Abstract
In the CNS, myelination and remyelination depend on the successful progression and maturation of oligodendroglial lineage cells, including proliferation and differentiation of oligodendroglial progenitor cells (OPCs). Previous studies have reported that Sox2 transiently regulates oligodendrocyte (OL) differentiation in the embryonic and perinatal spinal cord and appears dispensable for myelination in the postnatal spinal cord. However, the role of Sox2 in OL development in the brain has yet to be defined. We now report that Sox2 is an essential positive regulator of developmental myelination in the postnatal murine brain of both sexes. Stage-specific paradigms of genetic disruption demonstrated that Sox2 regulated brain myelination by coordinating upstream OPC population supply and downstream OL differentiation. Transcriptomic analyses further supported a crucial role of Sox2 in brain developmental myelination. Consistently, oligodendroglial Sox2-deficient mice developed severe tremors and ataxia, typical phenotypes indicative of hypomyelination, and displayed severe impairment of motor function and prominent deficits of brain OL differentiation and myelination persisting into the later CNS developmental stages. We also found that Sox2 was required for efficient OPC proliferation and expansion and OL regeneration during remyelination in the adult brain and spinal cord. Together, our genetic evidence reveals an essential role of Sox2 in brain myelination and CNS remyelination, and suggests that manipulation of Sox2 and/or Sox2-mediated downstream pathways may be therapeutic in promoting CNS myelin repair.
SIGNIFICANCE STATEMENT Promoting myelin formation and repair has translational significance in treating myelin-related neurological disorders, such as periventricular leukomalacia and multiple sclerosis in which brain developmental myelin formation and myelin repair are severely affected, respectively. In this report, analyses of a series of genetic conditional knock-out systems targeting different oligodendrocyte stages reveal a previously unappreciated role of Sox2 in coordinating upstream proliferation and downstream differentiation of oligodendroglial lineage cells in the mouse brain during developmental myelination and CNS remyelination. Our study points to the potential of manipulating Sox2 and its downstream pathways to promote oligodendrocyte regeneration and CNS myelin repair.
- myelination and remyelination
- oligodendrocyte differentiation
- oligodendrocyte regeneration
- oligodendroglial lineage progression
- oligodendroglial progenitor cells
- Sox2
Introduction
The transcription factor SRY (sex-determining region)-box 2 (Sox2) is a critical transcription factor in regulating the properties of stem cells, including neural stem cells (Zhang and Cui, 2014), and it is the key determining factor for in vivo reprogramming of differentiated neural cells into neural precursor cells (Niu et al., 2013; Heinrich et al., 2014). In the CNS, Sox2 was originally thought to inhibit the neuronal differentiation of neural stem/progenitor cells (NSPCs) (Graham et al., 2003). However, later genetic studies demonstrate that Sox2 positively regulates neuronal differentiation from NSPCs (Ferri et al., 2004; Episkopou, 2005).
In oligodendroglial lineage cells, Sox2 has been reported to be absent from in vitro oligodendroglial progenitor cells (OPCs) (Kondo and Raff, 2004; Lyssiotis et al., 2007). Recent studies demonstrate that in vivo OPCs constantly express low levels of Sox2 (Shen et al., 2008; Dai et al., 2015) and propose that Sox2 maintains OPC proliferation and plays an inhibitory role in oligodendrocyte (OL) differentiation and regeneration (Shen et al., 2008; Pedre et al., 2011). Until recently, genetic evidence suggests that Sox2 is an essential regulator of OL terminal differentiation, but dispensable for OPC proliferation and migration in the embryonic and perinatal spinal cord (Hoffmann et al., 2014). A subsequent study by Zhao et al. (2015) shows that Sox2 appears dispensable for developmental myelination in the postnatal spinal cord; instead, it has a crucial role in recruiting adult OPCs into the chemical-induced spinal demyelinating lesions during spinal cord myelin repair. These discrepant results strongly suggest that Sox2 may play a context-dependent role in regulating CNS oligodendrocyte development and regeneration. In this regard, the functions of Sox2 in brain myelination and remyelination have yet to be defined.
We found that Sox2 is expressed in all OPCs in the postnatal and adult CNS, and that Sox2 is transiently upregulated in newly differentiated OLs during developmental myelination and in newly regenerated OLs during remyelination. Using in vivo gene conditional knock-out (cKO), we demonstrate that Sox2 is essential not only for OPC proliferation and population expansion, but also for downstream OL differentiation during developmental myelination in the murine brain. We also demonstrate that Sox2 is required for OPC proliferation and OL regeneration after myelin damage in the adult brain and spinal cord. Our study suggests a context-dependent role of Sox2 in regulating CNS oligodendrocyte development and regeneration.
Materials and Methods
Transgenic mice.
All transgenic mice were maintained on a C57BL/6 background and covered by Institutional Animal Care and Use Committee protocols approved by University of California–Davis. The Cnp-Cre (Lappe-Siefke et al., 2003) (RRID:MGI_3051754) and Rosa-EYFP (RRID:IMSR_JAX:006148) transgenic mice were described in our previous study (Guo et al., 2012; Hammond et al., 2015). Sox10-Cre (RRID:IMSR_JAX:025807), Pdgfra-CreERT2 (RRID:IMSR_JAX:018280), Sox2-CreERT2 (RRID:IMSR_JAX:017593), and Sox2fl/fl (RRID:IMSR_JAX:013093) transgenic mice were purchased from The Jackson Laboratory. Both male and female mice were used in this study. We crossed Cre lines with Sox2fl/fl mice to generate Sox2 cKO mice, in which Cre transgenes were maintained as heterozygosity (Cre+/−). We used non-Cre Sox2fl/fl as Sox2 wild-type (WT) mice or non-Cre control mice. In our study, we referred to the Cnp-Cre, Sox2fl/fl mice as Cnp-Sox2 cKO mice, Sox10-Cre, Sox2fl/fl as Sox10-Sox2 cKO mice, and Pdgfrα-CreERT2, Sox2fl/fl mice treated with tamoxifen as Pdgfrα-Sox2 cKO mice. In Sox2-CreERT2 mice, the Cre transgene is homologously knocked in the endogenous locus of Sox2; therefore, Sox2-CreERT2, Sox2fl/+ mice would be Sox2 cKO mice after tamoxifen injection and were referred to as Sox2-Sox2 cKO mice.
Tamoxifen preparation and administration.
