Author contributions: H.T.: conception and design, manuscript writing, collection, data analysis and interpretation; N.T.-S. and K.N.: data analysis and interpretation; A.W.: provision of study material; H.A.: financial support, provision of study material; T.K.: conception and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSExpress October 16, 2008.
Multipotential neural stem cells (NSCs) in the central nervous system (CNS) proliferate indefinitely and give rise to neurons, astrocytes, and oligodendrocytes. As NSCs hold promise for CNS regeneration, it is important to understand how their proliferation and differentiation are controlled. We show here that the expression of sox2 gene, which is essential for the maintenance of NSCs, is regulated by the Gli2 transcription factor, a downstream mediator of sonic hedgehog (Shh) signaling: Gli2 binds to an enhancer that is vital for sox2 expression in telencephalic neuroepithelial (NE) cells, which consist of NSCs and neural precursor cells. Overexpression of a truncated form of Gli2 (Gli2ΔC) or Gli2-specific short hairpin RNA (Gli2 shRNA) in NE cells in vivo and in vitro inhibits cell proliferation and the expression of Sox2 and other NSC markers, including Hes1, Hes5, Notch1, CD133, and Bmi1. It also induces premature neuronal differentiation in the developing NE cells. In addition, we show evidence that Sox2 expression decreases significantly in the developing neuroepithelium of Gli2-deficient mice. Finally, we demonstrate that coexpression of Gli2ΔC and Sox2 can rescue the expression of Hes5 and prevent premature neuronal differentiation in NE cells but cannot rescue its proliferation. Thus these data reveal a novel transcriptional cascade, involving Gli2 → Sox2 → Hes5, which maintains the undifferentiated state of telencephalic NE cells. STEM CELLS2009;27:165–174
The central nervous system (CNS) contains multipotent neural stem cells (NSCs) that are capable of self-renewal and give rise to neurons, astrocytes, and oligodendrocytes [1, 2]. Both extracellular and intracellular factors help to maintain NSC proliferation and multipotentiality; however, details of the mechanisms involved are not known.
Sox2 is a member of the SRY-related box (Sox) gene family, which encodes transcription factors with a high-mobility group DNA-binding domain . Several lines of evidence suggest that Sox2 maintains the developmental potential and proliferation of NSCs [4–7]: neuroepithelial (NE) cells in the ventricular/subventricular zone (VZ/SVZ) express Sox2 but progressively lose it as the cells differentiate [4–7]. Overexpression of Sox2 in NE cells blocks differentiation, whereas overexpression of a dominant-negative form of Sox2 results in the loss of NSC markers and the acquisition of early neuronal markers [6, 7]. Together, these findings have suggested that Sox2 is required for NSCs to maintain their “stemness.”
The way in which sox2 expression is maintained in NE cells remains unclear. We previously identified an enhancer (designated R1) that is required for sox2 expression in telencephalic NE cells. This R1 sequence associates with Brm-containing SWI/SNF complex, and the factors within this complex induce sox2 expression in cultured NE cells . As it is unlikely that Brm recognizes a specific DNA sequence, we have attempted to determine what DNA-binding factor recognizes the R1 enhancer and regulates sox2 expression in telencephalic NE cells.
In the present study, we show that Gli2, a key mediator of sonic hedgehog (Shh) signaling, binds to the R1 enhancer in cultured NE cells, whereas two other mediators of Shh signaling, Gli1 and Gli3, do not. Overexpression of either a truncated form of Gli2 (Gli2ΔC) or Gli2-specific short hairpin RNA (Gli2 shRNA) in NE cells, in vivo and in vitro, inhibits the expression of Sox2 and other NSC markers, such as Hes1, Hes5, Notch1, CD133, and Bmi1; it also induces the premature differentiation of NE cells into βIII-tubulin-positive neurons in vivo. We present evidence that Sox2 expression is reduced significantly in the developing NE cells of Gli2-deficient mice. Moreover, we demonstrate that overexpression of sox2 in Gli2ΔC-expressing NE cells blocks premature neuronal differentiation and rescues the expression of Hes5, but not that of Hes1. Together, these findings reveal that Gli2 is a novel regulator of sox2 expression in telencephalic NE cells.
MATERIALS AND METHODS
Animals and Chemicals
Animals were obtained from the Laboratory for Animal Resources and Genetic Engineering at RIKEN Center for Developmental Biology. Gli2 mutant mice were provided by J. Motoyama . Chemicals were purchased from Sigma-Aldrich (St. Louis, MO http://www.sigmaaldrich.com), except where indicated.
