Neuron, Vol. 31, 881–884, September 27, 2001, Copyright 2001 by Cell Press
Radial Glial Cells: Are They Really Glia? John G. Parnavelas1 and Bagirathy Nadarajah Department of Anatomy and Developmental Biology University College London London WC1E 6BT United Kingdom
During the development of the cerebral cortex, radial glia serve as a scaffold to support and direct neurons during their migration. This view is now changing in the light of emerging evidence showing that these cells have a much more dynamic and diverse role. A recent series of studies has provided strong support for their role as precursor cells in the ventricular zone that generate cortical neurons and glia, in addition to providing migration guidance. Our traditional view of radial glial cells is that of specialized nonneuronal cells of the astroglial lineage found in the developing CNS. Radial glia have an ovoid cell body located near the ventricular surface and a characteristic bipolar form. They show a short endfoot directed toward the ventricular wall and an elongated radial process that terminates by conical endfeet at the pial surface. Similar to astrocytes, radial glial cells show electronlucent cytoplasm, abundance of intermediate filaments, and glycogen granules condensed at the endfeet. Immunocytochemical studies have shown that they express vimentin, nestin (Rat-401), RC2, and GFAP, the latter only in primate cortex (Rakic, 1995). These cells were first observed in the chick spinal cord shortly after the advent of the Golgi impregnation method and slightly later in the human fetal spinal cord and brain (for references, see Rakic, 1995). Although they were described under different terms (e.g., radial cells, fetal glia, epithelial cells, spongioblasts), they were undoubtedly the same morphologically distinct type of cell. The term “radial glia” was first introduced by Rakic in his classic descriptions of neuronal migration in the fetal primate neocortex (Rakic, 1972) and has since been widely adopted in developmental neurobiology. Although the morphological, biochemical, and functional properties of radial glia have been described in various vertebrate species, most studies have focused on their role in the developing mammalian telencephalon. It is in this system that the detailed electron microscopical analyses of Rakic in the 1970s (Rakic, 1972) provided support for a role of radial glia in neuronal migration. These studies strongly suggested that radial glial fibers provide a transient scaffold and impose radial constrains in the expanding cerebral wall that facilitate neuronal migration. Similar arrangements between migrating neurons and specialized glia have been described in a number of other systems (e.g., cerebellum, retina, hippocampus) (Rakic, 1995; Hatten, 1999). It should be mentioned that the geometric arrangement 1
Correspondence:
[email protected]
Minireview
of radial glial cells had prompted Ramo´n y Cajal to imply long before that these cells serve as scaffolding in the rapidly changing embryonic nervous system and direct neurons during their migration. More recent studies have implicated a number of molecules, including astrotactin, neuregulin, brain-lipid binding protein (BLBP), and tenascin, in the directed migration of young neurons along radial glial processes (for review, see Hatten, 1999). Although radial glia persist throughout life in most vertebrate species, evidence indicates that in the mammalian telencephalon they disappear by the end of the process of migration. Experiments in a variety of mammalian species that included examination of Golgi- and GFAP-stained sections, the uptake of fluorescent dyes by radial glial cells, and in vivo approaches have suggested that these cells transform into fibrous and/or protoplasmic astrocytes (for references, see Rakic, 1995). The cellular and molecular events that contribute to the transformation of radial glia into astrocytes are not known, but the morphological transformation appears to coincide with the loss of vimentin, RC2, and Rat-401 antigens and the acquisition of GFAP immunoreactivity. The development of the neuronal and glial lineages in the mammalian cerebral cortex has been the subject of investigation since the 1880s. In 1887, His described different precursor cells for neurons and glia in the germinal ventricular zone: germinal cells that