Oligodendrocyte precursor cells (OPCs) appear in the late embryonic brain, mature to become oligodendrocytes (OLs), and form myelin in the postnatal brain. Recently, it has been proposed that early-born OPCs derived from the ventral forebrain are eradicated postnatally and that late-born OLs predominate in the cortex of the adult mouse brain. However, intrinsic and extrinsic factors that specify the ability of self-renewing multipotent neural stem cells in the embryonic brain to generate cortical OL-lineage cells remain largely unknown. Using an inducible Cre/loxP system to permanently label Nestin- and Olig2-lineage cells, we identified that cortical OL-lineage cells start differentiating from neural stem cells within a restricted temporal window just prior to E16.5 through P10. We then showed, by means of electroporation of a Cre expression plasmid into the VZ/SVZ of E15.5 reporter mouse brains, that neural precursor cells in the dorsal VZ/SVZ are inhibited by Wnt signaling from contributing to cortical OLs in the adult brain. In contrast, neural precursor cells present in the dorsoventral boundary VZ/SVZ produce a significant amount of OLs in the adult cortex. Our results suggest that neural stem cells at this boundary are uniquely specialized to produce myelin-forming OLs in the cortex.
Multipotent and self-renewing neural stem cells (NSCs) present in the ventricular zone/subventricular zone (VZ/SVZ) of the developing mammalian brain are regionally specified along dorsoventral and anteroposterior axes (Zappone et al. 2000; Hitoshi et al. 2002) and are temporally regulated to produce neurons during early developmental stages and to produce glia during later stages in the developing brain (Okano and Temple 2009). Secreted morphogens, distributed in a concentration gradient, play critical roles in pattern formation in the developing brain and act as extrinsic factors to temporospatially regulate NSC differentiation. Neurons and astrocytes are generated from most, if not all, regions of the central nervous system. In contrast, oligodendrocytes (OLs) are only generated in several restricted areas; however, their precise places of origin and the mechanisms that define these regions remain to be determined.
Two populations of OLs, generated in a sonic hedgehog (Shh)-dependent and independent manner, exist in the mammalian central nervous system. Shh secreted from the floor plate of the developing spinal cord is indispensable for ventral patterning (Fuccillo et al. 2004). Shh also induces the basic helix-loop-helix transcription factors Olig1 and Olig2 in the pMN domain, from which a majority of spinal cord OLs originate (Lu et al. 2000; Takebayashi et al. 2000; Zhou et al. 2000). Later in development, another population of OLs originates from the dorsal portion of the spinal cord, independent of Shh signaling (Richardson et al. 2006). Similarly, oligodendrocyte precursor cells (OPCs) in the forebrain first appear around embryonic day (E) 12 in the ventral region that extends from the medial ganglionic eminence (MGE) to the anterior entopeduncular area (AEP) (Kessaris et al. 2001, 2006; Nakahira et al. 2006). These ventral forebrain-derived OPCs migrate tangentially to the cortex where they contribute to a subpopulation of myelinating OLs in the adult cortex (Nakahira et al. 2006), although most of them seem to be eradicated postnatally (Kessaris et al. 2006). Following the generation of MGE/AEP-derived OPCs, additional populations of OPCs derived from regions dorsal to the MGE/AEP are produced; these survive and become distributed in the adult cortex (Gorski et al. 2002; Kessaris et al. 2006). Although Gsh2- and Emx1-lineage OLs are thought to originate from the lateral ganglionic eminence (LGE) and cortex, respectively, their precursor cells or the subpopulation of NSCs they segregate from have yet to be characterized.
In order to understand intrinsic and extrinsic factors that specify self-renewing multipotent NSCs to acquire capability for the generation of cortical OLs, it is necessary to identify a temporally and spatially restricted niche, in which OL-producing NSCs are harbored. To track OL-producing precursor population, Olig2 could be a good marker because Olig2 is a key transcription factor for the generation of OLs, based on the observation that the number of OLs is markedly reduced in the spinal cord and forebrain of Olig2−/− mutant mice (Lu et al. 2002; Takebayashi et al. 2002; Zhou and Anderson 2002). Recently, it is suggested that Olig2+ cells exhibit NSC-like multipotent differentiation capability based on the observation of in vitro culture (Gabay et al. 2003). Indeed, Olig2-lineage cells that are traced by the inducible Cre/loxP system differentiate into neurons, OLs, and astrocytes in the forebrain (Ono et al. 2008).