NE cells were prepared from embryonic day (E) 14.5 mouse telencephalon and expanded in basic fibroblast growth factor (bFGF; 10 ng/ml; Peprotech, Rocky Hill, NJ, http://www.peprotech.com) as described before . Transfection of NE cells was performed as described previously . In brief, the cells (2 × 106) were transfected with the vector (10 μg) using the Nucleofector device (Amaxa Inc., Gaithersburg, MD, http://www.amaxa.com) and then cultured in bFGF. To examine the effects of the transgenes, the transfected cells were harvested, and 5 × 104 cells per well were recultured on poly-L-ornithine (15 μg/ml)- and fibronectin (1 μg/ml; Invitrogen, Carlsbad, CA, http://www.invitrogen.com)-coated cover glasses (f13 μm; Marienfeld, Lauda-Königshofen, Germany, http://www.marienfeld-superior.com).
In Utero Electroporation
In utero electroporation was performed using E14.5 ICR mice, as described before . In brief, 1–2 μg of plasmid was injected into the left side of the lateral ventricle of the embryonic brain and electroporated to the cerebral cortex with an Electro Square Porator BTX 830 (BTX Harvard Apparatus, Holliston, MA, http://www.btxonline.com).
The dissected mouse brains were fixed in 4% paraformaldehyde overnight at 4°C. After fixation, the brains were cryoprotected with 12%–18% sucrose in phosphate-buffered saline (PBS) and embedded in OCT (Sakura Finetek, Torrance, CA, http://www.sakura.com). Coronal sections (10 μm thick) were prepared from the cerebral cortex. The sections were pretreated with 0.3% Triton X-100 in PBS for penetration and then treated with blocking solution (2% skim milk, 0.3% Triton X-100; PBS) for 1 hour and incubated with primary antibodies for 16 hours at 4°C. Immunostaining of the fixed cells was carried out as described previously . The following antibodies were used to detect intracellular antigens: rabbit polyclonal anti-Sox2 (diluted 1:1,000; StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com) (specificity of the Sox2 antibodies is shown in supporting information Fig. 1), rabbit polyclonal anti-Hes1 (diluted 1:200; Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), rabbit polyclonal anti-Hes5 (diluted 1:100; Abcam, Cambridge, United Kingdom, http://www.abcam.com), rabbit polyclonal anti-Gli2 (diluted 1:200; Abcam), rabbit polyclonal anti-CD133 (diluted 1:100; Abcam), mouse monoclonal anti-βIII-tubulin (diluted 1:200; Sigma-Aldrich), mouse monoclonal anti-5-bromo-2′-deoxyuridine (anti-BrdU; diluted 1:400; G3G4; Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww), rat monoclonal anti-green fluorescent protein (anti-GFP; diluted 1:1,000; Nacalai Tesque Inc., Kyoto, Japan, http://www.nacalai.co.jp/en), and rabbit polyclonal anti-GFP (diluted 1:1,000; Abcam). The antibodies were detected with Alexa 568-conjugated goat anti-mouse or rabbit IgG (diluted 1:200; Molecular Probes, Eugene, OR, http://probes.invitrogen.com) or Alexa 488-conjugated goat anti-rabbit or rat IgG (diluted 1:200; Molecular Probes). To visualize all nuclei, the cells were counterstained with Hoechst 33342 (1 μg/ml). Fluorescence images were obtained by using a Fluoview FV1000 confocal laser-scanning microscope (Olympus, Tokyo, http://www.olympus-global.com), and Sox2 intensity (arbitrary units) was analyzed by FV10-ASW, version 01.06.01.00 (Olympus).
For BrdU staining, 500 μl of 10 μg/ml BrdU was injected into the pregnant mice intraperitoneally, and the pups were dissected 1 hour later. For the terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay, In Situ Cell Death Detection Kit TMR red was used according to the supplier's instructions (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com).
Luciferase assays were carried out as described previously . In brief, NE cells were transfected with 1 μg of sox2-enhancer-firefly luciferase expression vectors and 0.04 μg of the internal control vector pEF-Rluc (gift from S. Nagata and K. Shimozaki, Osaka University, Osaka, Japan) using the Superfect Transfection Reagent according to the supplier's instructions (Qiagen, Hilden, Germany, http://www1.qiagen.com). After 24 hours, the activities of the two luciferases were measured using the Dual-Luciferase Reporter Assay System according to the supplier's instructions (Promega, Madison, WI, http://www.promega.com).