Tamoxifen (T5648; Sigma-Aldrich) was prepared as described in our previous studies (Lang et al., 2013; Hammond et al., 2015). In the experimental designs of developmental myelination, Pdgfrα-CreERT2, Sox2fl/fl mice and Sox2fl/fl controls were intraperitoneally injected with tamoxifen at a dose of 100 μg/g body weight at time points indicated in the figures. In the experimental designs of demyelination/remyelination, adult (2–3 months old) Pdgfrα-CreERT2, Sox2fl/fl mice and Sox2fl/fl controls were intraperitoneally injected with 5 d course of daily tamoxifen at a dose of 1 mg per day.
BrdU or EdU preparation and administration.
BrdU (B5002, Sigma) or EdU (A10044, Thermo Fisher Scientific) was dissolved in 0.9% sterile saline at a concentration of 10 mg/ml. BrdU or EdU was intraperitoneally injected to Sox2 cKO and Sox2fl/fl control littermates at a dose of 100 μg/g body weight at time points indicated in the figures.
Myelin oligodendrocyte glycoprotein peptide 35-35-induced experimental autoimmune encephalomyelitis (MOG-peptide35-55 EAE) and cuprizone animal models of CNS demyelination.
The procedures of MOG-peptide35-55 EAE were conducted as described in our previous study (Guo et al., 2012). Complete Freund's adjuvant (CFA)-immunized mice were used as CFA controls for MOG-peptide35-55 EAE. In the MOG-peptide35-55 EAE animal model, multifocal inflammatory and demyelination lesions predominantly appear in the lumbar segment of spinal cord. Tamoxifen was intraperitoneally injected to Pdgfrα-Sox2 cKO and Sox2fl/fl mice for 5 consecutive days starting when mice showed a clinical score ≥2 (Guo et al., 2012). The cuprizone-induced demyelination model was conducted according to our published protocols (Hammond et al., 2015). In the cuprizone model, diffused demyelination lesions predominantly occur in the forebrain corpus callosum.
Antibodies and primers.
The antibodies used in immunohistochemical staining and Western blotting included the following: Olig2 (AF2418, RRID:AB_2157554, 1:100; R&D Systems), Olig2 (18953, RRID:AB_494617, 1:100; IBL), NG2 (AB5320, RRID:AB_91789, 1:200; Millipore), PDGFRα (sc-338, RRID:AB_631064, 1:150; Santa Cruz Biotechnology), O4 (MAB345, RRID:AB_94872, 1:200; Millipore), Sox2 (sc-17320, RRID:AB_2286684, 1:500; Santa Cruz Biotechnology), β-actin (3700, RRID:AB_2242334, 1:1000; Cell Signaling Technology), Sox10 (sc-17342, RRID:AB_2195374, 1:100; Santa Cruz Biotechnology), BrdU (sc-70441, RRID:AB_1119696, 1:100; Santa Cruz Biotechnology), Ki67 (9129, RRID:AB_10989986,1:200; Cell Signaling Technology), EYFP/GFP (06–896, RRID:AB_310288, 1:500; Millipore), TCF7l2 (2569S, RRID:AB_2199816,1:200; Cell Signaling Technology; sc-8632, RRID:AB_2199825, 1:100; Santa Cruz Biotechnology), MBP (NB600-717, RRID:AB_2139899, 1:200; Novus), SMI312 (SMI-312R, RRID:AB_2135329, 1:1000, Covance), active caspase-3 (G748A, RRID:AB_430875, 1:200; Promega), pan-oligodendrocyte marker Clone CC1 (OP80, RRID:AB_213434, 1:200; Calbiochem), and APC (sc-896, RRID:AB_2057493, 1:100; Santa Cruz Biotechnology). Our previous study (Lang et al., 2013) shows that the immunostaining patterns of APC and clone CC1 antibodies are different and that APC is transiently expressed in premyelinating oligodendrocytes. Bin et al. (2016) subsequently demonstrated that antibody clone CC1 binds Quaking 7, an RNA binding protein that is highly expressed in myelinating oligodendrocytes. All secondary antibodies were DyLight 488- or DyLight549-conjugated (Fab)2 fragments (from Jackson ImmunoResearch Laboratories). Brdu, Edu (Click-iT EdU imaging kits, Invitrogen C10339), and TUNEL (Promega, G3250) immunostaining was performed as in previous studies (Guo et al., 2011; Sohn et al., 2012). qPCR primers were from the PrimerBank at (pga.mgh.harvard.edu/primerbank/).
Primary OPC culture and in vitro differentiation, Western blot, immunohistochemistry, and qRT-PCR.
The procedures of the above-mentioned experiments were performed according to the published protocols in our previous studies (Guo et al., 2012; Lang et al., 2013; Hammond et al., 2015)
RNA sequencing and data analysis.
Total RNA was extracted from forebrains of Pdgfrα-CreERT2, Sox2fl/fl mice (n = 3, tamoxifen injection at P6 and P7, forebrain harvested at P14) and non-Cre littermate controls (n = 3, tamoxifen injection at P6 and P7, forebrain harvested at P14) using QIAGEN RNeasy for lipid tissues (catalog #74804) with on-column DNase I digestion. The quality of RNAs was determined by the Bioanalyzer 2100 system (Agilent Technologies). The cDNA library was prepared using the NEBNext Ultra Directional RNA Library Prep Kit (#E7420) for Illumina, and sequenced on the Illumina HiSeq 4000 sequencing platform. Single-end clean reads were aligned to the reference genome (mouse genome mm10) using TopHat version 2.0.12. Differentially expressed genes were analyzed using DESeq version 1.10.1, and p < 0.05 was considered as differentially expressed genes. Gene ontology analysis of the differentially expressed genes between Pdgfrα-Sox2 cKO and non-Cre controls was performed using the National Institutes of Health online tool DAVID (https://david.ncifcrf.gov/). In our RNA-seq results, the number of total clean reads was similar in Pdgfrα-Sox2 cKO (1.29E+08, n = 3) to those in non-Cre controls (1.20E+08, n = 3, two-tailed Student's t test, p = 0.63). Pearson correlation analysis showed that the intragroup and intergroup variations were neglectable and demonstrated by very high correlation coefficient (R2 = 0.986–0.996).
Image acquisition and in vivo Sox2 density quantification.
To quantify nuclear Sox2 density in OPCs and newly differentiated OLs, triple immunohistochemical images (Sox2, PDGFRα or NG2 and TCF7l2) (see Figs. 1K, 8D,G) projecting 10 μm optical thickness were obtained using Nikon Confocal C1 and imported to the National Institutes of Health ImageJ for subsequent quantification of Sox2 density. The nuclei of Sox2+/PDGFRα+ (or NG2+) OPCs and Sox2+/TCF7l2+ newly differentiated OLs were outlined by using the “freeform” selection tool of ImageJ. The nuclear Sox2 expression level was presented as Corrected Sox2 Fluorescence Density, which was calculated as the value of Integrated Density − (Area of Selected Nucleus × Mean Fluorescence of Background). The mean fluorescence of background was calculated by averaging at least four nonstaining locations on the same histological sections. At least 60 PDGFRα (NG2)+ OPCs or TCF7l2+ OLs from 3 animals were quantified in this study.