In this study, we show that Olig2 is not expressed by self-renewing NSCs, but by a more committed cell population, whereas Nestin is expressed by both self-renewing and committed neural precursor populations. By utilizing this difference, we analyzed the birth date and place of forebrain OLs.
Materials and Methods
Generation of Nestin Promoter/Enhancer-CreER™ Transgenic (Nestin-CreER tg) Mice
The expression plasmid was constructed by inserting a CreER™ cDNA, encoding Cre recombinase and a mouse estrogen receptor fusion protein, between the rat Nestin gene promoter and enhancer (Yamaguchi et al. 2000) (Supplementary Fig. S1A). The linearized transgene DNA was microinjected into one-cell stage fertilized mouse embryos obtained from superovulated C57BL/6J mice using standard procedures. All transgenic mice were hemizygous and were maintained on the C57BL/6J background. We initially obtained 4 transgenic mouse lines carrying CreERTM under the control of the Nestin promoter/enhancer; 3 lines were analyzed in detail for this study. All 3 Nestin-CreER tg lines demonstrated similar transgene expression patterns and recombination efficiency; thus, we showed results obtained from one line.
Nestin-CreER tg mice and Olig2-CreER (Olig2CreER) knockin mice (Takebayashi et al. 2002) were crossed to Z/EG reporter mice (Novak et al. 2000), which express GFP upon Cre-mediated recombination. Midday of the plug date was termed embryonic day (E) 0.5. In order to induce Cre-mediated recombination, 4-hydroxytamoxifen [4-OHT, dissolved in a dimethylsulfoxide/ethanol/sesame oil (4:6:90) mixture at a concentration of 10 mg/mL; Sigma] was administered intraperitoneally (2 mg/animal) to pregnant dams at E11.5 or E16.5 or subcutaneously (0.5 mg/animal) to neonatal double-heterozygous pups (Nestin-CreER;Z/EG or Olig2CreER;Z/EG). The mice were analyzed by a neurosphere assay at E14.5 or were subjected to immunohistochemical studies at postnatal day (P) 30 or P60. All experiments were carried out under the permission of the institutional Animal Research Committee.
A clonal colony-forming neurosphere assay has been described previously (Tropepe et al. 1999; Hitoshi et al. 2002). Double-heterozygous mice with Nestin-CreER:Z/EG or Olig2CreER:Z/EG were administrated with 4-OHT at E11.5 and were sacrificed at E14.5. Tissues dissected from the LGE and MGE were transferred to serum-free media (SFM) composed of a 1:1 mixture of Dulbecco's modified Eagle's medium (DMEM; GIBCO) and F-12 nutrient (GIBCO), 3 mm sodium bicarbonate (Sigma), and 5 mm HEPES buffer (Sigma). A defined hormone and salt mixture (Sigma) that included insulin (25 μg/mL), transferrin (100 μg/mL), progesterone (20 nm), putrescine (60 μm), and selenium chloride (30 nm) was used instead of serum. Tissues were mechanically dissociated into a cell suspension with a small-bore, fire-polished Pasteur pipette. Cells from the LGE and MGE of E14.5 embryos were cultured at 10 cells/μL in a 24-well plate (Falcon) in SFM containing 10 ng/mL FGF-2 (Sigma), 2 μg/mL heparin (Sigma), and/or 10 ng/mL EGF (Sigma). All remaining cells harvested from LGE and MGE were cultured in 6-cm culture dish to examine whether GFP+ neurospheres were formed or not. After 7 days in vitro, numbers of floating sphere colonies (neurospheres) in a 24-well plate possessing a diameter of >0.1 mm were counted.
