A Growth-Promoting Signaling Component Cyclin D1 in Neural Stem Cells Has Antiastrogliogenic Function to Execute Self-Renewal

Authors

  • Norihisa Bizen,

    1. Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan
    Search for more papers by this author
  • Toshihiro Inoue,

    1. Department of Ophthalmology, Faculty of Life Science, Kumamoto University, Kumamoto, Japan
    Search for more papers by this author
  • Takeshi Shimizu,

    1. Division of Neurobiology and Bioinformatics, National Institute for Physiological Sciences, Okazaki, Japan
    Search for more papers by this author
  • Kouichi Tabu,

    1. Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan
    Search for more papers by this author
  • Tetsushi Kagawa,

    Corresponding author
    1. Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan
    • Correspondence: Tetsuya Taga, Ph.D., Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan. Telephone/Fax: 81-3-5803-5814 and 5816; e-mail: taga.scr@mri.tmd.ac.jp; or Tetsushi Kagawa, Ph.D., Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan. Telephone/Fax: 81-3-5803-5814 and 5816; e-mail: kagawa.scr@mri.tmd.ac.jp

    Search for more papers by this author
  • Tetsuya Taga

    Corresponding author
    1. Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan
    • Correspondence: Tetsuya Taga, Ph.D., Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan. Telephone/Fax: 81-3-5803-5814 and 5816; e-mail: taga.scr@mri.tmd.ac.jp; or Tetsushi Kagawa, Ph.D., Department of Stem Cell Regulation, Medical Research Institute, Tokyo Medical and Dental University (TMDU), Tokyo, Japan. Telephone/Fax: 81-3-5803-5814 and 5816; e-mail: kagawa.scr@mri.tmd.ac.jp

    Search for more papers by this author

Abstract

Self-renewing proliferation of neural stem cells (NSCs) is intimately linked to the inhibition of neuronal and glial differentiation, however, their molecular linkage has been poorly understood. We have proposed a model previously explaining partly this linkage, in which fibroblast growth factor 2 (FGF2) and Wnt signals cooperate to promote NSC self-renewal via β-catenin accumulation, which leads to the promotion of proliferation by lymphoid enhancer factor (LEF)/T-cell factor (TCF)-mediated cyclin D1 expression and at the same time to the inhibition of neuronal differentiation by β-catenin-mediated potentiation of Notch signaling. To fully understand the mechanisms underlying NSC self-renewal, it needs to be clarified how these growth factor signals inhibit glial differentiation as well. Here, we demonstrate that cyclin D1, a NSC growth promoting signaling component and also a common component of FGF2 and Wnt signaling pathways, inhibits astroglial differentiation of NSCs. Interestingly, this effect of cyclin D1 is mediated even though its cell cycle progression activity is blocked. Forced downregulation of cyclin D1 enhances astrogliogenesis of NSCs in culture and in vivo. We further demonstrate that cyclin D1 binds to STAT3, a transcription factor downstream of astrogliogenic cytokines, and suppresses its transcriptional activity on the glial fibrillary acidic protein (Gfap) gene. Taken together with our previous finding, we provide a novel molecular mechanism for NSC self-renewal in which growth promoting signaling components activated by FGF2 and Wnts inhibit neuronal and glial differentiation. Stem Cells 2014;32:1602–1615

Introduction

Growth factor-mediated self-renewing proliferation of neural stem cells (NSCs) is maintained by coordination between proliferative signaling events and antidifferentiative effects of such a factor. Fibroblast growth factor 2 (FGF2; also known as basic FGF) promotes the proliferation of NSCs through the downstream pathways including PI3K-Akt and Ras-ERK pathways [1-4]. Canonical Wnt family members also promote NSC proliferation through inactivation of glycogen synthase kinase 3β (GSK3β) and resultant nuclear accumulation of β-catenin, which ultimately induces cyclin D1 expression via activation of lymphoid enhancer factor (LEF)/T-cell factor (TCF) family of transcriptional factors [5]. In addition, Wnt/β-catenin pathway is required for the development of the mammalian cerebral cortex and hippocampus through neural stem cell regulation [6-13]. Recently, we have found that FGF2 and Wnt3a cooperate in activating the common signaling pathway involving GSK3β inactivation, nuclear accumulation of β-catenin, and cyclin D1 expression to stimulate NSC proliferation [4]. Furthermore, stabilized β-catenin binds to the Notch1 intracellular domain and potentiates Hes1 expression that leads to inhibition of neuronal differentiation [4]. These findings provide a molecular basis for understanding the linkage between promotion of cell proliferation and inhibition of neuronal differentiation, which is a prerequisite for self-renewal of NSCs. Chen and Walsh reported that transgenic mice expressing stabilized β-catenin, of which the NH2 terminus was truncated, displayed enhanced expansion of the NSC population and attenuation of their cell cycle exit [14]. In the β-catenin conditional mutant mouse, spinal cord neural stem cells were not properly maintained and underwent precocious neuronal differentiation [8]. These reports support our model [4]; however, it needs to be clarified how this growth promoting pathway also inhibits glial differentiation to fully understand the molecular basis of NSC self-renewal. In this study, we demonstrate that cyclin D1, a major downstream effector of β-catenin, inhibits astroglial differentiation from NSCs. Interestingly, this antiastrogliogenic effect was mediated in a manner independent of its cell cycle progression activity, as is presented below.

Cyclin D1 forms a complex with cyclin-dependent kinase 4 and 6 (CDK4 and CDK6) and inhibits retinoblastoma protein, a negative regulator of cell cycle, resulting in promotion of the transition from G1 to S-phase [15, 16]. In the developing brain, cyclin D1 is mainly expressed in the ventricular zone (VZ) where NSCs are located [17, 18]. The central nervous system of the cyclin D1 deficient mouse showed hypoplastic retinas and cerebella and displayed the spastic leg-clasping reflex and abnormal behavior [19, 20]. Overexpression of cyclin D1 and CDK4 in NSCs shortens G1 phase, confirming their function as cell cycle regulators. In addition, their overexpression retards neurogenesis, suggesting that cyclin D1 and CDK4 inhibit neuronal differentiation [21]. Thus, cyclin D1 is involved not merely in the cell proliferation but also in proper neurogenesis.

The findings, in this study, of a novel role of cyclin D1 as a negative regulator of astroglial differentiation provides a missing piece of puzzle that fills the molecular basis of NSC self-renewal and propose a comprehensive model in which growth promoting signaling components activated by FGF2 and Wnts inhibit neuronal and astroglial differentiation during self-renewal.

