Fine structure of regenerated ependyma and spinal cord in Sternarchus albifrons.код для вставкиСкачать
THE ANATOMICAL RECORD 205:73-83 (1983) Fine Structure of Regenerated Ependyma and Spinal Cord in Sternarchus albifrons MARILYN J. ANDERSON, STEPHEN G.WAXMAN, AND MICHAEL LAUFER Department of Neurology, Veterans Administration Medical Center and Stanford University School of Medicine. Pa10 Alto, CA 94304 ABSTRACT The caudal-most regenerated spinal cord in Sternarchus albifrons consists solely of an ependymal tube. Ependymal cells are enlarged radially and are more numerous than in unregenerated cord. Projections of ependymal cell cytoplasm and Reissner’s fiber fill most of the central canal. Small groups of neurites and cell processes filled with dense-cored vesicles lie between abluminal processes of ependymal cells. Rostra1to this, additional cells appear dorsal and lateral to the inner ependymal layer. Some cell bodies contain numerous dense-cored vesicles. Larger bundles of neurites, some with synapses, are present. Invaginations of the peripheral edge of the cord create enclosed spaces lined with basal lamina. In the peripheral region, longitudinally oriented neurites extend through extracellular spaces or channels. The ventral portion a t some levels of regenerated cord is completely filled with neurites, processes containing dense-cored vesicles, and capillaries. Similar masses of neurites and processes containing dense-cored vesicles lie outside the cord proper, in or near the meningeal layer. In rostral-most sections, the organization of regenerated spinal cord approaches that of normal cord, with the regenerated cord exhibiting groups of myelinated axons, differentiated fibrous astrocytes and oligodendroglia,cell bodies containing dense-cored vesicles, and differentiated electromotor neurons. These observations indicate a degree of pluripotency in some of the ependymal cells in adult Sternarchus. Moreover, they are consistent with a role of ependymal cells in the guidance of regenerating neurites. It has long been recognized that the vertebrate central nervous system does possess some regenerative capability after injury. In mammals, limited functional recovery and regrowth of axons may occur after injury to the spinal cord (Bernstein and Bernstein, 1973; Guth, 1975; Puchala and Windle, 1977). Various experimental manipulations are reported to enhance regeneration of mammalian spinal axons: Application of an electromagnetic field (Wilson and Jagadeesh, 1976), grafting peripheral nerve into the injured cord (Kao et al., 1977; David and Aguayo, 1981), wrapping the cut end of the cord with millipore filter (Scott and Liu, 19641, and adding various chemical agents (McMasters,1962;Bernstein et al., 1978; Feringa et al., 1979).Despite these provocative approaches, regeneration in mammalian spinal cord is limited and the mechanism underlying it appears to be largely the regrowth of cut axons, rather than the generation of new nerve cell bodies. More complete regeneration of the spinal cord, in some cases including the generation of new neurons, occurs in lower vertebrates: teleosts, urodele amphibians, and lizards (Kirsche, 1951; Butler and Ward, 1967; Simpson, 1968; Bernstein and Gelderd, 1973). In lizards, the regenerated spinal cord is comprised of an ependymal tube and descending nerve fibers (Egar et al., 1970). In addition, in newt spinal cord new neuronal cell bodies are regenerated after amputation of the tail (Egar and Singer, 1972). In regenerated spinal cord of the teleost Sternarchus a1bifrons, new electromotor neurons are produced in even greater numbers than are present normally (Anderson and Waxman, 1981). One important problem concerns the differences between the inframammalian spinal cord, which readily regenerates, and mammalian Received April 29,1982;accepted September 28,1982. 0003-276X/83/2051-0073$03.500 1983 ALAN R. LISS, INC. 74 M.J. ANDERSON, S.G. WAXMAN, AND M. LAUFER cord, which does not, The structure and role of the ependyma are undoubtedly important in the ability of inframammalian spinal cord to regenerate. Enlargement and proliferation of ependymal cells has been noted as one of the first steps in regeneration after amputation of the tail in amphibians and lizards (Simpson, 1968; Nordlander and Singer, 1978) and also after transection of the spinal cord in larval lampreys (Wood and Cohen, 1981). The sequence of generation of new cells from the ependymal layer has been described in detail in amphibians (Egar and Singer, 1972; Nordlander and Singer, 1978) and lizards (Egar et al., 1970). In addition, the presence of ependyma is necessary for regeneration of spinal cord and