Immunological identification of a new 14×10³M_r membrane-bound protein in Torpedo electric organ

Torpedo electric organ has been proved to be invaluable material for the identification and purification of the essential components of peripheral cholinergic synapses. The electroplaques are derived from precursor myotubes (Ogneff, 1897; Fox and Richardson, 1979; Mellinger et al. 1978) and correspond to modified muscle fibres in which contractile elements have disappeared (Ranvier, 1878). They are highly polarized cells with a dorsal face exhibiting a dense network of tubular invaginations and possessing a high Na⁺,K⁺-ATPase activity (Zubler-Faivre and Dunant, 1976). The ventral face is richly innervated by cholinergic nerve endings (Feldberg et al. 1940; Feld-berg and Fessard, 1942), which cover its entire surface, and can be regarded as a large motor endplate. This disposition is very favourable for morphological studies. Indeed, a large number of synapses can be observed in every section of the electric organ. Moreover, it is possible to isolate single electroplaques by a rather simple dissection and to have en face views of the presynaptic apparatus on these thin flat cells.

In the present study, we report the characterization of a series of monoclonal antibodies to a 14×10³M_r membranebound polypeptide. This antigen is present at the membrane of various cell types in Torpedo electric organ and muscle, especially that of fibroblasts, which are immunolabelled very efficiently.

These antibodies were obtained in the course of the production of anti-proteolipid monoclonal antibodies (Morel et al. 1991). Balb/C mice were immunized with three intraperitoneal injections of large amounts (40 μg per injection) of delipidated proteolipids, extracted from Torpedo electric organ presynaptic plasma membranes (Israel et al. 1986). This proteolipid fraction contains a small amount of a 14×10³M_r polypeptide, which turned out to be much more immunogenic than the proteolipids (Morel et al. 1989). Screening for antibody-secreting hybridomas was performed on immunoblots of presynaptic proteolipids, as previously described (Morel et al. 1991).

SDS–polyacrylamide gel electrophoresis was carried out in 8% to 18% linear acrylamide gradient gels overlaid with 5% acrylamide stacking gels (Laemmli, 1970). Samples were solubilized in a 10% SDS lysis buffer. Electrophoresis was carried out at 40 mA constant current and stopped 15 min after Bromophenol Blue had run out of the gel.

Protein bands were transferred by electrophoresis onto 0.2 /an nitrocellulose filters (Schleicher and Schuell) according to the method of Towbin et al. (1979). Blots were saturated with Trisbuffered saline (TBS: 20 mu Tris buffer, pH 7.4, 200mM NaCl, 0.1% Tween 20) containing 1% bovine serum albumin (BSA). Antibodies diluted in TBS-BSA were allowed to bind overnight. Bound immunoglobulins were indirectly detected with peroxi-dase-conjugated anti-mouse IgG antibodies (Institut Pasteur Production, 1/1000 dilution in TBS-BSA for 2h). All washing steps were in TBS except the final one where NaCl was increased to 0.5 M. Peroxidase was revealed using diaminobenzidine and H₂O₂ as the substrates.

Molecular weight markers were purchased from BioRad Laboratories or Amersham (colored Rainbow markers).

Microtitration plates were coated with presynaptic plasma membranes (10 μg protein ml⁻¹ in 0.1 M sodium phosphate buffer, pH 8) for at least 48 h at 4°C. The antigen solution was then replaced with TBS-BSA. Monoclonal antibodies were allowed to bind for 3 h at room temperature. Bound immunoglobulins were detected using a second antibody conjugated to peroxidase (Institut Pasteur Production, 1/200 dilution in TBS-BSA). All washing steps were in TBS. Bound peroxidase activity was quantified using o-phenylenediamine and H₂O₂ as the substrates. The reaction was stopped with 3 M HC1.

Small pieces of Torpedo electric organ were fixed for 2 h in 2% formaldehyde in a phosphate-bufifered saline (PBS: 0.1M sodium phosphate buffer, 0.3 M NaCl, final pH 7.4). Isolated electroplaques were dissected out from fixed electric organ prisms. They were layered on gelatin-coated microscope slides in a 0.5 M Tris buffer, pH 7.5, drop and allowed to dry.

