified sulfatide in dot blots and on immunolabelling of TLC plates using lipid extracts from the Schwann cell line D6P2T, the two characteristic sulfatide bands are also recognized (Singh and Pfeiffer, 1985; Bansal and Pfeiffer, 1987). Results using polychlonal sulfatide antibodies, characterised using different methods, have, however, reported that, in both rat Schwann cells and oligodendrocytes, sulfatide appears on the cell surface after galactocerebroside (Ranscht et al. 1982), whereas 04 antigens appear before galactocerebroside (Sommer and Schachner, 1981; Schwab and Caroni, 1988; Wolswijk and Noble, 1989). We therefore decided to reinvestigate this point using extracts from sciatic nerves of newborn and adult rats to establish the identity of the lipid antigen recognized in the rat PNS.

Lipids extracted from newborn and adult sciatic nerves, and adult rat brain were analysed by immunolabelling of chromatograms with the monoclonal antibody 04.

It was clear that 04 antibody bound strongly to a glycolipid with the same chromatographic mobilities as purified sulfatide and not significantly to any other glycolipids isolated from these tissues (Fig. 6). In common with many glycolipids sulfatide runs as either one or two bands in most TLC solvent systems (see Figs 6, 7, 8). The bands correspond to differences in the ceramide moiety of the glycolipid. By chemical staining with orcinol (Fig. 6B), the molecular species of sulfatide corresponding to the lower band appears more abundant than the more diffuse upper band in the rat nervous tissues tested here, particularly in the newborn sciatic nerve when very little of the upper band is detectable. 04 monoclonal antibody binds to both sulfatide species although the affinity for the two forms appears to be slightly different depending on the loading of lipid on the chromatogram and the solvent system used. In Fig. 6A the heavier loadings of antigen used in lanes 1, 3 and 4 result in very low, pseudonegative staining of the lower band, which is, however clearly visualised by immunolabelling in Fig. 7C, and Fig. 8.

To further characterize this antigen, the glycolipids purified from different tissues were chromatographed in two different solvent systems and then either chemically stained with Azur A, which specifically detects sulfate groups on carbohydrate, or immunolabelled with 04 antibody. In both solvent systems (Fig. 7) the glycolipids recognized by the 04 antibody have the same chromatographic mobility as sulfatide when stained with Azur A, and the sulfatide is present in newborn sciatic nerve as well as adult nerve and brain. In both solvent systems, only one glycolipid is observed suggesting that in rat nervous tissues the 04 antibody detects only glycolipid migrating in the sulfatide positions. Furthermore, the active component is resistant to alkaline hydrolysis, and sensitive to treatment with HCI, which removes sulfate groups from carbohydrate (Fig. 8). The antigen molecule is present in both upper and lower phases from the Folch partition eluted from an anion exchange column with 0.1 M ammonium acetate. These properties all support the positive identification of the 04 binding antigen as sulfatide. In newborn sciatic nerve, the activity is associated mainly with the lower of the two ceramide forms, whereas in adult nerve and brain there appears to be a relative increase in the proportion of the upper ceramide form.

To test whether the 04 antibody recognises proteinassociated determinants, SDS-polyacrylamidegel electrophoresis of extracts of newborn and adult nerve, and adult brain were combined with immunoblotting. No positive bands were seen suggesting that the antibody does not recognize protein related epitopes in addition to sulfatide.

In the rat sciatic nerve, Schwann cells binding 04 or A007 sulfatide antibodies were first seen on cells removed at embryo day 16. At embryo day 18 between 30-50 % of cells bound the antibodies, and at birth essentially all Schwann cells were 04+ and A007⁺. Between embryo days 16 and 18, Schwann cells in the nerve surround large groups of axons and the 1: 1 segregation characteristic of later development of myelin Schwann cells has not yet occurred. Furthermore, galactocerebroside+ cells, which represent the first cells developing along the myelin pathway, are first detectable at embryo day 18, about two days after the first appearance of 04⁺ or A007⁺ cells. The advent of 04 and A007 sulfatide antibody binding therefore precedes the later developmental differentiation of the cells into either the myelin or non-myelin-forming pathways and is not specifically related to myelination. This conclusion is strengthened by two observations. First, although all Schwann cells bind 04 and A007 sulfatide antibodies at birth, these cells are a heterogenous population. Only about 60 % of them are galactocerebroside⁺ cells developing along the myelin pathway, while the remainder is a mixture of cells, some of which will subsequently be induced to myelinate and others of which will develop into non-myelin-forming Schwann cells, (Jessen et al. 1985). Second, in adult nerves, both non-myelin-formini and myelin-forming Schwann cells are 04⁺ and A007⁺ .

