Oligodendrocytes are the myelin-forming cells of the CNS. They differentiate from oligodendrocyte precursor cells (OPCs) that are produced from progenitors throughout life but more actively during the neonatal period and in response to demyelinating insults. An accurate regulation of oligodendrogenesis is required to generate oligodendrocytes during these developmental or repair processes. We hypothesized that this regulation implicates transcription factors, which are expressed by OPCs and/or their progenitors. Ascl1/Mash1 is a proneural transcription factor previously implicated in embryonic oligodendrogenesis and operating in genetic interaction with Olig2, an essential transcriptional regulator in oligodendrocyte development. Herein, we have investigated the contribution of Ascl1 to oligodendrocyte development and remyelination in the postnatal cortex. During the neonatal period, Ascl1 expression was detected in progenitors of the cortical subventricular zone and in cortical OPCs. Different genetic approaches to delete Ascl1 in cortical progenitors or OPCs reduced neonatal oligodendrogenesis, showing that Ascl1 positively regulated both OPC specification from subventricular zone progenitors as well as the balance between OPC differentiation and proliferation. Examination of remyelination processes, both in the mouse model for focal demyelination of the corpus callosum and in multiple sclerosis lesions in humans, indicated that Ascl1 activity was upregulated along with increased oligodendrogenesis observed in remyelinating lesions. Additional genetic evidence indicated that remyelinating oligodendrocytes derived from Ascl1+ progenitors/OPCs and that Ascl1 was required for proper remyelination. Together, our results show that Ascl1 function modulates multiple steps of OPC development in the postnatal brain and in response to demyelinating insults.
During development, the basic-helix-loop-helix transcription factor (TF) Olig2 is both necessary and, in some contexts, sufficient for oligodendrocyte precursor cell (OPC) generation (Lu et al., 2001; Zhou et al., 2001; Zhou and Anderson, 2002; Yue et al., 2006; Zhu et al., 2012). Despite this key role, Olig2 is however not a master gene, and it acts in concert with additional cofactors to regulate oligodendrocyte specification. For example, Olig2 must interact with signaling pathways controlling its phosphorylation state to acquire an oligodendrogenic activity (Li et al., 2011). The multiple focal pattern of oligodendrogenesis in the embryonic brain also suggests that Olig2 interacts with multiple TFs providing positional cues and regulating locally its action in progenitor cells (Spassky et al., 1998; Tekki-Kessaris et al., 2001; Miguez et al., 2012).
Ascl1 is another basic-helix-loop-helix factor playing a major role at different stages of neurogenesis and having potent proneural and reprogramming activities (Bertrand et al., 2002; Vierbuchen et al., 2010; Vierbuchen and Wernig, 2012). Ascl1 is also expressed by embryonic and neonatal OPCs (Kondo and Raff, 2000; Wang et al., 2001; Parras et al., 2004; Parras et al., 2007). Ascl1 operates in genetic interaction with Olig2 during OPC specification in the embryonic telencephalon, and the loss of Ascl1 function reduces embryonic oligodendrogenesis (Parras et al., 2007; Sugimori et al., 2007). Therefore, beyond its role in neurogenesis, Ascl1 is involved in the regulation of oligodendrogenesis. The importance of Ascl1 action in postnatal oligodendrogenesis and myelination as well as its requirement for proper myelin repair have however not yet been determined.
To ascertain its involvement in oligodendrogenesis during myelination and remyelination, we have examined Ascl1 expression and function in oligodendroglial cells during the postnatal period of development and during myelin repair in the adult brain. At all ages examined, Ascl1 expression was restricted to immature cell types (i.e., cortical subventricular zone [SVZ] progenitor cells and OPCs of gray matter and white matter). Using different genetic models to delete Ascl1 in progenitor cells or OPCs, and to trace the Ascl1-cell lineage, we have characterized Ascl1 requirement in oligodendrogenesis both under normal physiological conditions and after demyelinating lesions. We found that Ascl1 was required in SVZ progenitors for oligodendroglial cell fate decision as well as in OPCs for their proper differentiation into oligodendrocytes. After demyelinating lesions, Ascl1 was expressed by intralesional SVZ progenitors and OPCs, which regenerated myelin-forming cells and thus participated in myelin repair of demyelinated lesions. Although Ascl1 was dispensable for recruitment and maturation of intralesional progenitors and OPCs, we showed that it was required for a proper balance between the number of OPCs and oligodendrocytes. Finally, examination of human periventricular multiple sclerosis (MS) lesions confirmed that Ascl1 expression is a hallmark of OPCs involved in myelin repair, both in mice and humans. Together, our findings support a promoting role for Ascl1 during postnatal oligodendrogenesis and in the process of remyelination, both in progenitors for OPC specification as well as in OPCs for their differentiation.
Materials and Methods
Mouse breeding and genotyping.
The following transgenic lines were used and genotyped using protocols previously described: Ascl1Δ (Casarosa et al., 1999), Ascl1fl (Pacary et al., 2011), Ascl1::Cre, Ascl1::GFP (Parras et al., 2007), Rosa26stop-YFP (Srinivas et al., 2001), Rosa26stop-LacZ (Soriano, 1999), and PDGFRα::CreERT2 (Rivers et al., 2008). For simplicity, we renamed the different mice as follows: Ascl1GFP (for Ascl1::GFP), Ascl1Cre/YFP (for Ascl1::Cre; Rosa26stop-YFP), Ascl1Cre/LacZ (for Ascl1::Cre; Rosa26stop-LacZ), CkoOPC (for conditional knock-out in OPCs: PDGFRα::CreERT2; Ascl1fl/Δ; Rosa26stop-YFP), CtrOPC (for control for CkoOPC: PDGFRα::CreERT2; Ascl1fl/+; Rosa26stop-YFP), CkodSVZ (for dorsal Cre-electroporated Ascl1fl/Δ; Rosa26stop-YFP), and CtrdSVZ (for dorsal Cre-electroporated Ascl1fl/+; Rosa26stop-YFP). Both males and females were included in the study. All experiments were carried out in accordance to Inserm ethical committees (authorization #A75-13-19) and personal animal experimentation license A75-17-72 (C.P.).