are visible as mitotic figures lining the lumen, which give rise to neurons, and spongioblasts that appear to form a syncytium, from which neuroglial cells originate. Although opposing views have subsequently been voiced (see Levitt et al., 1981), more recent investigations have documented the heterogeneity of the cortical neuroepithelium and led to the prevailing view that neurogenesis and gliogenesis are distinct temporal events during corticogenesis, showing only some overlap (Figure 1). Further, lineage analyses have documented that, apart from relatively few exceptions, neurons and glia arise from separate lineages (Parnavelas, 1999, 2000). Thus, until recently, it has been widely believed that the germinal ventricular zone is primarily composed of neuroepithelial cells that give rise to distinct populations of neurons and glia, including radial glia, with the radial glial cells providing the substrate for migrating neurons. However, this concept of corticogenesis is gradually being eroded, particularly in the light of emerging evidence that ascribes a new and radically different role for radial glial cells in the developing cortex. Radial Glia as Neuronal Precursors Earlier reports had implied that radial glia of the mammalian cortex may be potential precursor cells since they were found to be immunopositive for nestin, an intermediate filament protein expressed in CNS precursor cells. In addition, radial glia have been shown to divide rapidly and undergo interkinetic nuclear migration, a feature unique to precursor cells in the germinal ventricular zone (Misson et al., 1988). However, it was Alvarez-Buylla and colleagues (1990) who first suggested that mitotically active radial glia may give rise to neurons. Using tritiated thymidine autoradiography in adult avian brain, these
Neuron 882
Figure 1. Three Models Illustrating Proposals on How Cell Diversity Is Generated in the Mammalian Cerebral Cortex from the Time of His (1887) to Present
authors showed that new neurons arise in the ventricular zone that was observed to be rich in mitotically active radial glia and hinted that radial glial cells may be the precursors of neurons. Consistent with these observations, subsequent lineage analyses carried out in chick optic tectum, mammalian neocortex, and striatum also suggested that radial glia might serve as possible neuronal precursors (Gray and Sanes, 1992; Halliday and Cepko, 1992). In these studies, clones of cells that were labeled with retrovirus were shown to contain neurons and many glia but only one radial glial cell, suggesting that the radial glia may have divided rapidly and asymmetrically to renew themselves and generate other neural cells. However, a recent series of elegant studies (Malatesta et al., 2000; Hartfuss et al., 2001; Miyata et al., 2001; Noctor et al., 2001) has provided the strongest challenge yet for the prevailing view of the role of radial glia in the developing cerebral cortex. In one of these studies, Malatesta et al. (2000) have provided compelling evidence that radial glia isolated from mouse cortex have the potential to produce neurons and glial clones in vitro. In this study, the authors have made use of a transgenic mouse line that expresses GFP under the human GFAP promoter to label radial glia whose identity was further confirmed by staining with RC2- or astrocyte-specific gluatamate transporter (GLAST) antibodies. Immunohistochemical characterization of the GFP-positive cells (E14-18) that were isolated by fluorescence-activated cell sorting showed that ⵑ80% of the GFP-positive sorted cells were radial glial precursors identified by staining for nestin, RC2, and GLAST. To investigate the composition of the progeny that arise from the radial glia precursors identified by GFP expression, sorted cells were cultured for 1 week and later characterized by immunohistochemistry. Intriguingly, the majority of precursors obtained from E14-16 transgenic mice gave rise to neurons, while those isolated from E18 animals differentiated into astrocytes. These observations were further confirmed by another method in which radial glia were labeled with fluorescent dyes applied to the pial surface. Thus, this study has demonstrated that precursor cells with long radial processes that express glial-specific markers can no longer be assumed to belong to the glial lineage.
