Proliferation and differentiation of ependymal cells in the injured carp spinal cord were studied by immunohistochemistry using proliferating cell nuclear antigen (PCNA) and 3H-thymidine (3H-TdR) autoradiography. A surge of proliferation of ependymal cells occurred in the lesion with a peak around 6th day after surgical operation (complete transection) of the spinal cord. The proportion of PCNA-positive ependymal cells at 6th post-operative day (6 POD) was over 14 times that in the normal state. Electron microscopic autoradiography revealed that most of the ependymal cells incorporating 3H-TdR at 5 POD, especially those located in the caudal side of the transection site, contained numerous free ribosomes in their apical portion, but the other organellae, such as rough endoplasmic reticulum and Golgi apparatus, were poorly developed in this portion. In three weeks thereafter, the ependymal cell layer was reconstructed through the lesion, and the apical part of the 3H-TdR-labeled ependymal cells became elongated and morphologically differentiated: that is, it had free ribosomes decreased and glial filaments increased in number. Many bundles of regenerating axons were observed to course within the reconstructed ependymal cell layer. These results may suggest that proliferation, differentiation and reconstruction of the ependymal cell layer following injury of the carp spinal cord are requisite to make the permissive milieu for elongation of the regenerating axons.
In the central nervous system (CNS) of the lower vertebrates such as amphibians and fishes, axonal regeneration occurs after injury (Coggeshall et al., 1982; Coggeshall and Youngblood, 1983; Stensaas, 1983; Reier et al., 1983; Davis et al., 1989). In the case of the spinal cord, the restoration of the central canal and the proliferation and migration of ependymal cells have been reported to take place at first after traumatic injury, i.e., transection (Nordlander and Singer, 1978; Michel and Reier, 1979; Singer et al., 1979; Simpson, 1983; Yamada et al., 1995). We have previously reported that there were numerous regenerated axons widely within the reconstructed ependymal cell layer at the repair stage of the carp spinal cord lesion caused by transection, and thus suggested that ependymal cells may offer the circumstances beneficial to elongation of the regenerating axons in the lower vertebrates (Yamada et al., 1995).
In the present study, we investigated in greater detail the proliferative response and morphological change of ependymal cells after transection of the carp spinal cord, and the possible roles of these cells in the axonal regeneration is discussed.
MATERIALS AND METHODS
Carps (Cyprinus carpio) of 18–21 cm in body length were used. They were anesthetized with 0.02% tricaine methanesulfonate (Nacalai Tesque), and laminectomy was performed at the caudal end of the dorsal fin. Then, the spinal cord was transected completely with a micro-dissecting scissors at the site between the vertebral segments 7 and 8 counted from the caudal end. The fish were kept in an ordinary water aquarium after the surgery.
PCNA (proliferating cell nuclear antigen) immunohistochemistry
At 3th, 6th and 10th post-operative day (POD), three fish per group were anesthetized, and the transection lesion (seg. 7–8) of the spinal cord was removed and was fixed by immersion in either Carnoy's fluid or methacarn for 4 hr at room temperature. Tissue was embedded in paraffin. Semi-serial 4 μm-thick transverse sections were cut from the caudal or rostral sides of the transection. Sections were incubated with anti-PCNA antibody (1:200 dilution by PBS, Novocastra Lab. Ltd., UK) for 2 hr at room temperature, and the immunoreaction product was visualized using immunostaining reagents (Vectastain ABC Kit: Vector Lab., Inc., USA) by the avidin-biotin-peroxidase complex (ABC) method. After counterstained with methyl green, sections were observed under a light microscope.
The total numbers of PCNA positive and negative ependymal cells were counted on every section (approximately 80 sections for each animal).
Light and electron microscopic 3H-TdR autoradiography
Flash labeling (2 hr exposure to 3H-TdR):
A total of eight fish were used, including two intact fish as controls. Two fish each were sacrificed at 1.5, 3 and 5 POD. All fish received an intraperitoneal injection of 3H-TdR (10–20 μCi/g b.w.; specific activity 3.3 TBq/mmol, DuPont/NEN) 2 hr before sacrifice. They were fixed by perfusion of a fixative containing 2.4% paraformaldehyde, 2.2% glutaraldehyde and 10 mM CaCl2 in 0.1 M cacodylate buffer (pH 7.4).
Pulse labeling (long survival period after labeling):
Pulse labeling (long survival period after labeling): Four fish received three successive injections of 3H-TdR injections (10 μCi/g b.w.) at 6-hr intervals at 5 POD. They were sacrificed and fixed at 21 and 56 days thereafter in the same manner as above.
