TISSUE-SPECIFIC STEM CELLS

Transcription

1 TISSUE-SPECIFIC STEM CELLS Epidermal Growth Factor Signaling Mediated by Grb2 Associated Binder1 Is Required for the Spatiotemporally Regulated Proliferation of Olig2-Expressing Progenitors in the Embryonic Spinal Cord YOSHIKA HAYAKAWA-YANO, a,b KEIGO NISHIDA, c SHINICHI FUKAMI, b YUKIKO GOTOH, d TOSHIO HIRANO, c,e TOSHIYUKI NAKAGAWA, a TAKUYA SHIMAZAKI, b,f HIDEYUKI OKANO b,f a Department of Neurobiology, Graduate School of Medicine, Gifu University, Gifu, Japan; b Department of Physiology, Keio University School of Medicine, Tokyo, Japan; c Laboratory for Cytokine Signaling, RIKEN Research Center for Allergy and Immunology, Yokohama, Japan; d Institute of Molecular and Cellular Biosciences, University of Tokyo, Bunkyo-ku, Tokyo, Japan; e Laboratory for Developmental Immunology, Graduate School of Frontier Biosciences, Osaka University, Osaka, Japan; f Core Research for Evolutional Science Technology, Solution-Oriented Research for Science and Technology, Japan Science and Technology Agency, Saitama, Japan Key Words. Epidermal growth factor Grb2 associated binder1 Olig2 Proliferation ABSTRACT Gab1 (Grb2 associated binder1) has been identified as an adaptor molecule downstream of many growth factors, including epidermal growth factor (EGF), fibroblast growth factor, and platelet-derived growth factor, which have been shown to play crucial roles as mitotic signals for a variety of neural progenitor cells, including stem cells, both in vitro and in vivo. Here, we show that Gab1 deficiency results in a reduction in the number of Olig2-positive (Olig2 ) progenitor cells in the developing mouse spinal cord after embryonic day 12.5 (E12.5), when gliogenesis starts in the pmn domain where the EGF receptor (EGFR) is expressed predominantly. Our in vitro analysis further revealed that Gab1 is essential for EGF-dependent proliferation of Olig2 progenitor cells derived from the E12.5 ventral and E14.5 Disclosure of potential conflicts of interest is found at the end of this article. dorsal but not ventral spinal cord, whereas Gab1 is always required for the activation of Akt1 but not of ERK1/2. Moreover, we found that the action of the Gab1/Akt pathway is context-dependent, since constitutively active Akt1 could rescue the proliferation defect only in the E12.5 spinal cord of the Gab1-deficient mouse in vitro. Finally, we demonstrated that EGFR-deficient mice and Gab1-deficient mice showed a similar reduction in the number of Olig2 progenitor cells in the developing spinal cord. These findings indicate that EGFR-mediated signaling through Gab1/Akt contributes to the sufficient expansion of Olig2 progenitor cells in a spatiotemporally regulated manner, which represents the origin of glial cells in the developing spinal cord. STEM CELLS 2007;25: INTRODUCTION Specification of progenitor cells in the embryonic spinal cord has been intensively studied [1]. Distinct types of cells are generated along the dorsoventral (D-V) axis of the neural tube, where progenitor domains are defined by a combination of homeodomain and basic helix-loop-helix (bhlh) transcription factors, whose expressions are precisely regulated by bone morphogenetic proteins, Wnt proteins secreted from the roof plate, and sonic hedgehog (SHH) derived from the notochord and floor plate [2 4]. Olig2 has been identified as a bhlh transcription factor expressed in the pmn domain where motoneurons and glia are sequentially generated under the control of the SHH signal [5 8]. More recently, it has been shown that a subpopulation of Pax7 (a paired homeodomain transcription factor)-expressing progenitor cells located in the dorsal half of the ventricular zone (VZ) of the mouse embryonic spinal cord also starts to express Olig2 after embryonic day 14.5 (E14.5) and may subsequently differentiate into oligodendrocytes [9, 10]. Analysis of mice lacking Olig2 has confirmed the essential role