Chromatin Immunoprecipitation Assay
Chromatin immunoprecipitation (ChIP) assays were carried out as described previously . In short, NE cells were transfected with a control vector, pcDNA3.1-HisB-mGli1, pCMV/FLAG/SV-Gli2 full (both HisB-mGli1 and FLAG/SV-Gli2 expression vectors were gifts from H. Sasaki, RIKEN CDB, Kobe, Japan), or pcDNA3-FLAG-Gli3 using a Nucleofector device (Amaxa). On the following day, transfected NE cells were fixed and sonicated. The cell extracts were harvested by centrifugation and immunoprecipitated with anti-FLAG M2 antibody (1:100 dilution) or anti-His antibody (1:100 dilution; Santa Cruz Biotechnology), and protein G-Sepharose (50 μl of a 50% suspension; GE Healthcare, Little Chalfont, U.K., http://www.gehealthcare.com). The precipitated DNA fragments were purified by phenol/chloroform extraction and used for polymerase chain reaction (PCR) with primer sets for R1, R3, and Patched1 promoter (PTCH1). The following oligonucleotides were synthesized to amplify PTCH1: the 5′ primer was 5-CGCTCGCAAAAGGCGTCTCG-3, and the 3′ primer was 5-AAAGGCTGGAACTCCCGCCC-3. PCR conditions were 30 seconds at 94°C, 30 seconds at 61°C, and 30 seconds at 72°C. Cycle parameters were 50 cycles for R1, R3, and PTCH1.
Reverse Transcription-PCR Analysis
Reverse transcription (RT)-PCR was carried out as described previously . The following oligonucleotides were synthesized. For sox2, the 5′ primer was 5-TTGATATCATGTATAACATGATGGAGAACGGAGCT-3, and the 3′ primer was 5-CTCCAGCCGTTCATCTGCGCGTA-3. For gapdh, the 5′ primer was 5-ACCACAGTCCATGCCATCAC-3, and the 3′ primer was 5-TCCACCACCCTGTTGCTGTA-3. For hes1, the 5′ primer was 5-CAGCCAGTGTCAACACGAC-3, and the 3′ primer was 5-TGATCTGGGTCATGCAGTTG-3. For hes5, the 5′ primer was 5-CGACTGCGGAAGCCGGTGGT-3, and the 3′ primer was 5-AGCAGCTTCATCTGCGTGTCG-3. For notch1, the 5′ primer was 5-AGCAGTCTGCCTGTGCACAC-3, and the 3′ primer was 5-AAATGCCTCTGGAATGTGGGT-3. For bmi1, the 5′ primer was 5-GGGTACTTCATTGATGCCAC-3, and the 3′ primer was 5-AGGTTCCTCTTCATACATGAC-3. For c-myc, the 5′ primer was 5-CACTGGTCCTCAAGAGGTGC-3, and the 3′ primer was 5-TGTCGACTTATGCACCAGAGTTTCGAAGC-3. For bcl2, the 5′ primer was 5-AGAATTCGCCACCATGGCGCAAGCCGGGAGAAC-3, and the 3′ primer was 5-TCTCGAGTCACTTGTGGCCCAGGTATGC-3. PCR conditions for sox2 and gapdh were 30 seconds at 94°C, 30 seconds at 58°C, and 40 seconds at 72°C, and cycle parameters were 45 cycles for sox2 and 35 cycles for gapdh. PCR conditions for hes1, hes5, notch1, and bmi1 were 30 seconds at 94°C, 30 seconds at 55°C, and 40 seconds at 72°C, with cycle parameters of 50 cycles for hes1, 40 cycles for hes5, 55 cycles for notch1, and 65 cycles for bmi1. PCR conditions for c-myc and bcl2 were 30 seconds at 94°C, 30 seconds at 59°C, and 50 seconds at 72°C, and cycle parameters were 40 cycles.
Statistical analyses of Sox2 intensity and proportion of Hes1, Hes5, βIII-tubulin, TUNEL, and BrdU-positive cells from immunohistochemistry were carried out by analyzing at least four slide glasses from one embryo. At least 100 cells were counted from each slide, and statistical analyses were performed using Excel Stacel2 (Microsoft, Redmond, WA, http://www.microsoft.com). Similar results were obtained from three independent embryos, and representative results are shown in each figure.
