ABSTRACT
To generate monoclonal antibodies, immunogen fractions were purified from embryonic chick retinae by temperature-induced detergent-phase separation employing Triton X-114. Under reducing conditions, the monoclonal antibody (mAb) 2M6 identifies a protein doublet at 40 and 46×103Mr, which appears to form disulfide-coupled multimers. The 2M6 antigen is regulated developmentally during retinal histogenesis and its expression correlates with Müller glial cell differentiation. Isolated glial endfeet and retinal glial cells in vitro were found to be 2M6-positive, identified with the aid of the general glia marker mAb R5. mAb 2M6 does not bind to any other glial cell type in the CNS as judged from immunohistochemical data. Cell-type specificity was further substantiated by employing retinal explant and single cell cultures on laminin in conjunction with two novel neuron-specific monoclonal antibodies. MAb 2M6 does not bind either to neurites or to neuronal cell bodies. Incubation of retinal cells in vitro with bromodeoxyuridine (BrdU) and subsequent immunodouble labeling with mAb 2M6 and anti-BrdU reveal that mitotic Müller cells can also express the 2M6 antigen.
To investigate whether Müller cell differentiation depends on interactions with earlier differentiating ganglion cells, transections of early embryonic optic nerves in vivo were performed. This operation eliminates ganglion cells. Müller cell development and 2M6 antigen expression were not affected, suggesting a ganglion-cell-independent differentiation process. If, however, the optic nerve of juvenile chicken was crushed to induce a transient degeneration/regeneration process in the retina, a significant increase of 2M6 immunoreactivity became evident. These data are in line with the hypothesis that Müller glial cells, in contrast to other distinct glial cell types, might facilitate neural regeneration.
Introduction
The central nervous system of vertebrates is an extremely complex network of millions of neurons whose functions are highly dependent on supportive glial cells. Various neuron-glial cell interactions have been suggested, such as regulation of neurite extension, cell differentiation, modification of homeostasis and indirect control of synaptic activity (Schnell and Schwab, 1990; Brew et al. 1986; Müller and Best, 1989; Hatten and Mason, 1986).
Early morphological investigations revealed different glial cell types (Ramon y Cajal, 1933). The advent of immunological markers allowed further discrimination of glial cell types and their different functions (Miller et al. 1989). However, the increasing number of subclasses suggests that comprehensive classification of glial cells would need substantially more markers specific for different cell types and their differentiation and functional states. The importance of characterizing glial specificities became evident when, for example, various different experimental approaches indicated that different glial cell types have opposing effects on neurite outgrowth. Schwann cells and semidifferentiated astrocytes provide permissive cell surfaces for outgrowing axons, whereas oligodendrocytes inhibit axon elongation (Schnell and Schwab, 1990; Smith et al. 1990; Kleitman et al. 1988). In view of the very limited regeneration capacity of the adult central nervous system, it is essential to understand in more detail the molecular differences between various glial cell types.
Müller cells have been identified as a distinct glial cell type specific for the neural retina (Ramon y Cajal, 1933). Their cell bodies are located in the inner nuclear layer, extending radially oriented processes towards the most inner and outer layers of the retina. In these regions, glial endfeet form the inner and outer limiting membranes, which separate the retina from the non-neuronal vitreous body and, in part, from the pigment epithelium. Because glial endfeet at the vitreous body contain more than 90% of all potassium channels of individual Müller cells, it has been speculated that a major Müller cell function in adults is homeostasis. Potassium originating from neural activity within the inner plexiform layer, where Müller cells form an elaborate system of fine cell processes, would be eliminated from the retina and pumped into the vitreous body (Brew et al. 1986).
Whether Müller cells serve distinct functions in embryogenesis and participate in neural pattern formation, such as lamination of the retina, has not been evaluated. Cell lineage analysis, however, suggests that Müller cells do not originate from a glial precursor cell, because neuronal siblings of Müller cells have been identified (Turner and Cepko, 1987). Instead, cell-cell interactions appear to determine cellular differentiation (McKay, 1989).
We have chosen the visual system of the chicken as a model system to study molecular mechanisms of cellular differentiation and regeneration. This system has a number of advantages. The retinal morphology has been revealed in much detail. The various neuronal cell types besides Müller cells, which represent the only glial cell type, have been identified (Ramon y Cayal, 1933). In addition, a number of elaborate in vivo and in vitro assays have been developed, which facilitate antigen characterization (Bonhoeffer and Huf, 1985; Thanos et al. 1984; Halfter et al. 1987; Thanos and Dütting, 1987; Walter et al. 1987). Using hybridoma technology, we sought to identify molecular components to gain new insight into Müller glial cell differentiation and responses to neuronal degeneration stimuli. For this report, three novel monoclonal antibodies have been employed, together with a number of in vivo and in vitro assays as well as biochemical tests, to characterize the Müller-glial-cell-specific 2M6 antigen.