Materials and Methods

Antibodies, Recombinant Proteins, and Chemical Inhibitors

Antibodies used in this study were: mouse monoclonal anti- glial fibrillary acidic protein (GFAP), anti-S100β, anti-FLAG, and rabbit polyclonal anti-β actin antibodies from Sigma-Aldrich (St. Louis, MO, http://www.sigmaaldrich.com); mouse monoclonal anti-Ki67 and rat monoclonal anti-PDGFRα antibodies from BD Pharmingen (San Diego, CA, http://www.bdbiosciences.com); rabbit polyclonal anti-Neuronal class III β-tubulin (Tuj1) antibody from Covance (Princeton, NJ, http://www.covance.com); rabbit monoclonal anti-cyclin D1 antibody from Lab Vision Corporation (Fremont, CA, http://www.thermoscientific.com); rabbit polyclonal anti-GFP antibody from MBL (Nagoya, Japan, http://www.mbl.co.jp); rat monoclonal anti-GFP antibody from Nacalai Tesque (Kyoto, Japan, http://www.nacalai.co.jp); rabbit polyclonal anti-STAT3 and anti-HA antibodies from Santa Cruz Biotechnology (Santa Cruz, CA, http://www.scbt.com); rabbit polyclonal anti-pSTAT3Y705 antibody from Cell Signaling Technology (Beverly, MA, http://www.cellsignal.com); rabbit polyclonal anti-β-gal antibody from Molecular Probes (Eugene, OR, http://www.lifetechnologies.com); rat-monoclonal anti-HA antibody from Roche Applied Science (Penzberg, Upper Bavaria, Germany, https://www.roche-applied-science.com); and mouse monoclonal anti-myc antibody from Calbiochem (San Diego, CA, http://www.merckmillipore.com). Chemical inhibitors used in this study were: SB216763 from Enzo Life Sciences (Farmingdale, NY, http://www.enzolifesciences.com); and CDK4 inhibitor from Calbiochem. Recombinant proteins used in this study were: recombinant mouse leukemia inhibitory factor (LIF) from Chemicon Intenational (Temecula, CA, http://www.millipore.com); recombinant human bone morphogenic protein 2 (BMP2) from R&D Systems (Minneapolis, MN, http://www.rndsystems.com); and recombinant human FGF-basic (FGF2) from Peprotech (Rocky Hill, NJ, http://www.peprotech.com).

Mice

For NSC preparation, pregnant ICR mice were purchased from Japan SLC, Inc (Shizuoka, Japan, http://jslc.co.jp). To monitor astroglial differentiation in vivo, GFAP reporter mice that carry nls-lacZ (nlacZ) gene knocked into the Gfap gene locus were used. These mice were killed by cervical dislocation for the assays in this study. The number of animals and their suffering were minimized in all analyses. All experiments were conducted in accordance with the guidelines of Tokyo Medical and Dental University (TMDU) Animal Care and Use Committee, and Kumamoto University Center for the Animal Resources and Development.

NSC Preperation

NSCs were isolated from embryonic day 14.5 (E14.5) ICR mouse telencephalons as described previously [4, 22]. The cells were plated on dishes that had been precoated with poly-l-ornithine (Sigma-Ardrich) and fibronectin (Invitrogen, Carlsbad, CA, http://www.lifetechnologies.com), and cultured for 4 days in N2 (25 µg/ml insulin; 100 µg/ml apotransferrin; 20 nM progesterone; 100 µM putrescine; 30 nM sodium selenite, Sigma-Aldrich; 1.27 mg/ml NaHCO3, Wako, Osaka, Japan, http://www.wako-chem.co.jp)-supplemented Dulbecco's modified Eagle medium/F-12 (DMEM/F12, pH7.2, Invitrogen) containing FGF2 (10 ng/ml). The cells were detached and replated on four well plates (Nunc, Penfield, NY, http://www.thermoscientific.com, 1 × 105 cells per well), 12 well plates (Nunc, 5 × 105 cells per well), and 60 mm dishes (Nunc, 2 × 106 cells per dish) for immunocytochemistry, luciferase assay, and Western blot analysis/chromatin immunoprecipitation (ChIP) assay, respectively.

RNA Interference

A synthetic double-stranded small interfering RNA (siRNA) for cyclin D1 (5′-ACUUGAAGUAAGAUACGGAGGGCGC-3′) and scramble siRNA as a negative control (GC 48%) were purchased from Invitrogen. Cyclin D1 siRNA or scramble siRNA (30 µM) was introduced into cultured NSCs using lipofectamine 2000 (Invitrogen) with pmax-GFP plasmids (0.5 µg/ml, Amaxa, Maryland, MD, http://www.lonza.com). Twenty-four hours after transfection, the cells were treated with SB216763 (1.5 µM) and LIF plus BMP2 (80 ng/ml each) for further 48 hours and then immunostained for GFAP, Ki67, and cyclin D1. For in vivo analysis, cyclin D1 siRNA or scramble siRNA (15 µM) was injected into lateral ventricle of E16.5 GfaplacZ/lacZ mice with pCAGGS-NLS-EGFP plasmids (0.5 µg/µl) and introduced into VZ cells by in utero electroporation. The coronal sections of P2 forebrains were immunostained for GFP and β-gal.

Retrovirus Preparation

Plat-E cells were plated on 10 cm dishes in DMEM with 10% fetal bovine serum. Human cyclin D1 cDNA [23] was inserted into pMY-FLAG-IRES-GFP [24], and the plasmids were transfected into Plat-E cells [25] using TransIT-293 (Mirus Bio, Madison, WI, https://www.mirusbio.com). Twenty-four hours after transfection, the medium was exchanged for fresh N2-suppremented DMEM/F-12 with FGF2. After further 24 hours, culture medium containing retroviruses was centrifuged at 6,000g for 16–20 hours and then the retrovirus pellets were resuspended in fresh N2-supplemented DMEM/F-12 containing bFGF. All the recombinant DNA experiments in this manuscript followed the guidelines by TMDU and Ministry of Education Culture, Sports, Science and Technology of Japan.

Luciferase Assay

The NSCs expanded by FGF2 for 4 days were replated on poly-l-ornithine and fibronectin-coated 12 well plate at 5.0 × 105 cells per well. The combination of a plasmid having the 2.5 kb GFAP promoter containing STAT3-binding consensus sequence, TTCCGAGAA, (GF1L-pGL3) or a modified STAT3-binding consensus sequence, CCAAGAGAA, (GF1L-SBSPM-pGL3) were transfected into NSCs using TransIT-LT1 (Mirus Bio) with pGL4.74-hRluc (Promega, San Luis Obispo, CA, http://www.promega.com); and pcDNA3.1-hcyclin D1 plasmids. One day after transfection, the cells were treated with LIF (80 ng/ml) and/or BMP2 (80 ng/ml) for 8 hours. Luciferase activity was measured using the Pikkagene dual luciferase assay system (Toyo Ink, Tokyo, Japan, http://www.toyo-b-net.co.jp) and Mithras LB 940 (Berthold Technologies, Bad Wildbad, Germany, https://www.berthold.com/).