In Utero Microinjection and Electroporation
In utero microinjection of FGF-2 was performed as described previously (Naruse et al. 2006). Timed pregnant mice were deeply anesthetized by intraperitoneal injection of 2% ketamine/0.2% xylazine in PBS (0.1 mL/10 g body weight). After cleaning the abdomen with 70% ethanol, a 1–2 cm midline laparotomy was made and the uterus was taken out. One microliter of 1% BSA/PBS solution containing 0.05% fast green with or without FGF-2 (100 μg/mL) was microinjected through the uterus into the lateral ventricles of E13.5 fetal brains using a glass micropipette (type G-1, Narishige). After the surgery, the uterus was placed back into the abdominal cavity, and the surgical incision in the mother was closed to allow embryonic development to continue until E15.5.
In utero electroporation using a Cre/loxP system was performed as described previously (Nakahira et al. 2006). Three micrograms of pCX-NLS-Cre plasmid, which encodes Cre recombinase tagged with a nuclear localization signal driven by the CAG promoter (Nakahira et al. 2006), or expression vectors for pCX-NLS-Cre (1 μg) and an active form of GSK-3β (GSK-3β S9A; 2 μg) (Tanji et al. 2002) or Dickkopf-1 (Dkk1; 2 μg) in 1 μL, was injected into the lateral ventricles of E15.5 Z/EG reporter mouse embryos through the uterus. After injection, embryos within the uterus were placed between the electrodes (CUY650P3, CUY661-3X7 with CUY661N; NEPA GENE). An electroporator (CUY21; NEPA GENE) was used to deliver five 50-ms pulses of 33–35 V, with 950-ms intervals. Brains were analyzed at P1, P3, or young adult (P30).
Coronal cryosections of 18-μm thickness and coronal vibratome sections of 40-μm thickness were immunostained for GFP or for GFP and cell-specific markers. We used rabbit anti-GFP (Invitrogen; 1:2000), rat anti-GFP (Nacalai Tesque, Japan; 1:2000), rabbit anti-GST-π (MBL, Japan; 1:400), mouse anti-APC (CC1) (Millipore; 1:400), rabbit anti-GFAP (Dako; 1:2000), mouse anti-NeuN (Chemicon; 1:1000), rabbit anti-NG2 (Millipore; 1:400), rabbit anti-Olig1 (Abcam; 1:200), rabbit anti-Olig2 (IBL, Japan; 1:200), rabbit anti-cleaved caspase 3 (Cell Signaling Technology, 1:200), and rabbit anti-Axin2 (Abcam, 1:400) antibodies. In situ hybridization was performed as described (Naruse et al. 2006), using Digoxigenin-labeled single-strand riboprobes (Roche) for the entire coding region of indicated gene cDNAs. The images were collected with an Olympus microscope (Olympus BX51; Olympus) and digital camera system (Olympus DP70; Olympus) or a confocal laser scanning microscope (LSM-510; Carl Zeiss).
Tracing Nestin- and Olig2-Lineage Cells
Self-renewing NSCs in the developing brain produce multipotent, non-self-renewing neural progenitor cells, and once neural precursors exit from the NSC self-renewing cycle, differentiation appears uni-directional and lineage-committed progeny never revert to the stem-cell stage in the normal brain (Alvarez-Buylla et al. 2001; Okano and Temple 2009). To examine when neural precursors exit the self-renewing stem cell population and begin to differentiate, we utilized 2 tamoxifen-inducible permanent-labeling systems: Nestin-CreER transgenic (tg) mice (encoding a Cre and estrogen receptor fusion protein driven by the Nestin promoter/enhancer; Supplementary Fig. S1A) and Olig2-CreER knockin (Olig2CreER) mice (Takebayashi et al. 2002). Tamoxifen treatment of mice generated by crossing CreER-expressing mice with a Cre-dependent Z/EG reporter line results in nuclear translocation of Cre recombinase and the subsequent labeling of cells by the expression of GFP. The recapitulation of the endogenous Nestin expression pattern by the CreER transgene was verified in embryonic and postnatal brains by in situ hybridization (Supplementary Fig. S1B–G).