Materials and methods
Materials
Embryonic retinae were dissected from White Leghorn chicken. Triton X-114 (Sigma) was precondensed as described (Bordier, 1981). Enzyme inhibitors from Sigma (2,3-dehydro-2-deoxy-N-acetylneuraminic acid, apro tinin, leupeptin, pepstatin and phenylmethylsulfonyl fluoride) were added to all homogenization and solubilization buffers as indicated recently (Schlosshauer, 1989). Concanavalin A-Sepharose, lentil lectin (lens culinaris agglutininj-Sepharose were from Pharmacia; peanut agglutinin and wheat germ agglutinin coupled to agarose from Sigma.
Rhodamine- and fluorescein-conjugated F(ab′)2 fragment goat anti-mouse IgG (heavy and light chain) as well as conjugated isotype-specific anti-IgM and anti-IgG (Jackson Immunoresearch Laboratories) were used for 1/2 h at a dilution of 1:100 at room temperature. HRP-conjugated F(ab′)2 fragment goat anti-mousè IgG and M (Jackson Immunoresearch Laboratories) were applied at room temperature at a dilution of 1:2000 for 5h for western blot analysis. Monoclonal antibody R5 specific for glial cells (Draeger et al. 1984) was kindly provided by U. Draeger, Harvard Medical School, USA. Mouse monoclonal antibodies (mAb), mAb 2A1 – specific for growing ganglion cell axons (Schlosshauer et al. 1990), mAb 1P1 – specific for the inner plexiform layer of the chick retina (Schlosshauer and Herzog, unpublished), mAb 314 – specific for retinal photoreceptors and mAb 2A10 – specific for neurons (Schlosshauer and Wild, unpublished) were all generated in our own lab and applied at an approx, concentration of 5 μg ml−1. The mAb specific for bromodeoxyuridine was purchased from Bio-Science Products, Switzerland, and employed according to the manufacturers’ instructions.
Methods
Immunogen purification/biochemical methods
Cytoplasmic membranes were enriched from neuronal tissue of embryonic chick by sucrose gradient centrifugation based on the method of Hoffman et al. (1982) and Walter et al. (1987). Solubilization of membranes with Triton X-114 was performed as described (Schlosshauer, 1989). After precipitation of insoluble material by ultracentrifugation, the clear supernatant was applied in sequence to mAb D3- and mAb Dl-columns against different isoforms of the neural cell adhesion molecule (N-CAM) (Schlosshauer et al. 1984; Schlosshauer, 1989) and thereafter to lectin affinity columns (ConA–sepharose, lentil lectin-sepharose, peanut agglutinin-agarose, wheat germ agglutinin-agarose) to eliminate glycoconjugates. Proteins that did not bind to any of the above affinity matrices were subjected to temperature-induced detergent-phase separation in two consecutive centrifugation steps as described earlier (Bordier, 1981; Schlosshauer, 1989).
Protein quantification was carried out according to Lowry et al. (1951). Analytical polyacrylamide gel electrophoresis (Laemmli, 1970), silver staining of SDS gels (Ansorge, 1985) and western blot analysis (Towbin et al. 1979; Hawkes et al. 1982) were employed for qualitative characterization of protein fractions. For SDS–gel electrophoresis, samples were solubilized in the presence or absence of 0.1 M dithiothreitol.
In vivo manipulations
For early embryonic optic nerve transection, chick eggs were windowed at embryonic day 3 (E3) (Hamburger-Hamilton stages 20–22; HH20–22) (Hamburger and Hamilton, 1951) and at E4 (HH24/25) the right optic nerve was transected just behind the bulbus with the aid of a pair of microscissors. The egg shell window was sealed with a small Petri dish and incubation continued until Ell (HH35–37) or E18 (HH41–43).
For optic nerve crush, White Leghorn chickens (150-500 g weight) were deeply anesthetized with 50 mg kg−1 Ketavet and 25 mg kg−1 Rompun (intramuscular) 3 to 8 weeks after hatching. An 8 mm incision was made in the skin and underlying connective tissue at the temporal edge of the eyeball and the optic nerve was exposed by transecting the lateral eye muscle. The nerve was crushed with jeweller’s forceps in its retrobulbar segment for about 10 s. The skin was sutured allowing quick recovery after the operation. Animals were killed 2–28 days later and the visual system processed for immunohistochemistry.
Antibody generation
Standard immunizations of Balb C mice were performed over a period of approx. 3 months using 50–500 μg protein per inoculation in conjunction with adjuvant for 5 consecutive intraperitoneal injections as detailed in Schlosshauer (1989).