Demyelinating mouse lesions.
Before surgery, adult (4–6 months) Ascl1GFP, Ascl1Cre/YFP, amd Ascl1Cre/LacZ or Cko/CtrOPC mice were weighted and anesthetized by intraperitoneal injection of mixture of ketamine (0.1 mg/g) and xylacine (0.01 mg/g). An analgesic (buprenorphine, 30 mg/g) was administered intraperitoneally to prevent postsurgical pain. Focal demyelinating lesions were induced by stereotaxic injection of 1.5 μl lysolecithin solution (LPC, Sigma, 1% in 0.9% NaCl) into the corpus callosum (CC; at coordinates: 1 mm lateral, 1.3 mm rostral to bregma, 1.7 mm deep to brain surface) using a glass-capillary connected to a 10 μl Hamilton syringe. Animals were left to recover in a warm chamber before being returned into their housing cages.
Postnatal brain electroporation (Boutin et al., 2008) was adapted to target the dorsal SVZ. Briefly, postnatal day 2 (P2) pups were cryoanesthetized for 3 min on ice, and 2 μl of plasmid solution (3 μg pCX-Cre) was injected into the ventricle using a glass capillary. pCX-Cre plasmid drive Cre expression under the CMV-modified strong and ubiquitous promoter (kindly gift from Alain Chedotal, Institut de la Vision, Paris). Electrodes (Nepagene CUY650P10) were positioned in the dorsoventral axis with the positive pole dorsal. Five electric pulses of 100 V, 50 ms pulse ON, 850 ms pulse OFF were applied using a Nepagene CUY21-SC electroporator. Pups were immediately warmed up in a heating chamber and brought to their cages at the end of the experiment.
Tamoxifen (Sigma) was dissolved in corn oil (Sigma) at 20 mg/ml. Postnatal mice received one subcutaneous injection of 400 μg (P5) or one intraperitoneal injection of 800 μg (P8/P9). Adult animals received one intraperitoneal injection per day of 200 μg for 5 consecutive days starting 7 d before LPC lesion.
MS and non-neurological control tissues.
Autopsy brain tissue samples from 4 patients with confirmed secondary progressive MS and two patients without neurological diseases (Table 1) were obtained from the United Kingdom MS tissue bank (Richards Reynolds, Imperial College, London). MS and control brain samples were excised from matching periventricular regions or white matter. Four MS tissue blocks containing active (n = 3) and chronic active lesions (n = 3), periventricular lesions, and periplaque white matter were selected for further analysis. MS lesions were classified as previously reported (Lassmann, 1998).
Ascl1Cre/LacZ mice (P120) were perfused with 4% PFA and 0.5% glutaraldehyde in PBS buffer. Brains were sectioned (300 μm) with a Tissue chopper (McIlwain) and subjected overnight to the X-gal reaction. Tissue was postfixed in 2.5% glutaraldehyde for 2 h. Regions containing the CC were dissected out under a binocular scope (MZ16, Leica). Tissue samples were osmicated, dehydrated in ethanol, and embedded in Epon. Ultrathin sections were made using a Leica electronic microscopy (EM) UC7 ultramicrotome, and micrographs were taken with a Philips CM120 electronic microscope.
We optimized immunostaining for Ascl1, which is very sensitive to fixation conditions. Mice were perfused with freshly made chilled 2% paraformaldehyde (Sigma) with the following amounts (P0-P10, 15 ml; P10-P16, 20 ml; adult, 25 ml), brain was dissected out, cryoprotected in PBS with 20% sucrose overnight, and included in OCT (BDH) before freezing and sectioning (14 μm thick) with a cryostated microtome (Leica). Sections were then either processed for immunohistochemistry or stored at −80°C. P0-P16 brain sections were postfixed with 4% PFA for 10 min. Sections were incubated in a blocking solution (PBS plus 10% normal goat serum, Vector Laboratories; and 0.1% Tween 20 or Triton X-100) and then with primary antibodies overnight at 4°C. The following primary antibodies were used: mouse monoclonal antibodies to APC (Calbiochem, 1:200), Ascl1 (BD Biosciences, 1:200), CNP (Millipore, 1:500), MOG (8-18C5, 1:5, Dr. C. Linington, Max-Planck Institute, Tuebingen, Germany), Olig2 (Ligon et al., 2006), Nkx2.2 (clone 74.5A5, Developmental Studies Hybridoma Bank, 1:5); rabbit polyclonal antibodies to Caspase3 activated (R&D Systems, 1:200), pan-Dlx (gift from G. Panganiban, 1:100), GFP (Invitrogen, 1:2000), Gpr17 (Cayman Chemical, 1:400), Gsx2 (gift from K. Campbell, 1:1000), Ki67 (Novocastra, 1:1000), myelin basic protein (MBP, Millipore, 1:1000), NG2 (Millipore, 1:1000), Olig1 (Mizuguchi et al., 2001), Olig2 (Millipore, 1:1000), Sox9 (R&D Systems, 1:500); guinea pig antibodies to Sox10 (gift from M. Wegner, 1:1000), Ascl1 (gift from J. Johnson, 1:1000); rat monoclonal antibodies to Ki67 (Dako, 1:50), PDGFRα (BD Biosciences, 1:800), and chicken polyclonal antibodies to GFP (Aves Laboratories, 1:2000) and MBP (Millipore, 1:50). Fluorescent secondary antibodies included the following: AlexaFluor-488, AlexaFluor-594, and AlexaFluor-647 Ig (1:1000, Invitrogen). Antigen retrieval for Ki67 antibody staining was performed using antigen retrieval solution (Vector, H-330), heating sections in citrate-based solution up to boiling and then allowing cooling down at room temperature for 20 min. Immunofluorescence was visualized with Leica TC SP2 confocal or Zeiss AxioImager-Apotome system microscopes. Pictures were taken as stacks of 5–10 μm with 0.5 μm between sections. Z-projections and orthogonal projections were done in ImageJ and processed with Adobe Photoshop. Figures were made using Adobe Illustrator.
