Advances in Cellular Neurobiology: Volume 3

Index

Section 1. Cell Differentiation and Interaction

Outline

Cell Division in the Normal Central Nervous System

Schwann Cells: An in Vitro Perspective

Molecular and Cell Biological Aspects of Learning: Toward a Theory of Memory

Cell Division in the Normal Central Nervous System

R.R. Sturrock,     Department of Anatomy, University of Dundee, Dundee, Scotland

Publisher Summary

This chapter provides an overview of cell division in the normal central nervous system (CNS). There is conclusive histological evidence that subependymal cells (probably glioblasts) and astrocytes continue to divide throughout life. There is less evidence that oligodendrocytes divide throughout life because it is possible that cells identified as recently produced oligodendrocytes in adult animals may have arisen from the differentiation of subependymal cells. It has been shown that in some areas of the CNS, oligodendrocytes that have begun myelination are capable of division and therefore, the division of mature oligodendrocytes cannot be excluded. Astrocytes and oligodendrocytes found undergoing mitotic division have been shown to have the capacity to do so without retracting their processes, thus retaining their premitotic anatomical relationships with other cells or myelin sheaths. Ependymal cells, choroid plexus cells, pericytes, and neurons probably do not divide once they are fully differentiated; however, it does seem a possibility that in the restricted regions of the mammalian CNS, even neurons may be capable of division.

I Introduction

II Neural Epithelium. Ventricular and Subventricular Layers

A Patterns of Proliferation

B Fate of Cells Produced

III Mitotic Cells of Ectodermal Origin

A Ependymal Cells

B Subependymal Cells

C Neurons

D Astrocytes

E Oligodendrocytes

F Choroid Plexus Epithelial Cells

IV Mitotic Cells of Mesodermal Origin

A Macrophages (Including Microglia)

B Neurolipomastocytoid Cells

C Pericytes

D Endothelial Cells

V Conclusions

References

I Introduction

Many histology textbooks continue to cite the central nervous system (CNS) as an example of a tissue in which cell division does not occur once the constituent cells have differentiated, although over the last two decades considerable evidence has accrued showing that cell division continues in the nervous system into old age. The misconceptions regarding proliferative activity in the CNS have arisen because until recently little attention has been paid to neuroglia in contrast to that given to neurons. The aim of this article is to try to redress the balance. While the main emphasis will be placed on cell division in the differentiated CNS, patterns of mitotic division during early development will also be considered because development is a continuing process that does not cease with differentiation.

It is not intended to review pathological changes that follow CNS damage, nor is it intended to describe in detail autoradiographic studies of cell cycle time since a recent review article by Korr (1980) has covered this aspect of cell proliferation in the brain in great detail.

Among the earliest studies of cell division in the postnatal CNS are those of Hamilton (1901) and Allen (1912), who concluded that cell division occurred even in the mature CNS. Very little further attention was paid to cell division in the normal CNS until the application of autoradiography to the CNS (Messier et al., 1958; Hain et al., 1960; Smart and Leblond, 1961; Smart, 1961; Altman, 1962; Hommes and Leblond, 1967). A major reason for the lack of interest in cell division in the CNS was that mitotic cells in the CNS are particularly sensitive to anoxia (Cavanagh and Lewis, 1969) and the lack of adequate fixation techniques hampered studies of cell division, particularly in larger brains. The routine use of perfusion fixation and autoradiography made possible the study of cell division in the adult nervous system. Studies of mitotic division of neuroglia were also hindered by the absence of a suitable technique for accurate identification of the dividing cell. Metallic impregnation methods tend to be time-consuming and capricious and usually only impregnate one type of cell. In any case they are far from ideal for studying mitoses. Paraffin sectioned material stained with hematoxylin and eosin, or similar nuclear stains, demonstrates mitosis very well but not the identity of the dividing cell.