Cell fusions were performed as described (Fazekas de St. Groth and Scheidegger, 1980) using spleen cells and NS-1 myeloma cells in a ratio 8:1. Fused cells of individual mice were plated in 12 microtiter plates containing macrophages from rodent intraperitoneum.
Primary cultures
Retinal single cell cultures from chick embryos on laminincoated (20 μg ml−1 for 16 h, 4 °C) 8-well slides (Flow) were performed as described earlier (Schlosshauer et al. 1988; Schlosshauer, 1989). For explant cultures, E6 chick retinae were flat mounted on nitrocellulose filters, cut into 0.3 mm strips and placed onto laminin-coated glass coverslips (Haleter et al. 1981) or laminin-coated coverslips on which retinal cells had been seeded up to 2 weeks earlier. After an additional 1–2 day incubation period, explant cultures were immunolabeled with or without prior fixation. The culture medium was based on F12 medium (Gibco) and contained serum as well as methylcellulose. Details are given in Walter et al. (1987).
For metabolic bromodeoxyuridine (BrdU) labeling, single cell cultures of retinal cells were incubated with 10 μM BrdU (Sigma) for 0.5–1 h and, after several washes in culture medium for approx. 0.5 h, incubation was continued for 1–48 h. Condensed chromosomes were found preferentially 6–8 h after BrdU-labeling.
Immunofluorescence assays
For screening of fusion cultures, special compound sections were created (Constantine-Paton et al. 1986; Schlosshauer and Wild, in preparation) to reveal antigen expression simultaneously in various tissues from three different developmental stages. Incubation with primary antibodies was performed in most cases overnight at 4 °C.
Isolation of the retinal inner limiting membrane containing glial endfeet was performed according to Halfter et al. (1987) using a poly-L-lysine-coated glass coverslip that was pressed against the inner surface of a flat-mounted retina. Glial endfeet adhering to the glass surface were immunolabeled after fixation in 4% paraformaldehyde in the same way as tissue sections.
Cell surface antigens were identified in vitro by addition of hybridoma supernatants to viable cells (Schlosshauer and Herzog, 1990). After washing and fixation, fluorescently labeled secondary antibodies were added. For immunodouble labeling, all antibodies were employed in sequence rather than in parallel, which was found to cause some artefacts.
Results
Antibody generation
Amphiphilic proteins were purified by temperature-induced phase separation. Increasing the temperature of homogeneous protein solution from 4 °C to 30 °C results in a phase separation with a lower phase comprising detergent Triton X-114, integral membrane proteins and traces of cytoskeleton components (Bordier, 1981). This fraction was used for subsequent immunization of mice to generate mAbs. In comparison to other methods employing different protein fractions for immunization, a surprisingly high number of gliaspecific mAbs could be generated with the aid of the Triton X-114 phase (Schlosshauer and Wild, in preparation). Most of these antibodies recognize different glia cell types, whereas the mAb 2M6, presented here, was found to be specific for Müller glial cells of the retina (see below). MAb 2M6 does not react with mice or goldfish antigens and appears to be species-specific.
Developmental expression
Developmental regulation of the 2M6 antigen is characterized by a permanent increase in antigen expression during embryogenesis.
At embryonic day 6 (E6), the chick retina is essentially unlaminated. At this stage only the optic fiber layer is clearly discernable from the layer of mainly undifferentiated cells, as revealed by staining ganglion axons with mAb 2A1 (Fig. 1B). This antibody binds to a membrane-associated cytoskeletal protein (Schlo-Bhauer et al. 1990), whereas no 2M6 immunoreactivity is evident at this stage (Fig. 1A).
2M6 antigen expression during retina development. Paraformaldehyde-fixed cryostat sections of chick retinae were immunolabeled with mAb 2M6 (A, D, G), mAb 2A1 (B), mAb 1P1 (E) and mAb 314 (H). Three pictures in a row represent one developmental stage of the retina with the corresponding phase-contrast figures (C, F, I): first row embryonic day 6 (E6), second row E10, third row E20. The 2M6 antigen is not expressed at E6, weakly at E10 and pronounced at E20. To illustrate the progression of tissue layer formation during histogenesis from the optic fiber layer towards the photoreceptor layer, monoclonal antibodies specific for the optic fiber layer (2A1), the inner plexiform layer (1P1) and photoreceptor layer (314) have been applied in parallel; IPL, inner plexiform layer; OFL, optic fiber layer; PR, photoreceptor layer. Calibration bar 50 μm.