Human brain tissues were immunostained as previously described (Nait-Oumesmar et al., 2007) with mouse monoclonal antibodies for human Ascl1 (Cosmo Bio, 1:500), CD68 (clone KP1, Dako, 1:100), rabbit antibodies for Olig2 (Millipore, 1:100), Olig1 (R&D Systems, 1:50), and goat antibody for Sox10 (R&D Systems, 1:50).
Quantification and statistical analysis.
Confocal images were quantified for expression of TFs by manual or automated counting using ImageJ software. To quantify Ascl1 staining intensity at different stages studied (P2, P14, adult control, and adult 5 d after LPC lesion), confocal pictures of brain sections, immunolabeled for Ascl1 and PDGFRα, were analyzed with ImageJ software to measure the mean nuclear Ascl1 intensity in CC OPCs (PDGFRα+ cells) versus SVZ (PDGFRα− cells). Mean and SEM values obtained counting several sections (n = 3–5) are shown as the ratio of Ascl1 intensity in CC OPCs versus SVZ cells. To estimate levels of TF expression at different stages of OPC differentiation, mean values of their relative intensity in cells at each stage were obtained using ImageJ from images (n = 3), such as those shown in Figure 2. Data are presented as a line chart in Figure 2J. Values range from 0 to 255, with values <50 considered as background levels. All values are shown as mean ± SEM. Statistical analysis was performed using Student's t test in Prism 4, GraphPad Software. p values in graphic are represented as follows: *p < 0.05, **p < 0.01, ***p < 0.001. Black bars in histograms correspond to SEM.
Immature oligodendroglial cells of neonatal cortex express Ascl1
Although Ascl1 is required for OPC generation in the embryonic forebrain and spinal cord (Parras et al., 2007; Sugimori et al., 2007), its roles during neonatal oligodendrogenesis and myelination are currently unknown. Postnatal OPCs derive from the cortical wall of lateral ventricles both in neonates (Gorski et al., 2002; Ivanova et al., 2003; Kessaris et al., 2006) and adults (Menn et al., 2006). To address Ascl1 function in postnatal oligodendrogenesis and myelination, we first examined its cortical expression in newborn mice (P2) by immunofluorescence. Ascl1 was detected in subsets of cells located in the cortical ventricle wall and in the CC (∼15% and 25% of total cells, respectively). The phenotype of Ascl1-expressing cells was characterized using antibodies against TFs expressed by oligodendroglial progenitors and precursors, including Olig1, Olig2, Sox9, Sox10, and Nkx2.2 (Qi et al., 2001; Sun et al., 2001; Zhou et al., 2001; Zhou and Anderson, 2002; Stolt et al., 2003, 2004; Arnett et al., 2004; Kitada and Rowitch, 2006; Liu et al., 2007; Cai et al., 2010). In the dorsal ventricle wall, Ascl1+ cells expressed Olig2 (85%; Fig. 1A,G) and Sox9 (80%; Fig. 1C,G), two TFs expressed by immature neural progenitor/stem cells (Hack et al., 2004; Scott et al., 2010). The majority of Ascl1+/Olig2+/Sox9+ cells expressed neither PDGFRα, the most specific marker for OPCs (Hall et al., 1996; Kessaris et al., 2006), nor other OPC markers, such as NG2 (data not shown), or TFs expressed in OPCs, such as Olig1, Sox10, and Nkx2.2 TFs (Fig. 1B,D,E,G). In the CC, however, Ascl1+ cells included equal populations (∼50%) of Olig2+/Sox9+/PDGFRα− progenitor cells (Fig. 1A–E,G) and PDGFRα+ cells (Fig. 1G,H). It is noteworthy that all Ascl1+/PDGFRα+ cells (∼10% of CC cells and ∼2% of SVZ cells) also expressed Olig2 (Fig. 1A,G,H) and both nuclear and cytoplasmic Olig1 (Fig. 1B,G,H), as well as Sox10 (Fig. 1D,G,H), Nkx2.2 (Fig. 1E,G,H), and Sox9 (Fig. 1C,C1,H). This phenotype suggests that all Ascl1+/PDGFRα+ cells are OPCs and a fraction of Ascl1+/PDGFRα− cells may correspond to immature neural progenitors in the process of generating OPCs (Fig. 1G,H). Supporting this possibility, we found that ∼48% of Ascl1+/PDGFRα− cells do not express either Dlx or Gsx2, two neurogenic cofactors (Petryniak et al., 2007; Waclaw et al., 2009) that are not expressed in OPCs (Fig. 1F–H). Therefore, Ascl1+/Olig2+/PDGFRα−/Dlx−/Gsx2− cells (corresponding to 5–7% of SVZ or CC cells) are likely immature oligodendroglial cells.
