Stage-specific regulation of oligodendrocyte development by Hedgehog signaling in the spinal cord

Abstract

Elucidation of signaling pathways that control oligodendrocyte (OL) development is a prerequisite for developing novel strategies for myelin repair in neurological diseases. Despite the extensive work outlining the importance of Hedgehog (Hh) signaling in the commitment and generation of oligodendrocyte progenitor cells (OPCs), there are conflicting reports on the role of Hh signaling in regulating OL differentiation and maturation. In the present study, we systematically investigated OPC specification and differentiation in genetically modified mouse models of Smoothened (Smo), an essential component of the Hh signaling pathway in vertebrates. Through conditional gain-of-function strategy, we demonstrated that hyperactivation of Smo in neural progenitors induced transient ectopic OPC generation and precocious OL differentiation accompanied by the co-induction of Olig2 and Nkx2.2. After the commitment of OL-lineage, Smo activity is not required for OL differentiation, and sustained expression of Smo in OPCs stimulated cell proliferation but inhibited terminal differentiation. These findings have uncovered the stage-specific regulation of OL development by Smo-mediated Hh signaling, providing novel insights into the molecular regulation of OL differentiation and myelin repair.

1. INTRODUCTION

Oligodendrocyte (OL) development in vertebrates is a highly orchestrated process involving several well-defined steps from specification of neural progenitor cells (NPCs) into proliferative oligodendrocyte progenitor cells (OPCs) to their differentiation into pre-oligodendrocytes followed by maturation into myelin-forming cells (Richardson et al., 2006). The specification and differentiation processes are controlled by a complex network of extracellular signals, transcriptional factors and epigenetic regulators (Emery, 2010). Elucidation of signaling pathways that control OL differentiation and myelin formation is a crucial prerequisite for developing novel strategies for myelin repair in several neurological diseases (Gallo & Deneen, 2014).

The Hedgehog (Hh) signaling pathway has been studied extensively in tissue development and demonstrated to play a vital role in OL development (Laouarem & Traiffort, 2018; Traiffort et al., 2016). To date, three distinct Hh genes, Sonic hedgehog (Shh), Indian hedgehog (Ihh) and Desert hedgehog (Dhh) have been identified in vertebrates (Varjosalo & Taipale, 2008). During development, OPCs arise from restricted domains of neuroepithelium of the ventricular zone (VZ) and then migrate to the surrounding white matter regions, where they subsequently differentiate into mature myelinating OLs (Miller, 2002; Rowitch, 2004). The Shh signaling pathway is notably implicated in the initial appearance of OPCs in the spinal cord and forebrain (Alberta et al., 2001; Nery et al., 2001; Orentas et al., 1999; Sussman et al., 2000), whereas the late population of OPCs arise from the dorsal spinal neuroepithelial cells independent of Hh signaling (Cai et al., 2005; Fogarty et al., 2005; Vallstedt et al., 2005). Indeed, both the concentration and timing of Shh exposure is important for ventral neurons and OL induction (Dessaud et al., 2007). Meanwhile, analysis of Ihh loss-of-function showed that this protein is also required for OPC specification from the pMN precursors (Chung et al., 2013). In spite of the crucial role of Shh and Ihh signaling in the induction of OPCs, Dhh primarily regulates spermatogenesis and is required for the development of peripheral nerve sheath (Bitgood et al., 1996; Parmantier et al., 1999). Smoothened (Smo), a member of the frizzled family of seven-transmembrane domain proteins, is essential for relaying Shh and Ihh signaling pathways across the plasma membrane (Zhang et al., 2001). In unstimulated cells, Smo is inhibited by the tumor suppressor membrane protein Patched1, thus ensuring that Hh pathway is repressed. Upon Hh stimulation, the secreted Hh ligand binds and inhibits Patched1, leading to accumulation of Smo in the cilium, which in turn triggers the downstream signal transduction events of the Hh pathway, ultimately causing activation of target gene transcription (Cohen, 2003; Huangfu & Anderson, 2006).

Despite the extensive work outlining the necessity of Hh signaling in the generation and commitment of OPCs, little is known about the role of Hh/Smo signaling in the passage of OPCs to myelinating OLs. The majority of evidence favors that Shh promotes OL specification and differentiation. For instance, in Shh^−/− or Smo^−/− mutants, generation of OPCs is found defective (Oh et al., 2005; Tan et al., 2006). Both in vitro culture and in vivo transplantation experiments showed that Shh promotes the maturation of neurosphere-derived OPCs from embryonic rat brain (Gibney & McDermott, 2007, 2009). Moreover, Smo inhibtion by cyclopamine impairs OPC differentiation to myelinating OLs and drug removal restores Mbp and MAG gene expression in primary OPCs (Wang & Almazan, 2016). On the contrary, in the transgenic mice that ectopically express Shh in the dorsal neural tube, spinal precursor cells are blocked in an undifferentiated state with elevated proliferative levels (Rowitch et al., 1999). Thus, the specific role of Hh signaling in OPC differentiation remains to be further determined.