Immunocytochemical methods have been developed that enable astrocytes (Dahl and Bignami, 1973a,b, 1976; Bignami and Dahl, 1974; Schachner et al., 1977; Bignami et al., 1980) and oligodendrocytes (Sternberger et al., 1978a,b) to be identified, and these methods have so far been used to confirm the division of differentiated astrocytes following a brain wound (Latov et al., 1979).

Identification of mitotic cells in the postnatal CNS, however, remains a difficult problem. Except in the vicinity of the subependymal plate, mitotic figures are very rare (Allen, 1912; Bryans, 1959; Smart, 1961; Lewis, 1968; Sturrock, 1979a) and the chances of finding a mitotic cell in a piece of tissue prepared for election microscopy are not very high. Semithin sections permit much larger areas of tissue to be scanned and also enable different glial types to be accurately identified (Griffin et al., 1972; Ling et al., 1973; Sturrock, 1976b). The question still remains as to how accurately mitotic cells can be identified in semithin sections. Examination of a large number of mitotic cells in different regions of the CNS at different stages of development and the correlation of semithin sections with electron microscopy indicates that the cytoplasmic constituents, and hence the cytoplasmic staining characteristics of neuroglia at different stages of differentiation, including mature cells, remain constant throughout the mitotic cycle, thus allowing mitotic cells to be accurately identified in semithin sections (Figs. 9–15).

Fig. 9 One-micron toluidine blue-stained section showing a mature mitotic astrocyte with a pale cytoplasm from the corticospinal tract of a 5-day postnatum mouse. × 2000. (From Sturrock and McRae, 1980.)

Fig. 10 One-micron toluidine blue-stained section showing a mature mitotic astrocyte from the anterior commissure of a 3-month-old mouse. (From Sturrock, 1976a.)

Fig. 11 One-micron toluidine blue-stained section showing a mitotic oligodendrocyte with a process (arrow) connected to a myelin sheath. (From Sturrock and McRae, 1980.)

Fig. 12 One-micron toluidine blue-stained section showing an oligodendrocyte in telophase with a process (arrow) connected to a myelin sheath. (From Sturrock and McRae, 1980.)

Fig. 13 One-micron toluidine blue-stained section showing a mitotic microglial cell from a 12-day postconception mouse embryo spinal cord. This cell can be identified as a microglial cell by the presence of small vacuoles and dense bodies (arrow).

Fig. 14 One-micron toluidine blue-stained section showing a mitotic macrophage containing large phagosomes from the ventral horn of a 14-day postconception mouse spinal cord.

Fig. 15 One-micron toluidine blue-stained section showing a mitotic glioblast from the subependymal layer of a 22-month-old mouse subependymal layer. (From Sturrock and Smart, 1980.)

It is therefore possible by direct observation of mitotic cells in semithin sections to determine which types of neuroglia undergo mitotic division. These observations are also supported by autoradiographic studies, but care must be exercised in the interpretation of autoradiographic findings since many autoradiographic studies are based on cells which are labeled many hours or even days after injection. Thus although the identity of the daughter cells may be indisputable it is by no means certain which type of precursor cell divided to produce them. In a recent paper Kaplan and Hinds (1980) demonstrated that astrocytes and oligodendrocytes are produced in the adult rat brain but since the animals were killed one month after injection with [³H]thymidine the identity of the cells at the time of division was unknown. It is possible that the original dividing cells were undifferentiated cells which underwent differentiation during the month after injection.

II Neural Epithelium. Ventricular and Subventricular Layers

A Patterns of Proliferation

The phenomenon of interkinetic nuclear migration during the mitotic cycle in the neural epithelium was proposed by F. C. Sauer (1935a,b, 1936, 1937) from light microscopic observations and has since been confirmed by cytophotometric measurements of DNA content (M. E. Sauer and Chittenden, 1959) and by autoradiography (M. E. Sauer and Walker, 1959; Sidman et al., 1959; Fujita, 1962, 1963, 1965, 1966; Atlas and Bond, 1965; Martin and Langman, 1965). It has also been studied both by transmission electron microscopy (Duncan, 1957; Fujita and Fujita, 1963; Lyser, 1964; Fujita, 1966; Fisher and Jacobson, 1970; Hinds and Ruffett, 1971) and by scanning electron microscopy (Seymour and Berry, 1975).