ChIP Assay

ChIP assay was performed according to the protocol recommended by Millipore (Billerica, MA, http://www.millipore.com) with minor modifications. Briefly, cultured NSCs were treated with LIF (80 nM) for 30 minutes and then fixed by formaldehyde at a final concentration of 1% in culture medium for 10 minutes. Cells were centrifuged at 700g for 4 minutes and suspended in lysis buffer (1% SDS, 10 mM EDTA, and 50 mM Tris-HCl [pH 8.1]) containing protease and phosphatase inhibitors (1 mM p-APMSF, complete EDTA-free protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany, http://www.roche-diagnostics.de), and 0.4 mM sodium orthovanadate) and then incubated for 10 minutes on ice for the lysis. Cell lysates were sonicated by Biorupter until the DNA fragments were in the 200–1,000 base pairs range. Chromatin samples were diluted 1:10 with dilution buffer (0.01% SDS, 1.1% Triron X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl [pH 8.1], and 167 mM NaCl). After pretreatment with Protein A agarose/Salmon sperm DNA, the clear lysates were immunoprecipitated with 1 µg of anti-phospho-STAT3 Tyr705 and anti-STAT3 antibodies (Cell signaling and Santacruz, respectively) or 1 µg of normal rabbit IgG (Santa Cruz and Millipore) at 4°C overnight with gentle rotation. Immune complexes were collected by 50 µl of salmon sperm DNA /protein A-agarose beads with gentle rotation for 1 hour at 4°C, followed by washing with 1 ml of low salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl [pH 8.1], and 150 mM NaCl), high salt immune complex wash buffer (0.1% SDS, 1% TritonX-100, 2 mM EDTA, 20 mM Tris-HCl [pH 8.1], and 500 mM NaCl), and LiCl immune complex wash buffer (250 mM LiCl, 1% Nonidet-P-40, 1% sodium deoxycholate, 1 mM EDTA, and 10 mM Tris-HCl [pH 8.1]) and with Tris-EDTA buffer (10 mM Tris-HCl [pH 8.0] and 1 mM EDTA) twice. Immune-complexes were then eluted by elution buffer (1% SDS, 100 mM NaHCO3, and 10 mM DTT). Crosslinks were removed at 65°C for 6 hours in the presence of NaCl. Samples were added with proteinase K at 3 µg/ml and incubated at 45°C for 1 hour. After purification by phenol extraction and ethanol precipitation, the DNA fragments were amplified by polymerase chain reaction (PCR) with the following primers: GSS (5′-TAAGCTGAAGACCTGGCAGTG-3′) and GSAS (5′-TGCTGAATAGAGCCTTGTTCTC-3′).

Reverse Transcription PCR

Cultured E14.5 NSCs were treated with 1.5 µM SB216763 for 24 hours, followed by LIF and BMP2 (80 ng/ml each) stimulation for 3 hours. After treatment, total RNA was extracted from the cells. Total RNA of 2 µg was reverse transcribed by Superscript III First-Strand Synthesis System (Invitrogen) according to the recommended manufacturer's protocol. The PCR for detecting GFAP, S100β, and Aqp4 consisted of 30 cycles of denaturation at 94°C for 20 seconds, annealing at 62°C for 20 seconds, and extension at 72°C for 30 seconds. GFAP specific primers: sense primer, 5′-GAAAGGTTGAATCGCTGGAG-3′; antisense primer, 5′-GCCACTGCCTCGTATTGAGT-3′. S100β specific primers: sense primer, 5′-AGAGGACTCCAGCAGCAAAGG-3′; antisense primer, 5′-AGAGAGCTCAGCTCCTTCGAG-3′. Aqp4 specific primers: sense primer, 5′-GTGTCTGTGGCAGCGAGATA-3′; antisense, 5′-GCATCTGCCTCAGAACATGA-3′. The PCR for detecting Actb (β-actin) consisted of 23 cycles of denaturation at 94°C for 30 seconds, annealing at 60°C for 30 seconds, and extension at 72°C for 40 seconds. Actb specific primers: sense primer, 5′-CCAGGGTGTGATGGTGGGAA-3′; antisense primer; 5′-CAGCCTGGATGGCTACGTACA-3′.

Immunocytochemistry, Immunohistochemistry, Immunoprecipitation, and Western Blotting

The details are described in the Supporting Information Materials and Methods.

Results

Anti-Astrogliogenic Effect of the GSK3β Inactivation-Mediated β-Catenin Stabilization Pathway

As we have demonstrated previously, activation of the β-catenin pathway by growth factors FGF2 and Wnt3a via GSK3β inactivation promoted proliferation of NSCs through cyclin D1 expression and inhibited neurogenesis through interaction of β-catenin and Notch1 intracellular region [4], suggesting counteraction of this pathway to astrogliogenesis. Therefore, we first examined whether the β-catenin stabilization pathway has an inhibitory effect also on astroglial differentiation to fully understand the molecular basis of NSC self-renewal. NSCs prepared from E14.5 mouse telencephalon were used for this purpose (see Material and Methods section). When the cells were treated with LIF and BMP2 for 48 hours, GFAP expressing astrocytes were detected (Fig. 1A, upper two images). However, the addition of 1.5 µM SB216763, a GSK3β inhibitor, dramatically decreased the number of GFAP-positive astrocytes in the culture containing LIF and BMP2 (Fig. 1A, bottom right; Fig. 1B). The LIF and BMP2-treatment did not significantly affect the number of Ki67-positive cells (Fig. 1C, 1D). SB216763-treatment upregulated the number of Ki67-positive cells as expected (Fig. 1C, 1D). SB216763 repressed the mRNA expression of astrocytic markers such as Gfap, S100β, and Aquaporin 4 (Aqp4) under the conditions with or without LIF and BMP2 (Fig. 1E). Taken together, it is suggested that the GSK3β inactivation-mediated β-catenin stabilization pathway counteracts the LIF/BMP2-induced astroglial differentiation and contributes to NSC proliferation. From these data, we further wanted to know which of the β-catenin downstream components plays a role in the inhibition of astroglial differentiation.

Figure 1.