Motor function assessment.
Motor function was tested by measuring accelerating Rotarod retention times. The parameter settings of the accelerating Rotarod in this study: starting speed = 4.0 rotations per minute (rpm), speed step = 1.3 rpm every 10 s, ending speed = 40 rpm. Mice were trained for three 5 min sessions daily for 2 consecutive days followed by data collection on the third day. The retention time of each mouse was calculated by averaging the retention duration on the Rotarod of 3 trials.
Toluidine blue staining and transmission electron microscopy.
Tissue processing for semithin sections of toluidine blue staining and ultrathin sections of transmission electron microscopy was adapted from our previous protocols (Guo et al., 2015; Sohn et al., 2017). In brief, postnatal 28 d Sox10-Cre, Sox2fl/fl and littermate Sox2fl/fl control mice were anesthetized with ketamine and xylazine mixture and perfused with 4% (w/v) PFA, followed by 3% (w/v) glutaraldehyde. Corticospinal tracts at the lumbar segments (∼1–2 mm thick) were dissected under the stereoscope for further process. Dissected specimens containing corticospinal tracts were washed with 0.2 m sodium cacodylate buffer, pH 7.2, postfixed in 2% (w/v) aqueous osmium tetroxide for 2 h, and dehydrated through ascending alcohols, followed by washing in propylene oxide. The resulting specimens were embedded in EMBed-812 resin. Semithin (500 nm) sections were cut on a Leica EM UC6 microtome and subjected for toluidine blue staining. Ultrathin (70–80 nm) sections were cut on a Leica EM UC7 microtome and collected on 12 mm Formvar-coated copper slot grids, double stained with uranyl acetate and lead citrate, followed by imaging on a CM120 electron microscope.
Experimental design and statistical analyses.
For immunohistochemical quantification, 10 μm optical thickness sections were obtained by confocal z-stacking and projected into one flattened image. The images were from the same anatomic locations of Sox2 cKO and control mice, and at least three histological sections were analyzed in each animal. In the MOG-peptide EAE experimental designs, due to the stochastic distribution of inflammatory infiltrations in the spinal white matter, we quantified marker-positive cells in the whole coronal sections of lumbar spinal cord, rather than the same anatomic locations. All the counting and calculation were performed using the confocal EZ-C1 viewer application (Nikon). For qRT-PCR quantification, the mRNA expression level of interested genes in each sample was normalized to the internal control, housekeeping gene Hsp90. The equation of 2∧(Ct of Hsp90 − Ct of interested gene) was used for gene expression calculation. We used National Institutes of Health ImageJ to quantify protein expression levels by analyzing the scanned grayscale films. The level of protein of interest was normalized to the internal control β-actin. Two-group comparisons were analyzed by two-tailed Student's t tests. Data were presented as mean ± SD. A p value <0.05 was considered as significant difference. Multiple comparisons were performed with one-way ANOVA, followed by Tukey's post hoc test to determine which two groups were significantly different. A p value <0.05 was considered as significant difference. For detailed information of statistical analyses, including sample size, p value, t value, and degrees of freedom, see Table 1.
Results
Sox2 is expressed in OPCs and transiently upregulated in newly differentiated, premyelinating OLs during oligodendroglial lineage progression and maturation
In the corpus callosum of neonatal pups of postnatal day 0 (P0) where no differentiated OLs appear yet, Sox2 was observed in PDGFRα+ OPCs (Fig. 1A, arrowheads) and brain lipid basic protein-positive astrocyte precursor cells (Fig. 1A, arrows). The observation of Sox2 expression in astroglial lineage cells was consistent with previous data (Guo et al., 2011; Zhao et al., 2015) and was beyond the scope of the current study. A previous study reported that Sox2 was expressed in early postnatal OPCs and absent from adult OPCs (Zhao et al., 2015). Similarly, our immunohistochemical data confirmed that early postnatal OPCs in the brain on P5 and P14 expressed Sox2 (Fig. 1B, arrowheads). We found that OPCs in the adult brain retained Sox2 expression (Fig. 1C, arrowheads, left). The expression of Sox2 in adult OPCs was unequivocally demonstrated by the absence of immunoreactive signals in the tamoxifen-induced Sox2 cKO brain of adult Pdgfra-CreERT2, Sox2fl/fl mice (Fig. 1B, arrows, right). Interestingly, Sox2 was also expressed in differentiated CC1+ OLs in the forebrain on P14 (Fig. 1D, left, arrowheads) and spinal cord on P10 (Fig. 1D, right, arrowheads), two representative time windows of active myelination in the murine CNS. The proportion of CC1+ OLs that were Sox2-positive in the corpus callosum progressively decreased during postnatal brain development (81.5 ± 7.0% at P7, 38.7 ± 5.8% at P13, 4.9 ± 1.7% at P60, mean ± SD). Particularly, most myelin MBP+ (Fig. 1E, left) and CNP+ (Fig. 1E, right) OLs expressed Sox2 in the forebrain on P7, a proximate time point that OL differentiation starts in the murine brain.
Using TCF7l2 to mark newly differentiated OLs (Hammond et al., 2015), we found that Sox2-expressing OLs (Sox2+/CC1+) were TCF7l2-positive in the forebrain (Fig. 1F, arrowheads) and spinal cord (Fig. 1G, arrowheads). Quantification data showed that >90% of TCF7l2+ newly differentiated OLs were Sox2-positive in the corpus callosum regardless of the time points we assessed (P7, P10, P14, and P60), although the density of TCF7l2+ newly differentiated OLs was decreased over time during developmental myelination, reflecting the gradually diminished rate of oligodendrocyte generation (Hammond et al., 2015).