Gli2 Is an Essential Factor for Sox2 Expression in Cultured NE Cells
We first examined whether the R1 enhancer contains any recognizable sequences for transcription factor binding and found two potential Gli-binding sites (Fig. 1A, GBS1 and GBS2). There are three mammalian Gli proteins (Gli1–Gli3) that are regulated by Shh signaling, which plays a crucial part in CNS development . It has been shown that Gli2 and Gli3 are expressed in the VZ/SVZ and are required for the proliferation of NE cells [13–15]; although Gli1-deficient mice are capable of developing normally, both Gli2-deficient and Gli3-deficient mice die during the perinatal period, with severe defects in the brain [14, 16–18]. Gli2 is a transcriptional activator, and Gli3 acts primarily as a repressor, although it is converted to a weak activator in the presence of Shh [12, 19, 20]. Moreover, we found that Gli2 is strongly expressed in NE cells, most of which are positive for Sox2, in culture and in developing VZ/SVZ (supporting information Fig. 2). Together, these findings suggested that Gli2 may be a primary regulator of sox2 expression in NE cells.
To test this possibility, we transfected NE cells with either a pCMS-enhanced green fluorescent protein (EGFP) control vector or a pCMS-EGFP-Gli2ΔC (Gli2ΔC) vector. One day after transfection, the cells were fixed and immunostained for Sox2 and GFP; more than 95% of control vector-transfected NE cells were labeled for Sox2, whereas less than 20% of the Gli2ΔC-expressing cells were so labeled (Fig. 1B, 1C). The proportion of Sox2-positive cells among the Gli2ΔC-transfected cells had decreased during the culture period (Fig. 1C). To confirm that overexpression of Gli2ΔC prevents sox2 expression, we transfected NE cells with either the control vector or the Gli2ΔC vector, purified GFP+ cells by flow cytometry, and then examined sox2 expression by RT-PCR. As is shown in Figure 1D, overexpression of Gli2ΔC significantly inhibited sox2 expression in the cultured NE cells. As Gli2ΔC has been shown to interfere with the function of Gli2 and Gli3 [12, 19], we constructed three other vectors: a control vector encoding GFP and a scrambled shRNA (scrambled), a Gli2-specific shRNA expression vector encoding GFP and a Gli2-specific shRNA (Gli2 shRNA), and a Gli3-specific shRNA expression vector encoding GFP and Gli3-specific shRNA (Gli3 shRNA) (specificity of Gli2 and Gli3 shRNAs is shown in supporting information Fig. 3). We transfected the NE cells with these vectors, used flow cytometry to purify GFP+ cells, and then analyzed GFP+ cells for sox2 expression by RT-PCR as described above. Overexpression of Gli2 shRNA significantly knocked down sox2 expression in cultured NE cells, whereas overexpression of either scrambled or Gli3 shRNA did not, indicating that Gli2, but not Gli3, is required for sox2 expression in cultured NE cells.
Gli2 Associates with the R1 Enhancer of the Sox2 Promoter in Cultured NE Cells
To address whether Gli2 directly regulates sox2 expression in cultured NE cells, Gli-binding sites (GBS; Fig. 1A, GBS1 and GBS2) in the R1 enhancer were first examined to determine which is critical for sox2 expression. GBS1 and GBS2 sequences were replaced with random sequences, making mR1-GBS1 and mR1-GBS2, respectively; they were inserted into the pGL3 promoter vector, which contained a minimal SV40 promoter upstream of a firefly luciferase gene (Fig. 2A, left schematic diagrams). The NE cells were transfected with R1, R3 (which has no enhancer activity), mR1-GBS1, or mR1-GBS2, along with a vector (pEF-Rluc) encoding a sea pansy luciferase that we used previously . Twenty-four hours after transfection, the cells were assayed for firefly luciferase activity and normalized against the sea pansy luciferase activity. As shown in Figure 2A (right panel), both the R1 and the mR1-GBS1 vectors produced significant luciferase activity, whereas neither the R3 nor the mR1-GBS2 vector did so, indicating that GBS2 is an essential site for sox2 expression in cultured NE cells. We also examined whether overexpression of Gli2ΔC inhibits the R1-dependent luciferase activity in cultured NE cells. The pGL3 control vector and R1 vector were transfected with either the pCMS-EGFP control vector or the Gli2ΔC vector into cultured NE cells, and their luciferase activity was assayed after 24 hours. As shown in Figure 2B, overexpression of Gli2ΔC strongly blocked the R1-dependent luciferase activity. Together, these results suggest that Gli2 targets GBS2 in the R1 enhancer and regulates sox2 expression in cultured NE cells.
