Among the interesting problems raised by demonstrations that adult mammals maintain populations of self-renewing, multipotent neural stem cells are whether embryonic neural precursors and adult neural stem cells have similar or different cell fate potential and whether molecular mechanisms that maintain and specify neural precursors early in development also operate in adult stem cells. During early development, the first neurons are produced from rapidly dividing neuroepithelial precursors. As neurogenesis proceeds, neuroepithelial cells transform to radial glia, which, like neuroepithelial cells, are connected to both the ventricular lumen and basement membrane at the pial surface, but have ultrastructural and gene expression characteristics similar to astrocytes (Mori et al., 2005). Radial glia give rise to large numbers of neurons and, near the end of embryogenesis, many lose their radial morphologies and differentiate as astrocytes (Rakic, 2003). Recent fate mapping work showed that radial glia also produce oligodendrocytes, the myelinating glial cells of the central nervous system (Fogarty et al., 2005; Ventura and Goldman, 2007).
The neural stem cells of the two documented sites of adult mammalian neurogenesis, the subventricular zone (SVZ) of the lateral ventricle and subgranular zone (SGZ) of the hippocampal dentate gyrus, have astrocytic characteristics (Doetsch et al., 1999; Garcia et al., 2004). Interestingly, the SVZ astrocytic neural stem cells maintain contacts with the ventricle (Doetsch et al., 1999), indicating that they retain some features of radial glia. Thus, neuroepithelial cells, radial glia, and subpopulations of astrocytes represent a continuum of neural precursors in the embryonic and adult CNS (Merkle and Alvarez-Buylla, 2006). Although other regions of the adult mammalian CNS are considered to be non-neurogenic, cultured spinal cord cells can give rise to neurospheres (Weiss et al., 1996), raising the possibility that cells with stem cell potential exist throughout the mammalian CNS.
Whereas radial glia, with the exception of cerebellar Bergmann glia and retinal Müller glia, are absent from adult mammals, large numbers of radial glia persist in adult reptiles, amphibians, and fish (Naujoks-Manteuffel and Roth, 1989; Garcia-Verdugo et al., 2002; Zupanc and Clint, 2003), which also have high rates of adult neurogenesis. In zebrafish, proliferative cells in the adult brain usually are close to the ventricle (Ekstrom et al., 2001; Zupanc et al., 2005). Recent fate mapping studies indicate that proliferative cells of the adult zebrafish brain have stem cell characteristics, in that they divide slowly and persistently and give rise to neurons and glia (Chapouton et al., 2006; Grandel et al., 2006). Interestingly, stem cells that occupy the boundary between midbrain and hindbrain express markers of radial glia and the Her5 transcription factor (Chapouton et al., 2006), raising the possibility that molecularly distinct subsets of radial glia produce different kinds of neurons and glia.
Here we describe a discrete population of spinal cord radial glia in zebrafish that arise during late embryogenesis and persist into adulthood. These cells express olig2, which encodes a bHLH transcription factor and, in post-embryonic stages, divide slowly and, apparently, asymmetrically to produce oligodendrocyte lineage cells but not neurons or astrocytes. Thus, olig2+ spinal cord radial glia may serve as a dedicated population of oligodendrocyte stem cells for the spinal cord of postembryonic zebrafish.
Oligodendrocyte Lineage Cells Express an olig2 Reporter Gene in Postembryonic Zebrafish
We previously reported that regulatory DNA from the olig2 locus drives expression of EGFP in pMN spinal cord precursors and their descendent motor neurons, interneurons and oligodendrocyte progenitor cells (OPCs), which migrate and divide, producing only cells of the oligodendrocyte lineage, in transgenic zebrafish embryos (Shin et al., 2003; Park et al., 2004, 2005). Near the end of embryogenesis, some EGFP+ cells assumed a radial morphology, contacting both the central canal and spinal cord pial surface, and were labeled by zrf-1 antibody, a marker of zebrafish radial glia (Park et al., 2004). Because radial glia function as neural precursors (Morest and Silver, 2003; Gotz and Huttner, 2005), we extended our investigation of olig2 reporter gene expression into postembryonic stages. Similar to 3 dpf (Fig. 1A), transverse sections of 20 dpf, 1 month post-fertilization (mpf) and 3 mpf Tg(olig2:egfp) fish revealed EGFP+ radial cells (Fig. 1B–D). At each stage, typically one or two EGFP+ cell bodies and radial fibers were evident on each side of the central canal. At 20 dpf, the weak EGFP fluorescence that at late embryonic stages marks neurons (Park et al., 2004) was no longer evident (Fig. 1B). However, several cells in both ventral and dorsal spinal cord, at positions characteristic of OPCs and mature oligodendrocytes, expressed EGFP at a high level (Fig. 1B). At 1 and 3 mpf, the number of EGFP+ cells had increased, raising the possibility that the olig2+ oligodendrocyte lineage expands as zebrafish grow to adulthood.