Muscle fibres were prepared from fixed muscles as previously described (Morel et al. 1989).

Electric lobe pieces were treated in two different ways. In some experiments, they were fixed as described above and then incubated in 0.5 M Tris buffer containing increasing sucrose concentrations (5, 10 and 20%). Tissue blocks were frozen in isopentane cooled by liquid nitrogen. Sections (10 μm) were mounted on gelatin-coated slides and stored at −70°C. In other experiments, paraffin-embedded sections (10 μm thick) were prepared from electric lobes fixed in Bouin’s solution. Paraffin was removed by several washes in toluene and ethanol and sections were progressively rehydrated.

Frozen sections of gills were prepared as described for electric lobe sections.

Antibodies diluted in 0.5 M Tris buffer containing 0.1% Triton X-100 and 1% BSA were allowed to bind overnight. Bound antibodies were visualized using fluorescein-copjugated antimouse IgG antibodies (Institut Pasteur Production, 1/500 dilution), or using biotinylated anti-mouse Ig antibodies and streptavidin-conjugated peroxidase (Amersham) and diaminobenzidine and H₂O₂ as the enzyme substrates.

Small pieces of electric organ were fixed for 3 h in 3% formaldehyde, 0.2% glutaraldehyde in PBS. After 1 h of incubation in 20 mM lysine in PBS, tissue pieces were soaked for 30 min at room temperature in 0.1% saponin in PBS containing 2% BSA. After washing out the detergent, monoclonal antibody 14K4 (1/4 dilution of a culture supernatant in PBS containing 2% BSA) was allowed to bind overnight. After extensive washing for 5–6 h, bound antibodies were indirectly visualized with goat antimouse Ig antibodies conjugated to colloidal 15 nm gold particles (purchased from Janssen, overnight incubation with a 1/10 dilution). All subsequent incubations were in 0.4M cacodylate buffer, pH 7.4. After a 3 h washing period, tissue pieces were fixed in 2% glutaraldehyde (Ih), postfixed in 2% OsO₄ (1h), dehydrated and embedded in Spurr’s resin. Ultrathin sections were stained with uranyl acetate and lead citrate.

Crude presynaptic plasma membranes (PSPM) were prepared from Torpedo electric organ according to a large-scale procedure (Morel et al. 1985b).

Protein was measured by the method of Lowry et ai. (1951) using BSA as a standard.

Monoclonal antibodies to a 14×10³M_r polypeptide were obtained in the course of the preparation of monoclonal antibodies to a 15×10³M_r proteolipid of the presynaptic plasma membrane of cholinergic nerve terminals (Morel et al. 1991). This 14×10³M_r polypeptide was detected in presynaptic plasma membrane fractions: in small amounts, when these membranes were prepared by subfractionation of isolated nerve endings (Morel et al. 1982; not shown); in larger amounts (Fig. 1) in crude presynaptic plasma membrane fractions prepared directly from frozen electric organ, according to a large-scale procedure (Morel et al. 1985b). The latter fractions contain membranes of all presynaptic elements (essentially nerve endings and Schwann cell membranes) and only small amounts of plasma membranes of the postsynaptic cells, which are recovered in a denser fraction, and of myelin (Morel et al. 1985b).

After an overnight incubation with 20 mM Chaps, a nondenaturing detergent, a substantial part of the antigen was recovered in a soluble form after centrifugation (160 000 g_max for 1h; Fig. 1A,B,C). It remained bound to the membranes at high ionic strength (1 h in the presence of 1 M MgCl₂; Fig. ID). After alkaline treatment of crude presynaptic membranes (1 h at pH 11 at room temperature), the antigen was not extracted and was found in the pellet (after a 1h centrifugation at 160 000 g_max; Fig. 1E,F,G). The antigen was split by various proteolytic enzymes such as α--chymotrypsin, proteinase K (Fig. 1H,I,K) and trypsin. Proteolysis products were still detected with monoclonal antibodies in membrane pellets. Partial delipidation of membranes resulted in an increased proteolysis of the 14×10³M_r antigen (Fig. 1J). Therefore, the 14×10³M_r antigen found in crude pre-synaptic plasma membrane fractions behaves as an integral membrane protein.