Further, radial glia have been shown to be heterogeneous in their molecular composition (Hartfuss et al., 2001). Specifically, three distinct subsets of radial glia that differ in morphology, molecular composition, and cell cycle characteristics have been identified. Interestingly, it appears that all radial glial cells are immunoreactive for RC2 but show heterogeneity in GLAST or BLBP expression. This finding suggested that only subsets of radial glia express glial characteristics during neurogenesis and that these subsets may be correlated to the generation of distinct progenies. For example, colocalization analysis has shown that RC2/GLAST⫹ precursor subtypes are restricted to the period of neurogenesis and disappear at the onset of gliogenesis. Taken together, the evidence from both studies (Malatesta et al., 2000; Hartfuss et al., 2001) has demonstrated that radial glial cells comprise a heterogeneous population of precursor cells that sequentially give rise to neurons and glia during corticogenesis. To investigate whether radial glia are neuronal precursors in vivo, Noctor et al. (2001) have labeled radial clones consisting of radial glia by injecting retroviral GFP into the lateral ventricles of embryonic rats (E1516). Early time points (24 hr after infection) revealed individual radial glial cells, while later time points showed clones that also included neurons arrayed along the radial glial fiber. In utero BrdU labeling demonstrated that the only dividing member of each clone was the radial glial cell, indicating that the radial clones were generated by mitotically active radial glia. Daughter neurons often had pyramidal morphology, and electrophysiological recordings showed that distinct neuronal membrane properties developed as neurons migrated into the cortex. In a series of time-lapse videomicroscopy experiments, the authors have shown that radial glia undergo interkinetic nuclear movements and divide asymmetrically to give rise to mitotic radial glia and postmitotic neurons. Daughter neurons then migrate along their own parent radial glial fiber, supporting a dual role for radial glia in both neurogenesis and migration guidance. The authors have suggested that familial relationships between radial glia and their daughter neurons might be involved in the establishment of cortical columns.
Minireview 883
A more recent study by Miyata et al. (2001) has further corroborated the earlier observations that radial glia are indeed neuronal precursors during early corticogenesis. By placing DiI on the pial surface of cortical slices, these investigators have identified two distinct populations of radially oriented cells that differed in morphology and somal position. They were either bipolar radial glia with their somata located in the ventricular or subventricular zones or pial-connected “radial neurons” with cell bodies in the subventricular or intermediate zones. During asymmetrical division of the radial glia, the newly generated radial neurons inherit the radial process, while their sibling radial glia grow new ascending processes. Moreover, it appears that radial neurons, which are initially bipolar and thus indistinguishable from their sibling radial glia, have to lose their ventricular attachments to exit from the proliferative zone. They also described a third population of radially oriented neurons located in the intermediate zone and having short processes, which they termed “short neurons.” Miyata et al. (2001) have further suggested that radial neurons migrate by long-range somal translocation as recently described by Nadarajah et al. (2001), while short neurons correspond to the locomoting population that relies on glial guidance. Taken together, the studies of Noctor et al. (2001) and Miyata et al. (2001), in agreement with the results of Tamamaki et al. (2001), provide compelling evidence that, during early corticogenesis, radial glia divide asymmetrically to generate neurons and mitotic radial glia. Together, the recent series of studies indicate that radial glia, while able to self renew, also generate neurons during early corticogenesis and astrocytes at later stages (Figure 1). However, these studies and earlier reports have typically identified the radial glial population by their bipolar morphology as well as by their antigenic properties that are not unique to these cells, as they appear to be common to neuroepithelial cells. When do neuroepithelial cells differentiate into radial glia? Are radial glia more restricted than the neuroepithelial stem cells in terms of their ability to self renew? To address such questions, it is imperative that future studies identify specific markers that distinguish