other structures of the tail in salamander (Goldfarb, 1909) and lizard (Kamrin and Singer, 1955; Simpson, 1964). The ependymal cells may also influence or guide the outgrowth of regenerating neurites through regenerating spinal cord. Ependymal cell processes are in close contact with outgrowing axons in regenerating Xenopus spinal cord (Michel and Reier, 1979) and may influence their growth by direct contact. In regenerating and embryonic spinal cords of some species, longitudinal channels, formed by spaces between ependymal cell processes, have been observed to anticipate the appearance of neurites. It has been proposed that such channels may guide neurite outgrowth through the cord (Singer et al., 1979; Nordlander and Singer, 1978, 1982). Inframammalian systems provide important models for the study of spinal cord injury and regeneration in several major ways. First, regeneration can initially be studied more easily in a system where it unequivocally occurs; results may then be extrapolated to systems (such as mammalian spinal cord) where regeneration occurs incompletelyif at all. Secondly, some of the basic processes of regeneration, e.g. regrowth of nerve fibers, are common to both mammals and inframammalian vertebrates. Thus, information gained from the study of spinal cord regeneration in lower vertebrates may be in some cases applicable to the mammalian (nonregenerative) nervous system. The teleost Sternarchus albifrons represents an especially advantageous system for study of both spinal cord and peripheral nerve regeneration. First, spinal cord regeneration, including the generation and differentiation of new neurons, occurs readily after amputation of the tail (Anderson and Waxman, 1981). Second, the spinal cord of Sternarchus contains a class of easily identified neuronal perikarya, whose axons peripherally comprise the electric organ of this weakly electric fish (Waxman et al., 1972; Pappas et al., 1975). The electrocyte axons run a highly distinctive and stereotyped course and have a very specific pattern of myelination which varies along the length of the axon. They also contain two types of nodes of Ranvier: normal, excitable nodes (Type I) and large, inexcitable (Type 11) nodes. Thus, the electrocyte axons provide a good system for studying the factors which guide nerve outgrowth and influence myelination. Moreover, the electrocyte axons are asynaptic, ending blindly in the electric organ. Thus, these cells are amenable to the investigation of the dependence of certain properties of neurons on synaptic connections, e.g. chromatolysis (Waxman and Anderson, 1982) and cell death (Anderson et al., 1982). Lastly, the production in regenerated spinal cord of significantly more electrocytes than normal suggests this system as a possible model in which to study the stimulation and control of cell division in response to injury. In light of the potential usefulness of this system of identifiable regenerating neurons, we have undertaken this description of the ultrastructure and relationship of neuronal and ependymal elements in the regenerating spinal cord of Sternarchus albifrons. MATERIALS AND METHODS To initiate regeneration, tails were amputated from Sternarchus anesthetized in a 1:15,000 solution of tricaine methanesulfonate (MS 2221, and the fish were returned to their tanks. After a regeneration period of 3 weeks, 4 months, or 1 year a t 26" C, the regenerated spinal cord was fixed either by perfusion or direct immersion in a solution of 2.5%glutaraldehyde in 0.1M sodium phosphate buffer with 2.5%dextrose and 0.01%CaC12, pH 7.35, 350 mOsm. The spinal cord was dissected in fixative, then immersed in a fresh vial of fixative, for a total of 3 hours fixation (at room temperature). The tissue was washed in 4 changes of the phosphate buffer over 45 minutes, and postfixed for 2 hours in 2% OsOlin the phosphate buffer, a t room temperature. The tissue was then washed for 2 hours in 0.2M acetateacetic acid buffer (pH 5) and stained en bloc with uranyl acetate. Tissue was embedded in Epon-araldite and thin sections were stained with lead citrate and uranyl acetate prior to observation on a JEOL 100 CX electron microscope. REGENERATED SPINAL CORD 75 Fig. 1. Transverse section of caudal regenerated ependyma, 8.5 mm from the site of transection in a 4-month regenerated fish. Central canal is largely filled with Reissner's fiber (R) and protrusions of ependymal cell cyto- plasm (arrows). N, ependymal cell nucleus; M, meningeal layer. Inset shows abluminal cell processes filled with densecored vesicles. x 5,400; inset, x 11,400. RESULTS radial axis, and are more numerous than in the intact spinal cord. We