We have obtained 11 different hybridomas secreting immunoglobulins that bind to a 14×10³M_r band in immunoblot experiments. The possibility that they bind to different 14×10³M_r antigens comigrating in SDS-PAGE was ruled out in immunoprecipitation experiments. After solubilization in Chaps, the antigen was affinity purified on 14K4 immunoglobulins coupled to CNBr-activated Sepharose 4B beads. The eluted 14×10³M_r antigen was recognized in immunoblotting experiments by the other monoclonal antibodies (experiments not shown). In order to determine if the different antibodies bind to the same or to different epitopes shared by the 14×10³M_r antigen, immunoblots of delipidated membranes, treated by a- chymotrypsin or trypsin, were incubated with the 11 different monoclonal antibodies we have prepared against the 14×10³M_r antigen. Different staining patterns were obtained (Fig. 2) that allowed the monoclonal antibodies (mAbs) to be classed in three groups. The first group (14K1, 14K3, 14K6, 14K8, 14K9, 14K10) bound to two polypeptide doublets (at about 12 and 6×10³M_r) after a- chymotrypsin digestion and to a single doublet after trypsin treatment. The second group is composed of mAbs 14K2, 14K5, 14K7 and 14K11. Finally, mAh 14K4, which stained only a 13×10³M_r band or doublet after proteolysis, constitutes the third group. In ELISA, using native membranes (without solubilization or proteolysis), none of the antibodies of the first group was able to bind to the antigen. The four antibodies of the second group gave three different binding curves (Fig. 2, bottom), mAbs 14K7 and 14K11 being identical and mAb 14K5 showing the highest binding. mAb 14K4 also showed strong binding to the native antigen. Taken together, these data show that at least three different epitopes are shared by the antigen.

All monoclonal antibodies binding in ELISA gave positive results by indirect immunofluorescence in morphological experiments, strongest staining being obtained with antibodies 14K4 and 14K5, which were used in all subsequent experiments.

After formaldehyde fixation, electroplaques were dissected out of the prisms, layered on a slide and incubated with monoclonal antibodies (14K4 in the experiments of Fig. 3). Antibody binding was indirectly visualized using fluorescein-labelled anti-mouse Ig antibodies. Nerve branches were stained (Fig. 3A), but, whereas large branches were intensely stained, terminal axonal ramifications and the presynaptic network were only faintly fluorescent. This variation in fluorescence intensity occurs abruptly. In another focal plane (Fig. 3C), satellite cells were also stained. These cells have long thin processes, with numerous varicosities and branchings, connected to processes of neighbouring cells in a large network. Occasionally, such cells, still interconnected by their processes, were tom away from the electroplaque surface. Blood capillaries running between electroplaques were also labelled by monoclonal antibodies to the 14×10³M_r polypeptide (not shown). Staining of large nerve branches, satellite cells and capillaries was of similar intensity. No staining of the electroplaques, the postsynaptic cells, was observed.

A very similar distribution of the 14×10³M_r antigen was observed in Torpedo muscles. Blood vessels and nerve branches running between muscle fibres were intensely stained by antibodies to the 14×10³M_r polypeptide. At neuromuscular junctions, identified using rhodamine-conjugated o-bungarotoxin to visualize dense spots of acetylcholine receptors (not shown), Schwann cells were also labelled. They were less brightly stained than the nerves and therefore were more easily observed when these nerve trunks were torn away (Fig. 3B). Satellite cells with long processes presenting branching and varicosities were stained by mAbs 14K4 and 14K5 (Fig. 3D). Such cells are adherent to the surface of the dissociated muscle fibres and are found throughout the fibre. In addition, the aponevrotic material that wraps around the muscle was also labelled by the anti-14×10³M_r polypeptide antibodies (not shown).