Here, we systematically investigated the role of Hh signaling in oligodendroglial development by inducing the hyperactivation of Smo at different stages through the conditional gain-of-function strategy. Specifically, we used Nestin^cre; SmoM2 animal model to constitutively activate Smo in NPCs, and Olig1^cre; SmoM2 and Pdgfrα^creER; SmoM2 to activate Smo in OPCs. Through analyses and comparison of OPC specification and differentiation among these animal models, we uncovered a stage-specific regulation of Smo-mediated Hedgehog signaling pathway on OL development.

2. METHODS

2.1. Mice

Smo^fx (no. 004526), SmoM2 (no. 005130), tdTomato (no. 007909), Nestin^cre (no. 003771) and Pdgfrα^creER (no. 018280) mice were purchased from Jackson Laboratories (Bar Harbor, Maine, USA) and maintained in C57BL/6 background. Nestin^cre and Olig1^cre knock-in line with Neo (Lu et al., 2002) were mated with SmoM2 mice to obtain double heterozygous transgenic mice. Pdgfrα^creER line which carried a tamoxifen-inducible cre gene under the control of Pdgfrα promoter were also crossed with Smo^fx and SmoM2 mice for intraperitoneal injection of tamoxifen. For mouse genotyping, genomic DNA was extracted from embryonic tissues or mouse tails and subsequently used for genotyping by polymerase chain reaction. All research procedures using animals were approved by the Institutional Animal Care and Use Committee at Hangzhou Normal University. All efforts were made to minimize the number of animals and their suffering. Animals of either sex were used for analyses.

2.2. Tamoxifen and 5-Bromo-2’-deoxyuridine (BrdU) treatments

Tamoxifen (T5648; Sigma-Aldrich, Darmstadt, Germany) was dissolved in an ethanol/sunflower seed oil (1:9) mixture at a concentration of 10mg/ml. From E14.5, the pregnant female parents were injected daily with tamoxifen (75 μg/g body weight). 5-Bromo-2’-deoxyuridine (BrdU) (B5002; Sigma-Aldrich) was prepared in sterile 1× phosphate buffer saline (PBS) at the concentration of 10mg/ml and administered to pregnant mice via intraperitoneal injection (100μg/g body weight) 2 hours before embryos were procured.

2.3. In situ RNA Hybridization (ISH)

After anesthesia, mice were perfused with 1×PBS and then with 4% paraformaldehyde (PFA) in 1×PBS. Brain and spinal cord tissues were collected, post-fixed in 4% PFA overnight, and cryo-protected in 30% sucrose (v/v) in 1×PBS at 4℃ overnight. Tissues were then embedded in optimal cutting temperature compound (OCT) medium and sectioned on a cryostat with 16–18-μm thickness. Section were subjected to ISH with digoxigenin-labeled riboprobes according to Schaeren-Wiemers and Gerfin-Moser (Schaeren-Wiemers & Gerfin-Moser, 1993) with minor modifications.

2.4. Immunofluorescent staining

Tissues for immunostaining were sectioned (14μm thickness) on a cryostat. The procedures for immunofluorescence staining has been described elsewhere (Xu et al., 2017). Tissue sections were firstly rinsed three times in PBS, blocked with 5% goat serum in PBS with 0.2% Triton-X-100 for 1 hour, and immediately incubated with primary antibody in blocking solution at 4℃ overnight. Sections were then washed three times in PBS and incubated with secondary antibodies (Thermo Fisher Scientific Inc., Rockford, IL, USA, 488/594 Alexa Fluor, 1:3000) for 1 hour at room temperature. After rinsed three times in PBS, sections were mounted in Mowiol Mounting Medium (MMM) with 4’,6-diamidino-2-phenylindole (DAPI). Fluorescent images were collected by Nikon Epifluorescence Microscope. Primary antibodies used were as follows: anti-Olig2 (Millipore, MA, USA, Cat# AB9610, 1:1000, RRID: AB_570666), anti-BrdU (DSHB, Cat# G3G4, 1:50, RRID: AB_1157913), anti-Ki67 (Abcam Cat# AB15580, 1:1000, RRID: AB_443209), anti-Caspase 3 (Millipore Cat#AB3623, 1:200, RRID: AB_91556), anti-Nkx2.2 (DSHB, Cat# 74.5A5, 1:30, RRID: AB_531794), anti-Cyclin D1 (Abcam Cat#ab16663, 1:500, RRDI: AB_443423). In BrdU immunostaining procedure, sections need to be incubated in 1M HCl for 1 hour at 45℃ before blocking for DNA denaturation. In Ki67 immunostaining procedure, sections need to be incubated in 10nM Sodium Citrate Buffer (Bio Basic Inc., Shanghai, China, Cat# 6132–04-3) for 30 min at 80℃ before blocking for antigen retrieval.