The cells of the neural epithelium (Fig. 1) are at first undifferentiated and all appear to be identical morphologically (Fujita and Fujita, 1963; Meller et al., 1966; Wechsler, 1966; Wechsler and Meller, 1967; Sturrock and Smart, 1980), but as migration from the ventricular layer occurs and a mantle and marginal layer begin to develop, it is possible to identify ependymoglial cells that are morphologically distinct from the undifferentiated precursor cells (Sturrock, 1981a). The undifferentiated cells are the most mitotically active, but mitotic ependymoglial cells are occasionally found.

Fig. 1 Electron micrograph showing a mitotic ventricular cell from a 14-day postconception mouse embryo. The plane of cleavage appears to be midway between tangential and radial. The cytoplasm is pale and contains numerous rosettes of free ribosomes. Bar = 1 μm. × 18,000.

As development proceeds, cells leave the ependymal layer and it has been suggested that these cells can be identified as they are in the process of dividing at the ventricular surface. Martin and Langman (1965) and Martin (1967) suggested that unlike the majority of cells, the plane of cleavage of cells leaving the ventricular layer is tangential to the ventricular surface, rather than radial. These authors found that as development proceeded, the proportion of tangentially cleaving cells increased. This was not supported by a series of quantitative studies of mitotic figures in the spinal cord (Smart, 1972a), diencephalon (Smart, 1972b), neocortex (Smart, 1973) and ganglionic eminences (Smart, 1976). In these studies around 90% of mitotic figures at the ventricular surface were observed cleaving radially throughout the major embryonic stages of development of each region and there was no evidence of a change of orientation of cleavage with age. The major differences between the various regions seemed to be in the number of mitotic figures observed deep to the ventricular surface.

Smart (1965) proposed a state of ependymal choke that occurs when the rate of proliferation is such that at any one time the ventricular surface is packed with mitotic cells and new mitotic cells are obliged to begin mitosis before the nucleus reaches the ventricular surface. These mitotic cells can be divided into cells that begin mitosis before reaching the ventricular surface (subsurface prophases) but that complete mitosis at the ventricular surface, and those that complete the entire mitotic cycle without ever reaching the ventricular surface (nonsurface mitotic figures) but that remain within the ventricular layer. The number of subsurface mitotic figures and the orientation of their planes of cleavage vary from region to region.

In the mouse spinal cord subsurface mitotic figures are most common at E12 in the dorsal region and appear to be randomly orientated (Smart, 1972a). In the diencephalon (Smart, 1972b) subsurface prophases were first observed at E11. By E12 and E13, as well as subsurface prophases, mitotic figures at all stages of mitosis were found scattered throughout the layer, with most of them at the periphery of the layer undergoing cleavage in a plane tangential to the ventricular surface. By E14 the ventricular layer has decreased in thickness and the nonsurface mitotic figures are less regularly orientated. In the telencephalon (Smart, 1973) the majority of nonsurface mitotic figures are located at the cortical plate surface of the ventricular layer and the majority are orientated with a plane of cleavage tangential to the ventricular surface. In the region of the ganglionic eminences the number of mitotic cells over a given area of surface is less than that along the ventricular roof (Smart, 1976; Sturrock, 1979a; Sturrock and Smart, 1980). The study of nonsurface mitotic figures in the region of the ganglionic eminences is complicated by the very early appearance of a subventricular (subependymal) layer in this region (Smart, 1976; Smart and Sturrock, 1979; Sturrock and Smart, 1980). In the subventricular layer there is no interkinetic nuclear migration as the cells comprising it do not retain any attachment to the ventricular surface. Quantitative studies of the changes in the number of ventricular layer and subventricular layer mitotic figures (Smart, 1976; Sturrock, 1979a) indicate that the peak of ventricular layer mitosis occurs slightly earlier than that of the subventricular layer, but the number of mitotic figures in each declines rapidly between E15 and birth in the mouse. The mitotic activity in the postnatal ependymal and subependymal layers is discussed below (Section III,A and B).