have counted up to 86 cells bordering the central canal in a regenerated spinal cord, as opposed to an average of 36 cells in normal, unregenerated ependyma. Lateral borders of the caudal regenerated ependymal cells are generally straight. Ependymal cell cytoplasm contains many ribosomes and polyribosomes, rough endoplasmic reticulum, mitochondria, Golgi apparatus, clear vesicles, and occasionally a multivesicular body. The Golgi apparatus is always located in the apical (luminal) half of the cell. The nuclei are large, usually oval, and have dispersed nucleoplasm or small patches of condensed chromatin. The regenerated ependymal cells are of one uniform cytoplasmic density. Large irregularly shaped projections of the apical border of caudal ependymal cells (Fig. 1) extend for distances up to 3 km into the central canal, where they fill much of the space. Reissner's fiber is present in the central canal, We have examined regenerated spinal cord in 15 fish, which had regenerated for periods ranging from 3 weeks to 1 year. This paper primarily presents results from one typical specimen, a 4-month regenerate. Sections a t different levels along the rostro-caudal axis of the regenerated spinal cord from this animal will be presented, as well as sections from two other fish, which regenerated €or different periods of time. Regeneration of the tail in Sternurchus occurs a t an initial rate of 0.6-1.7 mm per week at 26" C. After 3-4 months, regeneration is slower. Regenerated spinal cord close to the site of transection is of normal diameter, and decreases in diameter as the spinal cord progresses caudad. The caudal region of regenerated spinal cord in a fish with a four month regeneration period is presented in Figure 1.At this level (Fig. l), the cord consists solely of a tube of enlarged ependymal cells which are elongated along the 76 M.J. ANDERSON, S.G. WAXMAN. AND M. LAUFER along with many cilia extending from the ependymal cells. The cilia contain the typical 9 x 2 arrangement of doublet microtubules. Cross-striated rootlets with dense bands a t a periodicity of 600-700 A, are sometimes seen extending from the basal bodies of the cilia for several pm into the apical cytoplasm. Adjacent ependymal cells are joined a t their luminal edges by junctional complexes, followedby desmosomes along the lateral edges. Long desmosomes and series of 3-5 closely spaced desmosomes are common. Except for the terminal bar, tight junctions or gap junctions between ependymal cells are not seen in these caudal sections. Large cell processes packed with dense-cored vesicles occur in the abluminal region between some caudal ependymal cells (inset, Fig. 1). The membrane-boundvesicles are 800-1400 A in diameter. Some of the vesicles appear less dense than others. A few mitochondria, microtubules, and 100 A filaments are seen amidst the dense-coredvesicles in such processes. While caudal-most sections of regenerated ependyma have only a few of these vesicle-filledprocesses, they are more frequent as one progresses rostrally in the regenerated cord. Dense-cored vesicles are also frequently found in typical neurites within the regenerating ependyma, and in the meningeal layer adjacent to the ependyma. Individual fibers and small bundles of neurites are seen in the abluminal region of caudal ependyma (Fig. 2, inset). The neurites are in close proximity to adjacent ependymal cell processes. Large empty spaces between ependyma1 cells were not observed. Neurites in this region of regenerated cord are unmyelinated and are oriented parallel to the longitudinal axis of the ependyma. No synapses are seen between neurites in the caudal-most sections. Neurites can occur close to the central canal (within 1.5 pm), although they are more usually found near the periphery. Neurites are also frequently located outside the regenerating spinal cord in the meningeal layer and beyond it. Such extra-spinal neurites are always surrounded by glial processes, and are often grouped together inside a ring of fibroblastic processes and collagen. In sequential sections progressing caudally from the section shown in Figure 1,a large myelinated axon was seen first external to the meningeal layer, then inside the regenerating ependyma, lying between the abluminal portions of ependymal cells. Rostra1 to the section in Figure 1, regenerated ependyma has several layers of cells surrounding the central canal (Fig. 2). There is no clear demarcation separating the cells immediately lining the central canal and those more peripherally located. One of the peripheral cells in Figure 2 (asterisk) has a dispersed nucleoplasm and a cytoplasm distinctly different from that