To examine further whether Gli2 binds directly to the R1 enhancer in cultured NE cells, we used ChIP. NE cells were transfected with either a pcDNA3/FLAG control vector or a FLAG-tagged mouse Gli2 expression vector, and the cells were cultured for 24 hours. The cells were then fixed in formaldehyde and sonicated, cell extracts were incubated with an anti-FLAG antibody and protein G-Sepharose, and the precipitated DNA was analyzed using PCR. As shown in Figure 2C, FLAG-Gli2 associated with R1 but not R3. The other Gli members, Gli1 and Gli3, were also examined to determine whether they bind to the R1 enhancer in cultured NE cells as described above. Figure 2D shows that neither Gli1 nor Gli3 associated with the R1 enhancer in cultured NE cells, whereas all Gli proteins bind to PTCH1, which contains a definitive GBS . Together, these data indicate that some Gli2 binds to the R1 enhancer in cultured NE cells, whereas Gli1 and Gli3 do not.
Because we showed previously that Brm, an SWI/SNF chromatin remodeling factor, is recruited to R1 and induces sox2 expression in NE cells , we examined whether Gli2 associates with Brm in the cells. We sonicated NSCs in a cell lysis buffer, harvested the extracts, treated them with either anti-FLAG or anti-Brm antibodies followed by protein A-Sepharose, and then analyzed the immunoprecipitates by Western blotting using anti-FLAG and anti-Brm antibodies. As shown in supporting information Figure 4, anti-Brm antibodies precipitated endogenous Gli2, whereas anti-FLAG antibody did not, indicating that some Brm and Gli2 are in the same complexes in cultured NE cells.
Gli2 Regulates Sox2 Expression in Developing Telencephalic NE Cells
We tested whether Gli2 is involved in sox2 expression in NE cells in vivo. We electroporated either the control GFP-expression vector or the Gli2ΔC vector into the NE cells lining the ventricle in E14.5 mice, fixed the transfected brains, sectioned them, and immunostained the sections for Sox2 and GFP. One day after the transfection, the Sox2-positive VZ/SVZ layer became significantly thinner in the brain transfected with Gli2ΔC compared with those transfected with the control vector (Fig. 3A, left panels), and 2 days after transfection, the Sox2-positive layer became disorganized in the Gli2ΔC-transfected brain (Fig. 3A, right panels). The Sox2 intensity in the Gli2ΔC-expressing cells significantly decreased to less than 50% of that observed in the control vector-transfected cells (Fig. 3B). A similar result was also obtained when Gli2ΔC vector was electroporated into the NE cells of E16.5 mice (data not shown).
To confirm that Gli2 is essential for sox2 expression in vivo, we electroporated either the control scrambled vector, the Gli2 shRNA vector, or the Gli3 shRNA vector into the NE cells and labeled sections of the brain for Sox2 and GFP after 2 days. As shown in Figure 3C and 3D, overexpression of Gli2 shRNA significantly decreased Sox2 expression levels in the developing NE cells, whereas either scrambled or Gli3 shRNA did not. Collectively, these data indicate that Gli2 plays an important role in sox2 expression in developing telencephalic NE cells, as well as in cultured NE cells.
Overexpression of Gli2ΔC Induces Cell Cycle Arrest and Cell Death in the Developing Telencephalic NE Cells
To analyze the fates of Gli2ΔC-expressing telencephalic NE cells, the cultured NE cells were transfected with either the control vector or the Gli2ΔC vector, and then their proliferation and cell death were examined. The transfected cells were cultured with BrdU for 6 hours and then immunolabeled for BrdU and GFP. None of Gli2ΔC-transfected cells were positive for BrdU, whereas 38% of control vector-transfected cells were BrdU-positive (supporting information Fig. 5A). To detect cell death, we performed a TUNEL assay and found that more than 16% and 32% of Gli2ΔC-expressing GFP+ cells were TUNEL+ at 24 and 48 hours, respectively, whereas less than 3% and 2% of control vector-transfected cells were TUNEL+ (supporting information Fig. 5B–5D). Moreover, we found that Gli2ΔC-induced cell death was blocked significantly in the presence of the pan-caspase inhibitor Z-VAD (supporting information Fig. 5B–5D).