After fixation, the transection lesion of the spinal cord, the film terminale and the wall of the third ventricle were removed, cut into small pieces, and osmificated. They were dehydrated in ethanol and embedded in Epon 812. Semi-thin sections, about 1- μm-thick, were cut out of the Epon block of each material and processed for light microscopic autoradiography. For electron microscopic autoradiography, ultrathin sections were cut and stained with uranyl acetate, and then they were coated with llford L4 emulsion (1:2.5 dilution, llford Ltd., UK) using a small wire-loop. After exposure for 1–2 months, sections were developed with Microdol-X (Eastman Kodak, USA), fixed and stained with lead citrate. Ultrathin sections were observed under a JOEL JEM-1200 EX electron microscope.
PCNA is expressed in cell nuclei of the G1-S phases of proliferation cycle, and PCNA immunostaining has been proved to be useful to analyze the mode of cell proliferation (Hall et al., 1990). In the intact spinal cord, PCNA-positive ependymal cells were scarcely found: they constituted approximately 0.3% to 0.4% of all ependymal cells. At 3 POD, the ratio of PCNA-positive to all ependymal cells (PCNA/all ratio) became slightly elevated up to 1.0% in the sides both rostral and caudal to the transection. At 6 POD, a number of PCNA-positive ependymal cells were easily found in the caudal side of the transection (Fig. 1), and the PCNA/all ratio became markedly increased: the ratio in the caudal side was 6.4%, higher than that in the rostral side, 2.2%. At 10 POD, the number of PCNA-positive ependymal cells was decreased and the ratios in the caudal and rostral sides were 2.3% and 1.5%, respectively. Throughout the experimental period, the numbers of PCNA-positive ependymal cells counted in the regions 1.6 cm caudal and 2.0 cm rostral to the transection were almost the same as that of the normal spinal cord (Fig. 2).
Light and electron microscopic 3H-TdR autoradiography Flash labeling (2 hr exposure to 3H-TdR)
In all the unoperated control spinal cords, no 3H-TdR- labeled ependymal cells were found by light microscopic autoradiography using semi-thin sections, except that a few labeled ependymal cells were observed in the frontal wall of the third ventricle and in the film terminale, i.e., the anterior and posterior ends of the central canal.
At 1.5 and 3 POD, the apical part of most ependymal cells, which faces the central canal was elongated in the side caudal to the transection. A very few of these ependymal cells were found to be labeled with 3H-TdR on light microscopic autoradiography.
At 5 POD, in the side caudal to the transection, some 3H- TdR-labeled ependymal cells were observed by electron microscopic autoradiography. At this day, the ependymal cell layer of the caudal side was composed of tightly aligned ependymal cells, and the central canal was extremely narrow or obliterated (Fig. 3a). The apical part of the labeled ependymal cells contained numerous free ribosomes and a few mitochondria, but the other organelles such as rough endoplasmic reticulum and Golgi apparatus were poorly developed (Fig. 3b). Only in the processes of these cells, glial filaments, which are usually seen in the perikaryal cytoplasm of normal ependymal cells, were observed. In the rostral side of the transection, the central canal was dilated and ependymal cells were flattened in shape. Although the profiles of intracellular organelles of these cells were almost the same as those of the ependymal cells observed in the caudal side, bundles of glial filaments were also seen in the perikaryal cytoplasm (Fig. 3c).
Pulse labeling (labeling with 3H-TdR at 5 POD and long term survival up to 26 and 61 POD)
In the caudal side of the transection, the nuclei of most ependymal cells, usually spheroid in shape, were located far from the luminal surface of the central canal, which was still extremely narrow in diameter (Fig. 4a). Electron microscopic features of the 3H-TdR-labeled ependymal cells were of better differentiated cells: free ribosomes in the apical part of these cells were decreased in number, while glycogen particles were increased and bundles of glial filaments appeared (Fig. 4b). These cells had small Golgi apparatus, a few mitochondria and relatively well developed rough endoplasmic reticulum. In the rostral side, the central canal was widely dilated.
In both rostral and caudal sides, large bundles of regenerating axons, most of which were unmyelinated, were seen within the ependymal cell layer consisting of both 3H- TdR-labeled and non-labeled ependymal cells. Many small unmyelinated regenerating axons were also seen at the abluminal region of the ependymal layer (Fig. 5).