B Fate of Cells Produced

The fate of the mitotic cells in the ventricular and subventricular layers remains a matter of controversy. The view most widely quoted in standard textbooks is that of et al., , 1979; Sturrock and Smart, 1980).

It seems likely on morphological grounds that the subventricular layer around the lateral ventricles produces glial precursors prenatally, probably from the appearance of the incipient ganglionic eminences at E11 (Sturrock and Smart, 1980). Whether in the region of the ganglionic eminences neuron production is restricted to the ventricular layer is not known. From the time of its appearance the subventricular layer of the ganglionic eminences is made up mainly of small cells with a moderately dark cytoplasm (Sturrock and Smart, 1980); these are almost certainly glioblasts. This tends to support the view that the subventricular layer of the ganglionic eminences is mainly, if not solely, a glial precursor producer. The ventricular layer produces neurons, ependymoglial cells, and subventricular cell precursors. The neurons of the ganglionic eminences migrate (probably to the neostriatum), while the subventricular cells proliferate in situ, forming a large pool of glial precursors that migrate, mainly perinatally, throughout the forebrain (Smart and Leblond, 1961; Lewis, 1968; Blakemore, 1969; Privat and Leblond, 1972; Paterson et al., 1973; Smart and Sturrock, 1979; Sturrock, 1979a, 1980a; Sturrock and Smart, 1980).

Altman (1963, 1966, 1969) has proposed that the subependymal layer is the site of postnatal production of microneurons, particularly in the olfactory bulb and hippocampus (also Altman and Das, 1965, 1967). These studies, however, were carried out on paraffin sections in which identification of neurons and glia is equivocal. Doubt has been cast on the interpretation of these results by Korr (1980).

Apart from the forebrain, the CNS lacks a specialized nonventricular glial-producing germinal layer and glial precursors are produced prenatally by the ventricular layer. In the spinal cord, for example, the mechanism of postnatal gliogenesis differs from that in the forebrain since the spinal cord never has a subventricular or subependymal germinal zone (Smart, 1972a). The large increase in the number of glia that occurs in the spinal cord during the first 2 weeks after birth in rats (Matthews and Duncan, 1971) must be due to division of glia or glial precursors already in situ since there is insufficient postnatal ependymal cell division to account for such an increase (Sturrock, 1981a).

The commonly held view that neuron production precedes glial production is based on the misconception that glia, like neurons, lose the ability to divide once differentiated. This misconception leads to misinterpretation of autoradiographic data as Sidman (1970) has pointed out. It is possible to map neuron production accurately by [³H]thymidine autoradiography solely because differentiated neurons do not divide (but see Section III, C). The fact that glia continue to turn over throughout life (Smart and Leblond, 1961; Dalton et al., 1968; Korr et al., 1975; Sturrock, 1974b, 1979a; Korr, 1980) must be borne in mind when trying to interpret the results of autoradiographic studies of glial production.

From our own studies (, 1979; Korr, 1980) we have concluded that glial production in the forebrain occurs in concert with neuron production, but after neuron production ceases glia continue to increase in number, probably as a result of an increasing requirement by differentiated neurons for support, both physical and metabolic, including the production of myelin sheaths. The increase in glial number is brought about by division of cells in the subventricular/subependymal layer and their subsequent migration, and also by division of glia that have already migrated to other regions. In the mouse the major period of subependymal mitosis occurs from E14 to E15 and thereafter there is a rapid decrease in the number of mitotic figures in the subventricular/subependymal layer (Smart, 1976; Sturrock, 1979a). At the same time the cell cycle time is increasing (Korr, 1980) so that the major period of glial precursor production is prenatal in the mouse. Nevertheless, the number of subependymal and peripheral mitotic figures in postnatal mice indicates a continuing production of glia that is greatest during the first 2 postnatal weeks, but which continues at a low level throughout life (Dalton et al., 1968; Sturrock, 1979a; Korr, 1980).