We also examined the proliferation and cell death of Gli2ΔC-expressing NE cells in vivo. Two days after electroporation, BrdU was injected intraperitoneally into pregnant mice 1 hour before sacrifice, and brain sections of the embryo were immunostained for BrdU and GFP. As shown in Figure 4A and 4B, 69% of GFP+ cells were labeled for BrdU in the control vector-transfected brain, whereas 24% of GFP+ cells were labeled for BrdU in the Gli2ΔC-transfected brain. Moreover, the proliferating cell nuclear antigen (PCNA)-positive cell layer became thin in Gli2ΔC-transfected cells (supporting information Fig. 6A, 6B). Cell death in Gli2ΔC-expressing NE cells was examined using the TUNEL assay. As shown in Figure 4C and 4D, less than 0.2% of the control vector-transfected NE cells were TUNEL+, whereas 2.8% of Gli2ΔC-expressing cells were TUNEL+. These results suggest that overexpression of Gli2ΔC induces both cell cycle arrest and cell death in the telencephalic NE cells.
Overexpression of Gli2ΔC Induces Premature Neuronal Differentiation of Developing Telencephalic NE Cells
There is evidence that both Gli2 and Sox2 are involved in the expression of a number of NSC genes, including Bmi1 and Notch1 [22–24]. We therefore examined the relationship of Gli2, Sox2, and their target genes in NE cells. We transfected cultured NE cells with either the control EGFP-expression vector or the Gli2ΔC vector, cultured the cells for 2 days, purified GFP+ cells by flow cytometry, and then analyzed the expression of NSC genes in GFP+ cells by RT-PCR. As shown in Figure 5A, overexpression of Gli2ΔC significantly decreased the expression levels of sox2, hes1, hes5, notch1, and bmi1 in cultured NE cells. Using Gli2 shRNA and Gli3 shRNA, we confirmed that overexpression of Gli2 shRNA, but not that of Gli3 shRNA, inhibits the expression of same genes in cultured NE cells (Fig. 5B). This suggests that Gli2 may be coordinately stimulating the expression of a number of NSC genes.
We examined the differentiation of Gli2ΔC-expressing NE cells by immunolabeling in vitro and in vivo. Seven percent and 94% of control vector-transfected cultured NE cells were labeled for a neuronal marker, βIII-tubulin, and an NSC marker, Nestin, respectively, whereas 48% and 41% of Gli2ΔC-expressing cells were so labeled (supporting information Fig. 7A, 7B, 7D). In contrast, overexpression of Gli2ΔC did not have any significant effect on GFAP expression in cultured NE cells (3% of control vs. 4% of Gli2ΔC-transfected cells) (supporting information Fig. 7C, 7D). We confirmed that overexpression of Gli2ΔC induces premature neuronal differentiation in NE cells in vivo: 8% of control vector-transfected cells and 27% of Gli2ΔC-expressing cells were positive for βIII-tubulin (Fig. 6A, 6D). Moreover, we found that overexpression of Gli2ΔC decreased the expression of Hes1 and Hes5 in NE in vivo: 60% and 58% of control vector-transfected NE cells were labeled for Hes1 and Hes5, respectively, whereas 38% and 27% of Gli2ΔC-expressing cells were so labeled (Fig. 6B, 6C, 6E, 6F). In addition, we found that expression of CD133, a marker of NSCs, significantly decreases in Gli2ΔC-expressing NE cells (90% of control vs. 9% of Gli2ΔC-transfected cells) (supporting information Fig. 8). Taken together, these data suggest that overexpression of Gli2ΔC induces premature neuronal differentiation by inhibiting the expression of Notch effectors.
Reduction of the Sox2-Positive Layer in Gli2-Deficient Mice
To further confirm the results that the Gli2 function defects decrease Sox2 expression in telencephalic NE cells, we examined Sox2 expression in developing NE cells of Gli2-deficient mouse brain. We fixed E18.5 Gli2−/− brains, sectioned them, and immunostained them for Sox2, Hes5, and βIII-tubulin. As shown in Figure 7A, the Sox2-positive layer became extremely thin and the intensity of Sox2 immunostaining was significantly weak in Gli2-deficient brains, whereas Sox2 expression in Gli2+/− brains was similar to that in Gli2+/+ brains (data not shown). In addition, the βIII-tubulin-positive layer expanded into the VZ/SVZ of Gli2−/− brain, as seen in Gli2ΔC-transfected brains (Fig. 7A). The Hes5-positive layer also became noticeably thinner in Gli2−/− brain (Fig. 7B), consistent with premature neuronal differentiation in the VZ/SVZ of Gli2−/− brain. Thus, these data complement data that suggest that Gli2 is an important transcription factor for sox2 expression in telencephalic NE cells.