Binding of antibody 14K4 in Torpedo electric organ was detected at the electron-microscope level using anti-mouse Ig antibodies conjugated to 15 nm gold particles (Figs 4, 5, 6 and 7). As previously reported (Ranvier, 1878; Sheridan, 1965), myelinated axons are wrapped, in addition to the Schwann cells that cover them, in a cellular cuff (Figs 4 and 5). This second cellular envelope is separated from Schwann cells by an extracellular space containing numerous collagen fibrils. The picture shown in Fig. 4 was selected because nuclei of the Schwann cell and of the cuff cell were visible in the same section. That of Fig. 5 illustrates the usual presence of several flat cellular layers constituting the external envelope of the axons. The cellular cuff is present for some distance after the axons have lost their myelin sheath, but terminal branches are devoid of this structure (not illustrated). Cells constituting this external cellular cuff were labelled by monoclonal antibody 14K4 as shown by the presence of gold particles, mainly associated with cellular membranes (Fig. 5) but also evenly distributed throughout their cytoplasm (Fig. 4). In the latter case, no obvious association with any particular structure could be noticed.

Satellite cells, most probably fibroblasts (Sheridan, 1965), with long thin processes, were heavily decorated with gold particles (Fig. 6). Labelling was associated with the extracellular surface of their plasma membrane (see Fig. 6B,C), on the cell body as well as on the entire length of their processes. No intracellular staining could be detected, even after alteration of their plasma membrane by 0.5% Triton permeabilization (instead of 0.1%), or using 5nm gold particle-conjugated antibodies, whereas antibodies still bound to the surface of fibroblasts (experiments not shown). Blood capillaries in electric organ are covered on most of their surface by thin cellular processes presenting varicosities and branchings. Antibody 14K4 bound to the plasma membrane surface of these processes as well as to the cell body membrane of these blood capillary-associated fibroblasts (not shown).

Immunolabelling was less sensitive at the electronmicroscope level than in fluorescence. Binding of mAb 14K4 to blood capillary endothelial cells was only occasionally detected with gold particles by electron microscopy whereas it was always detected in immunofluorescence studies. Gold particles were associated with the luminal surface of the endothelial plasma membrane (Fig. 7A). A similar situation was observed for Schwann cells covering nerve terminals that were only labelled in favourable conditions (Fig. 7B).

Electric lobes contain the cell bodies of the electroneurons that innervate the electric organs. These cell bodies are up to 150 /an in diameter, closely packed, separated only by blood capillaries and neuronal fibres. In electric lobe sections, anti-14×10³M_r polypeptide antibodies bound only to capillary endothelium sections (Fig. 8A and B). No electroneuron labelling could be detected or staining of axons running between the electroneurons. The meningeal envelope was intensely stained (Fig. 8A). A similar staining pattern – labelling of blood capillaries and meninges and no staining of neuronal and glial cells – was found in other Torpedo brain regions (not shown).

Staining of blood capillaries and vessel walls has been observed in all Torpedo tissues tested so far and is illustrated here on a frozen section of Torpedo gills (Fig. 8C). The vascular arborization is brightly fluorescent whereas surrounding tissues and branchial epithelium were not stained.

A series of monoclonal antibodies has been obtained that bind to a 14×10³M_r polypeptide. This antigen presents a wide tissue distribution. In the present study, its cellular and subcellular distributions were studied in Torpedo electric organ. The localization of the 14xlO³Af_r antigen was first studied by immunofluorescence and then at the electron-microscope level using immunogold staining. Localizations of the antigen by both techniques were similar (in tissue distribution) even though immunofluorescence appeared to be more sensitive.

In the electric organ, the 14×10³M_r antigen was mainly found associated with fibroblasts, axonal cuff cells and capillary endothelial cells, and to a lesser extent with Schwann cells. A similar antigen distribution was found in Torpedo muscles, whereas in brain, labelling was restricted to blood capillaries and meninges. This distribution is very evocative of proteins associated with the extracellular matrix (see Carbonetto, 1984, for a short review). However, most (more than 90%) of the antigen was found associated with plasma membranes. It was sedimentable and membrane-bound after fractionation of electric organ in sucrose gradients. The membrane-bound antigen must be regarded as an integral membrane protein, since it was not extracted at alkaline pH or high ionic strength. In addition, the antigen was not released by incubation of membranes in the presence of 0.3 M lactose and β-mercaptoethanol (not shown), demonstrating that it does not belong to the soluble lactose-binding lectin family, a family of low molecular weight extracellular proteins (see Barondes, 1984, for a short review; Teichberg et al. 1975: in Electrophorus electricus electric organ; and Cooper and Barondes, 1990: in muscle). In accordance with biochemical data, no labelling of the extracellular matrix was noticed at the electron-microscopic level and antibody binding was restricted to the external surface of plasma membranes in all cell types that were labelled in Torpedo electric organ, with the exception of the nerve cellular sheath where part of the immunolabelling was intracellular. This latter finding suggests that, in addition to the membrane-bound antigen, a small proportion of the protein could exist, in Torpedo electric organ, in a cytoplasmic form.