III Mitotic Cells of Ectodermal Origin

A Ependymal Cells

Once the ependymal layer has differentiated and ceased to be a major center of cell production, very little mitotic activity is present in it. Even in hydrocephalus, which produces stretching of the ependymal epithelium, there is no evidence of mitotic division in the ependymal layer (Blackwood and Corsellis, 1976; Weller et al., 1978).

Westergaard (1970) in a comprehensive survey of the ependymal layer of adult rodents was unable to identify mitotic ependymal cells. A few mitotic ependymal cells were observed in adult mice during a quantitative study of cell division in the forebrain (Sturrock, 1979a), but some degree of caution must be exercised in interpretation of these results since the counts were carried out in 6-μm sections and electron microscopy demonstrates that occasionally the cytoplasm of adjacent ependymal cells may be stretched out in very thin layers so that mitotic figures that appear to belong to ependymal cells in the light microscope may in truth be subependymal cells. Korr (1980) came to the conclusion that there is very little, if any, mitotic activity in the ependyma of adult rodents, but Chauhan and Lewis (1979) found autoradiographic evidence for ependymal proliferation in rats up to 42 days postnatum, albeit very small in amount. This proliferation is greater in the lateral ventricles than in the third ventricle.

The central canal of the mouse spinal cord is lined with an ependymal epithelium whose morphology differs from that in other regions of the ventricular system (Sturrock, 1981a). These cells are extremely rich in microfilaments and glycogen and the canal surface consists of club-shaped microvilli and cilia. While a comprehensive survey of postnatal mitotic activity in these cells has yet to be carried out, well-differentiated ependymal cells have been observed undergoing mitosis in early postnatal animals (Sturrock, 1981a).

At present it seems likely from the available evidence that in the adult very little, if any, mitotic activity is present in the ependyma.

B Subependymal Cells

The presence of a mitotically active subependymal layer adjacent to the lateral ventricles was first described by Allen (1912) and this was confirmed by Bryans (1959). Smart (1961) examined the subependymal layer of the mouse using [³H]thymidine autoradiography and concluded that subependymal cell production continues into adult life. As one might expect, mitotic activity in the subependymal layer is greater prenatally (Smart, 1976; Sturrock, 1979a) and declines rapidly before birth in the mouse (Sturrock, 1979a). Nevertheless, mitotic figures can be found in the subependymal layer even in senile animals (Sturrock, 1979a; Korr, 1980; Fig. 15). Electron micrographic studies (Blakemore, 1969; Privat and Leblond, 1972; Sturrock and Smart, 1980) indicate that in mature rodents the majority of mitotic cells in the subependymal plate have the characteristics of immature glial precursors or glioblasts (Fig. 3).

Fig. 3 Electron micrograph showing a mitotic glioblast from the posterior column of a 17-day postconception mouse spinal cord. It is identifiable as a glioblast by its moderately electron-dense cytoplasm, numerous free ribosomes, and absence of glycogen granules. This cell is identical with the majority of dark mitotic glioblasts found in the subventricular and later subependymal layers of the mouse forebrain. Bar = 1 μm. × 12,600.

It seems probable that many of the cells produced by mitotic division in the subependymal layer migrate from the layer to maintain the neuroglial population throughout the forebrain (Smart, 1961; Smart and Leblond, 1961; Lewis, 1968; Hopewell, 1971; Paterson et al., 1973; Sturrock, 1979a; Korr, 1980). Maintenance of the glial population is also partly carried out by peripheral mitosis and mitotic figures are present throughout life in both grey and white matter (Sturrock, 1979a).

A true subependymal layer exists only around the lateral ventricles, and the transient subventricular zone in the diencephalon (, 1977) probably should not be equated with the subependymal layer.