GSK3β inactivation mediated β-catenin stabilization pathway inhibits leukemia inhibitory factor (LIF)/bone morphogenic protein 2 (BMP2)-induced astroglial differentiation. (A–D): Neural stem cells (NSCs) were cultured in the presence of 1.5 µM SB216763 (SB for short; diluted from 2.5 mM stock solution in dimethyl sulfoxide [DMSO]) or control DMSO alone (−), together with LIF plus BMP2 (80 ng/ml each) or medium alone (−), for 48 hours. The cells were then immunostained for glial fibrillary acidic protein (A, B) or Ki67 (C, D) (n = 3; *, p < .05; **, p < .01 by Student's t-test; Scale bars = 100 µm). (E) Reverse transcription polymerase chain reaction (PCR) analysis of LIF/BMP2-induced astrocyte marker mRNA in NSCs treated with SB216763. NSCs were cultured in the presence of 1.5 µM SB216763 (diluted from 2.5 mM stock solution in DMSO) or control DMSO alone (−) for 24 hours, during which LIF and BMP2 (80 ng/ml each) were added in the last 3 hours. Total RNA was recovered from the cells and subjected to reverse transcription and PCR with specific primers for Gfap, S100β, and Aqp4 transcripts. Abbreviations: BMP2, bone morphogenic protein 2; GFAP, glial fibrillary acidic protein; LIF, leukemia inhibitory factor.

Cyclin D1 Downregulation by RNAi Relieves the GSK3β Inactivation-Mediated Inhibition of Astroglial Differentiation

Cyclin D1, a critical downstream effector of the GSK3β inactivation-mediated β-catenin stabilization pathway, plays a pivotal role in the transition from G1 to S-phase of the cell cycle [15, 16]. SB216763 indeed induced cyclin D1 expression in NSCs (Supporting Information Figs. S1, S2A, S2B, S2D). To examine whether cyclin D1 is involved in the inhibition of astroglial differentiation, we performed knockdown of cyclin D1. SB216763-induced cyclin D1 expression in SB216763 treated NSCs was suppressed by RNAi to the control level (Supporting Information Fig. S2A, S2C, S2D). As shown in Figure 2A to 2D, cyclin D1 knockdown recovered the ratio of GFAP-positive cells that was decreased by SB216763, suggesting that the inhibition of astrogliogenesis by SB216763 involves cyclin D1. Interestingly, cyclin D1 siRNA did not significantly suppress the proliferation of NSCs (Fig. 2E–2H). This may be due to the function of other cell cycle regulating factors, such as cyclin D2 and D3, which are known to act in a redundant fashion. For example, inhibition of GSK3β activity in the cyclin D1 deficient mouse uterine epithelium exhibited cell proliferation through promotion of cyclin D2 nuclear translocation [26, 27]. Misexpression of cyclin D3 was induced in cyclin D1/D2 double deficient cerebella where a high level of cyclin D1 and a negligible level of cyclin D3 are normally observed [28].

Figure 2.

Cyclin D1 knockdown relieves the astroglial differentiation-inhibitory effect of GSK3β inactivation. Neural stem cells were transfected with either scramble siRNA or cyclin D1 siRNA (siCD1), each combined with the green fluorescent protein (GFP) expression plasmid, using lipofectamine 2000. Twenty-four hours after transfection, the cells were treated by SB216763 (SB) or control dimethyl sulfoxide (−) together with leukemia inhibitory factor and bone morphogenic protein 2 (80 ng/ml each) for 48 hours. The cells were immunostained for GFP and either glial fibrillary acidic protein (A–D) or Ki67 (E–H). Arrowheads indicate cells double-positive for GFP and the specific marker, whose ratio (in percentage) to the GFP-positive cells is shown in (D) and (H). (n = 3; *, p < .05; **, p < .01; n.s., not significant by Student's t-test). Scale bars = 50 µm. Abbreviations: GFAP, glial fibrillary acidic protein; GFP, green fluorescent protein; siScr, scramble siRNA.

Cyclin D1 Overexpression in NSCs Inhibits LIF and BMP2-Induced Astrogliogenesis

To further confirm the counteraction of cyclin D1 to the astrogliogenic signaling, we investigated the effect of cyclin D1 overexpression on astroglial differentiation by LIF and BMP2. Cyclin D1 or control retrovirus containing IRES-GFP was introduced into NSCs. On the following day, cells were replated, where LIF and BMP2 were added to induce astroglial differentiation. The forced expression of cyclin D1 significantly decreased the percentage of GFAP-positive cells among GFP-positive cells (Fig. 3A, 3F). Forced expression of cyclin D1 decreased the ratio of cells expressing S100β, another typical astrocyte lineage marker expressed earlier than GFAP, to the GFP-positive cells (Fig. 3B, 3F). The results indicate that cyclin D1 inhibits the LIF and BMP2-induced astroglial differentiation of NSCs. We then examined whether cyclin D1 turns NSCs towards the neuronal or oligodendroglial lineage under this culture condition at the expense of inhibition of the astroglial differentiation. The ratio of Tuj1-positive neurons to the GFP-positive cells was not increased and platelet-derived growth factor receptor α (PDGFRα)-positive oligodendrocyte progenitors did not become detectable by excessive cyclin D1 in this culture condition (Fig. 3C, 3D, 3F), indicating that cyclin D1 inhibits astroglial differentiation without affecting cell fate towards neurons or oligodendrocytes. The cyclin D1 overexpression increased the percentage of Ki67-positive proliferating cells among GFP-positive cells by approximately twofold (Fig. 3E, 3F). To examine the type of cells whose cell cycle progression is promoted by cyclin D1 overexpression, we applied triple-immunostaining for GFAP, Ki67 and GFP. Cyclin D1 overexpression significantly increased the percentage of both GFAP-positive and -negative proliferating cells, suggesting that cyclin D1 overexpression promotes both nonastrocytic (neural stem/progenitor) and astrocytic cell proliferation (Supporting Information Fig. S3). Importantly, cyclin D1 overexpression increased the proliferation of GFAP-negative cells to a somewhat greater extent than GFAP-positive cells because the percentage of proliferating GFAP-positive astrocytes among total proliferating cells was significantly decreased (Supporting Information Fig. S3B, number 5 pair of columns). Thus, these results could not rule out the possibility that the decreased percentage of GFAP-positive cells by cyclin D1 overexpression was partly due to the preferential increase in the proliferation rate of NSCs and progenitor cells.

Figure 3.

Cyclin D1 overexpression inhibits leukemia inhibitory factor (LIF)/bone morphogenic protein 2 (BMP2)-induced astroglial differentiation from neural stem cells (NSCs). Cyclin D1 or control retrovirus-infected NSCs were treated with LIF and BMP2 (80 ng/ml each) for 48 hours. The cells were immunostained for green fluorescent protein (GFP) and glial fibrillary acidic protein (GFAP) (A), S100β (B), Tuj1 (C), PDGFRα (D), or Ki67 (E). Scale bars = 100 µm. The vertical axis in (F) indicates the ratio of cells (in percentage) double-positive for GFP and each marker to the GFP-positive cells. (n = 4 for GFAP; n = 3 for S100β, Tuj1, PDGFRα, and Ki67; **, p < .01; n.s., not significant by Student's t-test; n.d., not detected). Abbreviations: GFAP, glial fibrillary acidic protein; GFP, green fluorescent protein.