the radial glial population from the rest of the neuroepithelium. Although the new findings do challenge the traditionally held view of radial glia, it is yet to be elucidated whether subsets of these cells or a common pool of radial glia sequentially generate neurons and astrocytes during corticogenesis as demonstrated in the in vitro experiments of Malatesta et al. (2000). In light of the finding that the vast majority of cortical interneurons are generated in the ventral telencephalon (Parnavelas, 2000), it is most likely that the neurons that arise from radial glia are pyramidal cells. However, it is yet to be shown whether all or subpopulations of pyramidal neurons are generated from radial glia. Earlier studies have shown that cortical precursor cells may divide symmetrically or asymmetrically to generate postmitotic neurons. Symmetric cell divisions occur during the early stages of neurogenesis and are essential for the expansion of the precursor population, while asymmetric divisions generate postmitotic neurons and maintain the progenitor pool. Although the precise mechanisms by which precursors switch from
symmetric to asymmetric division remain unknown, it has been suggested that during asymmetric cell division the localization of m-numb in the apical daughter cell results in the suppression of Notch1 activity and allows it to remain a progenitor (Zhong et al., 1997). However, in a more recent study, activated Notch1 signaling has been shown to promote the radial glial identity in the murine forebrain in vivo (Gaiano et al., 2000), lending further support to the notion that these cells may be neuronal progenitors. It is quite likely that the asymmetric distribution of activated Notch in dividing radial glial cells may favor an asymmetric cell output (neuron and progenitor) and continue to maintain their progenitor status (Nye et al., 1994). Together, the recent studies suggest that radial glia do not just provide a scaffold for migrating neurons, as envisaged in the past 30 years, but may be the principal component of the neuroepithelium that generates the neurons of the cerebral cortex. Radial glia have been studied predominantly in the developing brain, but it is likely that they continue to generate neurons in adult life, as demonstrated in the adult avian brain (Alvarez-Buylla et al., 1990). Interestingly, subventricular zone astrocytes in the postnatal brain have been identified as neural stem cells (Doetsch et al., 1999). Although, the origin of these astrocytes is not known, it is quite likely that they have arisen from radial glia. It may be speculated that radial glia also provide a continuity between embryonic and adult neural stem cells (Alvarez-Buylla et al., 2001). In the light of all the evidence in the last decade that supports new roles for radial glia in the developing and adult brain, should these cells be called glia? Selected Reading Alvarez-Buylla, A., Theelen, M., and Nottebohm, F. (1990). Neuron 5, 101–109. Alvarez-Buylla, A., Garcia-Verdugo, J.M., and Tramontin, A.D. (2001). Nat. Rev. Neurosci. 2, 287–292. Doetsch, F., Caille, L., Lim, D.A., Garcia-Verdugo, J.M., and AlvarezBuylla, A. (1999). Cell 97, 703–716. Gaiano, N., Nye, J.S., and Fishell, G. (2000). Neuron 26, 395–404. Gray, G.E., and Sanes, J.R. (1992). Development 114, 271–283. Halliday, A.L., and Cepko, C.L. (1992). Neuron 9, 15–26. Hartfuss, E., Galli, R., Heins, N., and Go¨tz, M. (2001). Develop. Biol. 229, 15–30. Hatten, M.E. (1999). Ann. Rev. Neurosci. 22, 511–539. His, W. (1887). Abh. kgl. sachs. Ges. Wissensch. math. phys. Kl. 13, 479–513. Levitt, P., Cooper, M.L., and Rakic, P. (1981). J. Neurosci. 1, 27–39. Malatesta, P., Hartfuss, E., and Go¨tz, M. (2000). Development 127, 5253–5263. Misson, J.P., Edwards, M.A., Yamamoto, M., and Caviness, V.S., Jr. (1988). Dev. Brain. Res. 38, 183–190. Miyata, T., Kawaguchi, A., Okano, H., and Ogawa, M. (2001). Neuron 31, 727–741 . Nadarajah, B., Brunstrom, J.E., Grutzendler, J., Wong, R.O.L., and Pearlman, A.L. (2001). Nat. Neurosci. 4, 143–150. Noctor, S.C., Flint, A.C., Weissman, T.A., Dammerman, R.S., and Kriegstein, A.R. (2001). Nature 409, 714–720. Nye, J.S., Kopan, R., and Axel, R. (1994). Develop. 120, 2421–2430.
Neuron 884
Parnavelas, J.G. (1999). Exp. Neurol. 156, 418–429. Parnavelas, J.G. (2000). Trends Neurosci. 23, 126–131. Rakic, P. (1972). J. Comp. Neurol. 145, 61–83. Rakic, P. (1995). In Neuroglial Cells, H. Ketterman and B.R. Ransom, eds. (New York: Oxford University Press), pp. 746–762. Tamamaki, N., Nakamura, K., Okamoto, K., and Kaneko, T. (2001). Neurosci. Res. 41, 51–60. Zhong, W., Jiang, M.-M., Weinmaster, G., Jan, L.Y., and Jan, Y.N. (1997). Development 124, 1887–1897.