of the ependymal cells, thus representing one of the first differentiating cells in the regenerating spinal cord. This cell also has dense-cored vesicles in its cytoplasm. Ependymal cells in this section are joined, as they are more caudally, by a terminal bar and series of desmosomes. Gap junctions are not seen between ependymal cells at this level of regenerating cord. Projections of ependymal cell cytoplasm into the central canal are smaller than in the more caudal section, while the large Reissner’s fiber occupies much of the space in the central canal. The peripheral edge of the regenerating ependyma (Fig. 3) is bounded by an external lamina. Outside the external lamina, a layer of longitudinally oriented collagen, 1-5 pm thick, lies between the ependyma and the surrounding meningeal cells. The basal portions of ependymal cells are convoluted, interdigitating among each other. Desmosomesjoin adjacent cell processes, even a t the very edge of the ependyma (Figs. 3, 4). Invaginations and infoldings of the extra-ependymal space, bounded by the external lamina, frequently occur at the edge of the cord (Fig. 3). In some cases the basal lamina appears attenuated in these invaginations. The spaces lined with basal lamina can be relatively large, up to 10 pm along their major axis. In more rostra1 sections, such lamina-lined spaces also occur more internally in the regenerated cord, and neurites have occasionally been seen in them. A large bundle of neurites from regenerated spinal cord is shown in Figure 4. Progressing rostrally in the regenerated cord, neurite bundles increase in size and may have some distinctly larger fibers at the edge of the bundle. A synapse, characterized by clear vesicles and a density a t the pre- and post-synaptic membrane, can be seen in the upper part of this figure. Dense-cored vesicles can occur in both pre- and post-synaptic processes. A second specialized contact, seen in the lower portion of the micrograph, may be an axo-glial synaptoid relationship. Clear vesicles joining with the plasma membrane, reminiscent of micropinocytosis, are seen at borders between cells and at the peripheral edge of the cord. As in the more caudal sections, neurites lie close to the surrounding ependymal cell processes. A section of regenerated spinal cord from a different fish (with a 3-week regeneration period) shows many layers of cell bodies lateral Fig. 2. Regenerated spinal cord 1.5 mm rostra1 to the section in Figure 1. Additional cells are present lateral to the immediate ependymal layer. One of these cells (asterisk) is differentiated morphologically from Lie ependymal cells, and contains dense-cored vesicles in its cytoplasm. Central canal with Reissner’s fiber at lower left; invaginations of peripheral border of cord (arrowhead, and Fig. 3); neurite bundle (N, and Fig. 4). Inset shows neurites extending in close contact with ependymal cell borders, located 1.6 pm from the central canal in a second regenerated spinal cord. X 5,000; inset, x 13,800. Fig. 3. Edge of regenerated spinal cord from Figure 2. An external lamina lines the periphery of the cord and the enclosed spaces (arrows). Outside this lamina lie longitudinally-oriented collagen fibrils (C). Cell processes at the edge of the cord are interdigitated and are frequently joined by desmosomes. x 30,100. 78 M.J. ANDERSON, S.G. WAXMAN, AND M. LAUFER Fig. 4. Neurite bundle from a section of regenerated spinal cord close to Figure ?,. Large neuritea surround the distal edge of the bundle. A synapse (wide arrow) is present near top. A second synaptoid profile in lower half (thin arrow) may represent an axo-glial contact. Note desmosomes near outer edge of spinal cord (top), and basal lamina (B) in enclosed space. Micropinocytotic-like vesicles join with plasma membranes a t the periphery (inset and arrowheads) and a t lateral cell borders (arrowhead). X 17,800;inset, x 49,400. 79 REGENERATED SPINAL CORD to the immediate ependymal layer (Fig. 5).Cells appear to progress radially away from the ependyma. The cells close to the ependymal layer resemble the ependymal cells in shape and scant cytoplasmic content. More peripheral cells are polygonal in shape. Some peripheral cells have an electron dense cytoplasm and resemble oligodendroglia; other cells contain prominent dense bodies or lysosomes. As in the 4-month regenerated ependyma (Fig. 2), there is no demarcation separating the immediate ependymal layer from the more peripheral cells. Near the edge of the cord in this specimen, groups of unmyelinated neurites extend through extracellular spaces (inset, Fig. 5 ) which resemble the channels described by Singer et al. (1979).A light