To investigate the identity of EGFP+ cells at postembryonic stages, we labeled transverse sections of 1 mpf Tg(olig2:egfp) transgenic fish with antibodies specific to neuronal and glial populations. During embryogenesis, spinal cord Sox10 expression specifically marks OPCs and oligodendrocytes in rodents and zebrafish (Kuhlbrodt et al., 1998; Park et al., 2002). Nearly every EGFP+ cell was Sox10+, with the exception of the radial cells and occasional cells located near the central canal and the dorsal medial septum (Fig. 2A,B), indicating that the majority of cells that express the olig2 transgene postembryonically belong to the oligodendrocyte lineage. Consistent with this, EGFP+ cells were dispersed throughout large tracts of anti-Myelin Basic Protein-positive (MBP+) and anti-acetylated Tubulin-positive axonal fibers (Fig. 2C–E).
To investigate whether olig2+ cells develop as neurons, we labeled sections of transgenic animals with anti-Hu antibody (Marusich et al., 1994). Despite examining many sections, we never observed convincing co-localization of EGFP and Hu (Fig. 2F). Thus, with the caveat that during embryogenesis both olig2 expression and Hu appear to label neurons only transiently (Park et al., 2004), we conclude that there is no appreciable neuronal expression of the olig2 transgene at postembryonic stages. We also labeled sections with anti-GFAP antibody, which is a canonical astrocyte marker but also labels cells that have neural precursor properties (Kimelberg, 2004; Gotz and Huttner, 2005). Peripheral regions of the spinal cord had numerous short, fine GFAP+ processes that we assume to be astrocytic (Fig. 2G–I). Additionally, several thick, long GFAP+ fibers extended radially from the medial septum to the pial surface. EGFP+ cell bodies were closely associated with both classes of GFAP+ processes in peripheral spinal cord but the fluorescent signals did not appear to colocalize (Fig. 2H, I), suggesting that they mark distinct cell populations. Similarly, neither the soma nor the processes of EGFP+ radial cells appeared to express high levels of GFAP.
Expression of BLBP marks radial glia that have precursor characteristics (Hartfuss et al., 2001; Gotz and Barde, 2005). Numerous cells lining the spinal cord central canal and medial septum were BLBP+ (Fig. 2J). Additionally, BLBP labeling was evident in several radial fibers. Both the soma and processes of EGFP+ radial cells were BLBP+ (Fig. 2K), suggesting that olig2 transgene expression marks a discrete population of spinal cord precursors.
Dividing olig2+ Radial Glia Have Asymmetric Features
The above data show that, in zebrafish larvae and adults, olig2 transgene expression marks a subpopulation of putative radial glial precursors and oligodendrocyte lineage cells, which increase in number over time. This raised the possibility that dividing olig2+ radial glia contribute cells to the oligodendrocyte lineage. To begin to investigate this, we first attempted to gain an impression of the cycling population of spinal cord cells over time by exposing animals to the thymidine analog BrdU, which serves as a permanent marker of cells in S-phase at the time of exposure. Similar to our previous observations (Park and Appel, 2003), a brief pulse of BrdU at 1 dpf, during embryogenesis, labeled numerous spinal cord cells (Fig. 3A), reflecting the presence of many rapidly cycling precursors. However, when labeling was performed at 3 dpf and later, substantially fewer cells were marked. The number of cells that incorporated BrdU remained low until about 15 dpf, when it began to slowly increase (Fig. 3A). Anti-phospho-Histone H3 antibody, a marker of M-phase cells, similarly labeled numerous spinal cord cells at 1 dpf and relatively few at 3 dpf and later (data not shown). These data indicate that dividing cells are present in the postembryonic zebrafish spinal cord. However, fewer cells divide in larval and juvenile fish relative to embryos, perhaps reflecting a lengthening of cell cycle time.