Cyclin D1 Inhibits Astroglial Differentiation in a Manner Independent of Cell Cycle Progression

Although cyclin D1 siRNA increased the ratio of LIF and BMP2-induced GFAP positive cells to the GFP-positive cells, it did not decrease the percentage of Ki67 positive cells (Fig. 2E–2H), which most likely owes to other cyclin species. Thus, cell cycle progression appears to dispensable for cyclin D1-mediated suppression of astroglial differentiation. To test this possibility, we examined whether forced expression of cyclin D1 inhibits astroglial differentiation in the presence of CDK4 inhibitor which disturbs the function of cyclin D1 as a cell cycle-promoting factor. Cyclin D1 retrovirus- or control GFP alone retrovirus-infected NSCs were treated with LIF and BMP2 together with or without CDK4 inhibitor for 48 hours. As we had expected, cyclin D1 forced expression reduced the ratio of GFAP-positive astrocytes to the GFP-positive cells independently of CDK4 activity (Fig. 4A, 4B). We confirmed that CDK4 inhibitor significantly decreased the percentage of Ki67 positive proliferating cells regardless of cyclin D1 overexpression (Supporting Information Fig. S4A, S4B). We further investigated whether cyclin D1 mutant (cyclin D1 K114E), which lacked CDK4 binding activity and had no effect on cell cycle progression [29], inhibits astrocyte differentiation. Overexpression of cyclin D1 K114E mutant significantly decreased the percentage of GFAP-positive cells among GFP-positive cells as similar to the wild-type cyclin D1-overexpressing cells, but did not affect the ratio of ki67-positive cells (Fig. 4C, 4D). These results suggested that the inhibition of astroglial differentiation by cyclin D1 is independent of CDK4-mediated cell cycle progression. Taken together, we conclude that decreased percentage of GFAP-positive cells by cyclin D1 expression is due to cyclin D1-mediated inhibition of LIF and BMP2-induced astroglial differentiation of NSCs, combined with increased proliferation of NSCs.

Figure 4.

Cyclin D1 overexpression inhibits leukemia inhibitory factor (LIF)/bone morphogenic protein 2 (BMP2)-induced astroglial differentiation from neural stem cells (NSCs) independent of CDK4-activating property. (A, B): Cyclin D1 or control retrovirus-infected NSCs were cultured with LIF and BMP2 (80 ng/ml each) in the presence of 75 nM CDK4 inhibitor (CDK4i; diluted from 2.5 mM stock solution in dimethyl sulfoxide [DMSO]) or control DMSO alone (equivalent volume) for 48 hours. The cells were then immunostained for green fluorescent protein (GFP) and either glial fibrillary acidic protein (GFAP). (C, D): Cyclin D1, cyclin D1 K114E, or control retrovirus-infected NSCs were treated with LIF and BMP2 (80 ng/ml each) for 48 hours. The cells were double-immunostained for GFAP or Ki67 together with GFP. Arrows indicate cells double-positive for GFP and each marker. The vertical axis in (B) and (D) indicates the ratio of cells double-positive for GFP and each marker to the GFP-positive cells (n = 3; *, p < .05; **, p < .01; n.s., not significant by Student's t-test). Scale bars = 100 µm (A) and 50 µm (C). Abbreviations: GFAP, glial fibrillary acidic protein; GFP, green fluorescent protein.

Cyclin D1 Suppresses STAT3-Mediated Transcriptional Activation of Gfap Promoter

Cyclin D1 plays a role in transcriptional regulation, in addition to cell cycle progression [15]. To investigate whether cyclin D1 repressed astroglial differentiation at the gene expression level, we examined whether cyclin D1 represses transcriptional activation using Gfap promoter. Astrogliogenic cytokines LIF and BMP2 have been demonstrated previously to directly and synergistically activate Gfap gene promoter through activation of respective downstream transcription factors STAT3 and Smad1, which form a fully competent transcription complex with p300 [22]. GF1L-pGL3 is a plasmid containing a part of the Gfap promoter, including STAT3 and Smad1 binding sites, which is fused to luciferase gene [22]. NSCs which had been transfected with the GF1L-pGL3 reporter plasmid together with or without cyclin D1 expression plasmid, and stimulated by LIF or/and BMP2 for 8 hours. As shown in Figure 5, the luciferase activity induced by LIF alone or LIF plus BMP2 was significantly repressed by cyclin D1 transfection, but that induced by BMP2 alone was not repressed. The results indicated that cyclin D1 inhibits STAT3-mediated transcriptional activation of Gfap promoter. We further performed luciferase assay for the Gfap promoter with base substitutions in the STAT3 binding consensus sequence [30]. As shown in Figure 5 (right half), no significant luciferase activity was observed. In addition, we confirmed the inhibitory effect of cyclin D1 overexpression on the expression of other astrocyte-related genes expressions, such as Gfap, S100β, and Aqp4 (Supporting Information Fig. S5; see also Fig. 1E). Cyclin D1 siRNA significantly increased the LIF-induced luciferase activity of GF1L-pGL3. pGL3-SBSPM-GF1L, in which STAT3 binding sites were mutated, almost completely abolished the response of Gfap promoter (Supporting Information Fig. S6). These results suggest that cyclin D1 disturbs the LIF-STAT3 signaling pathway that induces astrogliogenesis.

Figure 5.

Cyclin D1 overexpression suppressed the leukemia inhibitory factor (LIF)-mediated transcriptional activation of Gfap promoter. Cultured neural stem cells were transfected with GF1L-pGL3 or GF1L-SBSPM-pGL3 together with either control vector (white columns) or cyclin D1 expression vector (black columns). In all combinations, pGL4.74-hRluc vector was cotransfected. Twenty-four hours after transfection, the cells were treated with LIF alone, bone morphogenic protein 2 (BMP2) alone, or LIF plus BMP2 for 8 hours, and luciferase activities in the cell lysates were measured. The vertical axis indicates the fold increase of the luciferase activity in each condition compared with that in the unstimulated cells transfected with GF1L-pGL3 (n = 3; *, p < .05; **, p < .01; n.s., not significant by Student's t-test).