flocculant material is also present in some of these channels. The channels are located in the rostral portion of this 3 week-regenerated tail, 2 mm from the end of the tail. In the 4-monthregenerated spinal cord, many cell bodies and cell processes containing densecored vesicles are present. In some regions, virtually the entire ventral half of the cord is filled with neurites and processes containing densecored vesicles, interspersed with capillaries and epithelial cell processes (Fig. 6). Many of the processes are totally filled with clear and densecored vesicles, and abut directly upon the basal lamina of a capillary. However, no signs of vesicle depletion or secretion into the capillaries were observed. Dense-cored vesicles have not been observed in myelinated axons, nor in the central canal. The cell bodies containing dense-cored vesicles are 10-20 pm in diameter and very irregular in shape, often having bulbous or long thin projections which also contain the densecored vesicles. The vesicles are sometimes closely associated with the Golgi apparatus. Caudally, the cells containing dense-coredvesicles are always located near the periphery of the regenerated cord. In more rostral sections, they are found in many locations, including the area just lateral to the ependymal cells. The cells containing dense-cored vesicles receive multiple synapses from unmyelinated axons. The frequency of dense-cored vesicles in cells and processes in the regenerated spinal cord is especially striking in comparison to their infrequent occurrence in unregenerated spinal cord at the same spinal level and in cord anterior to the transection in the operated fish. In the latter two cases, one or two dense-cored vesicles are found in a few neurites, and no cell bodies containing dense-cored vesicles are seen (see Anderson and Waxman, 1982, pp. 85-92, this volume). Rostra1 sections of the 4-month regenerated spinal cord (1-5 mm from the site of transection) are characterized by the presence of many myelinated axons, large masses of unmyelinated neurites near the periphery, and by differentiated oligodendroglia,fibrous astrocytes, motor neurons and the differentiated electromotor neurons (Anderson and Waxman, 1981). The differentiated electrocytes are identified by their size, spherical shape, lack of dendrites, and their cytoplasm packed with mitochondria and ribosomes. Fibrous astrocytes are plentiful, and a well-defined glia limitans is present. Cells with electron-dense cytoplasm, presumably oligodendroglia, are frequently seen partially surrounding small-diameter myelinated axons. The ependymal layer is clearly distinct from adjacent cells and axons. Some ependyma1 cells have a long, fiber-filled process which extends deeply into the neuropil. Gapjunctions are seen between lateral borders of ependymal cells. In contrast to the more central position of cell bodies in normal, unregenerated cord, differentiated cell bodies of both the electrocytes and the cells containing dense-cored vesicles can occur a t the very edge of the regenerated cord. Although the number of cell bodies containing dense-cored vesicles decreases in rostral regenerated cord, degeneration or death of these cells has not been observed. Cell processes filled with both clear vesicles and densecored vesicles are seen in rostral regenerated cord, but are more frequent in the caudal regenerated cord. We have found no limit to the number of times the spinal cord will regenerate in one fish, nor to the distance rostrally that the cord will regenerate. Tails have been re-amputated from several fish 4 times each, over a period of one year, and regeneration proceeds as rapidly the fourth time as the first. The spinal cord and other structures of the tail still regenerate when a tail is amputated as far rostral as 5.5 cm from its caudal end, which can be virtually half the length of the fish. Although we have not amputated the spinal cord rostral to this level, it seems likely that it too would regenerate. DISCUSSION The caudal-most region of regenerated spinal cord in Sternarchus consists solely of the ependymal tube, as in regenerated spinal cord of lizard and newt (Egar et al., 1970; Egar and 80 M.J. ANDERSON, S.G. WAXMAN, AND M. LAUFER Fig. 5. Regenerated ependyma from a specimen with a 3-week regeneration period. Many cells extend radially from the central canal ( 0 , which contains cell processes and secreted material. Inset shows neurites in a channel between peripheral cell processes. Neurites are separated by cell processes with a darker cytoplasm. Basal lamina a t leR of inset pertains to adjacent space internal in the cord (not the cord periphery). x 