In the developing central nervous system, Olig2 is expressed not only in OPCs but also in multipotent neural precursor cells (Mukouyama et al. 2006; Ono et al. 2008). Indeed, CreER expression in the forebrains of Nestin-CreER;Z/EG and Olig2CreER;Z/EG embryos at E11.5, when OPCs have yet to be generated, was very similar and abundant in the ventral VZ/SVZ (Fig. 1A,B). If the administration of 4-OHT to E11.5 embryos labels the self-renewing neural stem cells, they would produce GFP+ neurospheres in the later developmental stages. We, therefore, performed the neurosphere assay 3 days after the 4-OHT administration and found many GFP+ neurospheres from the ganglionic eminence of E14.5 Nestin-CreER;Z/EG embryos (Fig. 1C,D,G). In contrast, no GFP+ neurospheres were generated in cultures of the ganglionic eminence VZ/SVZ cells from Olig2CreER;Z/EG mice embryos in either FGF-2 or EGF conditions (Fig. 1E–G). Consistent with this observation, 3 days after 4-OHT treatment at E11.5, while GFP+ cells were present in the ganglionic eminence VZ/SVZ as well as in the mantle zone of Nestin-CreER;Z/EG embryos (Fig. 1H,I), GFP+ cells were depleted from the VZ/SVZ and only found in the mantle zone of Olig2CreER;Z/EG embryos (Fig. 1J,K) despite that GFP+ cells were present in the VZ/SVZ at E12.5 (Fig. 1L), suggesting that GFP+ cells in the ganglionic eminence VZ/SVZ differentiated and migrated radially. The absence of GFP+ self-renewing neural stem cells in Olig2CreER;Z/EG embryos may be explained by lower recombination efficiency by CreER under Olig2 promoter than Nestin promoter/enhancer. However, we think this possibility unlikely because, 24 h after the 4-OHT treatment at E11.5, comparable number of GFP+ cells were detected in the whole as well as VZ/SVZ portion of ganglionic eminence of Nestin-CreER;Z/EG and Olig2CreER;Z/EG embryos (Fig. 1M). These data suggest that Nestin-lineage cells include self-renewing neural stem cells and that Olig2-lineage cells are devoid of this population.
Temporal Window of Cortical OL Lineage Cell Differentiation
Cre/loxP recombination, induced by 4-OHT (Fig. 2A, shown as a green vertical line), enables self-renewing NSCs and their progeny to be traced in Nestin-CreER;Z/EG mice (Fig. 2A, shown in red), and multipotent neural progenitor cells and OPCs to be traced in Olig2CreER;Z/EG mice (Fig. 2A, shown in blue). To determine the temporal window in which NSCs exit from the self-renewing cycle and start differentiating into cortical OLs in the adult brain, we injected 4-OHT into Nestin-CreER;Z/EG and Olig2CreER;Z/EG pregnant dams or neonates and examined the fate of labeled cells at P30. Based on morphology, GFP+ cells present in P30 brains were classified as OLs, astrocytes, and neurons (Fig. 2); the identity of some cells was further verified by immunostaining for cell-type-specific markers (Figs 3 and 4). OL-like cells had a smaller cell body than neurons with thin multiple processes, and GFP+ OL-like cells were also immunopositive for Olig1 (∼57%) and Olig2 (∼45%), which labels OPCs and immature OLs, Gst-π (∼35%) and CC1 (∼22%), which labels mature OLs (Fig. 3A–L). Consistent with these observations, it was reported that the cells labeled by Olig2CreER-mediated recombination at adult stages mostly produce OL-lineage cells and 75% of them are NG2+ cells in the cortex (Dimou et al. 2008). The ratios of Olig2+ and Gst-π+ cells to GFP+ OL-like cells were always comparable regardless of the CreER drivers (Fig. 4). GFP+ astrocyte-like cells had brushy morphology and the half of them expressed GFAP (Figs 3M–O and 4).