of this factor in the development of both motoneurons and oligodendrocytes [11 14]. Despite extensive analysis of the roles of Olig2, the effects of extracellular signalings other than SHH on Olig2 progenitors remain poorly understood. In the present study, we tried to examine the roles of epidermal growth factor (EGF) signaling mediated by Grb2 associated binder1 (Gab1) upon the proliferation of Olig2 progenitors in the embryonic spinal cord. Gab1, originally identified as a Grb2 (growth factor receptor bound protein 2) binding protein [15], is a member of the Gab/daughter of sevenless family of adaptor molecules. It is ubiquitously expressed and Correspondence: Takuya Shimazaki, Ph.D., Department of Physiology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo , Japan. Telephone: ; Fax: ; shimazak@sc.itc.keio.ac.jp; or Hideyuki Okano, M.D., Ph.D., Department of Physiology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo , Japan. Telephone: ; Fax: ; hidokano@sc.itc.keio.ac.jp Received September 19, 2006; accepted for publication February 18, 2007; first published online in STEM CELLS EXPRESS March 1, AlphaMed Press /2007/$30.00/0 doi: /stemcells STEM CELLS 2007;25:

2 Hayakawa-Yano, Nishida, Fukami et al plays a role as a common intracellular signaling mediator for various growth factor and cytokine receptors, including EGF receptor (EGFR), fibroblast growth factor receptors (FGFRs), gp130 (a common receptor subunit for the IL-6 cytokine family), platelet-derived growth factor (PDGF) receptor, and insulin-like growth factor-i receptor [15, 16]. Depending on the ligand stimulus, Gab1, which is recruited to activated receptors, assembles multimeric signaling complexes containing Grb2, SHP2, p85 phosphatidylinositol 3-kinase (PI3K), and phospholipase C- and serves as a signal amplifier [17]. These signals activate the Ras-mitogen-activated protein kinase (Ras/MAPK) and PI3K pathways, mainly through their association with Gab1 [18, 19]. It has been shown that Gab1 is actually involved downstream of various cytokine and growth factor signals in the proliferation and differentiation of various types of cells [20 22]. However, the involvement of Gab1 in the development of the central nervous system (CNS) has not yet been thoroughly studied, although the roles of Gab2, another member of the Gab family of proteins, in the fibroblast growth factor (FGF)2- dependent proliferation of neural stem cells and survival of differentiating neurons have recently been shown [23]. In this study, we attempted to elucidate the roles of Gab1 in CNS development by focusing on progenitor cell proliferation in the developing mouse spinal cord and found that EGF/Gab1/Akt pathway plays essential roles in the expansion of Olig2 progenitors in a context-dependent manner. MATERIALS AND METHODS Mice and Genotyping Wild-type ICR (CD-1) mice were purchased from CLEA Japan, Inc. (Tokyo, or Japan SLC (Shizuoka, Japan, Gab1-mutant mice were maintained in a C57BL6 background ( F5) and genotyped as described in a previous report [24]. Egfr-mutant mice were purchased from Jackson Laboratory (Bar Harbor, ME, and maintained in an ICR background ( F10) and genotyped as described in a previous report [25]. Noon of the day on which the vaginal plug was found was termed E0.5. Primary Cultures Spinal cords from E12.5 or E14.5 mouse embryos were dissociated mechanically and then cultured in a media hormone mix medium [26] in the presence or absence of growth factors (supplemental online Materials and Methods). Cells were seeded at cells per 0.75 cm 2 on 10-mm diameter coverslips (Matsunami, Osaka, Japan, coated with poly-l-ornithine (30 g/ml; Sigma-Aldrich, St. Louis, sigmaaldrich.com) and laminin (10 g/ml; Invitrogen, Carlsbad, CA, in a 48-well culture plate (Corning, Corning, NY, Immunohistochemistry and Immunocytochemistry Serial 12- m cryosections of the mouse embryo trunk were prepared and then processed for