2,700; inset, x 15,550. Fig. 6. Ventral region of a regenerated spinal cord, filled with neurites (wide arrows) and cell processes containing dense-cored vesicles (thin arrows). A capillary (Cp) with nucleated red blood cell is present in lower right comer. x 10.900. REGENERATED SPINAL CORD Singer, 1972). Rostra1 to this, ependymal cell mitoses generate numerous cells peripherally which differentiate into the various neuronal and glial cells of the spinal cord. This process occurs with some rapidity, producing new spinal cord segments a t a rate of about 1 mm per week. A similar sequence of spinal cord regeneration has been described in regenerating amphibian cord (Nordlander and Singer, 1978). The results presented here indicate that cells of the spinal cord s t a r t differentiating very early in the process of tail regeneration. Differentiating cells are present in the earliest sections where cells are noted peripheral to the ependyma. These cells, which contain dense-cored vesicles, are probably the cells of origin of the vesicle-packed processes seen in caudal sections of regenerated cord. In somewhat more rostral regenerated cord, oligodendroglia and fibrous astrocytes have differentiated. Even more rostrally, the large electromotor neurons characteristic of Sternarchus spinal cord differentiate and recapitulate their normal, distinctive morphology (Anderson and Waxman, 1981). The first stages of differentiation of the electromotor neurons have not yet been identified. Fully differentiated electrocytes are present in rostral regenerated cord within 3 months after amputation, but are not yet generated in the 3 week-regenerate. It is unclear whether some of the cells which initially contain dense-cored vesicles can later differentiate into electromotor neurons or any other cell type. We know that the cells containing dense-cored vesicles are among the first to differentiate, and are prevalent in more caudal sections of regenerated cord. In more rostral sections, the electrocytes are the prominent neuronal cell type, with fewer cells containing dense-cored vesicles in evidence. Since differentiation procedes in a rostro-caudal direction in regenerating spinal cord, with the rostral-most cord having regenerated the longest amount of time, and caudal-most cord the least, we are led to ask what happens to the cells containing densecored vesicles as the regenerating cord “matures.” If these cells do not represent an initial stage in the differentiation of some other cell type, then their absence in mature regenerated cord may be due to cell death or migration. No evidence of death or degeneration of the cells containing dense-cored vesicles was observed in this study. However, we cannot rule out this possibility, especially if the region of cell death is relatively localized. The second possibility, that the cells containing dense-cored vesicles are later converted to a different 81 cell type, also remains open. There is evidence for the conversion of adrenergic cells (exhibiting synapses characterized by dense-cored vesicles) to cholinergic cells (with synapses characterized by clear vesicles) both in vitro and in vivo (Patterson, 1978). Cell processes filled with both clear and dense-cored vesicles, which might be indicative of an intermediate stage in differentiation, do occur in the regenerated spinal cord in Sternarchus. However, these processes were not necessarily associated with other synaptic specializations. In addition, these processes are most frequently found in the caudal region of regenerated cord (where the cells and processes containing dense-cored vesicles are also most prevalent). It is also noteworthy that the cells containing dense-cored vesicles regenerate, after amputation of the tail, from rostral locations where normally none of these cells would be found. If regenerated spinal cord is generated from the extant ependyma alone, these results argue for the pluripotency of at least some of the rostral, “mature” ependymal cells in Sternarchus. The morphology of the cells containing densecored vesicles very closely resembles that of the neurosecretory, or “Dahlgren,” cells of the caudal neurosecretory system normally found in teleosts (Bern and Takasugi, 1962; and see Anderson and Waxman, 1982, pp. 85-92, this volume). Regeneration of caudal neurosecretory cells from the remaining ependyma after amputation of the tail has been reported in another teleost, Tilapia mossambica (Fridberg et al., 1966). The question arises whether the presence of neurosecretory cells in regenerated spinal cord is functionally significant. In Sternarchus, it is not yet known if the secretory products (dense-cored