C Neurons

It is generally accepted that neurons do not divide once they have begun to differentiate. The time of final neuron differentiation varies widely in different regions of the CNS. This has been fully discussed in a comprehensive review by Rodier (1980). The difference in timing of final differentiation varies between species (Dobbing and Smart, 1974) and when neuronal differentiation is being considered, the relative maturity of the CNS in different species at birth must be taken into consideration. In some ways birth is irrelevant when considering CNS development. This is obvious when one considers the relative stage of maturity of mice, lambs, deer, and human babies at birth, but it is surprising how often these obvious differences are ignored.

The cerebellum is an area of the brain in which neuron production occurs much later than most other regions and in which a nonventricular proliferative zone (the external granular layer) is known to give rise to neurons. This has been reviewed by (1974) and will not be further considered here except to emphasize that unlike the subependymal layer around the lateral ventricle, the external granular layer disappears completely once differentiation is complete and only produces microneurons (Lewis et al., 1977).

Other areas where late neurogenesis occurs are the dentate gyrus of the hippocampus and the olfactory bulb. There has been evidence of neurogenesis in adult rodents in both of these regions (Kaplan and Hinds, 1977) and in the visual cortex (Kaplan, 1981). It was not clear from the works of Kaplan and Hinds (1977) and Kaplan (1981) whether new neurons were produced by division of neuron precursors or by division of differentiated neurons, but their results supported earlier quantitative studies that appeared to show an increase in the total number of granule cells in the olfactory bulb with increasing age (Hinds and McNelly, 1977). The possibility cannot be excluded that in certain regions neuron production continues in the adult, leading either to an increase in neuron number or to replacement of existing neurons.

D Astrocytes

As long ago as 1894 Goodall noted the presence of karyokinetic changes in spider cells around a puncture wound of the brain. Del Rio-Hortega and Penfield (1927) observed an increase in astrocytes around a needle wound in the brain but found no evidence of mitotic division. Cavanagh (1970) found mitotic cells in the vicinity of a puncture wound of the rat brain. He classed them as astrocytes on the basis of the morphology of the mitotic figures, but because he used 10-μm paraffin sections, the reliability of identification is open to question. Mitotic astrocytes have, however, been observed in electron microscopical studies of degenerating nervous tissue by Bignami and Ralston (1969) and Skoff and Vaughn (1971) and recently Latov et al. (1979) have demonstrated division of fibrillary astrocytes using immunocytochemical techniques.

In the normal CNS mitotic astrocytes have been observed in the developing nervous system of the cat (Blunt et al., 1972), the mouse (Sturrock, 1974c), and the human fetus (Sturrock, 1975). The mitotic astrocytes had differentiated to the stage at which they contained microfilaments and glycogen granules (Fig. 4). Mitotic astrocytes have also been observed in an ultrastructural study of young adult rats (Mori and Leblond, 1969b).

Fig. 4 Electron micrograph showing a mitotic astrocyte from a 15-week human embryo optic nerve. This cell is rich in glycogen granules and microfilaments (mf). Bar = 1 μm. × 20,000.

In the human optic nerve astrocytes that appear to be well-differentiated at the ultrastructural level have a moderately dark-staining cytoplasm in semithin sections (Sturrock, 1975; Fig. 4–8). Astrocytes with a dark-staining cytoplasm have also been found in the developing mouse corpus callosum (Sturrock, 1976b) and the developing rat optic nerve (Skoff et al., 1980). This difference in cytoplasmic staining may be related to changes in the chemical structure of the microfilaments—an hypothesis supported by an immunocytochemical study by Bignami and Dahl (1974), who found that astrocytes do not show immunofluorescence until a relatively late stage of development. In the human fetal optic nerve mitotic astrocytes often appeared to retain their processes during division (Figs. 5–8). Autoradiographic studies confirm that differentiated astrocytes retain the ability to divide under normal conditions during development (Mori and Leblond, 1969b; Paterson et al., 1973; Lewis et al., 1977; Skoff et al., 1976a,b), into adult life and even senility (Korr, 1980). Pale-staining, and therefore almost certainly well-differentiated, astrocytes have been seen undergoing mitotic division in the early postnatal spinal cord (Sturrock and McRae, 1980) and in the adult anterior commissure (Sturrock, 1976a) of mice. Examples of mature mitotic astrocytes are shown in Figs. 9 and 10.