Cyclin D1 Reduces the STAT3-p300 Complex Formation and the Binding of STAT3 to Gfap Promoter

Next, we asked how cyclin D1 inhibits STAT3-mediated astroglial differentiation. LIF-stimulation leads to LIF receptor-gp130 heterodimer formation and activation of Janus kinase one (JAK1), followed by phosphorylation of STAT3 at tyrosine 705 (pSTAT3Y705) [22, 31]. The phosphorylated STAT3 forms the homodimer, which is translocated into the nucleus, and binds to Gfap gene promoter [22, 31]. First, we explored the interaction between cyclin D1 and STAT3. Coimmunoprecipitation analysis with tagged proteins expressed in HEK293 cells displayed the interaction between cyclin D1 and STAT3 (Fig. 6A). Moreover, the interaction between endogenous cyclin D1 and STAT3 in NSCs was also detected (Fig. 6B). We next investigated the effect of cyclin D1 overexpression on phosphorylation of STAT3 in NSCs. Western blot analysis with antiphosphorylated STAT3 antibody did not show any difference between LIF-induced phosphorylation of STAT3Y705 in cyclin D1 virus-infected NSCs and that in control virus-infected NSCs (Fig. 6C). We further examined whether cyclin D1 overexpression disturbs the translocation of STAT3 into the nucleus of cultured NSCs. After 30 minutes of LIF stimulation, no significant change was detected in the nuclear localization of pSTAT3Y705 (Supporting Information Fig. S7A, S7B).

Figure 6.

Cyclin D1 disturbs the STAT3-p300 complex formation and the binding of STAT3 to Gfap promoter. (A): Pull-down assays were performed to see interaction between cyclin D1 and STAT3. HEK293 cells were cotransfected with expression constructs encoding FLAG-tagged STAT3 and Myc-tagged cyclin D1. (B): The endogenous interaction between cyclin D1 and STAT3 in neural stem cells (NSCs) was examined. NSCs were stimulated with leukemia inhibitory factor (LIF; 80 ng/ml) for 30 minutes. The nuclear extracts were immunoprecipitated with STAT3 and then Western blotting was performed for cyclin D1 detection. (C): Western blotting analysis was performed to detect phosphorylation of STAT3 at Y705. NSCs that had been infected with either cyclin D1- or control-retroviral vector were stimulated by LIF and the cell lysates were subjected to this analysis. (D): Effect of cyclin D1 overexpression on the interaction between STAT3 and p300 in HEK293 cells was examined by pull-down assay. The cells were cotransfected with the indicated combinations (#1–#3) of expression constructs containing FLAG-tagged STAT3, HA-tagged p300, and Myc-tagged cyclin D1. (E): The intensity of each lane (for combinations #2 and #3) in C was measured by Image J (NIH, Bethesda, MD). The intensity of coimmunoprecipitated FLAG-STAT3 was normalized by that of immunoprecipitated HA-p300 and also by the expression levels of FLAG-STAT3 (the intensity of “input” in C). n = 4; *, p < .05 by Paired t-test. (F, G): Chromatin immunoprecipitation assay was performed to evaluate the binding activity of STAT3 to Gfap promoter. Cultured NSCs were infected with retrovirus encoding cyclin D1 (lane five) or control retrovirus (lanes 1–4). Forty-eight hours after infection, the cells were treated with LIF (80 ng/ml) for 30 minutes. Chromosomal DNA was coimmunoprecipitated with normal rabbit IgG (lanes 1 and 2 in (E) and (F)), anti-pSTAT3Y705 (lanes 3–5 in (E)), or anti-STAT3 antibodies (lanes 3–5 in (F)). Abbreviations: IP, immunoprecipitation; LIF, leukemia inhibitory factor.

STAT3 physically interacts with a transcriptional coactivator p300 to elevate its target promoter activation [22]. Interestingly, we found that cyclin D1 interacted with not only STAT3 but also p300 (Supporting Information Fig. S8). Therefore, we examined whether cyclin D1 affects the complex formation between STAT3 and p300 (Fig. 6D). FLAG-tagged STAT3 and HA-tagged p300 were coexpressed in HEK293 cells with or without Myc-tagged cyclin D1. In the presence of cyclin D1, the amount of FLAG-STAT3 protein coimmunoprecipitated with anti-HA antibody was partly but significantly reduced down to approximately 47% of the control (Fig. 6D, 6E). Because the reduction in STAT3-p300 complex formation should affect the STAT3 binding to Gfap promoter, we next tested the effect of cyclin D1 overexpression on the STAT3-binding activity to Gfap promoter. ChIP assay in Figure 6F showed that cyclin D1 overexpression suppressed the LIF-induced binding of pSTAT3 to the Gfap promoter, down to the level comparable to the basal level observed in the control virus-infected cells without LIF treatment. Similar results were obtained by the ChIP assay with anti-STAT3 antibody (Fig. 6G). Taken together, these results suggest that cyclin D1 suppresses the STAT3-mediated Gfap promoter activation partly by disturbing STAT3-p300 complex formation and the inhibition of the STAT3 binding to Gfap promoter in NSCs.

Cyclin D1 Knockdown Promotes the Emergence of GFAP-Positive Astrocytes In Vivo

To investigate whether cyclin D1 has the potential to inhibit astroglial differentiation in vivo, siRNA for cyclin D1 was introduced into the mouse fetal brain by in utero electroporation. To monitor astroglial differentiation, we used GFAP reporter mice that carried the nls-lacZ (nlacZ) gene knocked into the Gfap gene locus (Fig. 7A). The advantage of this system is that the nuclear β-galactosidase-positive cells are easy to be counted and detectable even before the GFAP filament formation in astrocytes. GAP43-GFP in this transgene is not expressed in the current experiment because of the lack of Cre gene in the mouse. Figure 7B confirmed nlacZ gene expression in GFAP-positive astrocytes in the white matter of P7 mouse cerebral cortex. In the experiments shown in Figure 7C to 7I, scramble or cyclin D1 siRNA together with the pCAGGS-NLS-EGFP plasmids were injected in utero into the lateral ventricle of the GfapnlacZ/nlacZ mice at E16.5, before the onset of astrogliogenesis, and electroporation was carried out. The coronal sections of the P2 mice forebrains were immunostained with anti-β-gal and anti-GFP antibodies. We found that cyclin D1 knockdown significantly increased the ratio of the cells double-positive for β-gal (Gfap-expression) and GFP (siRNA-introduction) to the GFP-positive cells in the lateral and dorsal white matter of postnatal cortex (Fig. 7C–7I and Supporting Information Fig. S9, respectively). To further investigate whether cyclin D1 inhibits astrocyte differentiation in a manner independent of cell cycle progression in vivo, cyclin D1, cyclin D1 K114E, and control plasmids were transfected into neural stem cells in the VZ of E16.5 mouse brains by in utero electroporation. The coronal sections of electroporated brains at P4 were immunostained for GFAP and GFP. The forced expression of both cyclin D1 and cyclin D1 K114E significantly decreased the ratio of the cells double-positive for GFAP and GFP to the GFP-positive cells in the mouse brains (Supporting Information Fig. S10). These results indicate that under in vivo physiological condition, cyclin D1 suppresses astrogliogenesis in a cell cycle progression ability-independent manner.