2M6 antigen expression during retina development. Paraformaldehyde-fixed cryostat sections of chick retinae were immunolabeled with mAb 2M6 (A, D, G), mAb 2A1 (B), mAb 1P1 (E) and mAb 314 (H). Three pictures in a row represent one developmental stage of the retina with the corresponding phase-contrast figures (C, F, I): first row embryonic day 6 (E6), second row E10, third row E20. The 2M6 antigen is not expressed at E6, weakly at E10 and pronounced at E20. To illustrate the progression of tissue layer formation during histogenesis from the optic fiber layer towards the photoreceptor layer, monoclonal antibodies specific for the optic fiber layer (2A1), the inner plexiform layer (1P1) and photoreceptor layer (314) have been applied in parallel; IPL, inner plexiform layer; OFL, optic fiber layer; PR, photoreceptor layer. Calibration bar 50 μm.
At E10, retina differentiation has proceeded further, leading to the formation of the inner plexiform layer (IPL), as indicated by mAb 1P1 staining (Fig. 1E). This antibody identifies a high molecular weight protein expressed exclusively in the IPL (Schlosshauer and Wild, unpublished). At E10, marginal 2M6 immunoreactivity is observed in perikarya containing nuclear layers, but hardly any is seen in the IPL (Fig. 1D).
At E20, one day before hatching, 2M6 immunolabeling becomes most pronounced in the retina, similar to the adult. The labeling suggests a radial orientation of the corresponding immunoreactive cells. This feature is reminiscent of Miiller glial cells that form elaborate arborizations in the IPL, engulf ganglion cells and their axons in the ganglion cell and optic fiber layer, and extend processes around photoreceptor perikarya in the outer nuclear layer. The outer segments of photoreceptors, however, are not surrounded by glial endfeet (Ramon y Cajal, 1933). The 2M6 staining pattern parallels these morphological characteristics of Miiller cells as shown by immunolabeling retina sections in conjunction with mAb 2M6 and the novel photoreceptor-specific mAb 314 (Fig. 1G,H).
Müller glia specificity
Immunohistochemical analysis suggested mAb 2M6 to be specific for retinal Miiller glial cells and their glial endfeet in the optic fiber layer. Because no rapid method has yet been established to purify Miiller glial cells specifically, we isolated glial endfeet by attaching the inner limiting membrane, which is composed mainly of glial endfeet, to a poly-L-lysine-coated glass surface, after removal of the vitreous body of the eyeball. Immunostaining of this preparation, which is free of neuronal elements (Halfter et al. 1987), clearly indicates that glial endfeet bear the corresponding antigen (Fig. 2).
2M6 antigen on glial endfeet. The inner limiting membrane of the retina containing endfeet from Müller glial cells was isolated on an adhesive glass coverslip, paraformaldehyde-fixed and immunolabeled with mAb 2M6. The antibody stains glial endfeet (arrows). Immunofluorescence (A), phase-contrast (B), calibration bar 30μm.
2M6 antigen on glial endfeet. The inner limiting membrane of the retina containing endfeet from Müller glial cells was isolated on an adhesive glass coverslip, paraformaldehyde-fixed and immunolabeled with mAb 2M6. The antibody stains glial endfeet (arrows). Immunofluorescence (A), phase-contrast (B), calibration bar 30μm.
This observation was further substantiated by investigations using single cell cultures. Fresh single cell cultures, originating from retinae of E6–E12, contain only a small percentage of glial cells, which can be increased by extended culturing. Neural cells die whereas presumptive glial cells appear to differentiate in vitro in a similar developmental sequence as in vivo.
Maintaining retinal cells from embryonic day 10 or older for 8 days in vitro, results in cultures of adherent flat cells. Double labeling with the glia cell marker R5 (Draeger et al. 1984) and mAb 2M6 reveals that all 2M6-positive cells of these retina cultures also express the R5 antigen (Fig. 3). Because Millier glial cells represent the only glia in the chick retina (Ramon y Cajal, 1933), these 2M6-positive cells must be classified as Müller glial cells.
2M6 antigen on glial cells of the retina. Single cell cultures of embryonic chick retinae were kept in vitro for 1–2 weeks, paraformaldehyde-fixed and immunodouble-labeled with mAb R5 (A) specific for glial cells and mAb 2M6 (B). Phase-contrast (C). mAb 2M6 marks retinal glial cells. Calibration bar 30 μm.
2M6 antigen on glial cells of the retina. Single cell cultures of embryonic chick retinae were kept in vitro for 1–2 weeks, paraformaldehyde-fixed and immunodouble-labeled with mAb R5 (A) specific for glial cells and mAb 2M6 (B). Phase-contrast (C). mAb 2M6 marks retinal glial cells. Calibration bar 30 μm.
Application of both antibodies to tissue sections of the chick brain of different developmental stages indicates unequivocally that (a) 2M6 and R5 must be different from each other and (b) that 2M6 is a true Müller glial cell marker: no immunolabeling was evident in the rest of the CNS using 2M6 (Fig. 4B) in contrast to mAb R5, which delineates either distal or all glial cell processes in the brain, depending on the developmental stage (Fig. 4A) (Vanselow et al. 1989).