immunohistochemistry. Sections were incubated overnight at 4 C or at room temperature with primary antibodies (supplemental online material) followed by incubation with Alexa dye-conjugated secondary antibodies (1:1,000; Molecular Probes, Carlsbad, CA, or the combination of biotinylated secondary antibodies (1:500; Jackson Immunoresearch Laboratories, West Grove, PA, jacksonimmuno.com), the VECTASTAIN Elite ABC kit (Vector Laboratories, Burlingame, CA, and the visualization by using the TSA Fluorescence System (PerkinElmer Life and Analytical Sciences, Boston, com). The cultured cells were fixed for 15 minutes in 4% paraformaldehyde in phosphate-buffered saline and processed for immunocytochemistry (ICC) as described above. Hoechst (Sigma- Aldrich) was used to detect cell nuclei for the quantitative experiments. Immunostained specimens were examined with either a universal fluorescence microscope (Axiophot 2) or a confocal laser scan microscope (LSM510) (Carl Zeiss, Jena, Germany, Immunoblot Analysis Cells were grown on 6-well culture plates (Corning) coated with poly-l-ornithine and laminin in the presence of 20 ng/ml FGF2 or EGF for 72 hours and then stimulated with 20 ng/ml EGF or FGF2 for 15 minutes. After the stimulation, the cells were lysed and subjected to immunoblot analysis. The immune complexes were visualized with a chemiluminescence system (ECL Western Blotting Detection Reagents; GE Healthcare Bio-Sciences Corp., Piscataway, NJ, and analyzed in an LAS-3000mini image analyzer (FUJIFILM, Kanagawa, Japan, Statistical Analysis The statistical significance of the variations was evaluated using an unpaired two-tailed Student s t test. RESULTS Reduction of Progenitor Cell Proliferation in the Developing Spinal Cord of Gab1-Deficient Mice After the Neurogenic Period Pattern formation and cytogenesis in the vertebrate spinal cord have been extensively analyzed [3, 4, 27]. To elucidate the role of Gab1 in CNS development, we conducted a study focusing on the proliferation of progenitor cells in the developing spinal cord in Gab1-deficient mice. Until E11.5, the development of the spinal cord in the Gab1 / mice was indistinguishable from that in the wild-type (wt) mice (supplemental online Fig. S1). In contrast, at E12.5, proliferative activity of progenitor cell, as determined by 5-bromo-2 -deoxyuridine (BrdU) incorporation and phosphohistone H3 (ph3) expression, was significantly lower in the Gab1 / mice than that in the wt mice (Fig. 1A, 1B, 1E, 1F, 1S). Moreover, in the pmn domain, where glial progenitors begin to appear at this stage [12, 28, 29], a significant reduction in the number of Olig2 progenitor cells (43.5%) was observed in the Gab1 / mice (Fig. 1I, 1M, 1S). The numbers of Olig1 and Mash1 cells, which are subpopulations of Olig2 cells in the pmn domain [30], were also decreased by 27.2% and 40.1%, respectively (Fig. 1J, 1L, 1N, 1P, 1S). In contrast to the dramatic changes in the pmn domain at this gliogenic stage, no differences in the number of Nkx2.2 cells were observed in either the p3 domain or the marginal zone (MZ) between the Gab1 / and the wt mice (Fig. 1K, 1O; data not shown). Interestingly, no changes in the numbers of Pax7 cells and Mash1 cells in the dorsal segment of the spinal cord were observed either in the Gab1-deficient mice (Fig. 1C, 1D, 1G, 1H; data not shown), despite the reduction in progenitor cell proliferation throughout the D-V axis. We did not detect any increase in cellular apoptosis as determined by terminal deoxynucleotidyl transferase dutp nick-end labeling assay and the presence of pyknotic nuclei (Fig. 1Q, 1R; data not shown). These findings indicate that Gab1 is involved in the proliferation of spinal cord progenitor cells, in particular that of Olig2 progenitor cells in the pmn domain after gliogenesis begins. Therefore, in the subsequent experiments, we focused on the roles of Gab1 in the proliferation of Olig2 progenitors.