We found that Sox2 peaks at the onset of OL differentiation. The Sox2 mRNA level in primary differentiating OLs was significantly increased compared with that in purified primary OPCs that had been isolated from neonatal mouse forebrains (Fig. 1H). The Sox2 protein level was also increased in differentiating OLs after OPCs were cultured in the differentiation medium for 1 d (Fig. 1I). The primary OPC culture and differentiation system were validated by the sharp increases in the expression levels of MBP mRNA (Fig. 1J) and protein (Fig. 1I) in differentiating OLs (MBP mRNA, ∼700-fold higher in D1 OLs vs OPCs, ∼20,000-fold higher in D2 OLs vs OPCs, and ∼80,000-fold higher in D4 OLs vs OPCs). Particularly, TCF7l2 was overlapped with Sox2 protein expression in differentiating OLs at D1 and D2 (Fig. 1I), consistent with the immunohistochemical observations (Fig. 1F,G). In line with the in vitro data, triple immunohistochemistry of Sox2, TCF7l2, and PDGFRα (Fig. 1K) showed that the corrected Sox2 fluorescence density was ∼4-fold higher in TCF7l2+ newly differentiated OLs than that in PDGFRα+ OPCs in the subcortical white matter tracts of P8 and P14 mice (Fig. 1L). Collectively, our data demonstrate that Sox2 is expressed in OPCs and transiently upregulated in newly differentiated, premyelinating OLs along the progression of oligodendroglial lineage during developmental myelination (Fig. 1M) and indicate that Sox2 may play a crucial role in coordinating multiple steps of oligodendroglial lineage progression.
Sox2 cKO inhibits developmental myelination and OL differentiation in the brain of Cnp-Cre, Sox2fl/fl mice
The Cre-LoxP-mediated genetic approach was used to study the role of Sox2 in the progression of oligodendroglial lineage cells. We observed prominent hypomyelination in the forebrain of Cnp-Cre, Sox2fl/fl (referred to as Cnp-Sox2 cKO) mice, compared with non-Cre Sox2fl/fl controls on P14 (Fig. 2A–C). Quantification data showed that the intensity of MBP-immunoreactive signals was significantly reduced in the Cnp-Sox2 cKO forebrains (Fig. 2D) at the histological level, which was confirmed by Western blot (Fig. 2E). In contrast, the intensity of SMI312+ axons was indistinguishable between Cnp-Sox2 cKO and control mice (Fig. 2D,F), indicating that the observed hypomyelination in the Cnp-Sox2 cKO brains is less likely due to a diminution of myelinable axons.
Previous studies demonstrate that Cnp-Cre-mediated gene deletion primarily occurs in the later stages of oligodendrocyte development (Dugas et al., 2010; Moyon et al., 2016; Zhao et al., 2016). Using a Cnp-Cre reporter system, we confirmed that >85% of PDGFRα+ OPCs were EYFP-negative in the subcortical white matter tract of P7 Cnp-Cre, Rosa26-EYFP reporter mice (Fig. 3A, arrowheads), and almost all EYFP+ cells were O4+ differentiation-committed late OPCs and/or differentiated OLs (Fig. 3B, arrowheads) (Miron et al., 2011). Consistent with the EYFP expression, Sox2 was intact in most PDGFRα+ OPCs (Fig. 3C, bottom, arrowheads) but primarily deleted in CC1+ OLs (Fig. 3C, bottom, arrows) in the subcortical white matter of Cnp-Sox2 cKO brains.
The hypomyelination phenotype (Fig. 2) and stage-specific Sox2 deletion (Fig. 3A–C) suggest that OL differentiation may be affected in the brain of the Cnp-Sox2 cKO mice. To this end, we used stage-specific markers to analyze OL differentiation: CC1 for OLs, PDGFRα for OPCs, and Olig2 for both OLs and OPCs. The density of Olig2+CC1+ differentiated OLs was significantly lower in the subcortical white matter of Cnp-Sox2 cKO mice, whereas the density of Olig2+PDGFRα+ OPCs was statistically similar between the two groups (Fig. 3D), suggesting a specific perturbation of OL differentiation in the Cnp-Sox2 cKO brains. We also found more than twofold reduction in the density of TCF7l2+ premyelinating OLs in the subcortical white matter of P14 Cnp-Sox2 cKO mice (Fig. 3E,F), indicating a diminished rate of OL generation in the Cnp-Sox2 cKO brains.
Together, by leveraging the stage-specific Cnp-Cre, Sox2fl/fl cKO system, our data demonstrate that Sox2 positively regulates OL differentiation during developmental myelination in the murine brain.
Conditional Sox2 ablation does not affect the density of oligodendrocytes in the postnatal spinal cord of P14 Cnp-Cre, Sox2fl/fl mice
Unlike impaired OL differentiation in the brain, the distribution of CC1+ differentiated OLs (Fig. 4A) and the density of Olig2+CC1+ differentiated OLs (Fig. 4B) were comparable in the spinal cord of Cnp-Sox2 cKO mutants from those in the non-Cre Sox2fl/fl controls at P14. Western blotting showed that MBP protein expression was similar between the two groups (Fig. 4C). Consistently, qRT-PCR quantification demonstrated that the mRNA levels of the major myelin genes, Mbp, Plp, Mag, and Mobp, and pan-oligodendroglial lineage cell marker Sox10 was statistically indistinguishable in the spinal cords of Cnp-Sox2 cKO and control mice (Fig. 4D). The unperturbed OL differentiation is unlikely due to the efficiency of Sox2 deletion because the levels of Sox2 protein (Fig. 4C) and mRNA (Fig. 4D) were significantly decreased in the spinal cord, and Sox2 expression was abolished in all CC1+ oligodendrocytes at the histological level (Fig. 4E). Our data collected from the P14 spinal cord of Cnp-Sox2 cKO mutants are in agreement with a recent study showing that Sox2 disruption elicited by the ubiquitously expressed CAG-CreERT2 does not affect the number of differentiated OLs in the postnatal spinal cord at P14 (Zhao et al., 2015).
Sox2 ablation specifically in OPCs reveals an essential role of Sox2 in OPC population expansion
The observation that Sox2 was expressed in OPCs from postnatal to adult CNS (Fig. 1) led us to hypothesize that Sox2 is additionally required for OPC population expansion. Because the Cnp-Sox2 cKO paradigm did not induce efficient Sox2 deletion in OPCs in the subcortical white matter, we used Pdgfrα-CreERT2, Sox2fl/fl cKO to ablate Sox2 specifically in OPCs.
Tamoxifen was injected intraperitoneally to the Pdgfrα-CreERT2, Sox2fl/fl mice (referred to as Pdgfrα-Sox2 cKO) and non-Cre Sox2fl/fl control mice on P6 and P7. We found that 88.9% (±9.4%, SD) OPCs in the subcortical white matter had no detectable Sox2 expression on P9, 2 d after the last tamoxifen treatment (Fig. 5A, right panels, arrowheads), in sharp contrast to the low level expression of Sox2 in all OPCs in control mice (Fig. 5A, left panels, arrowheads). On P14, 1 week after the last tamoxifen injection, the density of Sox10+PDGFRα+ OPCs (Fig. 5B) in the subcortical white matter was significantly lower in the Pdgfrα-Sox2 cKO mutants compared with non-Cre Sox2fl/fl controls (Fig. 5C, top). Consistent with the histological quantification, qRT-PCR results showed that the mRNA level of OPC marker PDGFRα was also significantly diminished (Fig. 5C, bottom).