2.5. Tdt-mediated dUTP Nick-End Labeling (TUNEL) assay

Apoptosis detection was performed according to the instruction of the DeadEnd™ Fluometric TUNEL system (Promega, WI, USA, Cat# G3250). Tissue sections were post-fixed in 4% PFA, followed by 20μg/ml Proteinase K treatment to prevent desorption. Sections were then rinsed in Equilibration buffer for 10 min at room temperature and subsequently incubated in Terminal deoxynucleotidyl Transferase (TdT) reaction mix according to manufacturer’s instructions. Finally, the reaction was terminated by high concentration of salt solution, like 2X Saline Sodium Citrate (SSC). After rinsed three times in PBS, sections were mounted in Mowiol Mounting Medium.

2.6. Experimental design and Statistical analyses

For each analysis, results from independent animals were treated as biological replicates (n≥3). Quantitative data are indicated as number of cells in mutant and control tissues, presented as mean ± SEM; n≧3 embryos per experimental point. Statistical significance of the difference was evaluated by Student’s t-test.

3. RESULTS

3.1. Hyperactivation of Smo in NPCs resulted in transient ectopic induction of oligodendrocytes

Previous research in Smo conditional knock-out (cKO) mice using Nestin^cre have demonstrated that Smo-mediated Hh signaling in neural progenitor cells (NPCs) is required for the generation of ventral OPCs (Yu et al., 2013). To examine the direct effect of Smo activation in NPCs, we generated Nestin^cre; SmoM2 (referred to as Nestin-SmoM2) mice to drive SmoM2 overexpression in all the NPCs. SmoM2 encoded the constitutively active W539L point mutation of Smo gene (Jeong et al., 2004). The resulting Nestin-SmoM2 transgenic mice died at birth with a domed head (data not shown), probably due to increased cell proliferation in the cortex. Efficacy of recombination was evaluated using tdTomato reporter mice crossed with Nestin-SmoM2 line. TdTomato fluorescence was widely visualized throughout the spinal cord during early stages (Figure 1a’), indicative of the wide and efficient recombination in Nestin-SmoM2 embryos. Analyses of myelin gene expression revealed that a small population of Mbp+ cells were induced in the ventral VZ of Nestin-SmoM2 spinal cords at as early as E12.5 (Figure 1b’). At later stages, ectopic Mbp+ and Plp1+ cells were found in more dorsal regions adjacent to the VZ (Figure 1c’, h’), and increased in numbers as development proceeded (Figure 1f). By E16.5, newly induced OLs were enriched around the entire VZ (Figure 1d’, i’). While this group of ectopic precocious OLs diminished at P0 (Figure 1e’, j’), the intrinsic OL differentiation in the white matter regions became apparent and a comparable number of Mbp+/Plp1+ OLs were found in Nestin-SmoM2 and control tissues (Figure 1f). Myrf is specifically upregulated in post-mitotic OLs and functions to activate OL differentiation (Emery et al., 2009; Huang et al., 2018). In Nestin-SmoM2 embryos, Myrf displayed a similar expression pattern to that of myelin genes (Figure 1l-o’, k) and ectopic Myrf+ OLs were also induced around the VZ (Figure 1m’, n’). These data thus indicated that hyperactivation of Smo in NPCs during early developmental stages was able to trigger precocious differentiation and transient accumulation of OLs in the central progenitor zone.

Figure 1.

Hyperactivation of Smo in NPCs triggered precocious OL differentiation. (a, a’) tdTomato fluorescence in SmoM2;tdTomato and Nestin^cre;SmoM2;tdTomato spinal cord at E12.5. Spinal cord sections from SmoM2 (control) and Nestin^cre; SmoM2 (Nestin-SmoM2) were subjected to Mbp (b-e’), Plp1 (g-j’) and Myrf (l-o’) ISH at indicated stages. (f, k) Quantification of Plp1+ and Myrf+ cells per section in control and Nestin-SmoM2 spinal cords at indicated stages. Error bars indicate SEM. *p<.05 and ***p<.001. Scale bar: 100μm.

We next compared the production and proliferation of OPCs between double transgenic and control embryos. At E12.5, the expression domain of Pdgfrα was expanded and more Pdgfrα+ OPCs were generated in Nestin-SmoM2 embryos (Figure 2a’). At E14.5, Pdgfrα+ cells increased dramatically (Figure 2b’, e), but slightly decreased at E16.5 (Figure 2c’, e). By P0, there was no significant difference in the number of OPCs between control and mutant embryos (Figure 2d, d’, e). Similarly, the number of Olig2+ cells in Nestin-SmoM2 embryos increased dramatically at E14.5 (Figure 2f, h, h’), suggesting that hyperactivation of Smo may promote the proliferation of OPCs. However, BrdU incorporation analysis revealed that overall cell proliferation level was unchanged between control and Nestin-SmoM2 spinal cords at E14.5 (Figure 2f, i, i’), and the percentage of Olig2+ cells that were also BrdU+ was comparable between control and transgenic lines (Figure 2g, j, j’). Thus, Smo activation in NPCs promoted the fate specification of OPCs instead of their proliferation.