When 4-OHT was administered at E12.5, GFP+ cortical OLs were only observed in Nestin-CreER;Z/EG mice but not in Olig2CreER;Z/EG mice (Fig. 2B,C), suggesting that neural progenitor cells that are fated to become cortical OLs in the adult brain have yet to be generated. Thus, GFP+ cortical OLs found in Nestin-CreER;Z/EG mice should be derived from neural progenitor cells generated from NSCs after E12.5. Contrary to this, 4-OHT treatments at E16.5 labeled cortical OLs in both Nestin-CreER;Z/EG and Olig2CreER;Z/EG mice, suggesting that cortical OL-producing neural progenitor cells are already present at E16.5 (Figs 2D,E and 4A,B). 4-OHT injection into P1 or P5 Nestin-CreER;Z/EG neonates mostly labeled, if not all, glial cells and a reduced number of total GFP+ cells in the adult cortex (Figs 2F,H and 4C,E). In the cortex of P30 Nestin-CreER;Z/EG mice, we observed a few GFP+ astrocytes (<100 cells in total) but no GFP+ OLs after tamoxifen treatment at P10 (Figs 2J and 4G). An abundance of GFP+ neurons in the olfactory bulb after treatment with 4-OHT at P10 confirmed efficient labeling of neural precursors in Nestin-CreER;Z/EG neonates (Fig. 2L). In addition, GFP+ OLs and other types of cells generated during E16 and P10 survived at least until P60 (Supplementary Fig. S2). GFP+ OL-like cells were also produced in the corpus callosum in the same period with cortical OL-cell production (Fig. 3P–R). These results suggest that neural progenitor cells, which subsequently differentiate into cortical OLs in the adult brain, are derived from self-renewing NSCs during the period just prior to E16.5 through P10.
Neural Precursor Cells at the Dorsoventral Boundary Produce Cortical OLs
We have previously shown that dorsal neural precursors from E12.5 mouse brains, labeled by in utero electroporation of a pCX-NLS-Cre expression plasmid into a ROSA-GAP43-EGFP reporter line, do not produce cortical OLs in the adult brain (Nakahira et al. 2006). These results were confirmed by electroporating pCX-NLS-Cre into the dorsal VZ/SVZ of Z/EG reporter mouse embryos at E15.5, when OL-lineage cells start segregating from NSCs (Fig. 5A). Many GFP+ cells were observed in the cortex of P30 brains (Fig. 5B); these consisted mostly of astrocytes (Fig. 5C) and less abundantly of neurons, but scarcely of OLs (Fig. 5D). To exclude a possibility that electroporation of NLS-Cre induces cell death, we analyzed the expression of cleaved caspase 3, a marker for apoptosis, at P1. We could hardly detect cleaved caspase 3-positive cells 5 days after the electroporation (Supplementary Fig. S3). Thus, most, if not all, of neural precursors present in the dorsal VZ/SVZ of E16.5 brains are not the source of cortical OLs in the adult brain.
Our current and previous results appear inconsistent with a recent report showing that the majority of cortical OLs in the adult brain originate from the Emx1+ dorsal forebrain and that the minority of them originate in the Gsh2+ LGE (Kessaris et al. 2006). Emx1 mRNA is abundant in the dorsal VZ/SVZ of the E16.5 mouse brain, where it is co-expressed with Nestin, and extends to the cortico-striatal boundary (Supplementary Fig. S4A,B). Gsh2 mRNA is present in the VZ/SVZ of the LGE, which borders the Emx1+ domain (Supplementary Fig. S4C). Therefore, it is possible that the cortico-striatal boundary of the developing brain represents a common point of origin for the second and third waves of cortical OLs in the adult brain (Kessaris et al. 2006). To test this possibility, we electroporated a pCX-NLS-Cre expression vector directly into the cortico-striatal boundary of E15.5 Z/EG brains (Fig. 5E). Analysis of the distribution of GFP+ cells at P3 verified the precise electroporation of the cortico-striatal boundary (Fig. 5F). Some GFP+ cells had migrated laterally through the corpus callosum and were positive for Olig2 (Fig. 5G–I). At P30, many GFP+ cells were observed in the corpus callosum and the lateral portion of the cortex (Fig. 5J); based on morphology, these were identified as mature OLs, astrocytes and neurons (Fig. 5K,L). GFP+ cells were further characterized by double immunolabeling with cell-type-specific markers (Fig. 5M–U). These results suggest that dorsoventral boundary of the late embryonic brain is a source of cortical OLs in the adult brain.