Ascl1 in OPC is downregulated upon oligodendrocyte differentiation
We then investigated Ascl1 expression during OPC transition into myelinating oligodendrocytes. In the CC, when many OPCs start to differentiate during the second postnatal week, Ascl1 is still expressed in all PDGFRα+ OPCs (Fig. 2A), but APC/CC1-expressing maturing oligodendrocytes (Bhat et al., 1996) were rarely Ascl1+ (Fig. 2B,B1) suggesting that maturing OPCs downregulate Ascl1 expression. Indeed, using the Ascl1GFP transgenic line, in which GFP remains expressed longer than Ascl1, we found that most APC+ cells at this stage were GFP+ (Fig. 2B).
To assess Ascl1 coexpression with other TFs involved in OPC differentiation, we performed immunofluorescence for different TF combinations and for markers of different stages of oligodendroglial lineage, including Gpr17, which is expressed during OPC differentiation (Chen et al., 2009; Boda et al., 2011). At P14, when many oligodendroglial cells are at different stages of differentiation in the CC, Ascl1 is coexpressed with Sox9, Sox10, Olig2, Olig1, and Nkx2.2 in PDGFRα+ cells (OPCs; labeled #1 in Fig. 2C,D,J,K). According to the expression levels of these TFs and oligodendroglial markers, we could divide premyelinating oligodendrocytes in two stages (Fig. 2C–K): (1) premyelinating oligodendrocyte stage I (mOLI) characterized by downregulation of Ascl1, Olig1, and PDGFRα and upregulation of Nkx2.2/Gpr17/APC/Sox9/Sox10 (labeled #2 in Fig. 2C–K); and (2) premyelinating oligodendrocyte stage II (pre-mOLII), implicating the downregulation of Nkx2.2/Gpr17 and upregulation of Olig1-cytoplasmic expression while presenting the highest levels of Sox9/Sox10/Olig2 (labeled #3 in Fig. 2C–K). Ascl1GFP transgene was still expressed at these premyelinating stages characterized by coexpression of high levels of Nkx2.2/Olig2 or Sox10/Olig2, indicating that Ascl1-expressing OPCs do differentiate into maturing oligodendrocytes (Fig. 2I,K). Eventually, APC cells starting to express myelin proteins, such as MOG, present a larger cytoplasm and intermediate levels of Olig2/Sox9/Sox10 expression with persistent Olig1 cytoplasmic expression (labeled #4 in Fig. 2H,J,K). Therefore, combinatorial expression of Ascl1 and other oligodendrogenic factors defines more precisely the successive stages of oligodendrogenesis. The initial stage of OPC differentiation is marked by a transient upregulation of Nkx2.2/Gpr17 correlating with a downregulation of Ascl1/Olig1 (Fig. 2J,K), whereas further steps of transition toward a myelin-forming cell correlate with Nkx2.2/Gpr17 downregulation together with Olig1 cytoplasmic expression (Fig. 2J,K).
Ascl1 is required for OPC generation from SVZ progenitor cells
Given the expression of Ascl1 in SVZ-progenitors and in OPCs during the postnatal wave of oligodendrogenesis described above, we first investigated its role in cortical SVZ progenitors and neonatal oligodendrogenesis. For this purpose, we deleted Ascl1 in dorsal SVZ cells in a temporally controlled manner by targeting a Cre-expressing plasmid to the dorsal SVZ at P2 of Ascl1fl/Δ; Rosa26stop-YFP (mutant, CkodSVZ) or Ascl1fl/+; Rosa26stop-YFP (control, CtrdSVZ) mice, using a modified postnatal electroporation technique (Boutin et al., 2008). In control experiments using a GFP-expressing plasmid, many GFP+ SVZ cells showed radial glial-like processes, thus demonstrating that dorsal neural progenitor/stem cells were successfully targeted by our electroporation technique (Fig. 3B). In brains electroporated with Cre-expressing plasmids, Ascl1 protein was detected only in CtrdSVZ but not in CkodSVZ cells (Fig. 3C,C1,C′,C′1), indicating that Ascl1 could be efficiently deleted by Cre recombination. To assess the fate of electroporated cells, we analyzed the brains at 30 d postelectroporation (dpe). The majority of CtrdSVZ-YFP+ cells were oligodendrocytes (62.5 ± 5.6%, n = 5, APChigh), some were astrocytes (32.1 ± 5.9%, n = 5, GFAP+/APClow), and few were OPCs (9.3 ± 0.8%, n = 3, PDGFRα+) (Fig. 3D,D1,E). In contrast, CkodSVZ-YFP+ cells were mostly astrocytes (82.9 ± 4.1%, 2.7-fold), only 15.3 ± 1.8% were oligodendrocytes (3.9-fold reduction), and 1.6 ± 0.5% were OPCs (5.5-fold reduction) (Fig. 3D′,D′1,E), indicating that loss of Ascl1 inhibits oligodendrogenesis for the benefit of astrogenesis. We did not observe a selective cell death of Ascl1-deficient cells at 3 dpe (as shown by the absence of activated caspase3 expression; data not shown). There was also a 1.7-fold increase in the number of Ascl1-deficient GFAP+ cells in the SVZ compared with controls (88.9 ± 11.1% vs 48.3 ± 4.7%; p = 0.0082, t test, n = 4; Fig. 3D,D1). Ascl1 is therefore required for the proper generation of oligodendrocytes from the cortical SVZ; and, in the absence of Ascl1, the SVZ progenitor's fate is predominantly astroglial.