Figure 2.

Overexpression of Smo in NPCs did not promote OPC proliferation. (a-d’) Spinal cord sections from control and Nestin-SmoM2 were subjected to Pdgfrα ISH at indicated stages. (e) Quantification of Pdgfrα+ cells in control and Nestin-SmoM2 spinal cords at indicated stages. (h-j’) Spinal cord sections from control and Nestin^cre; SmoM2 were subjected to Olig2 (Red) and BrdU (Green) IF at E14.5. (j) and (j’) are the higher magnifications of the boxed areas in (i) and (i’), respectively, with BrdU and Olig2 double immunostaining. Arrows pointed to BrdU+/Olig2+ cells. (f) Quantification of Olig2+ and BrdU+ cells in control and Nestin-SmoM2 spinal cords after a 2-hr pulse of BrdU at E14.5. (g) Quantification of the proportion of BrdU+/Olig2+ cells over the Olig2 population in control and Nestin-SmoM2 spinal cords after a 2-hr pulse of BrdU at E14.5. (k-l’) Spinal cord sections from control and Nestin-SmoM2 were subjected to TUNEL assay at E14.5. (l) and (l’) are the higher magnifications of the boxed areas in (k) and (k’), respectively, with Fluorescein-dUTP and Olig2 double immunostaining. Fluorescein-dUTP+/Olig2+ cells are represented by arrows. (m) Quantification of Fluorescein-dUTP+ cells per section in control and Nestin-SmoM2 spinal cords at indicated stages. Error bars indicate SEM. *p<.05 and ***p<.001. n.s.: no significant difference. Scale bar: 100μm.

Meanwhile, TUNEL assay revealed that at E14.5, the apoptotic level was much higher in Nestin-SmoM2 embryos (Figure 2k, k’), but diminished gradually with time and returned to normal by P0 (Figure 2m). Unlike the even distribution in control embryos, most apoptotic cells in Nestin-SmoM2 embryos were located in the central gray matter (Figure 2k’) and co-labeled with Olig2 (Figure 2l’), suggesting that the ectopic OPCs or OLs might be gradually eliminated by cell apoptosis in the absence of axonal contacts.

3.2. Smo specifically induced OL differentiation in NPCs through co-expression of Olig2 and Nkx2.2.

In previous studies, we have proposed that OLs might be specified from Nkx2.2+/Olig2+ progenitor cells (Fu et al., 2002). In ovo electroporation of chicken embryonic spinal cords revealed that co-expression of Nkx2.2 and Olig2 was able to induce precocious and ectopic OL differentiation (Zhou et al., 2001), similar to our findings in Nestin-SmoM2 embryos. To understand the mechanisms underlying the precocious differentiation triggered by active Smo expression in NPCs, we analyzed the co-expression of Nkx2.2 and Olig2 in Nestin-SmoM2 spinal cords. As previously reported, before OPC specification, Nkx2.2 and Olig2 expression were restricted in adjacent non-overlapping domains of ventral neuroepithelial cells (Figure 3a). In Nestin-SmoM2 spinal cords, the number of Nkx2.2+ cells was significantly elevated (Figure 3c). Accordingly, the expression domains of Nkx2.2 and Olig2 were both expanded laterally, and dramatic increase in Nkx2.2+/Olig2+ cells was observed around the adjacent region (Figure 3b, d). Sox10 has been proven to be a major determinant in terminal differentiation of OLs (Stolt et al., 2002). At E12.5, comparison between the Nkx2.2/Olig2 and Sox10 expression in immediately adjacent spinal cord slices revealed that the number and distribution of Nkx2.2+/Olig2+ cells were comparable to that of Sox10+ cells (Figure 3e, f, h). Ectopic Mbp+ OLs were also located in the same region (Figure 3g). Thus, it is reasonable to speculate that precocious myelin gene expression and OL differentiation in Nestin-SmoM2 spinal cords are triggered by the co-expression of Olig2 and Nkx2.2. Hyperactivation of Smo in NPCs promoted the co-expression of Olig2 and Nkx2.2 in adjacent regions. These Nkx2.2+/Olig2+ OPCs initiated the expression of Sox10 and proceeded to terminal differentiation.

Figure 3.

The specific induction of OLs in Nestin-SmoM2 spinal cords were due to co-expression of Nkx2.2 and Olig2. (a,b) Spinal cord sections from SmoM2 and Nestin^cre; SmoM2 were subjected to anti-Nkx2.2 (Red) and anti-Olig2 (Green) double-immunostaining. Arrows pointed to Nkx2.2/Olig2 double positive cells. (c) Quantification of Nkx2.2+ cells per section in SmoM2 and Nestin^cre; SmoM2 mouse spinal cords at E10.5. (d) Quantification of Nkx2.2/Olig2 double positive cells per section in SmoM2 and Nestin^cre; SmoM2 mouse spinal cords at E10.5. (e-g’) Successive sections from Nestin^cre; SmoM2 was subjected to Nkx2.2/Olig2 double-immunostaining, Sox10 and Mbp ISH at E12.5. (h) Quantification of Nkx2.2/Olig2 double positive cells and Sox10+ cells in successive sections from Nestin^cre; SmoM2. Error bars indicate SEM. *p<.05 and **p<.005. n.s.: no significant difference. Scale bar: 100μm.