3 1412 Role of Gab1 in Spinal Cord Development Figure 1. Significant reduction in the number of Olig2 progenitors in the E12.5 Gab1 / spinal cord. Cross sections of E12.5 wt and Gab1 / embryos were immunostained with antibodies directed against BrdU ([A, E]; red), ph3 ([B, F]; red), Pax7 ([C, G]; green), Mash1 ([D, H, L, P]; red), Olig2 ([I, M]; green), Olig1 ([J, N]; green), Nkx2.2 ([K, O]; red), and TUNEL labeled ([Q, R]; red). Panels (D) and (H) represent higher-magnification micrographs of the dorsal spinal cord. Panels (I P) represent higher-magnification micrographs of the ventral spinal cord. Quantification of the cells is shown in panel (S). Five sections (12 m) from each embryo were counted (three embryos for each genotype). The absolute numbers of immunoreactive cells are represented as mean SEM (blue bars: wt, red bars: Gab1 / ;, p.05;, p.01 vs. wt). Scale bars: 50 m (shown in [G] for [A C] and [E G] and in [R] for [D, H R]). Abbreviations: BrdU, 5-bromo-2 -deoxyuridine; D, dorsal; TUNEL, terminal deoxynucleotidyl transferase dutp nick-end labeling; V, ventral. Gab1 Is Essential for EGF-Dependent Proliferation of Olig2 Progenitors Derived from the E12.5 Spinal Cord To determine what kind of growth factor signal Gab1 mediates in the proliferation of the Olig2 progenitors, we first carried out primary cultures of the E12.5 spinal cord from Gab1-mutant mice. Whole spinal cord cells from E12.5 embryos were dissociated and grown in the presence or absence of EGF, FGF2, or PDGF-AA, which are known mitogens for neural stem/progenitor cells (NSPCs) [31 33]. Transforming growth factor- (which can substitute the EGF function), FGF2, and PDGF-A are expressed in the developing spinal cord [34 36]. One micromolar of BrdU was added to the cultures to label the newly generated cells. Seventy-two hours after plating, the cultured cells were fixed and processed for immunostaining for Olig2, BrdU, and O4 expressed in the late oligodendrocyte precursor cell (OPC) [37]. As shown in Figure 2, significant differences were detected between the wt and Gab1 / cells only when they were exposed to EGF. The total number of cells and number of Olig2 cells per field, percentage of Olig2 cells relative to the total cell population, and the percentage of BrdU-incorporated Olig2 cells were all significantly lower in the Gab1 / cultures than in the wt cultures (49.4%, 76.6%, 67.3%, and 64.3% reduction, respectively) and close to those in control cultures grown in the absence of any growth factor (designated as GF ). On the other hand, the percentage of O4 oligodendrocytes in the Gab1 / cultures was not significantly different from that in the wt cultures under any condition. Since the number of Olig2 cells per field 2 hours after plating ( 10 cells; data not shown) was even less than that of GF cultures at 72 hours after plating

4 Hayakawa-Yano, Nishida, Fukami et al Figure 2. Deficiency of Gab1 resulted in a reduction of the EGF-dependent proliferation of Olig2 progenitors in vitro. Whole spinal cord cells from E12.5 Gab1 mutant mice were cultured for 72 hours in the absence of any growth factor or presence of 20 ng/ml EGF, 20 ng/ml FGF2, or 10 ng/ml PDGF-AA and then processed for double immunocytochemistry (ICC) for Olig2/O4 or Olig2/BrdU. Representative micrographs show the results of Olig2/BrdU ([A, B]: Olig2, green; BrdU, red; Hoechst, blue) and Olig2/O4 ICC ([C]: Olig2, green; O4, red) in cells grown in the presence of EGF or GF-. The absolute numbers of Olig2 