Neural Precursor Cells at the Medial Dorsoventral Boundary Overcome the Suppression by Wnt Signaling and Produce Cortical OLs
Next, we examined why dorsal neural precursor cells does not produce cortical OLs. Wnt signaling is known to inhibit OPC induction in the embryonic forebrain and spinal cord (Shimizu et al. 2005; Ye et al. 2009; Langseth et al. 2010), and Wnts are present in the developing brain where they are distributed in a gradient with higher concentrations dorsomedially and lower concentrations ventrolaterally (Langseth et al. 2010). Accordingly, it is possible that high concentrations of Wnts may suppress the generation of OLs from neural precursors in the dorsal VZ/SVZ. To address this possibility, we selectively suppressed Wnt signaling in dorsal neural precursor cells. We electroporated an active form of GSK-3β (GSK-3β S9A) that blocks the transduction of Wnt signaling or Dkk1, an inhibitor of Wnt, together with an NLS-Cre expression vector into the dorsal VZ/SVZ of E15.5 Z/EG reporter mice (Fig. 6A), and analyzed the fate of recombinant GFP+ cells at P30 (Fig. 6B–E). Consistent with the findings by Langseth et al., GFP+ OL-like cells appeared in cortex by inhibition of Wnt signaling (0.81 ± 0.50% after the electroporation of control vector; 20.31 ± 6.34% after that of GSK-3β S9A, t(7) = 2.71, P = 0.030; 18.35 ± 3.65% after that of Dkk1, t(4) = 7.29, P = 0.002, Fig. 6E), and substantial portion of GFP+ cells co-expressed the OL marker GST-π (Fig. 6C,D) (Langseth et al. 2010). Our results suggest that while the potential of NSCs present in the dorsal VZ/SVZ to differentiate into OLs is preserved, the production of OLs from dorsal NSCs is suppressed by Wnt signaling.
We then examined the capability of NSCs at the medial dorsoventral boundary, where Wnt signaling is the strongest in the developing brain due to the secretion of Wnt by the cortical hem (Shimogori et al. 2004; Langseth et al. 2010). Surprisingly, when the pCX-NLS-Cre vector was electroporated into the cortico-septal boundary of E15.5 brains (Fig. 6F), a substantial number of GFP+ OLs were also found in the medial portion of the adult cortex at P30 (Fig. 6G–I). These results can be explained if an inhibitor for Wnt signal is present in the VZ/SVZ of cortico-septal boundary of E15.5 brain. However, we think this possibility unlikely because we found abundant expression of Axin2, which represents the activity of Wnt signaling, at the medial boundary (Fig. 6J,K) (Jho et al. 2002). The gradient of Wnt signal persists in the postnatal brain, since more Axin2+ cells were present in the medial portion than the lateral portion of P14 brain (Supplementary Fig. S5). Because Wnt signal is known as a negative regulator of OL maturation (Fancy et al. 2011), the high Wnt activity in the medial cortex may suppress the maturation of the cortico-septal boundary-derived OLs. Indeed, we found more NG2+ OPCs in the medial portion than the lateral portion of P14 brain (Supplementary Fig. S5).
It is possible that inducing factor(s) exist at the cortico-septal boundary, which overcomes the suppression of OL induction by Wnts. We considered FGF signaling as a candidate for such factors because we have previously shown that FGF-2 injection into the lateral ventricle of E13.5 brain induced ectopic OPCs in the dorsal VZ near the cortico-striatal boundary (Naruse et al. 2006). We therefore examined effects of FGF-2 microinjection at E13.5 on the medial side of boundary and observed that Olig2+ cells in the VZ distributed more dorsally at E15.5 as compared with control brains (arrowhead, Fig. 6L,M). These results suggest that NSCs at the cortico-septal boundary are resistant to the suppression by Wnt signaling and are allowed to be competent by some inducing factors such as FGF-2 for generating cortical OL-lineage cells.