3.3. Endogenous Smo activity in OPCs is not required for oligodendroglial development.

Despite extensive research on the importance of Hh signaling in the control of OPC specification from NPCs (Laouarem & Traiffort, 2018), little is known about its role in the differentiation of OPCs into pre-OLs and myelinating OLs. The RNA-Seq expression data of Hh signaling genes in oligodendroglial cells from both embryonic and postnatal stages showed relatively low levels of Smo in OPCs and none in OLs (Fig. 4a) (Zhang et al., 2014), implying that Hh/Smo signaling is inactivated after OPCs are specified and continuous Hh signaling activity is not required for their subsequent proliferation and differentiation. To test this hypothesis, we performed Smo loss-of-function analyses using inducible Pdgfrɑ^creER line. Tamoxifen was administered from E10.5 for seven days to completely block Smo expression in newly specified OPCs, and spinal cord tissues from Pdgfrɑ^creER;Smo^fx/fx (referred to as Pdgfrɑ-Smo cKO) and control mice were harvested at E18.5 (Fig. 4b). Double-labeling confirmed the overlapping signals of tdTomato fluorescence and Olig2 (Fig. 4c), indicating efficient recombination in OPCs. At E18.5, there was no difference in the number and distribution of OLs between control and Pdgfrɑ-Smo cKO embryos, based on in situ hybridization for mature OL markers Mbp and Plp1 (Fig. 4d-e’, h). The number and distribution of OPCs were also comparable in control and mutant embryos by detection of Olig2 (Fig. 4f, f’, i). Immunofluorescence staining for proliferative marker Ki67 revealed a similar number of Ki67+ cells between the control and mutant tissues (Fig. 4g, g’, j). Thus, in keeping with its low-level expression in OPCs, Hh/Smo signaling in specified OPCs is dispensable for later oligodendroglial development.

Figure 4.

Endogenous Smo activity was not required for oligodendroglial development. (a) Average Smo FPKM values in astrocytes, neurons, OPCs, newly formed OLs, myelinated OLs, microglia and endothelial cells from P7 mouse brains (Zhang, et al., 2014). (b) Pdgfrα^creER; Smo^fx/+ (control) and Pdgfrα^creER; Smo^fx/fx (Pdgfrα-Smo cKO) mice were administrated with tamoxifen from E10.5 for 7 days and harvested at E18.5. (c) Representative tdTomato (Red) and Olig2 (Green) double positive cells in E18.5 white matter of Pdgfrα-tdTomato spinal cord. (d-g’) Spinal cord sections from control and Pdgfrα-Smo cKO were subjected to Mbp (d, d’), Plp1 (e, e’) ISH and Olig2 (f, f’), Ki67 (g, g’) IF at E18.5. (h-j) Quantification of Plp1+, Olig2+ and Ki67+ cells per section in E18.5 spinal tissues. Error bars indicate SEM. n.s.: no significant difference. Scale bar: 100μm.

3.4. Overexpression of Smo in OPCs stimulated cell proliferation but inhibited their differentiation.

The inactivation of Hh/Smo signaling after OPC specification implies that continuous Hh signaling may interfere with their subsequent proliferation and differentiation. To test this hypothesis, we utilized Pdgfrα^creER mice to drive SmoM2 overexpression in OPCs in a time-specific manner. Tamoxifen was administered from E14.5 for 4 days, and embryos were collected at E16.5 and E18.5 for detection of OPC proliferation and OL differentiation, respectively (Fig. 5a). As expected, tdTomato fluorescence was detected in the majority of Olig2+ cells in Pdgfrα^creER; SmoM2 (referred to as Pdgfrα-SmoM2) spinal tissues (Fig. 5b). Contrary to the scenario in Nestin-SmoM2 spinal cords, the number of Mbp+ or Plp1+ OLs was significantly reduced at E18.5 in Pdgfrα-SmoM2 (Fig. 5c-d’, f). In contrast, the numbers of Pdgfrα+, BrdU+ and Ki67+ cells were all significantly increased in E16.5 transgenic tissues (Fig. 5e-h’, j). Double immunostaining of Ki67 and Olig2 detected a higher percentage of Olig2+ cells that were also Ki67+ (Fig. 5i, i’, j). Therefore, continuous activation of Smo in OPCs promoted their proliferation and inhibited OL differentiation.

Figure 5.