cells (green bars) and the total number of cells (blue bars) are shown at the bottom in panels (A). The percentages of Olig2 /BrdU cells (yellow bars) and BrdU cells (red bars) and of Olig2 /O4 cells (yellow bars) and Olig2 cells (green bars) are shown at the bottom in panels (B) and (C), respectively. Ten visual fields in each condition were randomly chosen and counted. Each bar represents mean SEM (n 3 embryos for each genotype;, p.05;, p.01 vs. wt). (D): Developmental changes in the EGF receptor expression in the mouse spinal cord. Confocal micrographs of Olig2 (green) and EGFR (red) immunohistochemistry in the thoracic spinal cord at E12.5, E13.5, and E14.5 are shown. Right panels are higher-magnification pictures corresponding to the squares (dotted lines) in left panels. Expression of EGFR was not detected until E11.5. At E12.5 and E13.5, EGFR expression was detected predominantly in the Olig2 -pmn domain. At E14.5, EGFR expression was also detected in the dorsal progenitor domain, including Olig2 cells. Scale bars: 50 m (shown in the bottom left panel for all left panels); 10 m (upper left panel and shown in bottom right panel for both bottom and middle right panels). Abbreviations: BrdU, 5-bromo-2 -deoxyuridine; E, embryonic day; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; FGF, fibroblast growth factor; GF, without growth factor; PDGF, platelet-derived growth factor. (an average of 16 cells in wt cultures), the difference between wt and Gab1 / cultures in the presence of EGF could not be caused by a selective cell death. Thus, although Gab1 is essential for EGF-dependent proliferation of the spinal cord Olig2 progenitors, it may not mediate either FGF2 or PDGF-AA signaling in this context. It is worthy of note that, in the wt cultures, the vast majority of BrdU cells grown in the presence of EGF were Olig2 cells (87.4%), whereas proliferation of mainly Olig2 cells (44.1%) was induced in the presence of FGF2, suggesting that the EGF signal induces a selective proliferation of Olig2 progenitor cells. In fact, when we separately cultured the dorsal and ventral halves of the E12.5 mouse spinal cord in the presence or absence of EGF for 72 hours, EGFdependent expansion of Olig2 cells was observed almost exclusively in the ventral cord cultures (14.2-fold more Olig2 cells in the ventral cord than in the dorsal cord cultures after 72 hours of incubation in the presence of EGF) (supplemental online Fig. S2). This is consistent with a report by Gabay et al. [38]. They showed that FGF2 but not EGF induces Olig2 progenitors from Pax7 /Olig2 dorsal progenitors derived from E14 rat dorsal spinal cord, which corresponds to that of mouse E12 E13. Finally, we analyzed EGFR expression in the developing spinal cord by immunohistochemistry. According to a previous report, EGFR expression cannot be detected by reverse transcription-polymerase chain reaction until E11.5 in the mouse spinal cord; however, thereafter, it is gradually upregulated and becomes immunohistochemically detectable at E14.5 in the ventral region of the generative zone [39]. In this study, we succeeded in raising the sensitivity of EGFR immunohistochemistry using a different source of the antibody, which allowed us to analyze its expression more precisely. As seen in Figure 2D, EGFR expression was restricted to Olig2 cells in the pmn domain at E12.5. Thus, EGF signaling may induce a selective proliferation of Olig2-expressing progenitor cells from the pmn domain, at least in vitro.