We next asked if EGFP+ cells of Tg(olig2:egfp) animals divide. As noted above, a single exposure to BrdU at postembryonic stages labeled only a few cells but, occasionally, these included olig2+ radial glia (Fig. 3B). To attempt to label the entire proliferative spinal cord population, we exposed fish between 15 and 30 dpf to BrdU repeatedly. This procedure labeled many cells near the central canal and medial septum, including olig2+ radial glia (Fig. 3C, D). Additionally, some cells in both the gray and white matter of the spinal cord were BrdU+, including olig2+ oligodendrocyte lineage cells (Fig. 3D). Because fish were exposed to BrdU repeatedly, these cells could have incorporated BrdU during a precursor division near the central canal or medial septum and subsequently migrated radially, or during a progenitor division during or after radial migration. To help distinguish between these possibilities, we labeled Tg(olig2:egfp) fish with anti-PCNA antibody, which marks proliferative cells. Some olig2+ cells in gray and white matter were PCNA+ (Fig. 3E). These data suggest that olig2 transgene expression marks both oligodendrocyte precursor and progenitor cells as zebrafish grow into 1-month-old juvenile fish.
olig2+ radial glia divide as zebrafish grow into adulthood, yet their number in the transverse plane of the spinal cord remains the same whereas the number of OPCs and oligodendrocytes increases. This raised the possibility that olig2+ radial glia divide asymmetrically to produce one cell that remains as a radial glial precursor and one that enters the oligodendrocyte lineage. Many precursor cells that divide asymmetrically have polarity revealed by apical localization of PAR complex proteins (Wodarz and Huttner, 2003). Thus, we labeled Tg(olig2:egfp) fish with antibody that recognizes zebrafish aPKC, a PAR complex component (Horne-Badovinac et al., 2001). At 15 dpf, aPKC was localized to a rim surrounding the spinal cord central canal, including the membrane of olig2+ radial glia (Fig. 4A). We also found evidence of apparent asymmetric cell division patterns. In BrdU-labeled animals, we sometimes found two olig2+ BrdU+ cells next to each other near the central canal, suggestive of recently divided sibling cells (Fig. 4B). We then labeled transgenic animals with anti-phospho-Histone H3 antibody, which reveals chromosomes of mitotic cells. Occasionally, we found olig2+ radial glia in which the mitotic spindle was oriented perpendicularly to the central canal (Fig. 4C), consistent with an asymmetric division. Taken together, these data indicate that postembryonic olig2+ radial glia divide slowly and asymmetrically at the same time that olig2+ oligodendrocyte lineage cells increase in number. Thus, olig2+ radial glia may serve as dedicated precursors for oligodendrocyte lineage cells in larvae and adults.
In the spinal cords of vertebrate embryos, Olig2+ pMN precursors produce motor neurons, oligodendrocytes, interneurons, ependymal cells, and astrocytes (Lu et al., 2000; Takebayashi et al., 2000; Zhou et al., 2000; Park et al., 2004; Masahira et al., 2006). In the absence of Olig2 function, neither motor neurons or spinal cord oligodendrocytes form (Lu et al., 2002; Park et al., 2002; Takebayashi et al., 2002; Zhou and Anderson, 2002). Thus, during embryogenesis, Olig2+ precursors are multipotent.
In adult rodents, slowly dividing neural stem cells of the SVZ, called type B cells, produce rapidly dividing transit amplifying type C cells (Doetsch et al., 1999). Most type C cells produce interneurons that migrate anteriorly to the olfactory bulb through the rostral migratory stream (RMS) but more recent studies showed that they also give rise to OPCs that migrate radially to populate nearby regions of the brain where they differentiate as oligodendrocytes (Hack et al., 2005; Menn et al., 2006). Interestingly, a subset of type C cells, as well as a small fraction of cells with type B characteristics, express Olig2 (Hack et al., 2005; Menn et al., 2006). Forced expression of Olig2 by retroviral delivery to the SVZ blocked interneuron development and promoted formation of oligodendrocytes in adult mice (Hack et al., 2005) and both oligodendrocytes and astrocytes in neonatal rats (Marshall et al., 2005). Thus, in rodents, transition from embryonic stage to postnatal stage correlates with an apparent restriction in the fate of proliferative Olig2+ cells. Whereas during embryogenesis Olig2+ precursors produce both neurons and glia, adult SVZ cells that express Olig2+ are restricted to glial fate.