OL differentiation was inhibited and OPC proliferation was elevated in Pdgfrα-SmoM2 spinal cords. (a) SmoM2 (control) and Pdgfrα^creER; SmoM2 (Pdgfrα-SmoM2) mice were administrated with tamoxifen at E14.5 and harvested at E16.5 and E18.5, respectively. (b) Representative tdTomato (Red) and Olig2 (Green) double positive cells in E18.5 gray matter of Pdgfrα-tdTomato spinal cord. (c-e’) Spinal cord sections from control and Pdgfrα-SmoM2 were subjected to Mbp (c, c’), Plp1 (d, d’), Pdgfrα (e, e’) ISH at E18.5. (f) Quantification of Plp1+ and Pdgfrα+ cells per section in E18.5 spinal tissues. (g-h’) Spinal cord sections from control and Pdgfrα-SmoM2 were immunostained by BrdU (g, g’) and Ki67 (h, h’) at E16.5 for cell proliferation analyses. (i) and (i’) are the higher magnifications of the boxed areas in (h) and (h’), with Ki67 and Olig2 double immunostaining. Arrows pointed to the Ki67/Olig2 double positive cells. (j) Quantification of BrdU+ and Ki67+/Olig2+ cells per section in E16.5 spinal cords. Error bars indicate SEM. *p<.05 and **p<.005. Scale bar: 100μm.

3.5. Sustained expression of Smo in Olig1-positive cells produced stage-dependent effects on OL differentiation.

As Olig1 was firstly expressed in NPCs before gliogenesis and later predominantly expressed in OPCs and OLs (Lu et al., 2002), we next generated the Olig1^cre;SmoM2 mice (referred to as Olig1-SmoM2) to confirm the opposing roles of Smo in OPC proliferation and differentiation. As expected, Olig1^cre was able to induce tdTomato signal in the ventral progenitor cells of spinal cord at E12.5 and later in oligodendroglial cells at P0 (Fig. 6a, b). The resulting Olig1-SmoM2 mice died neonatally without apparent deficiency. At early stages of oligodendrogenesis, the generation of Olig2+ OPCs from the pMN domain was comparable in both genotypes, although Olig2+ cells in the ventral neuroepithelium were more spread out dorsally in transgenic tissues (Fig. 6c, c’, e). Noticeably, a small fraction of Olig2+ cells were observed in the dorsal-most region of VZ at E14.5 in Olig1-SmoM2 tissues (Fig. 6d’, e) due to the dorsal expression of Olig1 at this stage (Cai et al., 2005). Contrary to the scenario in Nestin-SmoM2 spinal cords, ectopic expression of Mbp and Plp1 was detected around the central canal in the dorsal, but not ventral, region at E16.5 (Fig. 6i’, k’). Further detection of Nkx2.2 expression revealed that, unlike Olig2, the number of Nkx2.2+ cells was significantly increased at early stages (Fig. 6f, f’, h). Double immunostaining assay discovered a small number of Nkx2.2/Olig2 double positive cells in the dorsal part of Olig1-SmoM2 spinal cords (Fig. 6g), co-localized with ectopic Mbp+/Plp1+ cells (Fig. 6i’, k’). Thus, Smo activation in dorsal neural progenitor cells is sufficient to induce the co-expression of Olig2 and Nkx2.2 genes followed by ectopic OL differentiation. Surprisingly, there was no overlapping expression of Nkx2.2 and Olig2 (Fig. 6g’), and no ectopic Mbp and Plp1 expression in the ventral spinal cord (Fig. 6i’, k’).

Figure 6.

Sustained expression of Smo in Olig1-positive cells produced stage-dependent effects on OL differentiation. (a, b) tdTomato fluorescence in Olig1^cre;tdTomato spinal cord at indicated stages. (c-d’) Spinal cord sections from SmoM2 and Olig1^cre; SmoM2 were subjected to Olig2 IF. The white bracket in c and c’ delineated the Olig2 expression domain in ventral VZ. Arrow in d’ indicated the ectopic induction of Olig2 in the dorsal-most region of VZ. (e) Quantification of Olig2+ cells per section in control and Olig1-SmoM2 mouse spinal cords at indicated stages. (f, f’) Spinal cord sections from SmoM2 and Olig1^cre; SmoM2 were subjected to Nkx2.2 IF at E14.5. g and g’ are the higher magnifications of the boxed areas in f’, with Nkx2.2 and Olig2 double immunostaining. Arrows pointed to Nkx2.2/Olig2 double positive cells. (h) Quantification of Nkx2.2+ cells per section in control and Olig1-SmoM2 mouse spinal cords at indicated stages. (i-l’) Spinal cord sections from SmoM2 and Olig1^cre; SmoM2 were subjected to Mbp and Plp1 ISH at indicated stages. Error bars indicate SEM. *p<.05, **p<.005 and ***p<.001. n.s.: no significant difference. Scale bar: 100μm.