OPC population is mainly expanded during the first postnatal week in the murine brain. To determine the role of Sox2 in OPC expansion during this early postnatal stage, we deleted Sox2 in OPCs on P1, P2, and P3 by tamoxifen injections and analyzed OPC population on P8 (Fig. 5D). Our data showed that the density of Sox10+PDGFRα+ OPCs in the subcortical white matter (Fig. 5E) was significantly diminished in the Pdgfrα-Sox2 cKO mice compared with that in the non-Cre controls (Fig. 5F).
We speculate that diminished OPC population size observed in the Pdgfrα-Sox2 cKO brains will result in decreased OL differentiation and myelination. RNA sequencing was used to identify differentially expressed genes at the transcriptional level between Pdgfrα-Sox2 cKO and control forebrains on P14 (tamoxifen injections on P6 and P7). Gene ontology analysis of the differentially expressed genes (Table 2) showed significant enrichments in the biological processes of OL development, differentiation, and myelination (Fig. 5G). qRT-PCR quantification confirmed that the mRNA levels of myelination- and OL differentiation-related genes were significantly reduced in the Pdgfrα-Sox2 cKO forebrains (Fig. 5H). The density of Sox10+CC1+ differentiated OLs was also significantly attenuated at the histological level (Fig. 5I). Together, our data suggest that Sox2 is required for OPC population expansion in the postnatal murine brain.
Sox2 controls OPC population supply by regulating OPC cell proliferation but not survival
OPC homeostasis is regulated by OPC proliferation, survival, or both. We analyzed proliferation rate of brain OPCs of P14 Pdgfrα-Sox2 cKO and control mice, both of which had received tamoxifen on P6 and P7. Two hours of EdU pulse labeling demonstrated that the number of Sox10+EdU+ proliferating OPCs (Fig. 6A) was significantly less in the Pdgfrα-Sox2 cKO brain than that in controls (Fig. 6B, left). The decreased proliferative rate was confirmed by another proliferation-related antigen Ki67 (Fig. 6B, right). The survival of OPCs, however, was not affected as demonstrated by no significant differences in the number of active Caspase3+Sox10+ (Fig. 6C,D, left) and TUNEL+Sox10+ (Fig. 6D, right) apoptotic oligodendroglial lineage cells between the two groups. We also did not observe differences in the density of Caspase3+Sox10+ apoptotic oligodendroglial lineage cells in the subcortical white matter near the subventricular zone in the forebrain of P8 Pdgfrα-Sox2 cKO that had been administered tamoxifen on P1, P2, and P3 (4.6 ± 0.8/mm2 in Pdgfrα-Sox2 cKO vs 5.4 ± 1.2/mm2 in Sox2fl/fl, n = 3 each group).
Previous study suggests that Sox2 critically regulates neural stem cell survival in the subgranular zone (SGZ) (Favaro et al., 2009) and SVZ (Feng et al., 2013) in the murine brain. We ablated Sox2 in all Sox2-positive cells (including Sox2+ neural stem cells and oligodendroglial lineage cells) by using Sox2-CreERT2, Sox2fl/+ cKO system (the Sox2-CreERT2 is a homologous knock-in transgene). Tamoxifen was administered on P6 and P7, and cell survival was analyzed on P14. Our quantification data showed that the number of active Caspase3+ apoptotic cell in the SGZ and SVZ was significantly increased in Sox2-CreERT2, Sox2fl/+ mice (referred to as Sox2-Sox2 cKO) (156.6/mm2 in Sox2-Sox2 cKO SGZ vs 11.4/mm2 in control SGZ, n = 3, p < 0.0001; 24.7/mm2 in Sox2-Sox2 SVZ vs 4.2/mm2 in control SVZ, n = 3, p = 0.0021, two-tailed Student's t test). The altered cell survival detected in the SGZ and SVZ stem cell niche of the Sox2-Sox2 cKO mice is consistent with previous publications (Favaro et al., 2009; Feng et al., 2013), which also supports the effectiveness of our cell survival analysis. Double immunohistochemistry of active Caspase3 and lineage-specific markers demonstrated that apoptotic cells in the SGZ (Fig. 6G) and SVZ (data not shown) were GFAP+ neural stem cells but not doublecortin+ neuroblasts nor HuC/D+ neurons.
In agreement with the cell survival analysis in the Pdgfrα-Sox2 cKO system (Fig. 6C,D), we did not notice any changes in the number of active Caspase3+Sox10+ apoptotic oligodendroglial lineage cells in SGZ and SVZ between Sox2-Sox2 cKO and non-Cre controls (Fig. 6E,F, right panels). These data suggest that Sox2's role in cell survival is cell type-dependent: it regulates neural stem cell survival, but it is dispensable for oligodendroglial lineage cell survival.
Oligodendrocyte differentiation and myelination are impaired in the Sox10-Cre, Sox2fl/fl mutants, even in the later stages of brain development
Using stage-specific Sox2 cKO paradigms, our experimental data suggest that Sox2 regulates brain myelination by coordinating upstream OPC proliferation and downstream OL differentiation. Therefore, we predict that Sox2 cKO in all oligodendroglial lineage cells (both OPCs and OLs) will lead to the inhibition of OPC expansion, OL differentiation, and brain myelination, even in the later stages of brain development. To support this prediction and also to strengthen our conclusion drawn from Cnp-Sox2 cKO and Pdgfrα-Sox2 cKO mutants, we used Sox10-Cre (Matsuoka et al., 2005) to ablate Sox2 in all oligodendroglial lineage cells and analyzed brain myelination and OL differentiation at later developmental ages.
The Sox10-Cre, Sox2fl/fl (referred to as Sox10-Sox2 cKO) mice developed severe tremors and ataxia, typical phenotypes reminiscent of CNS hypomyelination, by the third postnatal weeks P21 (Movie 1), whereas littermate control mice did not show any of the aforementioned behavioral phenotypes (Movie 2). Behavioral testing demonstrated that Sox10-Sox2 cKO mice displayed severe motor function impairment, evidenced by significantly less retention time on the accelerating rod (129.2 s, non-Cre Sox2fl/fl control mice vs 2.1 s, Sox10-Sox2 cKO mice) (Fig. 7A).