The ability of Schwann cells to bind 04 or A007 antibodies appears to be controlled by axonal contact in a very similar way to that seen in the well-established axonal regulation of galactocerebroside and myelin protein expression, (Mirsky et al. 1980; Jessen et al. 1987b; Lemke and Chao, 1988; Trapp et al. 1988). Schwann cells from embryo day 15 to 16 sciatic nerve will not acquire the ability to bind 04 or A007 antibodies in the absence of axons, and 04⁺ Schwann cells lose the ability to bind 04 antibody if removed from axonal contact either in situ after denervation or in culture. Similar results were obtained with A007 sulfatide antibody. Low levels of sulfatide synthesis are, however, detectable in long-term cultures of secondary Schwann cells indicating that synthesis of sulfatide is not completely abolished in the absence of axons (Fryxell, 1980). This is completely consistent with the present observation that at high magnification traces of 04 binding were still detectable on denervated Schwann cells in distal stumps of transected nerves. If crushed nerves are allowed to regenerate, the Schwann cells once more become 04⁺ indicating that, in regeneration as in development, axonal contact up-regulates levels of 04 binding on the Schwann cell surface.

There is evidence that the effect of axon-associated signals on the Schwann cell phenotype can in some circumstances be mimicked by agents that elevate intracellular cAMP levels (Baron van Evercooren et al. 1986; Sobue et al. 1986; Lemke and Chao, 1988; Shuman et al. 1988). Forskolin and cAMP analogues will, for instance, induce re-expression of surface galactocerebroside in cultured neonatal Schwann cells that have lost surface expression of the molecule due to removal from axonal contact. Our results show that expression of the antigen(s) recognized by the 04 or A007 sulfatide antibodies can also be induced by agents that mimic or generate high levels of intracellular cAMP. In the case of the neonatal cells, this represents re-expression of antibody binding, previously lost due to culturing without axons. In the embryo day 15 cells, however, cAMP induced 04 or A007 sulfatide antibody binding for the first time, since the cells were removed from the nerve before 04⁺ or A007⁺ cells appear during development. In being inducible by cAMP, the antigen(s) recognized by the 04 and A007 sulfatide antibodies behaves like other axonally induced Schwann cell molecules, including galactocerebroside and the myelin proteins.

One developmental event to which the appearance of 04 and A007 binding might be linked is the acquisition of a discrete basal lamina by Schwann cells. The appearance of the Schwann cell basal lamina has not been timed carefully in the rat sciatic nerve but the available evidence indicates that this occurs at about the time that all Schwann cells bind 04 and A007 sulfatide antibodies. The antibody binding studied in the present experiments is most likely to be essentially due to sulfatide (see below), and sulfatide has been reported to bind selectively to laminin (Roberts et al. 1985), a component of the Schwann cell basal lamina (Cornbrooks et al. 1983; Bunge et al. 1986; Eldridge et al. 1989). It seems possible that the appearance of sulfa tide in the Schwann cell membrane could be related to stabilization and formation of the basal lamina.

The 04 antigen expressed by Schwann cells is very likely to be sulfatide since no other antigen was recognized in either TLC or SDS-PAGE immunoblotting of neonatal or adult nerves and since another characterized sulfatide antibody, A007, gave similar results in the immunofluorescence experiments. 04 antibodies show some cross-reactivity with seminolipid in immuno-dot blot assays when the lipid is used at high concentrations (Bansal et al. 1989). When embryo day 15 Schwann cells were examined with anti-seminolipid antibodies (Goujet-Zalc et al. 1986), 55 % of the cells were immunolabelled, and showed a characteristic labelling pattern with a few fluorescent dots per cell; similar results were obtained with embryo day 18 cells (data not shown). Neither 04 nor A007 label any Schwann cells at embryo day 15 and at embryo day 18 the majority of cells are labelled, showing, particularly in the case of 04, a densely speckled immunofluorescence pattern. The contribution of seminolipid to 04 or A007 binding in the present experiments must therefore be minimal, if any. Although the results of labelling with 04 and A007 antibodies are in broad agreement some small differences can be detected between them. Both antibodies detect positive cells at embryo day 16, but the number of 04⁺ cells is always slightly greater than the number of A007⁺ cells in parallel experiments. The labelling with 04 antibodies is also stronger. Most probably the 04 antibody is binding with greater affinity than the A007 antibody. It is, however, always possible that although the major antigen detected in Schwann cells by both antibodies is sulfa tide another unidentified antigen is being recognized in addition by the 04 antibody.