The brain is devoid of 2M6 antigen. The tectum opticum was paraformaldehyde-fixed after hatching. Cryosections were immunodouble-labeled with mAb R5 (A) and 2M6 (B). Immunofluorescence pictures (A, B), phase-contrast (C). R5 labels distal glial cell processes (arrows in A), whereas essentially no 2M6-immunoreactivity is evident. Calibration bar in C for A, B and C 50 μm.
The brain is devoid of 2M6 antigen. The tectum opticum was paraformaldehyde-fixed after hatching. Cryosections were immunodouble-labeled with mAb R5 (A) and 2M6 (B). Immunofluorescence pictures (A, B), phase-contrast (C). R5 labels distal glial cell processes (arrows in A), whereas essentially no 2M6-immunoreactivity is evident. Calibration bar in C for A, B and C 50 μm.
Neither of the above histochemical nor in vitro experiments could evaluate whether neural cells or their processes also express the 2M6 antigen. Therefore, two additional in vitro approaches were employed. Retinal tissue was aseptically dissected out of the eye and immobilized on adhesive nitrocellulose filters, cut into 300 μm strips and explanted onto laminin, which results in extensive axonal outgrowth after 1–2 days in vitro. Ganglion cell axons were identified with the aid of mAb 2A1 (Schlosshauer et al. 1990) (Fig. 5A). As outlined above, in the second assay E6–E9 retinal tissues were processed to gain single cell suspensions, which were plated onto laminin for a short period. Using this approach, neurons were identified by mAb 2A10 which marks not only axons but all neuritic processes and the corresponding neural cell bodies in contrast to mAb 2A1 (Schlosshauer and Wild, unpublished) (Fig. 5B). Double labeling using mAb 2M6 also revealed that neither ganglion cell axons (Fig. 5C) nor neuronal cells (Fig. 5D) express the 2M6 antigen. Instead, in single cell cultures only flat, 2A10-negative cells were found to be 2M6-positive. These data further demonstrate the high specificity of mAb 2M6 for Müller glial cells and the absence of cross-reactivity with neuronal cells.
Retinal neurons are devoid of 2M6 antigen. Tissue strips (A, C, E) or cell suspensions (B, D, F) of chick retinae were cultured on glass coverslips coated with laminin. After fixation, in vitro cultures were immunodoublelabeled with mAb 2M6 and mAb 2A1 specific for axons or mAb 2A10 specific for neurons. E is the corresponding phase-contrast picture for A and C; F for B and D. Arrows in B and F mark a neuron located on a Müller glial cell. MAb 2M6 does not stain either neuronal cell processes or perikarya of neurons. Calibration bar in E for A, C and E 30 μm, in F for B, D and F 20 μm.
Retinal neurons are devoid of 2M6 antigen. Tissue strips (A, C, E) or cell suspensions (B, D, F) of chick retinae were cultured on glass coverslips coated with laminin. After fixation, in vitro cultures were immunodoublelabeled with mAb 2M6 and mAb 2A1 specific for axons or mAb 2A10 specific for neurons. E is the corresponding phase-contrast picture for A and C; F for B and D. Arrows in B and F mark a neuron located on a Müller glial cell. MAb 2M6 does not stain either neuronal cell processes or perikarya of neurons. Calibration bar in E for A, C and E 30 μm, in F for B, D and F 20 μm.
Cell-cycle-independent antigen expression
The high proliferation rate in early embryonic chick retinae (Dütting et al. 1983) in conjunction with the absence of 2M6 immunoreactivity observed on retinal tissue sections of early chick embryos, could be attributed to the fact that presumptive glial cells are highly mitotic and therefore differentiation antigens like the 2M6 protein would not yet be expressed. Consequently, we tried to evaluate whether mitotic as well as nonmitotic glial cells were capable of expressing the 2M6 antigen. Retinal cells were cultured as single cells and metabolically labeled with bromodeoxyuridine. After fixation and permeabilization, BrdU was detected with the aid of an anti-BrdU mAb. The same cultures were counterstained with mAb2M6. As shown in Fig. 6, BrdU-negative and -positive cells express the 2M6 antigen suggesting a cell-cycle-independent antigen expression. Because mitotic glial cells are also capable of expressing the 2M6 antigen, it is likely that the enhanced immunoreactivity at advanced stages of retinal histogenesis is based on a distinct differentiation program rather than simply on the completion of cell division.
2M6 antigen in mitotic glial cells. Retinal cells cultured on poly-L-lysine/laminin were metabolically labeled with bromodeoxyuridine (BrdU) and after fixation immunodouble-labeled with anti-BrdU and mAb 2M6. Immunofluorescence pictures (A, B), phase-contrast (C). A 2M6-positive glial cell (arrow in B) reveals BrdU-labeled condensed chromosomes during mitosis (A). Calibration bar 30 μm.