Figs. 5–8 One-micron toluidine blue-stained sections of 15-week human embryo optic nerve. × 1800.

Fig. 5 Astrocyte in prophase. At this age astrocyte cytoplasm stains moderately darkly despite the well-differentiated appearance of these cells in electron micrographs. Note the continued presence of processes (arrows).

Fig. 6 This figure shows a mitotic endothelial cell (e) as well as an astrocyte in anaphase complete with processes (arrows).

Fig. 7 Astrocyte in early telophase still complete with processes (arrows).

Fig. 8 Late telophase (T1; T2). The chromatids of the daughter cells are much thicker in appearance than those of the prophase shown in Fig. 5.

E Oligodendrocytes

When one considers the complex connections of oligodendrocytes with myelin sheaths, it might seem unlikely that differentiated oligodendrocytes undergo division. The relationship of oligodendrocytes to myelin sheaths has recently been elegantly demonstrated by Sternberger et al. (1978a), who reconstructed oligodendrocyte connections from sections stained immunocytochemically. Skoff et al. (1976a) observed a few labeled oligodendrocytes having presumptive connections with myelin sheaths 1 hr after injection with [³H]thymidine, suggesting that myelinating oligodendrocytes in the developing rat optic nerve retain the capacity to divide. Their study did not, however, demonstrate that these oligodendrocytes retain their connections during division, and indeed the work of Ludwin (1979) suggested that in pathological conditions oligodendrocyte proliferation does not occur until demyelination is complete.

Mori and Leblond (1970) found [³H]thymidine uptake by cells identified as oligodendrocytes in the corpus callosum of young rats, but Paterson et al. (1973) believed that only a few immature oligodendrocytes divided. Following an autoradiographic study of the developing rat corpus callosum, Imamoto et al. (1978) concluded that oligodendrocytes do not divide under normal conditions.

In the early postnatal mouse spinal cord, however, dark-staining cells connected to myelin sheaths (Figs. 11 and 12) have been observed in all phases of the mitotic cycle (Sturrock and McRae, 1980) and these authors concluded that the mitotic cells were oligodendrocytes that continued to remain in contact with their myelin sheaths throughout division. Since then electron microscopic evidence confirming mitotic division of oligodendrocytes in the developing mouse spinal cord has been presented (Sturrock, 1981b; Fig. 16).

Fig. 16 Electron micrograph showing a mitotic oligodendrocyte with a process (arrow) connected to a myelin sheath. Parts of the nuclear membrane are still present, indicating that this cell is in prophase. Bar = 1 μm. × 12,600. (From Sturrock, 1981b.)

The absence of mitotic oligodendrocytes from the corpus callosum and their presence in the optic nerve and spinal cord may be a reflection of the differences in gliogenesis in different regions of the CNS and may be related to the presence or absence of a subependymal layer.

As yet no positive identification of a mitotic oligodendrocyte has been made in the adult CNS under normal conditions, although Korr (1980) claims to have observed [¹⁴C]thymidine-labeled oligodendrocytes in metallic stained preparations at 18 months of age in the mouse brain. Counts of mitotic figures in various regions of the mouse forebrain (Sturrock, 1979a) showed that the mitotic index in white matter is much lower than in gray matter. It was suggested that this was tentative evidence that most nonsubependymal mitotic cells are astrocytes, which form a much larger proportion of the glial population in grey matter, but because these counts were performed on 6-μm sections no positive identification was possible.