Postmitotic OLs are produced during late embryonic and neonatal stages in the developing mammalian brain and mature to form myelin (Zhou et al. 2000). To understand the molecular mechanisms that specify the generation of OPCs from NSCs, the migration and proliferation of OPCs, and the differentiation of OPCs into OLs, it is crucial to elucidate when and where OPCs originate in the developing brain. However, the fact that OPCs vigorously migrate and divide shortly prior to terminal differentiation makes a birth-dating study difficult (Zhou et al. 2000). To clarify the timing of when OL-lineage cells segregate from the self-renewing NSC pool, we utilized inducible Cre/loxP systems, which permanently label Nestin-lineage, or Olig2-lineage cells. Olig2+ cells are tri-potential in the developing brain, producing neurons, astrocytes, and OLs (Ono et al. 2008) although it remains to be clarified whether single Olig2+ cells proliferate and differentiate into more than one fate or whether Olig2+ cells represent a heterogeneous population consisting of uni-potential neuronal and glial precursor cells. Our data suggest that Olig2+ cells in the VZ/SVZ of embryonic brains are devoid of neurosphere-initiating NSCs and that Olig2 is expressed in multipotent but non-self-renewing neural progenitor cells and differentiating OPCs. Consistent with this idea, Olig2+ cells only generate OLs in the spinal cord during late embryonic stages when neural stem/progenitor cells disappear (Mukouyama et al. 2006).
By differentially labeling NSCs and neural progenitor cells/OPCs using Nestin-CreER and Olig2CreER mice, respectively, we have identified the temporal window during which OL-lineage cells begin to differentiate from NSCs; this occurs between E12.5 and P10. Our previous study showed that very few GFP-labeled OLs were present in the cortex of adult Olig2CreER mice that had received tamoxifen injection at E14.5 (Ono et al. 2008), suggesting that the majority of cortical OLs should begin to segregate from NSCs shortly before E16.5 and beyond. Indeed, NSCs produce a large volume of neurons during early development, and then switch to become gliogenic in the late embryonic brain (Okano and Temple 2009). Consistent with our findings, LGE/CGE-derived Gsh2-lineage OPCs emerge in the E16.5 telencephalon and persist in the adult cortex (Kessaris et al. 2006). It is important to note that our experimental paradigm should not trace MGE/AEP-derived Nkx2.1-lineage OPCs or OLs because these cells are eradicated from the cortex postnatally (Kessaris et al. 2006).
Our current and previous studies show that, despite the extensive investigation, no GFP+ OLs were observed in the adult cortex following labeling of neural precursor cells in the dorsal VZ/SVZ, excluding the dorsoventral boundary, by electroporation at E12.5∼E16.5 (Nakahira et al. 2006). These findings appear inconsistent with the previous results that cortex-derived Emx1-lineage OLs predominate in the adult cortex (Kessaris et al. 2006). However, this inconsistency is resolved if we assume that NSCs at the lateral and medial dorsoventral boundary, which express either Emx1 or Gsh2, yield most, if not all, of the cortical OLs. Alternatively, OPCs born in the dorsal VZ/SVZ could be trapped in the corpus callosum formed during late embryonic stages thus preventing their spread into the cortex. However, we view this possibility as unlikely given that the corpus callosum is absent or very sparse at E15.5 and because few, if any, GFP+ OLs were detected in the corpus callosum at P30 after electroporation of the Cre plasmid into dorsal VZ/SVZ at E15.5. Moreover, dorsal neural precursors, in which Wnt signaling was attenuated by expressing the active form of GSK-3β or Dkk1, were able to migrate radially and were distributed in the P30 cortex. It is also possible that NSCs born in the Emx1+ dorsal VZ migrate ventrally to the boundary region before E15.5. Indeed, some Emx1-lineage NSCs can be detected in the ventral germinal zone (Willaime-Morawek et al. 2006). In either case, the VZ/SVZ at the dorsoventral boundary provides a specialized niche that uniquely enables boundary NSCs to contribute to the neurogenesis of the piriform cortex and amygdala during early embryonic stages (Carney et al. 2006). These NSCs subsequently acquire the ability to generate cortical OLs during later embryonic stages.