The gap caused by transection was already connected with the regenerated nervous tissue, through which the central canal was reconstructed. Elongated ependymal cells made up the ependymal layer of this reconstructed central canal. Numerous microvilli and a few cilia were seen on the surface of each ependymal cell (Fig. 6a). The overall morphological appearance of the 3H-TdR-labeled ependymal cells was similar to that observed in the fish at 26 POD (Fig. 6b), and organellae such as Golgi apparatus, mitochondria and rough endoplasmic reticulum were well developed (Fig. 6c). Many regenerating axons, both myelinated and unmyelinated, were seen within the ependymal cell layer.
In the present study, using PCNA immunohistochemistry and light and electron microscopic 3H-TdR autoradiography, it was clearly demonstrated that ependymal cells of the carp spinal cord proliferate actively in response to transection injury.
The proliferative reaction of the carp ependymal cells was weak up to 3 POD and became very intense at 6 POD. This is in contrast to the general belief that mature ependymal cells of the human spinal cord have little, if any, capacity for reactive proliferation (Duchen, 1984). Ependymal cells of the rodent spinal cord have also been reported to show strong proliferation in response to injury (Adrian and Walker, 1962; Kraus-Ruppert et al., 1975; Gilmore and Leiting, 1980), our preliminary study with BrdU immunohistochemistry of the rat spinal cord showed that the number of BrdU-labeled ependymal cells was increased up to about 7 times with a peak at 10th day after hemisection of the spinal cord (our unpublished data). On the other hand, the peak of proliferation of GFAP (glial fibrillary acidic protein)-positive astrocytes in the cortex of rat cerebrum was observed 3 to 4 days after injury (Miyake et al., 1988). Proliferative response of ependymal cells in the carp and rat spinal cords seem to occur more slowly compared with that of astrocytes in rat brains.
The PCNA/all ratio of ependymal cells in the caudal side of the transection was about 3 times higher than that in the rostral side at 6 POD, and the ependymal cells were usually cylindrical in shape in the caudal side while they were flattened in the rostral side. These differences might depend on some chemical trophic factors such as NGF and would be worth another study. It seems to safe that no fine structural observation can answer this question.
At 5 POD, the central canal in the caudal side of the transection was seen to be obliterated, and elongated ependymal cells were arranged tightly with each other, forming a thick ependymal cell layer. The apical part of these ependymal cells contained numerous free ribosomes but few other intracellular organellae. Although this electron microscopic feature was similar to that of immature cells (matrix cells) of the early embryonic brain (Fujita, 1963), the ependymal cells observed in this study characteristically contained glial filaments in their cytoplasmic processes. This means that these cells were not immature matrix cells but were more mature cells differentiated into ependymal cells. Some of these ependymal cells have proliferative activity, as revealed by the flash labeling with 3H-TdR.
Neurogenesis of the lower vertebrates continues even in adulthood, as demonstrated, for example, in the optic tectum of goldfish (Raymond and Easter, 1983), the telencephalon of lizards (Perez-Sanchez et al., 1989) and the hypothalamus of frogs (Chetverukhin and Polenov, 1993; Polenov and Chetverukhin, 1993). Anderson and Waxman (1985) reported that some ependymal cells of the spinal cord of teleosts, Sternarchus albifrons retained their capability to divide, and these cells proliferate to generate nerve and glial cells following amputation of the spinal cord. Furthermore, Anderson and Waxman (1985) detected signal of 3H-TdR on nerve cells after long term exposure to 3H-TdR of the cultured spinal cord explant. In the present study, the cells in the ependymal cell layer which had proliferated in response to injury maintain the characteristic figure of glial nature such as glial intermediate filaments, and we could not find any 3H-TdR-labeled nerve cells. This fact suggests that the production of nerve cells does not occur in the adult carp spinal cord.
Numerous regenerated axons were observed in the ependymal cell layer which was reconstructed after transection of the carp spinal cord. In rats as well as in fish, proliferation of the ependymal cells occurs after spinal cord injury as mentioned above. However, the gap of the rat spinal cord caused by transection is not connected with regenerated axons (our unpublished data). Thus, proliferation of ependymal cells by itself is likely to be insufficient for axonal regeneration. Astroglial processes are known to be the major impediment to axonal regeneration in the CNS of higher mammals (Reier et al., 1983; Liuzzi and Lasek, 1987; Stensaas et al., 1987). However, we have previously shown that the processes of ependymal and astroglial cells, which have common fine structural characteristics, do not prevent elongation of regenerating axons in the spinal cord of carp (Yamada et al., 1995). The key to axonal elongation might be found in the physicochemical differences of the cell membrane of ependymal and glial cells between higher and lower vertebrates.