Overexpression of Sox2 Rescues Hes5 Expression in the Gli2ΔC-Expressing Telencephalic NE Cells
The finding that Sox2 is one of the downstream factors of Gli2 in telencephalic NE cells led us to examine whether Sox2 is involved in the expression of Gli2 target genes, including Bmi1 and Bcl2 [22, 25]. We transfected NE cells with either the scrambled vector or a sox2 shRNA that encodes sox2-specific shRNA and GFP, cultured the cells for 2 days, purified the GFP+ cells by flow cytometry, and then analyzed gene expression by RT-PCR. As shown in Figure 5B, overexpression of sox2 shRNA prevented the expression of sox2, hes1, and hes5 but not that of notch1, bmi1, or bcl2. Thus, Sox2 does not appear to regulate expression of the putative Gli2 target genes bmi1 and bcl2.
We next examined whether overexpression of sox2 could rescue the expression of Hes1 and Hes5 in Gli2ΔC-expressing NE cells. We electroporated pCMS-EGFP-Gli2ΔC-IRES-Sox2 (Gli2ΔC-Sox2) vector into developing NE cells and immunolabeled the brain sections for Hes1 and Hes5. Overexpression of sox2 slightly rescued the expression of Hes1 in Gli2ΔC-expressing NE cells; whereas 39% of Gli2ΔC-expressing cells were positive for Hes1, 43% of Gli2ΔC-Sox2-expressing cells were positive for Hes1 (Fig. 6B, 6E). In contrast, overexpression of sox2 strongly increased the expression of Hes5 in the transfected NE cells; whereas 25% of Gli2ΔC-expressing cells were positive for Hes5, 97% of Gli2ΔC-Sox2-expressing cells were found to be positive for Hes5 (Fig. 6C, 6F). Consistent with Hes5 upregulation in Gli2ΔC-Sox2-expressing cells, only 1% of Gli2ΔC-Sox2-expressing NE cells were positive for βIII-tubulin, compared with 26% for control vector-transfected cells (Fig. 6A, 6D). Moreover, we confirmed that overexpression of Hes5 is sufficient to block premature neuronal differentiation in Gli2ΔC-expressing NE cells (supporting information Fig. 9).
We also immunolabeled the brain sections for PCNA to examine proliferation of Gli2ΔC-Sox2-expressing NE cells. It was found that overexpression of sox2 could enhance the proliferative defects of Gli2ΔC-expressing NE cells (supporting information Fig. 6). Sox2, therefore, appears to play an important part in keeping NE cells undifferentiated, but it is not sufficient, in the absence of Gli2, to rescue the proliferative defect observed.
Many transcription factors have been shown to be involved in the maintenance of NSCs; however, the precise function of each one remains uncertain. We show here that expression of Sox2, an essential transcription factor for NSC maintenance, is regulated by Gli2, also an important transcription factor in NSCs, in the telencephalic NE cells. Since all three Gli family members have been shown to recognize the GBS and share redundant functions, it was speculated that Gli1 and Gli3 might also be involved in sox2 expression. However, our data clearly show functional differences between Gli2 and other Gli members in sox2 expression in telencephalic NE cells: Gli2 binds to the R1 enhancer that is essential for sox2 expression in telencephalic NE cells, whereas Gli1 and Gli3 do not. Overexpression of either Gli2ΔC or Gli2 shRNA decreased sox2 expression in telencephalic NE cells, whereas overexpression of Gli3 shRNA did not change sox2 expression. Furthermore, sox2 expression is greatly reduced in the developing telencephalic NE cells of Gli2-deficient mice. Many lines of evidence also suggest that each of the Gli members has specific functions. For instance, Gli1 and Gli2 are transcriptional activators, whereas Gli3 primarily acts as a repressor, although it can act as a weak activator in the presence of Shh [12, 19, 20]. Gli1-deficient mice are capable of developing normally, whereas both Gli2-deficient and Gli3-deficient mice die during the perinatal period with severe brain defects [14, 16–18]. Moreover, expression of both copies of Gli1 in place of Gli2 in mice cannot completely rescue the prenatal death observed in Gli2-deficient mice . Thus it is of great interest to investigate the specific functions of each Gli transcription factor in NE cells.