The 14×10³M_r antigen was found associated with the cellular sheath that surrounds the axonal branches. This structure was described in detail in Torpedo electric organ by Ranvier (1878) at the optical level and more recently, by electron microscopy, by Sheridan (1965) and Prado-Figueroa and Barrantes (1989). This cellular cuff, which appears to be made of several layers of flat squamous cells, is separated from the Schwann cell, which closely envelops the axon by a space containing extracellular material such as collagen fibrils. This was taken as an argument in favour of a fibroblastic origin for these cells (Sheridan, 1965). However, convincing morphological evidence was presented, in various other animal species to show, that this nerve envelope was a perineural epithelium, in continuity with the leptomeninges (Shanthaverappa and Bourne, 1962). In accordance with this view, meninges and arachnoid spaces were strongly labelled by the anti-14×10³M_r antibodies in our experiments. After having lost their myelin sheaths, axons are still enveloped for some distance by the cellular cuff, which becomes very thin and disappears. Terminal axonal branches are surrounded only by a Schwann cell. Fluorescence intensity decreases abruptly when axons lose their external cellular sheath, residual staining corresponding to labelling of Schwann cells covering distal axonal branches and nerve endings.

Satellite cells with thin, long processes, probably fibroblasts (Sheridan, 1965), were heavily stained. Gold particles were bound to the external surface of their plasma membrane. The antigen appears to be an abundant constituent of the fibroblast plasma membrane, since gold particle densities of 150–200 particles urrT² were usually observed. A homogeneous antigen density was observed on the cell body membrane and along the entire cell processes. Immunofluorescence labelling permitted the demonstration that these fibroblast processes are interconnected and form a loose mesh network extending between electroplaques. Fibroblasts adherent to the surface of muscle fibres were also visualized by immunofluorescence and present a very similar morphology to that of electric organ fibroblasts. They were observed in synaptic and extrasynaptic areas of muscle fibres.

We do not know yet what possible role the 14×10³M_r protein might play, but preliminary experiments of immunoprecipitation by mAb 14K4 of solubilized membrane proteins suggest that the 14×10³M_r antigen has affinity for two other proteins present in Torpedo electric organ membranes. One of these proteins is the 15×10³M_r proteolipid mediatophore (Israel et al. unpublished data) while the other appears to correspond to the 68×10³M_r glycoprotein previously identified using mAb Cl-8 (Morel et al. 1985a). These proteins have a more restricted and different tissue distribution than the 14×10³M_r protein. By immunofluorescence using either polyclonal (Morel et al. 1989) or monoclonal antibodies (Morel et al. 1991), the 15×10³M_r proteolipid mediatophore was detected in the nerve endings of Torpedo electric organ or neuromuscular junctions. No staining of blood capillaries, nerve branches or Schwann cells was noticed. In Torpedo electric lobes, the antigen was only visualized inside electroneurons. The 68×10³M_r glycoprotein appeared to be located only at zones of synaptic contact in Torpedo electric organ and neuromuscular junctions (unpublished results). The coprecipitation of these proteins with the 14×10³M_r antigen suggests that this 14×10³M_r antigen might be involved in cell-cell or cell-extracellular matrix interactions.

We are grateful to Dr R. Manaranche for initiating the morphological experiments, to Dr S. O’Regan for improving our manuscript. Guy Brochier is a fellow of the Association Française contre les Myopathies. This work was supported by an INSERM no. 886009 grant to N. Morel.