Assuming that the main 04 antigen is sulfatide, it is perhaps surprising that 04 binding is detectable on the cell surface 2 days before galactocerebroside, since sulfatide is synthesized from galactocerebroside using the enzyme galactosyl sulfotransferase. There is considerable evidence, however, for the existence of two separate pools of galactocerebroside within cells (Benjamins et al. 1982; see Bansal and Pfeiffer, 1987 for discussion on this point). One of these pools is sulfated in the Golgi apparatus to form sulfatide which is then transported via vesicles to the cell surface (Townsend et al. 1984), and the second by-passes the Golgi and reaches the cell surface via a different mechanism. Our tests have not revealed why a discrepancy exists between studies using monoclonal antibodies recognizing sulfatide, including 04 antibody and A007 monoclonal anti-sulfatide, and those using rabbit antibodies to sulfatide. Monoclonal antibody studies suggest that sulfatide appears in development in the cell surface membrane of Schwann cells or oligodendrocytes before galactocerebroside, while rabbit antibody studies indicate that it appears later. One possibility is that the rabbit antibodies used in previous studies recognized primarily the upper of the two sulfatide species visible on the TLC plates, whereas the 04-antibody recognizes both the upper and lower species of sulfatide. The neonatal nerve contains very little of the upper species relative to the lower species. Thus neonatal Schwann cells would react poorly with antibodies recognizing primarily the upper species of sulfatide whereas cells from nerves of older animals where higher amounts of the upper species are present, would be more reactive in immunofluorescence tests. An alternative possibility could be that the monoclonal antibodies are of higher affinity than the polyclonal antibodies and thus would be capable of detecting relatively lower levels of sulfatide in the cell membrane in immunofluorescence tests.

Other lipid differentiation antigens in addition to sulfatide, including galactocerebroside and the 08 and 09 antigens, are present on mature Schwann cells in both myelinated and unmyelinated fibres (Jessen et al. 1985; 1987a; Eccleston et al. 1987). This suggests that the plasma membranes of the two adult Schwann cell variants differ less in lipid composition than in protein composition where there is a distinct difference between the two cell types. Thus non-myelin-forming Schwann cells in adult nerves express the cell adhesion molecules N-CAM and Ll/NgCAM, the cell surface proteins 217c(Ran-l), A5E3 and Ran-2, in addition to the intermediate filament protein GF AP (Jessen and Mirsky, 1984; Jessen et al. 1984; Mirsky and Jessen, 1984; Mirsky et al. 1985; 1986; Jessen et al. 1990). These molecules are down-regulated in myelin Schwann cells shortly after they start to express the major myelin glycoprotein P₀, myelin basic protein and MAG (Martini and Schachner, 1986; Jessen et al. 1987a; 1990). Therefore, it appears that it is differences in the regulation of protein expression in the myelin Schwann cell pathway rather than differences in lipid expression that generate and maintain the two distinct phenotypes (Jessen et al. 1987a; 1990).

We show here that 04 and A007 sulfatide antibodies label Schwann cells very early in nerve development (two days prior to galactocerebroside appearance) most probably due to binding to cell surface sulfatide. The ability of Schwann cells to bind significant levels of these antibodies depends on the cells retaining contact with axons. This indicates that expression of surface sulfatide in Schwann cells is up-regulated by axonSchwann cells signals, as previously shown for galactocerebroside and the myelin proteins.

Expression of myelin proteins in cultured Schwann cells can be induced by elevation of intracellular cAMP levels and it has been suggested that cAMP is involved as a second messenger in the axonal induction of myelination. Elevation of cAMP, however, also induces Schwann cells to express surface galactocerebroside and, as indicated in the present work, sulfatide. Neither molecule is specifically associated with the myelin phenotype. Therefore we suggest that if cAMP has a role in Schwann cell development that role is unlikely to be restricted to myelination, but is probably a more general one extending to the maturation of both Schwann cell types. The observation that cAMP analogues also increase the synthesis of basal lamina components in cultured Schwann cells further strengthens this conclusion.

We would like to thank Dr J P Brockes, Dr P A Eccleston, Dr KL Fields, Dr C Goridis, Dr B Ranscht, Dr M Schachner, Dr I Sommer and Dr K Stefansson for gifts of antibodies and Ms Katayoun Rezvani for help with experiments on embryonic Schwann cells. This work was supported in part by grants from the Medical Research Council of Great Britain, the Multiple Sclerosis Society, Action Research for the Crippled Child and Association Fran aise contre les Myopathies.