However, at later stages, the expression of myelin genes in the white matter region was significantly decreased in double transgenic embryos (Fig. 6j’, l’), similar to the phenotype observed in Pdgfrα-SmoM2 spinal tissues. Compared to littermate control, a mild but significant increase of Pdgfrα+ cells was observed in the double transgenic tissues (Fig. 7a-b’, c). Cell proliferation analyses by BrdU incorporation demonstrated that the cell proliferation level was elevated in Olig1-SmoM2 spinal cords (Fig. 7d, d’, f). Immunostaining for apoptosis marker Caspase3 showed a comparable number of apoptotic cells in both genotypes (Fig. 7e, e’, f), excluding the possibility that the loss of OLs was caused by increased cell death.

Figure 7.

Overexpression of Smo in Olig1+ cells enhanced OPC proliferation. (a-b’) Spinal cord sections from SmoM2 and Olig1^cre; SmoM2 were subjected to Pdgfrα ISH at indicated stages. (c) Quantification of Pdgfrα+ cells per section in spinal tissues at indicated stages. (d-e’) Spinal cord sections from SmoM2 and Olig1^cre; SmoM2 were subjected to BrdU and cCaspase3 IF. (f) Quantification of BrdU+ and cCaspase3+ cells per section in spinal tissues at indicated stages. Error bars indicate SEM. *p<.05, **p<.005 and ***p<.001. n.s.: no significant difference. Scale bar: 100μm.

To explore the possible mechanisms underlying Smo enhancement of OPC proliferation, we next examined the expression of cell cycle-related gene Cyclin D1 which has been implicated in OPC proliferation (Casaccia-Bonnefil, 2003). At E14.5, it was found that a cluster of Cyclin D1+ cells were observed in the dorsal VZ in Olig1-SmoM2 spinal cords (Fig. S1b, b’), at the same position of ectopic Olig2+ cells (Fig. 6d’). However, double-labeling study detected that most Cyclin D1+ cells were not Olig2+ OPCs (Fig. S1b’). The overall percentage of Cyclin D1/Olig2 double positive cells among Olig2+ cells was relatively lower in Olig1-SmoM2 embryos than that in littermate control (Fig. S1e), suggesting that the ectopic induction of Olig2 was not due to over-proliferation of OPCs. At E16.5, the expression level of Cyclin D1 and Olig2 were both elevated in Olig1-SmoM2 embryos (Fig. S1c, d). Quantitative analysis revealed that the percentage of Olig2+ OPCs co-expressing Cyclin D1 was dramatically increased in Olig1-SmoM2 embryos at this stage (Fig. S1e), indicative of increased OPC proliferation. Therefore, sustained expression of Smo in Olig1+ cells exerts stage-dependent effects on OL development: (1) inducing transient OPC generation and OL differentiation in the dorsal part at earlier stages when it is expressed in neural progenitor cells; (2) increasing the proliferation of OPCs and inhibiting OL differentiation in the white matter at later stages when it is expressed in OPCs.

3.6. The generation and maturation of motor neurons were unaffected in Olig1-SmoM2 embryos.

Noticeably, it was shown that Olig1^cre induced tdTomato fluorescence in other cell types, including motor neurons (MNs) (Fig. 6a, b). Hedgehog signaling has been demonstrated to play pivotal roles in the control of MN-OPC fate switch (Traiffort et al. 2016), prompting us to speculate that the neonatal death of Olig1-SmoM2 might be related to defective MN development. To evaluate the MN development in Olig1-SmoM2 embryos, we examined the expression of motor neuron specific markers, Islet1 and choline acetyltransferase (ChAT). The generation of MNs was unaffected by detection of Iselt1 during early embryogenesis (Fig. S2a-b’). At later stages, mature motor neurons labeled by ChAT immunostaining were detected normal in Olig1-SmoM2 embryos as compared to control littermates (Fig. S2c-d’). Thus, overexpression of Smo in Olig1+ cells did not perturb motor neuron development, suggesting that the neonatal death of the mutants are caused by other unknown developmental defects. The lack of early motoneuron phenotype in Olig1cre-SmoM2 spinal tissue could be due to the delayed expression of Cre in NPCs under the Olig1 promoter.

4. DISCUSSION

In this study, we provided genetic and molecular evidences that Smo-mediated Hh signaling regulates the development of oligodendrocytes in a stage-dependent manner. There are several novel findings in our study: (1) Activation of Smo in NPCs is able to trigger transient ectopic OPC generation and precocious OL differentiation; (2) Endogenous Smo-mediated Hh signaling is not absolutely required for the terminal differentiation of OPCs in vivo; (3) After the commitment of OL-lineage, sustained expression of Smo in OPCs promotes the proliferation but negatively regulates their differentiation into OLs; (4) The induction of precocious OL differentiation in NPCs by Smo was closely related to co-expression of Olig2 and Nkx2.2. Based on our findings, we propose a model of stage-specific regulation of OL development by Smo-mediated Hh signaling (Fig. 8). Before OPC specification, increased Smo-mediated Hh signaling in neural progenitor cells induces the expression of Olig2 and Nkx2.2 transcription factors. Those OPCs that express Olig2 only proliferate to expand the pool of OPCs, whereas those that express both Nkx2.2 and Olig2 differentiate prematurely into premyelinating OLs. At later stages when OPC lineage is specified, continuous activation of Smo in OPCs would stimulate their proliferation and inhibit their differentiation. Thus, Smo-mediated Hh signaling might need to be attenuated for the terminal differentiation OL during development.