Figure 7.

Cyclin D1 knockdown promotes the appearance of glial fibrillary acidic protein (GFAP)-positive astrocytes in vivo. (A): Schematic representation of the structure of GFAP reporter gene in mice used in this experiment. The detailed construction of targeting vector and the generation of the knock-in mouse will be described elsewhere (T. Kagawa et al., manuscript in preparation). It should be noted that GAP43- green fluorescent protein (GFP) was not expressed in this experiment due to the absence of Cre. The NLS-EGFP expression plasmid used in (C–H) was also depicted. (B): X-gal staining and immunostaining for GFAP were performed on the cryosection of P7 GfapnlacZ/+ mouse brain, confirming the specific expression of β-galactosidase in the GFAP-positive astrocyte nuclei. (C–I): One-microliter of either cyclin D1 siRNA or scramble siRNA (0.2 µg each) and pCAGGS-NLS-EGFP (0.5 µg) were injected into the lateral ventricle of E16.5 GfapnlacZ/nlacZ mouse brains and transferred into the ventricular zone cells of the lateral wall by in utero electroporation. The coronal sections of P2 forebrains were immunostained with anti-β-gal and anti-GFP antibodies. The mean ratio of the cells double-positive for β-gal (GFAP-expression) and GFP (siRNA-introduced) to the GFP-positive cells is shown in (I). The sections containing equivalent parts of the brain were selected for this analysis. All the detectable GFP-positive cells on the sections (>250 cells) were analyzed (three pups each for scramble siRNA and cyclin D1 siRNA). Yellow arrows indicate β-gal and GFP double-positive cells. Scale bars = 50 µm (B) and 100 µm (C–H). (J): A model of the signaling networks explaining neural stem cell self-renewal based on the linkage between promotion of cell proliferation and inhibition of neurogenesis and astrogliogenesis. The details are described in the main text. Abbreviations: CP, cortical plate; FGF2: fibroblast growth factor 2; FGFR, fibroblast growth factor receptor; GFAP, glial fibrillary acidic protein; GFP, green fluorescent protein; LEF, lymphoid enhancer factor; NSC, neural stem cell; siRNA, small interfering RNA; TCF, T-cell factor; WM, white matter.

Discussion

Self-renewing proliferation and cell-fate determination of NSCs are indispensable steps in proper formation and maintenance of brain architecture and function. Elucidation of the molecular mechanisms governing these steps will help us to better understand the brain development and provide us with a clue to identify therapeutic targets for neural diseases. An important molecular mechanism underlying NSC self-renewal is the coordination between promotion of cell proliferation and inhibition of neuronal and glial differentiation. In this study, we demonstrated counteractive effect of cyclin D1, a common downstream effector of the growth- promoting signaling pathways initiated by FGF2 and Wnt, on LIF-induced astroglial differentiation (Fig. 7J; illustrated in red). The left half of this figure summarizes what we reported previously [4], in which β-catenin, a common downstream signaling component of the FGF and Wnt pathways, potentiates the Notch-mediated antineurogenic activity (illustrated in blue). Thus, our study provides a molecular basis underlying the NSC self-renewal by demonstrating how FGF2 and Wnt signals block the neuronal and astroglial differentiation while promoting proliferation (Fig. 7J).

Although cyclin D1 is well known to promote the transition from G1 to S-phase of cell cycle, cyclin D1 also has other multiple biological functions such as the regulation of cellular apotosis, survival, migration, differentiation, and metabolism [15]. Therefore, it is considered that cyclin D1 is indispensable for the development. In fact, cyclin D1 has been reported to play an important role in proper construction of mammalian central nervous system: Cyclin D1 deficient mice show neurological abnormality such as the abnormal leg-clasping reflex and the hypoplasia of retinas and cerebella [19, 20, 32]. Similarly with the expression pattern of the FGF receptor 1 protein, which is a high-affinity receptor for FGF2 in mouse fetal brain [33], cyclin D1 expression is observed in the VZ where NSCs exist during the early embryonic stages [17, 18]. The expression levels of the cyclin D1 protein in the VZ of mouse telencephalon decrease as the development proceeds [17]. Thus, taken together with our study, the decreased amount of cyclin D1 may trigger the onset of astrogliogenesis during brain development [34, 35].

Recent reports demonstrated that cyclin D1 inhibited the transcriptional activity of STAT3 in HepG2 hepatocellular carcinoma cells and MCF7 breast cancer cells [36, 37]. Bienvenu et al. reported that cyclin D1 inhibited the recruitment of CBP histone acetyltransferase and RNA polymeraseII to the p21waf1 promoter, and thus repressed its gene expression [37]. Consistent with this article, we here demonstrate that cyclin D1 at least in part disturbed the STAT3-p300 complex formation and inhibited the STAT3-mediated transcriptional activation of the Gfap promoter (Figs. 5, 6). In addition, we newly found that cyclin D1 has a potential to disrupt the interaction between STAT3 and the Gfap promoter in NSCs by ChIP analysis (Fig. 6). It is interesting to note that cyclin D1 had no effect on the STAT3 binding ability to p21waf1 promoter in HepG2 cells [37]. Similarly to these previous reports [36, 37], we did not detect a significant inhibitory effect of cyclin D1 on the in vitro binding of STAT3 to the synthetic DNA containing three tandem STAT3 binding elements (Supporting Information, Fig. S11). Even in this experiment (Supporting Information Fig. S11), the existence of cyclin D1 in the DNA-protein complex containing STAT3 was indicated as predicted by Figure 6A. We, therefore, speculate that the significant reduction in the amount of the STAT3-binding element-containing Gfap promoter in the anti-STAT3 immune complex observed with native NSCs may be due to the stoichiometric difference between the native chromatin and synthetic DNA, the former of which has one STAT3 binding site while the latter has three. Also in the former, the STAT3-DNA binding appears to be weak and requires smad proteins and p300 for the efficient binding and promoter activation [22]. The environments around the STAT3 binding site in the respective promoters in NSCs, hepatoma cells, and breast cancer cells may also be different. Cyclin D1 was reported to recruit the histone deacetylase (HDAC) and histone methyltransferase Suv39H1 to the local chromatin of the murine lipoprotein lipase (LPL) promoter, and inhibit LPL gene expression [38]. Thus, surrounding environments may largely involve the function of cyclin D1 on STAT3. Interestingly, it was reported that orphan nuclear receptor TLX (also known as NR2E1) promoted the adult neural stem cell proliferation and self-renewal through Wnt/β-catenin pathway [13], indicating that TLX-inducing cyclin D1 expression may contribute to inhibition of astrogliogenesis. Furthermore, TLX directly bound to the gfap promoter and repressed astrocyte differentiation [39]. Because both TLX and cyclin D1 reduced the transcription of target genes through the recruitment of HDAC [38, 40], it is tempting to speculate the interaction between TLX and cyclin D1.