Our data indicate that, in the zebrafish spinal cord, olig2 expression marks cells that transform from rapidly dividing, multipotent precursors in embryos to slowly and asymmetrically dividing precursors in larvae and adults that exclusively produce oligodendrocytes. Our previous fate mapping showed directly that individual olig2+ neuroepithelial precursors can give rise to both neurons and oligodendrocytes in embryos (Park et al., 2004). Although we are unable to perform similar fate mapping experiments at later stages, we took advantage of the stability of EGFP fluorescence in transgenic animals to investigate the characteristics of olig2+ cells in postembryonic animals. We learned that a discrete subset of olig2+ cells have features suggestive of stem cell properties. First, from late embryogenesis into adulthood, olig2+ cells in contact with the spinal cord central canal had radial glial morphologies and expressed BLBP, a marker of radial glia that function as neural precursors (Gotz and Barde, 2005). Second, the olig2+ radial glia population persisted into adulthood, dividing slowly but continuously. The long-term maintenance of the population requires that at least some divisions are self renewing. Third, the PAR protein aPKC was localized to the apical membrane of olig2+ radial glia. In flies, asymmetric divisions of neuroblasts requires apical localization of PAR complex proteins (Betschinger and Knoblich, 2004). Finally, in some dividing olig2+ radial glia, the mitotic spindle was oriented perpendicularly to the central canal. Potentially, these represent asymmetric divisions wherein only the cell remaining in contact with the central canal persists as a radial glia.
As the Tg(olig2:egfp) fish grew through larval stage into adulthood, we noted a continual increase in the number of EGFP+ spinal cord cells. Most of these cells were dispersed throughout the axon-rich white matter and nearly all expressed the oligodendrocyte lineage cell marker Sox10. We found no evidence that olig2+ radial glia produced neurons or astrocytes. Thus, olig2+ spinal cord precursors have more restrictive fate at postembryonic stages than during embryogenesis, similar to the Olig2+ cells of the adult rodent SVZ.
Wild-type AB and Tg(olig2:egfp)vu12 (Shin et al., 2003) fish were used for this study.
BrdU Labeling and Immunohistochemistry
Twenty- to 30-dpf fish were incubated in a 0.5% solution of BrdU (Roche) in embryo medium (EM) (15 mM NaCl, 0.5 mM KCl, 1 mM CaCl2, 1 mM MgSO4, 0.15 mM KH2PO4, 0.05 mM NH2PO4, 0.7 mM NaHCO3) for 2 or 4 days at 28.5°C and then anesthetized until movement had ceased. The brains, internal organs, and some of the trunk muscles were dissected in EM and fixed in 4% paraformaldehyde overnight. For the quantitative data shown in Figure 3C, all BrdU incubations were 24 hr, except for the 24-hpf treatment, which was a 12-hr incubation. For immunohistochemistry, we used the following primary antibodies: mouse anti-BrdU (G3G4, 1:1,000, Developmental Studies Hybridoma Bank [DSHB], Iowa City, IA), mouse anti-HuC/D (16A11, 1:20, Molecular Probes) (Marusich et al., 1994), rabbit anti-phospho-Histone-H3 (1:1,000, Upstate Biotechnology, Charlottesville, VA), rabbit anti-BLBP (1:1,500) (Feng et al., 1994; Chapouton et al., 2006), rabbit anti-Sox10 (1:1,000) (Park et al., 2005), rabbit anti-MBP (1:100) (Lyons et al., 2005), mouse anti-GFAP (G-A-5, 1:100, Sigma) (Zupanc et al., 2005), mouse anti-acetylated Tubulin (1:100, Sigma), rabbit anti-PCNA (1:200, Santa Cruz Biotechnology, Inc.), rabbit anti-aPKC (1:200, Santa Cruz Biotechnology, Inc.) (Horne-Badovinac et al., 2001). For fluorescent detection of antibody labeling, we used Alexa Fluor 488, Alexa Fluor 568, Alexa Fluor 647 goat anti-mouse, or goat anti-rabbit conjugates (1:500, Molecular Probes). Fluorescence images were collected using a Zeiss LSM 510 laser scanning confocal microscope.
We thank Dr. N. Heintz and Dr. W. Talbot for their generous gifts of anti-BLBP antibody and anti-MBP antibody, respectively. The anti-BrdU antibody, developed by S. J. Kaufman, was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. Confocal microscopy was performed through the use of the VUMC Cell Imaging Shared Resource (supported by NIH grants CA68485, DK20593, DK58404, HD15052, DK59637 and EY08126).