OLs derived from neural precursor cells at the lateral and medial dorsoventral boundary distribute to the lateral and medial portion of the adult cortex, respectively, but not to the dorsal most portion. It is possible that other dorsoventral boundary regions are the source of OLs for this region, such as rostral or caudal regions or both, which we could not label by electroporation. Alternatively, LGE/CGE-derived Gsh2-lineage OPCs could differentiate from NSCs shortly before E16.5 and migrate to the dorsal most portion of the adult cortex. Indeed, our previous study showed that neural precursor cells in the ventral VZ/SVZ of E12.5 embryos migrate and produce OLs in the whole cortex (Nakahira et al. 2006). In spinal cord, ventrally derived OLs vigorously spread in the whole spinal cord, whereas dorsally derived OLs show limited migration to the dorsal and dorsolateral funiculi of the spinal cord (Tripathi et al. 2011). Currently, it is not clear whether OLs from different origins possess distinct function or not. It would be intriguing to examine electrical properties and myelination of lateral and medial boundary-derived OLs because the latter OLs appear to contribute to myelin in the corpus callosum than the former OLs.
Wnt signaling strongly suppresses the generation of OLs from neural precursor cells (Langseth et al. 2010 and this study). Consistent with this idea, neural precursor cells in the cortico-striatal boundary, which receive less Wnt signals than those in the dorsal area, produce large quantities of cortical OLs. However, this model cannot explain the fact that neural precursor cells in the cortico-septal boundary also generate cortical OLs despite the fact that the medial portion of the developing brain, including the cortico-septal boundary, is under the influence of strong Wnt signals secreted from the cortical hem (Shimogori et al. 2004). Inducing signals able to overcome the suppressive Wnt effects may exist in the cortico-septal boundary of the developing brain. We have previously shown that intraventricular microinjection of FGF-2 at E13.5 induces neural precursors in the VZ/SVZ and extending dorsally to express the OL-lineage cell markers PDGFRα and O4 (Naruse et al. 2006), and conversely, the disruption of FGF signaling through Fgfr1 and Fgfr2 attenuates the generation of OPCs (Furusho et al. 2011). Interestingly, FGF-8 is expressed in a restricted area in the medial portion of the developing brain (Smith et al. 2006). Indeed, Olig2 expression was observed in cortico-septal boundary at late embryonic stages. Furthermore, overdose of FGF-2 expanded the area of Olig2 expression more dorsally but could not induce the Olig2 expression in the dorsal most region despite that FGF-2 and FGF receptors are expressed in both dorsal and ventral forebrain in rodents (Raballo et al. 2000; Bansal et al. 2003). Those observations suggest that FGF-2 could not inhibit Wnt signals and that additional factor(s) are required to fully induce OPCs. Alternatively, FGF signaling could act as one of OL-inducing factors in subpopulation of NSCs, which reside in the dorsoventral boundary region and are primed to acquire capability to produce OLs.
Transplantation of OLs derived from ES or iPS cells into lesions may be useful for the treatment of chronic demyelinating diseases. However, despite the successful induction of NSCs from ES/iPS cells using several different protocols, mostly via the formation of embryoid bodies, the efficiency in obtaining OLs is relatively unsatisfactory (Tokumoto et al. 2010). Thus, the elucidation of transcription factors or gene expression patterns that would confer neural precursors with the characteristic ability of dorsoventral boundary NSCs to differentiate into OLs will be highly beneficial.
This work was supported by Grants-in-Aid for Scientific Research on Priority Area (20019032) (K. I.), Scientific Research (B) (19300136) (S. H.), and Scientific Research on Innovative Areas (25123706) (S. H.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan and a Grant from Daiichi-Sankyo Foundation of Life Science (S.H.).
We thank Drs. A. Kikuchi and K. Kaibuchi for plasmids and K. Inaba and E. Seki for technical helps. Conflict of Interest: None declared.