2M6 antigen in mitotic glial cells. Retinal cells cultured on poly-L-lysine/laminin were metabolically labeled with bromodeoxyuridine (BrdU) and after fixation immunodouble-labeled with anti-BrdU and mAb 2M6. Immunofluorescence pictures (A, B), phase-contrast (C). A 2M6-positive glial cell (arrow in B) reveals BrdU-labeled condensed chromosomes during mitosis (A). Calibration bar 30 μm.
2M6 antigen expression independent of ganglion cells
Ganglion cells are the first neuronal cells to differentiate during neuroembryogenesis of the retina (Prada et al. 1981). As indicated by cell lineage analysis in vivo (Turner and Cepko, 1987) as well as by in vitro assay systems investigating different neuronal and nonneuronal cell types (Hatten and Mason, 1986; Schlosshauer and Herzog, 1990), cell differentiation appears to be dependent on direct cell-cell interactions rather than early embryonic cell determination. Therefore, it was of interest to investigate whether the first cell type to be generated in the retina would affect Müller glia cell differentiation and 2M6 antigen expression. The idea was to eliminate ganglion cells in vivo and subsequently follow possible 2M6 antigen expression in situ.
The optic nerve was transected at embryonic day 4 before most of the ganglion cell axons have emerged from the retina. During the following days of incubation, ganglion cell axons leave the retina at the optic disc and grow aberrantly around the eye ball up to Ell, but do not penetrate to the brain (Fig. 7B). The lack of synaptic target leads finally to degeneration of ganglion cells, which is evident from the missing ganglion cell and optic fiber layer at E18 (Schlosshauer et al. 1990). Comparison of the 2M6 antigen expression between control and operated sides of chick embryos of two developmental stages (Ell and E18) demonstrates no reproducible difference (Fig. 7), in contrast to degeneration events occurring after completion of embryogenesis (see below). Therefore, we conclude that the 2M6 antigen expression and possibly Müller glial cell differentiation in general, proceeds independently of prolonged interactions between ganglion cells and presumptive glial cells.
2M6 antigen expression independent of ganglion cells. In order to eliminate ganglion cells as the earliest retinal neurons to differentiate, the right optic nerves of E4 chick embryos were transected. Subsequently retinal cryosections were immunolabeled with mAb 2M6 at Ell (C, D) and E18 (E, F). Control side (A, C, E), operated side (B, D, F). Toluidine/cresyl violet staining at E 11 (A, B). Before the lack of the synaptic target leads to ganglion cell death, their axons grow initially around the eyeball (arrows in B), instead of forming a regular optic nerve as on the control side (A). Elimination of ganglion cells is demonstrated by the absence of the optic fiber layer (arrowheads in E and F). 2M6 immunoreactivity between control and transected side does not reveal a reproducible difference, suggesting a ganglion-cell-indepcndent 2M6 antigen expression. IPL, inner plexiform layer; OFL, optic fiber layer; ON, optic nerve; RET, retina. Calibration bar in B for A and B 400 μm, in F for C, D, E and F 50 μm.
2M6 antigen expression independent of ganglion cells. In order to eliminate ganglion cells as the earliest retinal neurons to differentiate, the right optic nerves of E4 chick embryos were transected. Subsequently retinal cryosections were immunolabeled with mAb 2M6 at Ell (C, D) and E18 (E, F). Control side (A, C, E), operated side (B, D, F). Toluidine/cresyl violet staining at E 11 (A, B). Before the lack of the synaptic target leads to ganglion cell death, their axons grow initially around the eyeball (arrows in B), instead of forming a regular optic nerve as on the control side (A). Elimination of ganglion cells is demonstrated by the absence of the optic fiber layer (arrowheads in E and F). 2M6 immunoreactivity between control and transected side does not reveal a reproducible difference, suggesting a ganglion-cell-indepcndent 2M6 antigen expression. IPL, inner plexiform layer; OFL, optic fiber layer; ON, optic nerve; RET, retina. Calibration bar in B for A and B 400 μm, in F for C, D, E and F 50 μm.
Degeneration response after hatching
Various functions have been attributed to glial cells such as facilitation of neurite extension during neuroembryogenesis and regeneration (Kleitman et al. 1988). To determine whether induced ganglion cell degeneration would affect 2M6 antigen expression after the development of the visual system is complete, the optic nerves of juvenile chickens were crushed just behind the eye balls. A few days after the operation, ganglion cell axons had degenerated as demonstrated by degeneration plaques in the optic nerve (Fig. 8B). Degeneration varies depending on the extent of axon damage and the percentages of axons that have been affected. Two days after the crush, no obvious changes were evident, whereas 11 days after the operation 2M6 immunoreactivity was significantly increased (Fig. 8). As has been shown above, ganglion cell degeneration during embryogenesis does not affect 2M6 antigen expression as it does in hatched chicken. One explanation of these results would be that Müller cell differentiation has to proceed beyond the period of embryogenesis to allow a glial response to degeneration signals. This might include a developmentally regulated expression of distinct cell surface components of glial cells (Schlosshauer et al. 1988; Baehr and Schlosshauer, 1989).