Ascl1 promotes the balance between OPC differentiation and proliferation
We next examined Ascl1 function specifically in neonatal OPCs by performing a temporal and conditional Ascl1 deletion in OPCs of CkoOPC and CtrOPC genotypes using PDGFRα::CreERT2 mice (Rivers et al., 2008; see Materials and Methods). We induced Ascl1 deletion at early postnatal stages when OPC numbers peak in the brain (tamoxifen injection at P5 or P8-P9) and assessed their proliferative and differentiation capacity. Ki67 expression in OPCs was used as marker of proliferative OPCs. Seven days post-tamoxifen injection (dpi) CkoOPC YFP+ OPCs showed a 1.7-fold increase in the proportion of Ki67+ cells (36.4 ± 5.2%, Fig. 4A′,D) compared with CtrOPC (21.3 ± 0.7%, Fig. 4A,D), indicating an increase of OPC proliferation in the absence of Ascl1. Similar results were obtained when tamoxifen was injected at P5 or P8-P9 (Fig. 4D). Therefore, OPC proliferation increases in the absence of Ascl1.
Neonatal OPCs have been shown to divide either symmetrically, generating pairs of OPCs (OPC/OPC pairs) or mature oligodendrocytes (APC+ cells, named OL/OL pairs), or asymmetrically to generate one OPC and one oligodendrocyte (OPC/OL pairs) (Sugiarto et al., 2011; Zhu et al., 2011). To assess whether the difference in proliferation rate observed in the absence of Ascl1 could result from a change in OPC mode of division, we quantified the different types of YFP+ cell pairs in the CC at 7 dpi. CtrOPC (tamoxifen at P5) showed 23.1 ± 4.0% of OPC/OPC pairs, 29.0 ± 3.5% of OPC/OL pairs, and 47.8 ± 3.7% of OL/OL pairs (Fig. 4A,A1,A2,F). In CkoOPC, we observed a twofold increase in OPC/OPC pairs (48.4 ± 2.7%; Fig. 4A′,A′1,F), a twofold decrease in OPC/OL pairs (14.6 ± 3.1%), and no change in numbers of OL/OL pairs (37.0 ± 5.9%; Fig. 4A′,A′2,F), suggesting that, in the absence of Ascl1, OPCs that would divide asymmetrically (OPC/OL pairs) divide instead symmetrically to generate two OPCs (Fig. 4F). Similar results were obtained when tamoxifen was injected at P5 or P8-P9 (Fig. 4F). Ascl1 thus regulates and promotes asymmetric divisions of OPCs. Consequently, the ratio of OPC/oligodendrocytes within the YFP+ population was increased in CkoOPC cells (1.4-fold) compared with CtrOPC cells (CkoOPC: 57.1 ± 1.5% APC+ cells vs 42.8 ± 1.5% PDGFRα+ cells; n = 5; CtrOPC: 47.6 ± 1.8% APC+ cells vs 52.4 ± 1.5% PDGFRα+ cells; n = 3, Fig. 4E), indicating that loss of Ascl1 in OPCs favors proliferation over differentiation. The differentiation defect of Ascl1-mutant cells appeared more pronounced regarding their expression of myelin proteins, such as CNP (Fig. 4C–C′2,E). To assess whether these differences were maintained at later stages, we studied the proportion of OPCs and oligodendrocytes present at 28 dpi. A similar ratio of 4:1 oligodendrocytes per OPC was found in CkoOPC and CtrOPC cells (CkoOPC: 79.6 ± 0.9% APC+ cells vs 20.4 ± 0.8% PDGFRα+ cells, n = 5; CtrOPC: 79.8 ± 1.8% APC+ cells vs 20.1 ± 1.8% PDGFRα+ cells, n = 4; Fig. 4B,B′,G), suggesting that other factors regulating OPC proliferation and differentiation compensate for the loss of Ascl1 function over time.
Demyelination upregulates Ascl1 levels in adult OPCs
Because Ascl1 expression is lost with oligodendrocyte differentiation, we then extended our study to adulthood to determine whether Ascl1 was still expressed in OPCs under normal and demyelinating conditions. Using an optimized protocol for Ascl1 immunofluorescence (see Materials and Methods), we detected intermediate to low levels of Ascl1 in young (P70) and aged (P200) brain, in virtually all OPCs (99%) in the CC, whereas high levels were observed in the neurogenic regions (i.e., SVZ/rostral migratory stream [RMS]; Fig. 5A,B,G). Ascl1 expression was also detected at low levels in a large majority of OPCs (>90%) in other brain regions, including in the gray matter (in the hippocampus, thalamus, olfactory bulb; Fig. 5C; and data not shown) and the white matter (in the fimbria, striatum, cerebellum, brainstem; data not shown). In the cortex, Ascl1 expression was detectable in a large fraction (67%) of OPCs (Fig. 5D). These results are in accordance with the expression of the GFP reporter in Ascl1GFP mice in the adult brain (Fig. 5B) (Parras et al., 2007).