vesicles) or the physiological function are the same as those of caudal neurosecretory systems of other fish (Anderson and Waxman, 1982, pp. 85-92, this volume). No signs of depletion or secretion of dense-cored vesicles (as are seen in Tilapia) were seen in the neuro-hemal areas of regenerated Sternarchus spinal cord. It is possible that the neurosecretory cells (cells containing dense-cored vesicles) are regenerated simply because their normal location is at the end of the tail and that is where regeneration is occurring. The means of encoding and recognizing such positional information would be of great interest. What signals spinal cord in the middle of the fish to differentiate with characteristics appropriate to the end of the fish, after amputation of the tail? It may be that amputation, formation of the blastema, or multiplication of 82 M.J. ANDERSON, S.G. WAXMAN, AND M. LAUFER the ependymal cells triggers a recapitulation of embryological development,which results in regeneration of the caudal neurosecretory system. The grouping of longitudinally oriented neurites in extracellular spaces near the periphery of regenerating Sternarchus spinal cord corresponds to the outgrowth of neurites in channels between ependymal cell processes in regenerating amphibian and reptile cords (Egar et al., 1970; Simpson, 1968). The outgrowth of regenerated spinal neurites in Sternarchus differs from that proposed in the “blueprint hypothesis” of Singer et al. (1979) in that no evidence was found for channels present in advance of neurite outgrowth. In the caudal sections of regenerating Sternarchus spinal cord, individual neurites and groups of neurites extend in close contact t o ependymal cell borders. Only in more rostral sections of regenerating cord were large extracellular spaces with neurites in them observed. Axonal regeneration without preexisting ependymal channels has also been observed in transected and r e d spinal cord of Xenopus tadpoles (Michel and Fkier, 1979). The unmyelinated neuritic processes observed in early regenerating ependymal tube are probably axonal rather than dendritic, since they are usually found in groups and are parallel to each other. These processes presumably extend posteriorly from cell bodies located in the unregenerated part of the spinal cord rostral to the transection, or from differentiated cells in the rostral part of the regenerated cord. The fact that individual neurites do also grow through the ependyma indicates that the grouping of neurites is not necessary for their outgrowth. Likewise, axons growing outside of the regenerating cord are often in groups, but single axons (surrounded by glial processes) are also common. The possible role of astrocytes in retarding or encouraging regeneration has been commented upon by a number of workers. The presence of numerous fibrous astrocytes in regenerating Sternarchus spinal cord is noted as evidence that fibrous astrocytes do not, per se, inhibit regeneration. Also addressing this point, b i e r (1979)has shown that regenerating optic axom can penetrate astrocytic scars. While the regeneration of spinal cord in Sternarchus is in large part similar to spinal cord regeneration previously reported in other animals (Simpson, 1968; Nordlander and Singer, 19781, Sternarchus has several outstanding differences which may be functionally important to an interpretation of the process of spinal cord regeneration. These include the lack of obvious channels in advance of neurite outgrowth, the frequent invaginations of adluminal ependymal cell surface, the presence of numerous cells and processes containing densecored vesicles, and the (previously noted) generation of more electrocyte cell bodies than are normally present. ACKNOWLEDGMENTS This work was supported in part by the Medical Service, Veterans Administration, and by grants from the National Institutes of Health and the Paralyzed Veterans of America. We are grateful to Mary E. Smith and Marsha s.Kantor for their excellent technical support. LITERATURE CITED Anderson, M.J., and S.G. Waxman (1981) Morphology of regenerated spinal cord in Stemrchw albifins. Cell Tiss. Res. 219:1-8. Anderson, M.J., and S.G. Waxman (1982) Caudal spinal cord of the teleost Sternarchus albifrons resembles regenerating cord. Anat. Rec., 205:85-92. Anderson, M.J., S.G. Waxman, and C.H. Tadlock (1982) Cell death of asynaptic neurons in regenerating spinal cord. J. Cell Biol., 95:32a. Bern, H.A., and N. Takasugi (1962)The caudal neurosecretory system of fishes. Gen. Comp. Endocrinol. 