The efficiency of Sox2 ablation in the Sox10-Sox2 cKO CNS was nearly 100% in the oligodendroglial lineage cells, including OPCs (Fig. 7B) and OLs (data not shown) assessed on P21. The density of MBP+ myelin fibers in the Sox10-Sox2 cKO brains was reduced by 50% of that in littermate controls, whereas the density of SMI312+ axons was similar between the two groups (Fig. 7C). Western blotting (Fig. 7D, left) demonstrated that the protein levels of MBP, CNP, and MAG were all significantly decreased in the Sox10-Sox2 cKO brains compared with controls (Fig. 7D, right). In line with the Western blotting data, qRT-PCR quantification showed that the transcription levels of Mbp and mature OL-specific Plp isoform (exon3b containing Plp, Plp-E3b) were reduced by >50% of those in the control brain on P21 (Fig. 7E). Immunohistochemical analysis demonstrated that the numbers of CC1+ mature OLs (Fig. 7F) and PDGFRα+ OPCs were significantly decreased in the corpus callosum of P21 Sox10-Sox2 cKO mice (Fig. 7G). Haploinsufficiency of Sox2 in regulating brain oligodendroglial development was not observed, evidenced by the unaltered numbers of OPCs and OLs in brains between Sox2fl/fl and Sox2 one-allele cKO (Sox10-Cre, Sox2fl/+) mice (Fig. 7G).
Consistent with previous data derived from the embryonic and early postnatal spinal cord (Hoffmann et al., 2014), the distribution of Olig1+ cells was similar between Sox10-Sox2 cKO and Sox2fl/fl littermate controls within the spinal cord cross sections (Fig. 7H), but the density of Olig1+ cells decreased by 30% (625 ± 66/mm2 in Sox10-Sox2 cKO, 430 ± 47/mm2 in controls, two-tailed Student's t test, p = 0.0029, t = 4.82, df = 6) at the weaning age of P21, suggesting a dispensable role of Sox2 in oligodendrocyte migration (Hoffmann et al., 2014) and an essential role in oligodendrocyte production. Our quantification data showed that the densities of PDGFRα+ OPCs and CC1+ OLs were significantly reduced in the Sox10-Sox2 cKO spinal cord than those in Sox2fl/fl controls on P21 (Fig. 7I). Toluidine blue staining of semithin sections (Fig. 7J) and transmission electron microscopic images of ultrathin sections (Fig. 7K) showed that the myelinated axons were substantially fewer in the corticospinal tract of Sox10-Sox2 cKO mutants compared with littermate controls on P28, which is in line with the decreased density of mature OLs in the Sox10-Sox2 cKO spinal cord.
Collectively, the in vivo data derived from the Cnp-Sox2 cKO, Pdgfrα-Sox2 cKO, and Sox10-Sox2 cKO paradigms unequivocally demonstrate that Sox2 plays an essential role in regulating oligodendroglial lineage progression and maturation during brain developmental myelination.
Sox2 expression in oligodendroglial lineage cells during remyelination after chemical-induced and autoimmunity-induced demyelination
We assessed Sox2 expression in both cuprizone-induced demyelination and MOG-peptide35-55-induced EAE animal models (Guo et al., 2011). In the cuprizone-induced demyelinated corpus callosum, the density of Sox2+ cells was substantially increased during the time window of active oligodendrocyte regeneration (Hammond et al., 2015), for example, at 1 week after withdrawal of 6 week cuprizone treatment (6 + 1 week) as shown in Figure 8B, compared with the normal-chow controls (Fig. 8A). Consistent with the observations from developmental myelination (Fig. 1), Sox2 was also expressed in most TCF7l2+ newly regenerated premyelinating OLs (Hammond et al., 2015) (Fig. 8C, arrowheads). Triple immunohistochemistry of Sox2, NG2, and TCF7l2 (Fig. 8D, top) showed that the level of Sox2 in TCF7l2+ newly regenerated OLs was significantly higher than that in NG2+ OPCs in the corpus callosum of mice maintaining on 6 consecutive weeks of 0.25% cuprizone diet (Fig. 8D, bottom). In the MOG-EAE spinal cord in which OPC proliferation and OL regeneration are consistently observed (Tripathi et al., 2010; Guo et al., 2011), more Sox2+TCF7l2+ newly regenerated OLs were observed (Fig. 8F, arrowheads) compared with scarce TCF7l2+ OLs in the CFA control spinal cord (Fig. 8E, arrowheads) at day 21 (D21) post-MOG immunization (Fig. 8G). Similarly, Sox2 expression levels in TCF7l2+ newly regenerated OLs were significantly higher than that in NG2+ OPCs in the D21 spinal cord treated with MOG-peptide35-55 (Fig. 8H,I). These data suggest that Sox2 may play a role in oligodendrocyte regeneration and remyelination during remyelination.
Sox2 is essential for brain remyelination after cuprizone-induced myelin damage
Cuprizone-induced demyelination/remyelination in murine corpus callosum is a well-established animal model for studying molecular mechanisms underlying OL regeneration and myelin repair. Because constitutive Sox2 cKO affected brain developmental myelination, we used the time-conditioned, tamoxifen-inducible Pdgfrα-CreERT2, Sox2fl/fl (Pdgfrα-Sox2 cKO) to study the role of Sox2 in OL regeneration and remyelination.
Adult Pdgfrα-Sox2 cKO and non-Cre Sox2fl/fl control mice were intraperitoneally injected three times of 5 day course of tamoxifen and BrdU starting from the third week of cuprizone diet maintenance (for experimental designs, see Fig. 9A). Brain tissues were analyzed at the end of 1 week after returning to normal diet (Fig. 9A), a time point of active OL differentiation (Hammond et al., 2015). In the adult corpus callosum, all PDGFRα+ OPCs express Sox2 in Sox2fl/fl control mice (Fig. 9B1), and the tamoxifen paradigm used in the study (Fig. 9A) resulted in ∼90% efficiency of Sox2 knock-out in OPCs. As shown in Figure 9B2, all EYFP-fate-mapped, PDGFRα+ OPCs had no detectable Sox2 expression (PDGFRα+Sox2−EYFP+) in the corpus callosum of Pdgfrα-CreERT2, Sox2fl/fl, Rosa-EYFP triple transgenic mice.