In the adult CNS, demyelination results in the upregulation of several TF involved in oligodendrogenesis, including upregulation of Olig2, Nkx2.2, and Sox10 transcripts (Fancy et al., 2004, 2009; Watanabe et al., 2004). To determine whether Ascl1 expression is also altered in response to demyelination, we used adult Ascl1GFP animals to perform LPC-induced focal demyelinating lesions in the CC, as previously reported (Nait-Oumesmar et al., 1999). At 5 d postlesion (dpl), cells expressing high to intermediate levels of Ascl1 and Ascl1GFP (Ascl1high/Ascl1GFP+ cells) were detected both in the SVZ and within and around the lesion, which was characterized by a very dense cellularity (Fig. 5E–E2). Ascl1high/Ascl1GFP+ cells were either OPCs (PDGFRα+) or more immature precursors (PDGFRα− cells), the latter having (as shown by GFP expression) characteristics of migrating cells (Fig. 5E1). Ascl1 expression in OPCs (relative to its expression in SVZ cells) was upregulated to levels similar postnatal OPCs (Fig. 5G). Moreover, Ascl1GFP+ cells in the lesion (Fig. 5F) and in the SVZ (data not shown) coexpressed high levels of Olig2 and Nkx2.2, suggesting that all three TFs are upregulated in progenitors and nearby OPCs involved in remyelination.
Ascl1 expression was not induced in oligodendrocytes upon/after demyelination given that Ascl1 protein was not detected in Olig2+/APC+ cells (Fig. 5H, arrowheads). However, the GFP transgene was expressed in maturing oligodendrocytes of Ascl1GFP animals (cells expressing high levels of Olig2 and Sox10, Fig. 5I,I1), as we previously observed at postnatal stages (Fig. 2). These data indicate that, in both the postnatal brain and in response to demyelination, a similar combination of TFs, including Ascl1 is induced in OPCs/progenitors and likely involved in (re)myelination.
Ascl1-expressing cells repopulate demyelinated lesions
Ascl1+ progenitor/OPCs are localized in and around focal LPC lesions, suggesting that they may differentiate locally into myelinating oligodendrocytes and participate to myelin repair. In Ascl1GFP mice, GFP expression persists long after that Ascl1 is downregulated, which allows cell fate mapping of Ascl1 progeny. We have examined the contribution of Ascl1-expressing cells to oligodendrocyte regeneration and lesion repair in Ascl1GFP mice. The demyelinated lesion can be recognized by a dense cellularity, the absence of myelin proteins (such as MBP; Fig. 6A), and oligodendrocyte depletion (APC+ cells, Fig. 6C,E). At 2 dpl, there were many GFP+ cells in the lesion, most of them being OPCs (79%; PDGFRα+/Olig2+ cells, Fig. 6D) and the remaining cells being immature progenitors. At 5 dpl, the number of GFP+ cells in the lesion was increased sevenfold, ∼60% being OPCs (Fig. 6B,D). At both time points, all OPCs were GFP+, indicating that OPCs present within the lesion derive from Ascl1-expressing cells. In addition, at 5 dpl, 25% of GFP+ cells in the lesion were APC+ oligodendrocytes (Fig. 6C,E), suggesting that these cells had differentiated locally from Ascl1+ OPCs/progenitors.
OPCs contributing to myelin lesion repair in the CC may derive from progenitors located in the SVZ (Nait-Oumesmar et al., 1999, 2008; Aguirre et al., 2007; Franklin and ffrench-Constant, 2008; Staugaitis and Trapp, 2009). LPC-induced Ascl1GFP mice showed OPCs associated with GFP+/PDGFRα− progenitors in the RMS, displaying a bipolar shape morphology and with the main cell body axis oriented toward the CC where the lesion is located (Fig. 6F,F1). This observation suggests that Ascl1+ progenitors and OPCs generated in the SVZ/RMS neurogenic zone can migrate away from the RMS toward CC lesions, likely recruited with signals delivered from the injured tissue.
Ascl1 progenitors differentiate into remyelinating oligodendrocytes
To assess the contribution of Ascl1+ progenitors to the remyelination process, we used the Cre/loxP recombination system to trace the Ascl1-cell lineage in LPC-demyelinated mice. LPC-induced lesions were performed in the CC of adult Ascl1Cre/YFP mice, and the presence of intralesional APC+/YFP+ oligodendrocytes was investigated at 5, 14, and 28 dpl. At all stages examined, the majority (67%) of intralesional PDGFRα+ OPCs and APC+ oligodendrocytes expressed YFP and belong to the Ascl1 cell lineage (Fig. 7A,D), with a similar proportion of YFP+ oligodendrocytes present in nonlesion territories (Fig. 7D). At 14 dpl, intralesional APC+/YFP+ cells expressed myelin proteins, such as MOG (Fig. 7B,C,C1,C2), suggesting that these cells were maturing into myelinating oligodendrocytes. We also used EM to identify intralesional β-galactosidase+/X-gal+ myelinating oligodendrocytes using Ascl1Cre/LacZ mice. X-gal-reaction deposits were detected within the lesion area, in cells with ultrastructural characteristics of newly formed oligodendrocytes enwrapping multiple axons (Fig. 7E,E1,E2). These observations confirmed that Ascl1+ progenitors/OPCs generate myelin-forming cells that participate in myelin repair.
Ascl1 is required for normal timing of OPC differentiation during remyelination
To determine Ascl1 requirement for proper remyelination, we deleted Ascl1 expression specifically in OPCs before inducing CC myelin lesion. We assessed whether YFP+-CkoOPC cells were able to migrate into lesions, proliferate, and differentiate in the absence of Ascl1. At 8 and 15 dpl, remyelinating lesions in both CkoOPC and CtrOPC displayed YFP+ cells being PDGFRα+ OPCs and APC+ oligodendrocytes (Fig. 8A,A′,B,B′). This result indicates that Ascl1 is dispensable for OPC recruitment and subsequent oligodendrocyte maturation in demyelinated lesions. At 8 dpi, the proportion of YFP+ OPCs and YFP+ oligodendrocytes was similar in CkoOPC and CtrOPC. Nevertheless, a relatively large number of YFP+-CkoOPC cells were negative for either oligodendroglial (PDGFRα or APC) or astroglial (GFAP) markers at this stage (Fig. 8B,B′). Some of them had typical oligodendrocyte morphology despite not expressing APC (Fig. 8B′), most likely corresponding to cells with abnormal differentiation. In contrast at 15 dpl, YFP+-CkoOPC cells had a smaller ratio of oligodendrocyte to OPCs than CtrOPC cells (Fig. 8C,C′,E), whereas the glial marker-negative population was very much reduced (Fig. 8). Moreover, we found fewer YFP+ cells expressing CNP myelin protein in CkoOPC than in CtrOPC (Fig. 8D,D′,E). Therefore, these data strongly suggest that Ascl1 is also required for normal OPC differentiation during remyelination and, as during myelination, other TFs most likely compensate its loss of function at later time points of the differentiation process.