Axotomy-induccd induction of 2M6 antigen expression after hatching. Optic nerves of juvenile chickens were crushed in vivo and retinal cryosections immunolabeled with mAb 2M6 11 days after the operation. Toluidine blue/ cresyl violet staining of the optic nerve head (A, B). Immunofluorescence staining (C, D), corresponding phasecontrast (E, F). Control side (A, C, E), operated side (B, D, F). The crush induces degeneration plaques in the optic nerve (arrows in B) and an enhanced 2M6 immunoreactivity in the retina (D). ON, optic nerve; RET, retina. Calibration bar in B for A and B 500 μm, in F for C, D, E and F 70 μm.
Axotomy-induccd induction of 2M6 antigen expression after hatching. Optic nerves of juvenile chickens were crushed in vivo and retinal cryosections immunolabeled with mAb 2M6 11 days after the operation. Toluidine blue/ cresyl violet staining of the optic nerve head (A, B). Immunofluorescence staining (C, D), corresponding phasecontrast (E, F). Control side (A, C, E), operated side (B, D, F). The crush induces degeneration plaques in the optic nerve (arrows in B) and an enhanced 2M6 immunoreactivity in the retina (D). ON, optic nerve; RET, retina. Calibration bar in B for A and B 500 μm, in F for C, D, E and F 70 μm.
Biochemical analysis
Using coarse membrane preparations in conjunction with SDS gel electrophoresis and western blot analysis, it was found that mAb 2M6 recognizes a proteinaceous antigen. Two major bands at approx. 40×103 and 40×103Mr could be revealed. Depending on the redox state of proteins, a triplet in the range of 150×103Mr. was observed if native disulfide bridges of retinal proteins were left intact (Fig. 9). A possible explanation of this result might be that, under physiological conditions, a minor antigen fraction exists, consisting of a mixed population of covalently coupled multimers such as 3×46×103+l×40×103Afr, 2×46×103Mr, 2×40×103Mr etc.
Biochemical characterization. Chick retina samples were subjected to western blot analysis using mAb 2M6 and a horseradish peroxidase-conjugated secondary antibody. Under nonreducing conditions protein bands at about 150×103, 46×103 and 40×103 are apparent. Reduction eliminates the high molecular weight components. –SH, reduced sample; S–S, nonreduced sample. Relative molecular masses are given on the right-hand side.
Biochemical characterization. Chick retina samples were subjected to western blot analysis using mAb 2M6 and a horseradish peroxidase-conjugated secondary antibody. Under nonreducing conditions protein bands at about 150×103, 46×103 and 40×103 are apparent. Reduction eliminates the high molecular weight components. –SH, reduced sample; S–S, nonreduced sample. Relative molecular masses are given on the right-hand side.
Discussion
Müller glia specificity
Five lines of evidence suggest strongly that mAb 2M6 is indeed specific for retinal Müller glial cells. (1) In the retina, mAb 2M6 labels radially oriented cells with cell bodies located in the inner nuclear layer and cell processes extending to the inner and outer limiting membrane. The latter is located between the inner and outer segments of photoreceptors. Such a cell morphology is reminiscent of Müller cells as described by Ramon y Cajal (1933). (2) In preparations of the inner limiting membrane (Halfter et al. 1987), glial endfeet of Müller cells are marked by mAb 2M6. (3) Double labeling of single cell cultures with mAb 2M6 and mAb R5, whose corresponding antigen is expressed in different types of macroglia (Draeger et al. 1984), reveals that mAb 2M6 binds to R5-positive retinal cells. Because the chicken retina is devoid of glial cells with the exception of Müller cells, retinal cells identified by these antibodies are unequivocally Müller glial cells. (4) The absence of 2M6 immunoreactivity in other regions of the brain, where the presence of glial cells can be demonstrated by mAb R5 (Vanselow et al. 1989), suggests that probably mAb 2M6 does not bind to any other glial cell type. (5) No cross-reactivity of mAb 2M6 exists with neuronal cell perikarya or cell processes that are identified by mAb 2A1 and mAb 2A10. Therefore mAb 2M6 is specific for Miiller cells. To our knowledge, this is the first immunological marker with such a distinct specificity.