One week after returning to normal diet, the densities of CC1+Olig2+ OLs (Fig. 9C) and PDGFRα+ OPCs were significantly diminished in the corpus callosum of Pdgfrα-Sox2 cKO mice compared with non-Cre control (Sox2 WT) mice (Fig. 9D). BrdU is incorporated into proliferative OPCs (but not postmitotic OLs) upon administration, and CC1+BrdU+ OLs are representatives of newly regenerated OLs that inherited BrdU from BrdU+ OPCs during the time window between BrdU treatment and tissue analysis. Our quantification data showed that CC1+BrdU+ OLs were fewer in Pdgfrα-Sox2 cKO mice than that in non-Cre Sox2fl/fl controls (Fig. 9D). We also observed that Pdgfrα-Sox2 cKO resulted in fewer Ki67+PDGFRα+ proliferative OPCs (Fig. 9D). Using APC and TCF7l2 to label newly regenerated OLs (Fig. 9E) (Hammond et al., 2015), we found that the density of APC+TCF7l2+ newly regenerated OLs was significantly lower in cuprizone-treated corpus callosum of Pdgfrα-Sox2 cKO mice than that of non-Cre control mice (Fig. 9F). Immunohistochemistry of MBP and pan axonal marker SMI312 (Fig. 9G) showed that MBP+ myelin density was reduced by 50% in the Pdgfrα-Sox2 cKO corpus callosum, whereas SMI312+ axonal density was indistinguishable (Fig. 9H). Together, our data suggest that Sox2 is required for remyelination in the adult corpus callosum after chemical-induced demyelination.
Sox2 regulates OPC proliferation and OL regeneration in the spinal cord after inflammation-induced demyelination
In the MOG-peptide35-55-induced EAE model, inflammation-induced demyelination in the spinal cord elicits a robust augment of OPC proliferation and a modest increase of oligodendrocyte regeneration (Tripathi et al., 2010), although the extent of remyelination is much less than that in cuprizone-induced demyelination model (Jones et al., 2008; Constantinescu et al., 2011). Therefore, we used MOG-peptide35-55-EAE animal model and Pdgfrα-CreERT2, Sox2fl/fl transgenic mice to study the role of Sox2 in regulating OPC proliferation and OL regeneration in the spinal cord after inflammation-induced demyelination.
We showed that all NG2+ OPCs in the adult spinal cord expressed Sox2 (Fig. 10B1). To avoid potential effects of Sox2 ablation before injury on OPC response to subsequent inflammation, we administered tamoxifen Pdgfrα-Sox2 cKO and non-Cre Sox2fl/fl (Sox2 WT) mice after massive CNS inflammation occurred, typically at day 9–12 after MOG immunization (Fig. 10A). Sox2 was ablated in virtually all NG2+ OPCs in the Pdgfrα-Sox2 cKO spinal cord (Fig. 10B3, arrowhead and boxed area), in sharp contrast to Sox2 WT spinal cord in which all NG2+ OPCs had Sox2 expression (Fig. 10B2, arrowheads and boxed area). In agreement of our previous report (Guo et al., 2011), MOG-peptide immunization resulted in a fourfold increase in the density of Sox10+NG2+ OPCs (Fig. 10C1) in the lumbar spinal cord of Sox2 WT mice treated with MOG (Sox2 WT + MOG) compared with Sox2 WT mice treated with CFA control (Sox2 WT + CFA) (Fig. 10C3). However, Sox2 deletion significantly reduced the OPC density in the spinal cord of Pdgfrα-Sox2 cKO mice treated with MOG (Sox2 cKO + MOG) (Fig. 10C2,C3), indicating that Sox2 is required for OPC population expansion after inflammatory insults. Notably, the similar distribution patterns of OPCs within inflammatory lesions were observed in the spinal cords between Pdgfrα-Sox2 cKO and Sox2 WT mice with MOG (Fig. 10B2 vs B3, C1 vs C2), indicating that Sox2 appears dispensable for OPC recruitment into the inflammation-induced demyelination lesions. The number of Ki67+Olig2+ proliferating OPCs (Fig. 10D1,D2) was significantly decreased in the spinal cord of Pdgfrα-Sox2 cKO mice treated with MOG (Sox2 cKO + MOG) compared with Sox2 WT mice treated with MOG (Sox2 WT + MOG) (Fig. 10E).
Our previous study reports that the number of TCF7l2+APC+ newly regenerated OLs increases in the spinal cord of MOG treatment (Fig. 10F) (Hammond et al., 2015). Nevertheless, Sox2 deletion resulted in significant decrease in the generation of TCF7l2+APC+ OLs in the spinal cord of Pdgfrα-Sox2 cKO mutants treated with MOG (Fig. 10F). Together, these data suggest that Sox2 is essential for OPC proliferation and OL regeneration in the spinal cord after inflammation-induced demyelination.
Discussion
There are several novel findings in this study: (1) Sox2 is upregulated in newly differentiated OLs during developmental myelination and in newly regenerated OLs during remyelination; (2) Sox2 is essential for brain developmental myelination by regulating OPC proliferation and OL differentiation; and (3) in the context of myelin repair, Sox2 is required for OPC proliferation and/or OL regeneration not only in autoimmune-induced spinal cord demyelination lesions but also in chemical-induced brain demyelination lesions.
In this study, we found that OPCs retain Sox2 expression in the adult murine CNS. Previous study reports that Sox2+Olig2+ cells (presumably Sox2+ OPCs) can be immunohistochemically detected in the adult human white matter and that these Sox2/Olig2-positive cells can be differentiated into mature oligodendrocytes in vitro (Oliver-De La Cruz et al., 2014). This report, together with our finding in the murine CNS, suggests that Sox2 expression in adult OPCs is well conserved from rodent to human CNS. However, the function of Sox2 in adult OPCs under normal conditions has not been defined. Our study demonstrates that Sox2 is required for OPC proliferation and population expansion during developmental myelination (Fig. 5A–F). Inspired by these data, we hypothesize that Sox2 is required for adult OPC homeostasis and brain myelin turnover and/or remodeling (Zhang et al., 1999; Yeung et al., 2014) and is essential for OPCs' response to various demyelinating insults, the latter of which is in agreement with a recent report showing that Sox2 deletion before lysolecithin-induced focal demyelination reduces progenitor response and recruitment to demyelinating lesions (Zhao et al., 2015).
Interestingly, we also found that Sox2 is upregulated in a subpopulation of differentiated OLs during developmental myelination and remyelination in the rodent CNS. This upregulation is counterintuitive to the established concept that Sox2 maintains the “stemness” of neural stem/progenitor cells and is downregulated upon progenitor differentiation (Graham et al., 2003). Specifically, Sox2 is transiently upregulated in the differentiated oligodendrocytes that are TCF7l2-positive, which we have previously identified as postmitotic, newly formed (regenerated) OLs (Hammond et al., 2015). Although the number of Sox2+ differentiated OLs are reduced over time during developmental myelination, the percentage of Sox2+ cells among TCF7l2+ newly formed OLs did not change (>90% at all time points we assessed). Given the asynchronous properties of in vivo OL differentiation, our quantitative data suggest that all OLs express Sox2 at certain developmental stages during OL lineage progression. Sox2 upregulation is necessary for OL lineage progression and maturation, as Cnp-Sox2 cKO resulted in diminished myelin formation (Fig. 2) and OL differentiation (Fig. 3) in the brain. From a pretranslational perspective, it would be very important and interesting to test whether dose-controlled, transient Sox2 overexpression in postmitotic OPCs and/or differentiating OLs is sufficient to promote OL differentiation and/or (re)myelination.