Ascl1 is expressed in active MS lesions
In the mouse brain, Ascl1 expression is a hallmark of OPCs and precursor cells involved in remyelination. We hypothesized that, in humans, immature oligodendroglial cells in MS lesions may also express Ascl1. To test this hypothesis, we examined Ascl1 expression in four MS and two non-neurological control postmortem brain samples. Lesions were first characterized using Luxol fast blue/cresyl violet (Fig. 9B), MOG, and CD68 immunolabeling (Fig. 9C). The expression of Ascl1 and of oligodendroglial markers (Olig1 and Sox10) was examined in actively remyelinating borders (patchy MOG staining/CD68+), in the chronic core of the lesions (MOG−/CD68−; Fig. 9C), and in normal-appearing white matter (Fig. 9B,C). Ascl1+ cells were more numerous in active borders of MS lesions compared with chronic silent core or normal-appearing white matter. Moreover, the number of Ascl1+ cells was reduced in the core of chronic lesions compared with the other regions studied. Interestingly, Ascl1+ cells were more concentrated within the CD68+ active borders of periventricular MS lesions (Fig. 9D) compared with other regions studied. Indeed, in other brain regions, including cortex and the normal-appearing white matter, Ascl1 expression level was very weak, most likely resulting from poor Ascl1 detection due to the fixation constraints of human brain samples. Nevertheless, the majority of Ascl1-expressing cells in active MS periventricular lesions were Olig1+ and Sox10+ (Fig. 9E,F), indicating that upregulation of Ascl1 in the endogenous OPCs may play a role in oligodendrocyte regeneration under pathological conditions. Overall, our findings on MS tissue are in agreement with our data on animal models of demyelination and demonstrate that Ascl1 is expressed at higher levels in oligodendroglial cells in MS plaques. Ascl1 may therefore be involved in oligodendrocyte differentiation and myelin repair in demyelinating pathology.
In this paper, we show that Ascl1 is expressed during brain myelination and remyelination in SVZ progenitors and OPCs, playing a cell-autonomous role both in OPC specification and in fine-tuning of OPC proliferation versus differentiation (Fig. 10). First, we demonstrate that Ascl1 is expressed in neonatal cortical SVZ progenitors and that its conditional deletion strongly suppress OPC generation, thus demonstrating that Ascl1 is required for postnatal oligodendrocyte specification. Second, Ascl1 is expressed in all OPCs in most brain regions and its conditional ablation in OPCs favors proliferative over differentiative OPC divisions, leading to a reduced number of oligodendrocytes. Third, Ascl1 is not expressed in differentiating oligodendrocytes, but lineage-tracing studies show that Ascl1-expressing cells differentiate into myelinating and remyelinating oligodendrocytes. Fourth, Ascl1-expressing cells (OPCs and immature progenitors) are found in and around active lesions both in mouse and MS brain, Ascl1 being required for normal OPC differentiation during remyelination in mice and most likely implicated in human remyelination.
Ascl1 promotes OPC specification
Ascl1-conditional deletion in dorsal SVZ cells strongly reduces the generation of oligodendroglial cells in the dorsal telencephalon and the majority of Ascl1-deficient cells become instead astrocytes. This effect could be the result of the following: (1) selective cell death of Ascl1-deficient oligodendroglial cells; (2) reduced proliferation of oligodendroglial cells in the absence of Ascl1; (3) blocking of Ascl1-deficient SVZ cells at immature progenitor stage; or (4) specification switch from oligodendrocyte to astrocyte fate in the absence of Ascl1. We did not find preferential cell death of Ascl1-deficient SVZ cells or of their progeny. Also, Ascl1-deficient OPCs proliferate more than controls. In contrast, we found a 1.7-fold increase in the number of Ascl1-deficient GFAP+ cells in the SVZ compared with controls. Therefore, Ascl1 is most likely required for OPC specification in the postnatal SVZ; and in the absence of Ascl1, the majority of SVZ cells become astrocytes or remain undifferentiated progenitors. In vivo clonal studies would be required to discriminate between these two possibilities. Indeed, a role in oligodendrocyte specification is supported by previous reports on in vitro experiments of Ascl1 loss- and gain-of-function in progenitors. Clonal neurosphere cultures of Ascl1-deficient progenitors display a reduced capacity to generate oligodendrocytes (and neurons) and differentiate instead into astrocytes (Parras et al., 2004), whereas Ascl1-overexpressing progenitors favor oligodendrocyte differentiation at the expense of an astrocytic fate (Sugimori et al., 2007).