Figure 8.

Schematic representation of the stage-dependent regulation of OL development by Smo. Before OPC specification, intrinsic (black) Smo-mediated Hh signaling in NPCs functions to promote the specification of OPCs through inducing Olig2 expression. After OPCs are specified, they differentiate into premyelinating OLs. Hyperactivation of Smo (red) in NPCs induces co-expression of Olig2 and Nkx2.2, and ectopic OL differentiation. However, constitutive activation of Smo (red) in OPCs further stimulates OPC proliferation and inhibits OL differentiation.

Nkx2.2 and Olig2 are previously identified as the direct targets of Hh signaling (Lu et al., 2000; Qi et al., 2001). Prior to oligodendrogenesis, Nkx2.2 and Olig2 repress each other in the ventral spinal cord to form distinct neural progenitor domains (Sun et al., 2003). During the differentiation stage of oligodendrogenesis, Nkx2.2 is upregulated in OPCs, and co-expression of Olig2 and Nkx2.2 is responsible for triggering the terminal differentiation of oligodendrocytes in embryonic spinal cords (Fu et al., 2002; Qi et al., 2001; Zhou et al., 2001, Zhu et al., 2014). Consistently, in Nestin-SmoM2 spinal cords, the Nkx2.2/Olig2 co-expressing domain (Fig. 3e) was found at the position of ectopic precocious OL differentiation (Fig. 3g). Therefore, we postulate that SmoM2 expression in neural progenitors induces ectopic and precocious OLs by inducing Olig2-Nkx2.2 co-expression.

The complex phenotypes in Olig1-SmoM2 embryos (Fig. 6) are likely caused by the unique temporal-spatial expression pattern of Olig1 in the developing cords. Olig1 is initially expressed in the ventral pMN domain (at E9.5-E10.5) prior to oligodendrogenesis, and later in dorsal neural progenitor cells (at around E14.5) with a lower level based on immunostaining and in situ hybridization (Cai et al., 2005; Vallstedt et al., 2005). In the dorsal spinal cord of Olig1-SmoM2 embryos, Olig1^cre-driven weak expression of Smo led to a lower level of Hh activity, causing increased proliferation of ventricular progenitor cells (Fig. S1B), and a fraction of them became OLs. As a result, co-expression of Olig2 and Nkx2.2 was induced (Fig. 6D’, G’) and ectopic precocious Mbp+/Plp1+ OLs were observed in the dorsal spinal tissues (Fig. 6I’, K’). However, in the ventral part of Olig1-SmoM2 tissues, there was a lack of Olig2 and Nkx2.2 co-expression (Fig. 6G’). One plausible explanation is that a high level of Hh activity in ventral ventricular zone induced Nkx2.2 expression, which then inhibited Olig2 expression due to cross inhibition and repressed oligodendrogenesis, resulting in reduction of OLs. Indeed, the number of Olig2+ cells was markedly reduced in the ventral region, while that of Nkx2.2+ cells significantly increased (Fig. 6E, H). This could explain the absence of ectopic precocious Mbp+/Plp1+ OLs in Olig1-SmoM2 ventral spinal cords (Fig. 6I-L’).

Intriguingly, cross-repression of Olig2 and Nkx2.2 that is normally observed in NPCs and early OPCs (Sun et al., 2003) is lost in differentiating OPCs and SmoM2-expressing NPCs (Fig. 3). The underlying mechanism is not known at this stage. Recently, nuclear factor of activated T cells (NFAT) proteins have been identified to be essential for the terminal differentiation of OLs (Weider et al., 2018). Functional analyses revealed that NFAT proteins cooperate with Sox10 to relieve the cross-repression of Olig2 and Nkx2.2 in differentiating OPCs at later stage. In addition, previous findings demonstrated that Nfat factors are also expressed in neural progenitor cells, and act to promote the proliferation and differentiation of NPCs (Serrano-Perez et al., 2015). The interplay between Hh signaling and Nfat proteins might disrupt the mutual repression of Nkx2.2 and Olig2 proteins in NPCs when Smo activity is high in transgenic tissues, resulting in their co-induction and the formation of ectopic precocious OLs. Thus, the co-repression of Nkx2.2 and Olig2 transcription factors in the developing spinal tissues could be a dynamic phenomenon, depending on developmental stages and cell context.

Main points:

• Activation of Smo in NPCs is able to trigger transient ectopic OPC generation and precocious OL differentiation;

• After the commitment of OL-lineage, sustained expression of Smo in OPCs promotes the proliferation but negatively regulates their differentiation into OLs;

• The induction of precocious OL differentiation in NPCs by Smo was closely related to co-expression of Olig2 and Nkx2.2.