Developmental regulation
During the first week of incubation no immunoreactivity is evident in the retina. In the second week, faint labeling is observed, followed by a drastic increase during the last week, of embryogenesis, which remains at this high level in adulthood. This developmental profile of 2M6 antigen expression roughly parallels the maturation of Müller cells as revealed by the glial marker enzyme glutamine synthetase, which is induced as late as embryonic day 15 (Moscona and Linser, 1983). Because the 2M6 antigen can be detected 4-5 days earlier, it can be considered as an early marker of differentiation during retinal gliogenesis.
The development of some glial cells has been demonstrated to be influenced by neurons of the mammalian CNS (Levine, 1989). Heterotypic cell-cell interactions are thought to have a regulatory influence on Müller cells, as suggested by results from rotation-mediated suspension cultures of reaggregating retinal cells. In these cultures, lectins as well as a distinct mAb specific for neuronal cell surface antigens, appear to block glial differentiation (Linser and Perkins, 1987). Although it has not been resolved which retinal cell type influences Müller cell differentiation in this model assay, a hierachical ordered differentiation system would assume that more advanced cell types partly control less differentiated cells to form finally a functional neuronal network. Ganglion cells are the first to differentiate in the retinal cell population and therefore might fulfill a master function during histogenesis.
We have approached this question in vivo by transecting the optic nerve to prevent innervation of the tectum opticum by retinal ganglion cell axons. Target deprivation results in ganglion cell death at a developmental period when Müller cell differentiation is about to start (Moscona and Linser. 1983). However, despite the absence of ganglion cells in these specimens, Müller cell differentiation does not seem to be impaired since MAB R5 staining as well as 2M6 antigen expression are not affected. Therefore, our data do not support the interpretation of Linser and Perkins (1987) with regard to a potential interaction between Muller and ganglion cells during gliogenesis.
Molecular identity
Western blot analysis revealed that under reducing conditions mAb 2M6 recognizes a protein doublet at 40/46×103Mr. When disulfide bridges of the antigen are left intact, an additional triplet at 150×103Mr. becomes apparent, suggesting the existence of multimers under physiological conditions. Comparison with a number of identified proteins that are predominantly expressed in Müller glial cells, indicates that the 2M6 antigen is most likely to represent a novel component.
The intermediate filament protein vimentin with a relative molecular mass of 53×103 has been found in chicken Müller glia. Although in the adult retina 2M6- and anti-vimentin-staining coincide, only vimentin is expressed by retinoblasts at early stages of retinal histogenesis. During differentiation, vimentin becomes restricted to Müller cells (Lemmon and Rieser, 1983). Similar features have been observed for the actin-binding protein filamin. The additional localization of filamin in vascular endothelial cells in the tectum opticum of the chicken as well as the relative molecular mass of 250×103 (Lemmon, 1986) clearly distinguishes this intracellular protein from the 2M6 antigen.
The acidic calcium-binding protein S-100 has been regarded as a marker for Müller glial cells, too. However, S-100 is unmistakably different from the 2M6 antigen, because S-100 is localized also in astrocytes surrounding blood vessels and therefore is widely distributed in the brain (Kondo et al. 1983). In addition, expression in retinal glia as well as in brain astrocytes have been found for the intracellular 7G4 antigen (relative molecular mass 61×103) (Lemmon, 1985), the R5 antigen (Draeger et al. 1984) and the glial fibrillary acidic protein (relative molecular mass 51×103), which is also found in retinal Müller glial cells (Björklund et al. 1985). In contrast, 2M6 appears to be specific for Müller glia.
Two enzymes that subserve distinct glial functions have been also regarded as marker proteins for Müller cells. Glutamine synthetase plays a key role in recycling the neurotransmitters glutamate and GABA and subsequently supplying glutamine for neurons (Vardimon et al. 1988). The developmental expression of glutamine synthetase parallels that of the 2M6 antigen, because the enzyme becomes induced only after glucocorticoid hormone secretion of the adrenal gland at the end of the second week of incubation. However, the histological distribution in the central nervous system of enzyme and 2M6 antigen reveal clear differences, because glutamine synthetase is also found in astrocytes, oligodendrocytes and choroid cells (Moscona and Linser, 1983). Carbonic anhydrase, which regulates the CO2 concentration and, consequently, the tissue pH, differs from the 2M6 antigen in its developmental regulation, because it is already expressed in early embryonic retina cells, as well as being found in other tissues (Moscona and Linser, 1983).
The spatiotemporally regulated expression of the 2M6 antigen during development and regeneration suggests that the antigen may be a novel intracellular protein playing a crucial role in these processes.
ACKNOWLEDGEMENTS
We are grateful to Dr F. Bonhoeffer who provided research facilities. The authors acknowledge the excellent technical assistance of M. Wild and inspiring discussions with Drs F. Bonhoeffer, U. Draeger, C. Stürmer, T. Voigt and P. M. Whitington.