Cell-autonomous Ascl1 function in oligodendrogenesis
It has been suggested that Ascl1 plays an indirect role in oligodendrogenesis by Notch-mediated repression of the neurogenic determinants Dlx1/2 (Petryniak et al., 2007). However, in contrast with the embryonic expression of Ascl1 in a subset of OPCs (Parras et al., 2007; Sugimori et al., 2007), we demonstrate here that Ascl1 protein is present in the majority of postnatal brain OPCs, suggesting that Ascl1 plays a cell-autonomous function in oligodendrogenesis. Some immature progenitors coexpress Ascl1 and Dlx in oligodendrocyte-generating regions of embryonic (Petryniak et al., 2007) and postnatal telencephalon (current study), leaving space for Dlx inhibition of OPC specification. Nevertheless, Ascl1 conditional deletion in either SVZ cells or OPCs not only alters oligodendrocyte specification, but OPC proliferation versus differentiation balance, demonstrating a cell-autonomous role of Ascl1 at different stages of oligodendrogenesis. It is worth noting that Ascl1 and Olig2 are coexpressed in immature SVZ, suggesting that OPC specification likely requires the downregulation of Dlx1/2 in their progenitor cells.
Ascl1 function in OPC proliferation versus differentiation balance
Even though Ascl1 is expressed in proliferating cells, its conditional deletion in OPCs increased proliferation as measured by Ki67 expression. This could result from the fact that Ascl1 activates target genes involved either in proliferation or in cell cycle exit depending on the context, as recently reported for neurogenesis (Castro et al., 2011). In OPCs, the presence of inhibitors, such as Id2/4 (Wang et al., 2001; Yun et al., 2004) and other cofactors involved in proliferation and differentiation, could balance Ascl1 activation of proliferation versus cell cycle exit. Supporting this hypothesis, Ascl1 genome-wide chromatin immunoprecipitation data (ChIP-seq) in purified OPCs suggest that Ascl1 binds regulatory regions of both positive and negative factors controlling cell cycle (Clavairoly A, Parras C, unpublished observations). Ascl1 deletion in OPCs also decreases differentiation during myelination (postnatal) or remyelination (adult). Indeed, one or 2 weeks after Ascl1 deletion, more Ascl1-deficient cells are still OPCs and fewer have started differentiation. Our data suggest that Ascl1 cooperates with other factors to activate OPC differentiation and myelin protein expression. Nevertheless, Ascl1 requirement is compensated at later stages, and no myelination phenotype is found in Ascl1-mutant oligodendrocytes. Although Ascl1 is known to regulate MBP transcription in vitro (Gokhan et al., 2005), a genome-wide analysis of Ascl1 targets in oligodendrogenesis will be important to identify the Ascl1-target genes involved in the different stages of oligodendrocyte development (specification, proliferation, division mode, differentiation, and myelination).
Similar mechanisms of oligodendrogenesis during myelination and remyelination
Demyelination has been shown to upregulate Olig2, Nkx2.2, and Sox10 transcript levels in and around lesion and also many other factors (Fancy et al., 2004, 2009), raising the question of the similarities and differences between oligodendrogenesis during myelination and remyelination. Our detailed characterization of the expression of Ascl1 and other TFs involved in oligodendrogenesis (Figs. 2 and 5) indicates that the same types of progenitors and OPCs, characterized by the same combined expression of markers, are involved in oligodendrogenesis during myelination and remyelination. After demyelination, we found that Ascl1 was upregulated in OPCs at levels found during myelination. Moreover, the levels of Ascl1, Olig2, Nkx2.2, and Sox9/10 expression at different stages of oligodendrogenesis were similar during remyelination and myelination (Fig. 5; and data not shown). Therefore, the upregulation of these factors during active remyelination likely results from the rejuvenation of a developmental genetic program enabling OPCs to actively proliferate, migrate, and differentiate into new remyelinating oligodendrocytes.
Demyelination induces axonal functional and structural loss, leading to major neurological and motor dysfunctions. In MS, despite the recent therapeutic advances aimed at reducing the inflammatory component and stopping demyelination, treatments to improve myelin repair are lacking because of the poor understanding of mechanisms leading to effective remyelination. In this context, our finding on the role of Ascl1 during myelin formation and in particular after demyelination, and future work characterizing its target genes, may pave the way toward a better understanding of the key players involved in oligodendrogenesis after demyelination and therefore help identifying new therapeutic targets aimed at favoring myelin repair.
This work was supported by Institut National de la Santé et de la Recherche Médicale Grant R05245DS to C.P. and H.N., National Multiple Sclerosis Society RG 4318A1/1 to H.N. and E.M., Fondation pour l'Aide à la Recherche sur la Sclerose En Plaques (H.N., H.H.), Neuropôle de Recherche Francilien/Région de l'Ile-de-France (H.H.), Fondation de France (C.P., H.N.), Université Pierre et Marie Curie-Paris 6 (A.C.), and “Investissements d'Avenir” ANR-10-IAIHU-06 at the Institut du Cerveau et de Moelle épinière (C.K., B.N.-O.). We thank Dr. W.D. Richardson for the PDGFRα::CreERT2 mice; A. Chedotal, J. Johnson, K. Campbell, G. Panganiban, M. Okabe, and M. Wegner for reagents; the UK Multiple Sclerosis Tissue Bank (Pr R. Reynolds, Imperial College, London), funded by the Multiple Sclerosis Society of Great Britain and Northern Ireland, registered charity 207495, for brain tissue samples; and A. Dauphin and D. Langui from the Plate-forme d'Imagerie Cellulaire de la Pitié-Salpêtrière for their advice, notably with EM analysis.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Carlos Parras, Université Pierre et Marie Curie-Paris 6, Centre de Recherche de l'Institut du Cerveau et de la Moelle épinière, Hôpital de la Pitié-Salpêtrière 83, Boulevard de l'Hôpital ICM, 4ème étage (4.007), 75013 Paris, France.