SB216763 and cyclin D1 overexpression decreased the expression of all tested astrocytic genes, Gfap, S100β, and Aqp4. However, among them, S100β gene expression was not induced but rather reduced by LIF and BMP2 (Fig. 1E; Supporting Information Fig. S5). LIF and BMP2 may preferentially induce specific astroglial subtypes because it has been reported that there are some astroglial subtypes such as S100β-positive, GFAP-positive, and S100β /GFAP-double positive astrocytes in adult mouse brains [41, 42]. The number of S100β-positive astrocytes is larger than that of GFAP-positive astrocytes in the cortical plate region where most of “protoplasmic” astrocytes are located, whereas there are more GFAP-positive astrocytes compared with S100β-positive astrocytes in the corpus callosum (white matter) where many ‘fibrous' astrocytes exist [41]. Interestingly, LIF induces elongated bipolar or tripolar GFAP-positive cells which are similar to fibrous astrocytes, while BMPs induce astocytes to display highly branched protoplasmic-like stellar morphology [43, 44]. These reports suggest that each of LIF and BMP may induce the differentiation of distinct astroglial subtypes. Importantly, the cyclin D1-mediated reduction in all tested astrocytic gene expression suggests that cyclin D1 suppresses pan-astrocytic differentiation programs.

We have demonstrated here that cyclin D1 downregulation promotes the astrogliogenesis of at least fetal NSCs in vitro and in vivo (Figs. 2A–2D, 7C–7I; Supporting Information Fig. S9). It would have been interesting if a previous article dealing with cyclin D1 deficient mice [45] reported anything about the abnormality in the brain in view of astrogliogenesis in the late gestational stages where astroglial differentiation normally starts. However, no such observation was documented, except for the reduction in cell proliferation in the dentate granule zone and the subventricular zone in the cyclin D1 deficient brain at 4–5 weeks postnatally. In that article, instead, it was documented that NSCs prepared from the brains of postnatal day 1 to day 3 cyclin D1 deficient mice and expanded by FGF2 and epidermal growth factor (EGF) displayed the decreased number of GFAP-positive astrocytes in the astrogliogenic culture condition with LIF and BMP2 [45] compared with those prepared from the brains of wild type and cyclin D1 heterozygous mice, suggesting that cyclin D1 rather promotes astroglial differentiation in the postnatal NSCs at least in vitro. A possible cause for this discrepant observation may be the difference in the characters of postnatal [45] and embryonic (our current study) NSCs. Alternatively, the discrepancy may be caused by the difference in the use of growth factors for the NSC culture: Ma et al. [45] cultured postnatal cyclin D1 deficient NSCs with FGF2 and EGF, but we cultured E14.5 NSCs with FGF2 only. Both FGF2 and EGF are known to stimulate different populations of NSCs, but FGF2 promotes the proliferation of mid-gestational NSCs that show preference to differentiate into neurons and EGF promotes the proliferation of late-gestational NSCs that have astrogliogenic competence [34, 46, 47]. Thus, developmental stages and growth factors may change the character of NSCs, resulting in the opposing action of cyclin D1 on astroglial differentiation. Recently, cyclin D1 was shown to promote the Notch1 gene expression via the recruitment of CBP to this promoter in mouse retinas [48], suggesting that cyclin D1 functions as a transcriptional activator at least for Notch1. Thus, cyclin D1 may have dual functions as a transcriptional repressor and activator in a context-dependent manner. It is intriguing to elucidate the compositional change of the transcriptional complex containing cyclin D1 and STAT3 and/or chromatin structure of astrocyte-specific promoters in between embryonic and postnatal NSCs. Talking about context-dependency, contrary to our study with fetal telencephalic NSCs, a recent report showed that cyclin D1 promotes neurogenesis from NSCs in the embryonic spinal cord [49].

Altogether, our results provide a model explaining how self-renewing NSCs do not easily undergo astroglial differentiation even in the presence of astrogliogenic signaling cues such as LIF and ciliary neurotrophic factor in the developmental brain. Because cyclin D1 is also an important regulator of cell cycle progression in NSCs, it is interesting that cyclin D1 itself plays a role as an ‘interface' of the coordination between the promotion of cell proliferation and the inhibition of differentiation in NSCs during central nervous system development.

Conclusion

In summary, we show that cyclin D1, common downstream effector of the FGF2 and Wnt that cooperatively promote NSC self-renewal, inhibits astroglial differentiation while accelerating cell cycle progression. We further demonstrate that cyclin D1 reduces the STAT3-p300 transcriptional complex formation and the STAT3 binding to the Gfap promoter, resulting in the inhibition of astroglial differentiation. Taken together with our previous finding, we provide a novel molecular mechanism for NSC self-renewal in which growth promoting signaling components activated by FGF2 and Wnts inhibit neuronal and glial differentiation.

Acknowledgments

We thank Takeshi Kawano and Hisashi Yamada (The Jikei University school of Medicine Institute of DNA Medicine) for the gift of human cyclin D1 cDNA; Shinji Fukuda (Ehime University), Taichi Kashiwagi (Tokyo Medical University), and Yutaka Yoshinaga (Saiseikai Kumamoto Hospital) for helpful discussions; Rie Taguchi and Yuko Saiki for their technical assistance; Mako Fushimi and Michiko Teramoto for their secretarial assistance; and Tokyo Medical and Dental University (TMDU) Center for Experimental Animal for use of facilities. This work was supported by Grants-in-Aids 17500255 (to T.K.) for Scientific Research (C), 20300129 (to T.K.) and 24300119 (to T.T.) for Scientific Research (B), 20022034 (to T.T.) for Scientific Research on Priority Areas from the Japan Society for the Promotion of Science, and Takeda Science Foundation (to T.T.), the Joint Usage/Research Program of Medical Research Institute, TMDU.

Author Contributions

N.B.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; T.I. and T.S.: collection and/or assembly of data, data analysis and interpretation; K.T.: provision of study material; T.K.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; T.T.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

Disclosure of Potential Conflict of Interests

The authors indicate no potential conflicts of interest.

Ancillary