F Choroid Plexus Epithelial Cells

The mitotic pattern in choroid plexus epithelial cells has been largely neglected in the literature. In human embryos Zand (1930) found no mitoses in the choroid plexus epithelium, although some mitotic activity was present at the junction of the choroid plexus with the ependyma. Ariëns Kappers (1958) and Boyd (1958) were also unable to detect mitotic activity in choroid plexus epithelium. Knudsen (1964), however, found mitoses in both the choroid plexus epithelium and connective tissue of prenatal mice. Most of the epithelial mitoses were situated near the root in all four ventricles, but the mitoses in the connective tissue were found throughout the plexus. This distribution of choroid plexus epithelial mitoses was supported by a morphological study of choroid plexus development in the mouse (Sturrock, 1979c) that indicated that the choroid plexus undergoes its greatest development between 11 and 14 days postconception. An example of fetal choroid plexus mitosis is shown in Fig. 18.

Fig. 18 One-micron toluidine blue-stained section showing choroid plexus of a 13-day postconception mouse embryo with two mitotic plexus epithelial cells, one at the base and the other further along the plexus. × 1000. (From Sturrock, 1979c.)

Altman (1969) considered that the choroid plexus epithelium proliferates up to the end of the second week of life in rodents and this was confirmed by Korr (1978), but Chauhan and Lewis (1979) found a few [³H]thymidinelabeled cells in the rat choroid plexus up to 42 days postnatum.

In an attempt to learn whether choroid plexus epithelium displayed any mitotic activity in the adult, the complete choroid plexus in the lateral ventricle of mice aged 6 and 22 months was examined in 6-μm serial sections. In neither animal was a single convincing mitotic figure found anywhere in the choroid plexus (R. R. Sturrock, unpublished results). This tends to confirm the available evidence that in mature rodents the choroid plexus is not mitotically active. Kaplan (1980) has, however, shown that choroid plexus epithelial cells continue to divide in the adult primate.

IV Mitotic Cells of Mesodermal Origin

A Macrophages (Including Microglia)

A number of types of macrophage are associated with the CNS. Within the ventricular system of the brain macrophages can be found lying on the ventricular surface of the ependyma and also on the choroid plexus. Those lying on the ependyma have been referred to as supraependymal cells (Clementi and Marini, 1971; Noack et al., 1972; Allen and Low, 1973; Coates, 1973a,b; Bleier, 1975; Walsh et al., 1978; Sturrock, 1978b), whereas those on the surface of the choroid plexus have been named either epiplexus or Kolmer cells (Kolmer, 1921; Carpenter et al., 1970; Chamberlain, 1974; Peters, 1974; Sturrock, 1978; Ling, 1979b). There seems no doubt that morphologically and developmentally (Sturrock, 1978b, 1979b; Ling, 1979b) epiplexus cells and supraependymal cells are identical and there is no good reason for referring to them by different names. During fetal and early postnatal life intraventricular macrophages are often seen in different stages of mitosis (Fig. 17; Sturrock, 1978a,b; Ling, 1979b). While no actual counts of mitotic figures have been carried out, one has the impression that mitotic macrophages are more often present on the surface of the choroid plexus than on the ependymal surface. Macrophages are also normally present in the subarachnoid space (Weed, 1917, 1938; Essick, 1920; Woollard, 1924; Kubie and Schultz, 1925; Wislocki, 1932; Pease and Schultz, 1958; Waggener and Beggs, 1967; Morse and Low, 1972; Cloyd and Low, 1974; Allen and Low, 1975; Malloy and Low, 1976). These macrophages are morphologically identical to those found within the ventricular system (Malloy and Low, 1976). In prenatal animals, at least, mitotic macrophages are present in the subarachnoid space (Sturrock, 1981d). They can increase in number rapidly when stimulated (Merchant and Low, 1977), but whether this increase in number is due to division of macrophages in situ or to an invasion of new macrophages from the subarachnoid vessels is not yet known.

Fig. 17 Six-micron section stained with hematoxylin and eosin showing a mitotic intraventricular macrophage in close proximity to the choroid plexus. ×