Our study suggests a previous unappreciated concept that Sox2 regulates CNS myelin formation and repair in CNS region and context-dependent manners. Using oligodendroglial-specific Sox2 cKO mice, Hoffmann et al. (2014) demonstrate that Sox2 is dispensable for OPC proliferation and migration; instead, it is required for OL differentiation in the embryonic and perinatal spinal cord (P3). More recently, Zhao et al. (2015) used the chicken β-actin promoter and CMV enhancer-driven Cre to ubiquitously ablate Sox2 (Cag-CreERT2, Sox2fl/fl) and found that Sox2 plays a minor, if any, role in OL differentiation and myelination in the postnatal spinal cord. In our study, we used Cnp-Cre to conditionally ablate Sox2 in the later stages of OL development (Dugas et al., 2010; Moyon et al., 2016; Zhao et al., 2016) and found a CNS region-dependent role of Sox2 in regulating OL differentiation: it is required for brain OL differentiation (Fig. 3) and appears to play a minor role in OL differentiation (or regulate the timing of OL differentiation) during developmental myelination in the postnatal spinal cord (Fig. 4). The mechanisms underlying the CNS region-dependent role of Sox2 are unclear. Previous study shows that another SoxB1 family member Sox3 functions redundantly with Sox2 in regulating spinal cord OL differentiation (Hoffmann et al., 2014). It is plausible that Sox3 may compensate the loss of Sox2 during postnatal spinal cord development. However, the compensatory effects of Sox3 on Sox2 loss of function are less likely to occur in the brain myelination, as Sox10-Cre, Sox2fl/fl mutant mice display severe defects in motor function, OL differentiation, and brain developmental myelination (Fig. 7). Our results also indicate that, in the context of remyelination, Sox2 is required for myelin repair in both adult brain (Fig. 9) and spinal cord (Fig. 10) through regulating OPC population supply and/or OL differentiation. Our remyelination study did not provide definitive evidence that Sox2 is required for remyelination by directly regulating OL differentiation, as it does in brain myelination. In this regard, time-conditioned, stage-specific Sox2 cKO or conditional overexpression paradigms are needed to unequivocally define the role of Sox2 in regulating OL differentiation itself after myelin damage.
Oligodendrocyte number was not affected in the spinal cord of the Cnp-Sox2 cKO mutants at P14 (Fig. 4) but was impaired in the spinal cord of the Sox10-Sox2 cKO mutants even at later ages (Fig. 7H–K). This discrepancy presumably reflects the stage-specific Sox2 cKO in oligodendroglial lineage cells. In the Sox10-Sox2 cKO mutants, Sox2 was initially ablated in the upstream proliferating OPCs, whereas in the Cnp-Sox2 cKO mice, primarily in the differentiation-committed, late-stage OPCs and OLs. The reduced number of differentiated OL observed in Sox10-Sox2 cKO spinal cord is likely due to the reduced proliferation and supply of OPCs as we noticed diminished density of OPCs in the Sox10-Sox2 cKO (but not in Cnp-Sox2 cKO at P14) spinal cord. These data are in line with our conclusion that Sox2 is required for OPC proliferation and population expansion.
The working model in which Sox2 coordinates OPC proliferation and OL differentiation is compatible with the one derived from NSPCs (Episkopou, 2005) and with the expression patterns of Sox2 along the progression of oligodendroglial lineage (Fig. 1M). Previous studies have documented that Sox2 genetic knock-out (or knockdown) diminishes the proliferation of NSPCs under self-renewal conditions and reduces neuronal differentiation from NSPCs under neurogenic conditions (Ferri et al., 2004; Cimadamore et al., 2013). Mechanistically, Sox2 interacts with Chd7, a chromatin remodeler to directly regulate a variety of downstream signaling pathways, including Shh and Notch signaling pathways in NSPCs (Engelen et al., 2011), both of which are important regulators of oligodendroglial development (He and Lu, 2013; Wheeler and Fuss, 2016). Interestingly, a recent study shows that Chd7 regulates the onset of CNS myelination and remyelination through interacting with Sox10 (He et al., 2016). Biochemical evidence suggests that Sox2 interacts with Sox10 to regulate Schwann cell development in the peripheral nervous system (Arter and Wegner, 2015). Our unpublished data show that Sox2 is colabeled with Chd7 in oligodendroglial lineage cells at the histological level and displays similar expression dynamics to that of Chd7 (He et al., 2016). Therefore, it is tempting to hypothesize that Sox2 may regulate brain developmental myelination and CNS remyelination through binding to Chd7-Sox10 and/or through the downstream signaling pathways targeted by the putative Sox2-Chd7-Sox10 complex. Further studies are needed to support or falsify this hypothesis.
In conclusion, a series of genetic experiments in our study demonstrate that Sox2 plays a crucial role in regulating OPC proliferation and OL differentiation during postnatal brain development, although it appears to play a minor, transient role in OL differentiation during postnatal spinal cord development. Furthermore, we show that Sox2 is required for myelin repair by regulating OPC proliferation and/or OL differentiation in both the demyelinated spinal cord and brain, suggesting a common mechanism. Our study suggests that Sox2 may be a therapeutic target that can be experimentally manipulated to promote OL regeneration and CNS remyelination. Considering the established protocols of expressing Sox2 in the stem cell biology field, it would be interesting to investigate whether Sox2 expression through genetic and/or viral-mediated approaches is sufficient to promote endogenous CNS remyelination after myelin damage.
Footnotes
This work was supported by the National Institutes of Health Grants R01NS094559 and R21NS093559 to F.G., National Institutes of Health Grant R01NS100761 to A.W., and Shriners Hospitals for Children research grants to F.G. and postdoctoral fellowship grant to S.Z. We thank Dr. Q. Richard Lu (University of Cincinnati, Cincinnati) for critical comments on the manuscript; and Daffcar Erol (University of California–Davis) for the editions.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Fuzheng Guo, Department of Neurology, University of California–Davis School of Medicine, c/o Shriners Hospitals for Children, 2425 Stockton Blvd, Sacramento, CA 95817. fzguo{at}ucdavis.edu