2:96-110. Bernstein, J.J., and M.E. Bernstein (1973) Neuronal alteration and reinnervation and sprouting in the mammalian cord. Brain Behav. Evol., 8:135-161. Bernstein, J.J., and J.B. Gelderd (1973) Synaptic reorganization following regeneration of goldfish spinal cord. Exp. Neurol., 41:402-410. Bernstein, J.J., M.R. Wells, and M.E. Bernstein (1978) Effect of puromycin treatment on the regeneration of hemisected and transected rat spinal cord. J . Neurocytol., 7215-228. Butler, E.G., and M. Ward (1967)Reconstitution of the spinal cord &r ablation in adult Tri t um.Dev. Biol., 15:464-#36. David, S.,and A.J. Aguayo (1981)Axonal elongation into peripheral nervous system “bridges” after central nervous system injury in adult rats. Science, 214:931-933. Egar, M., S.B. Simpson, and M. Singer (1970) The growth and differentiation of the regenerating spinal cord of the lizard, Anolis carolinensis. J. Morphol., 131 :131-152. Egar, M., and M. Singer (1972) The role of ependyma in spinal cord regeneration in the urodele, Triturw. Exp. Neurol., 37:422-430. Feringa, E.R., K.R. Nelson, H.L. Vahlsing, and R.C. Dauser (1979) Spinal cord regeneration in rats made immunologically unresponsive to CNS antigens. J. Neurol. Neurosurg. Psych., 42:642448. Fridberg, G., R.S. Nishioka, H.A. Bern, and W.R. Fleming (1966) Regeneration of the caudal neurosecretory system in the cichlid teleost Tilapia mossambica. J. Exp. Zool., 162:311-336. Goldfarb, A.J. (1909) The influence of the nervous system in regeneration. J. Exp. Zool., 7:643-722. Guth, L. (1975) History of central nervous system regeneration research. Exp. Neurol., 48(No.3, Part 2):3-15. Kamrin, R.P., and M. Singer (1955) The influence of the spinal cord in regeneration of the tail of the lizard, Anolis carolinensis. J . Exp. Zool., 128:611427. Kao, C.C., L.W. Chang, and J.M.B. Bloodworth (1977) Axonal regeneration across transected mammalian spinal REGENERATED SPINAL CORD cords: an electron microscopic study of delayed microsurgical nerve grafting. Exp. Neurol., 54:591-615. Kirsche, W. (1951) Uber die Neubildung von ganglienzellen nach durchtrennung des ruckenmarkes bei erwachsenen Teleostiern. Erg. H. zu Anat. Anz. 97:127-132. McMasters, R.E. (1962) Regeneration in the spinal cord of the rat. Effects of Piromen and ACTH upon the regenerative capacity. J . Comp. Neurol., 116:113-125. Michel, M.E., and P.J. Reier (1979) Axonal-ependymal aasociatiom during early regeneration ofthe transected spinal cord in Xenopus Zueuis tadpoles. J. NeuFocytol., 8:529-548. Nordlander, R.H., and M. Singer (1978) The role of ependyma in regeneration of the spinal cord in the urodele amphibian tail. J . Comp. Neurol., 180:349-374. Nordlander, R.H.,and M. Singer (1982) Spaces precede axon8 in Xenopus embryonic spinal cord. Exp. Neurol., 75~221-228. Pappas, G.D.,S.G. Waxman, and M.V.L. Bennett (1975) Morphology of spinal electromotor neurons and presynaptic coupling in the gymnotid Sternarchus albifrons. J. Neurocytol., 4:469478. Patterson, P.H. (1978) Environmental determination of autonomic neurotransmitter functions. Ann. Rev. Neurosci., 1:l-17. Puchala, E., and W.F.Windle (1977)The possibility of structural and functional restitution after spinal cord injury. A Review. Exp. Neurol., 55:1-42. Reier, P.J. (1979) Penetration of grafted astrocytic scars by regenerating optic nerve axons in Xenopus tadpoles. Brain Res.. 164:6148. 83 Scott, D.,Jr., and C.N. Liu (1964) Factors promoting regeneration of spinal neurons: Positive influenceof nerve growth factor. Prog. Brain Res., 131:127-150. Simpson, S.B. (1964) Analysis of tail regeneration in the lizard Lygosoma laterale. 1. Initiation of regeneration and cartilage differentiation: the role of ependyma. J. Morphol., 114:425436. Simpson, S.B. (1968) Morphology of the regenerated spinal cord in the lizard Anolis carolinensis. J. Comp. Neurol., 134:193-210. Singer, M., R.H. Nordlander, and M. Egar (1979) Axonal guidance during embryogenesis and regeneration in the spinal cord of the newt: the blueprint hypothesis of neuronal pathway patterning. J. Comp. Neurol., 1851-22. Waxman, S.G., and M.J. Anderson (1982) Retrograde axon reaction following section of asynaptic nerve fibers. Cell Tissue Res., 223:487492. Waxman, S.G., G.D.Pappas, and M.V.L. Bennett (1972) Morphological correlates of functional differentiation of nodes of Ranvier along single fibers in the neurogenic electric organ of the knife fish Sternarchus. J. Cell Biol., 53:210-224. Wilson, D.H., and P. Jagadeesh (1976) Experimental regeneration in peripheral nerves and the spinal cord in laboratory animals exposed to a pulsed electromagnetic field. Paraplegia, 14.12-20. Wood, M.R., and M.J. Cohen (1981) Synaptic regeneration and glial reactions in the transected spinal cord of the lamprey. J . Neurocytol., 10:57-79.