The brain phenotype of Gli2-deficient mice is much milder than that seen in Sox2-deficient mice; the Sox2-deficient brain has cerebral malformations with parenchymal loss and ventricle enlargement, whereas the Gli2-deficient brain has thin telencephalic vesicles and a reduced tectum and cerebellum [14, 27]. These results suggest that transcription factors other than Gli2 are also involved in sox2 expression in telencephalic NE cells. Indeed, as shown in Figure 7, one cell layer lining the ventricle in Gli2-deficient brain is still immunolabeled for Sox2, although the signal is much weaker than that in Gli2+/− brain. In addition, it was shown that wnt signaling and the POU-domain transcription factors Brn1 and Brn2 also regulate sox2 expression in NE cells [28, 29]. Since sox2 expression in the developing CNS depends on a patchwork of separate enhancers, one of which contains the R1 enhancer , it is of interest to determine transcription factors that regulate sox2 expression in other brain regions.
We also showed that overexpression of Gli2ΔC in telencephalic NE cells decreases the expression of other NSC factors, including Notch1, Hes1, Hes5, and Bmi1, and induces premature neuronal differentiation. Moreover, coexpression of sox2 with Gli2ΔC in telencephalic NE cells rescues the expression of Hes5, but not Hes1, and prevents neuronal differentiation; this gene combination, however, does not rescue NE cell proliferation, suggesting that although Sox2 is crucial for the maintenance of NE cells, it might not be directly involved in their proliferation. This result is consistent with previous reports that enforced expression of sox2 in the developing NE cells maintains the characteristics of NSCs but also decreases their proliferation, although the way in which sox2 overexpression decreases cell proliferation remains unclear .
It has been shown that overexpression of Sox2 in NE cells can induce the expression of Notch1, which regulates Hes5 expression through the activation of RBPjk [23, 24]. However, coexpression of Gli2ΔC and sox2 in telencephalic NE cells was found to activate Hes5 expression but not Notch1 expression (our unpublished observations). Since the mouse Hes5 promoter contains a putative Sox2-binding site (−123 CTTTGTG −117), Sox2 might directly induce Hes5 expression in telencephalic NE cells. Together, these findings suggest that Gli2 stimulates the self-renewal of NSC by inducing the expression of Bmi1 and Hes1 [31, 32] and maintains the undifferentiated state of NSC by inducing the expression of Sox2.
There is growing evidence that all three Gli proteins are strongly expressed in malignant brain tumors [33, 34]. There is also evidence that brain tumors contain cancer stem cells, which are essential for both the development and recurrence of the tumors [35–38]. A recent study has revealed that cyclopamine, a chemical inhibitor of Shh signaling, can block the growth of gliomas, suggesting that Shh-Gli signaling could potentially be major targets for glioma therapy . In addition, it was found that an active form of Gli2 induced sox2 gene expression in a keratinocyte cell line, which was used as a model for basal cell carcinoma . It will, therefore, be of great interest to investigate whether Sox2 plays an important part in gliomagenesis.
For the success of CNS regenerative medicine, it is important to understand how proliferation and differentiation of NSCs are controlled. In the present study we have shown that the expression of sox2 gene, which is essential for the maintenance of NSCs, is regulated by the Gli2 transcription factor. Moreover we demonstrated that Sox2 negatively regulates neuronal differentiation of telencephalic NE cells but is not involved in their proliferation. Thus, we revealed a novel transcriptional cascade, Gli2 → Sox2 → Hes5, which is essential for maintaining undifferentiated state of telencephalic NE cells.
We thank M. Raff for critical reading and comments on the manuscript; R. Lovell-Badge for mouse Sox2 5′ genomic DNA; J. Motoyama for Gli2-deficient mice; K. Shimozaki and S. Nagata for the pEF-Rluc vector; H. Sasaki for the pcDNA3.1-HisB-mGli1 and pCMV/FLAG/SV-Gli2 full vectors; K. Nakashima for technical advice on mouse NE cell preparation; T. Suetsugu, D. Konno, and F. Matsuzaki for technical advice on in utero electroporation; K. Miwa for purification of GFP+ cells by flow cytometry; H. Hiraga for critical reading of the manuscript; and members of our laboratory for their discussion and encouragement. H.T. was supported in part by the Uehara Memorial Foundation in Japan.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.