5 1414 Role of Gab1 in Spinal Cord Development Table 1. In vitro properties of Olig2 progenitors derived from E12.5 spinal cord Condition EGFR Nestin NG2 Mash1 Nkx2.2 BrdU O4 GF EGF FGF E12 mouse spinal cords were dissociated and cultured in the presence or absence of EGF or FGF2 (20 ng/ml) for 72 hours followed by fixation and double immunocytochemistry of Olig2 with several progenitor markers. The expansion was confirmed by BrdU labeling, in which 1 M BrdU was administered after plating. Values are percentages of immunopositive cells in the Olig2 cells (mean SEM, n 3)., p.05;, p.01; and, p.001 vs. GF cultures. Abbreviations: BrdU, 5-bromo-2 -deoxyuridine; EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; FGF, fibroblast growth factor; GF, without growth factor. To clarify the cellular phenotype of the EGF-responsive Olig2 progenitors derived from E12.5 spinal cord in vitro, we assessed the expression of several progenitor markers by double ICC and compared it with that of the Olig2 population proliferating in response to FGF2 (Table 1 and supplemental online Fig. S3). In contrast to the GF, exposure to EGF or FGF2 for 72 hours significantly increased the percentage of Olig2 progenitors coexpressing EGFR and markers of undifferentiated progenitors including glial progenitors, Nestin, and NG2 [40], indicating that Olig2 cells still maintain their undifferentiated state in the presence of EGF or FGF2. In the developing spinal cord, after motoneuron generation ceases, some of the Olig2 cells in the pmn domain acquire expression of Mash1 [30], then begin to express Nkx2.2 and subsequently differentiate into oligodendrocytes [29, 41]. Additionally, Olig2 OPCs may also be recruited from other progenitor domains (e.g., the p3 domain expressing Nkx2.2) [28, 29]. On the other hand, some of these Olig2 cells in the pmn domain finally differentiate into astrocytes [8]. Therefore, we conducted further investigation to determine the expressions of Mash1 and/or Nkx2.2 in the Olig2 progenitors, in order to clarify what type of Olig2 progenitors proliferate in response to EGF in vitro (Table 1 and supplemental online Fig. S3J S3R). Approximately one-half of the Olig2 progenitors coexpressed Mash1 under all the culture conditions employed, although no Mash1 single-positive cells were detected. In contrast, the percentage of Olig2 /Nkx2.2 cells significantly differed depending on the growth factor. Exposure to EGF resulted in a decrease in the proportion of Nkx2.2 population, whereas that to FGF2 rather increased it. There was no significant difference in the proportion of O4 late OPCs in any condition. Note that there were no Olig2 cells expressing -III-tubulin, a marker for postmitotic neuron, or glial fibrillary acidic protein, a marker for astrocyte (data not shown), in these cultures. Taken together, the results suggest that EGF signal is considerably specific to the induction of the proliferation and self-renewal of undifferentiated or nonoligodendrocytic lineage Olig2 progenitors compared with FGF signal. On the other hand, the absolute number of Olig2 /Nkx2.2 OPCs per field in the EGF cultures was significantly higher than that in GF culture (GF, 2 1; EGF, ; p.0012; n 3), whereas the total cell number was much higher in the EGF culture (GF, ; EGF, ; p.001; n 3), suggesting that EGF may possibly be a weak mitogen and/or survival factor for some of them expressing EGFR. Alternatively, their EGF-dependent increase may have resulted from spontaneous differentiation of undifferentiated Olig2 /Nkx2.2 progenitors during their expansion. Since there is virtually no Olig2 /Nkx2.2 cells in the mouse spinal cord at E12.5 yet ([28] and data not shown), it is possible that Olig2 /Nkx2.2 OPCs found in our cultures, if not all, were derived from a certain population of Olig2 /Nkx2.2 progenitors. Otherwise, they may have been originated from Nkx2.2 single positive cells as in vivo [28]. In that case, EGF signal may not be involved in the gain of Olig2 expression, because EGFR expression is restricted to the Olig2 cells at E12.5 (Fig. 2D). The fact that the proportion of Olig2 /Nkx2.2 OPCs within Olig2 cells derived from Gab1-deficient E12.5 spinal cords after 72 hours of culture in the presence of EGF was not different from that in GF cultures and was similar to that in GF cultures of wt spinal cord (supplemental online Fig. S4) suggests that Gab1 is required not only for proliferation but also for self-renewal of those Olig2 progenitors. Alternatively, a certain level of mitotic activity may be essential for the maintenance of undifferentiated state of Olig2 /Nkx2.2 progenitors. In fact, the proportion of Nestin cells within Gab1- deficient Olig2 cells in EGF or GF cultures was also the same level as that in wt GF cultures. Again, in case that Olig2 /Nkx2.2 OPCs are mainly originated from Nkx2.2- single positive cells in these cultures, the result that EGF did not induce the reduction in the proportion of Olig2 /Nkx2.2 OPCs within Olig2 cells in Gab1 / cultures may be simply reflecting the reduction of proliferation of Olig2 /Nkx2.2 progenitors. Spatiotemporal Differences in the Requirement of Gab1 for the Expansion of Olig2 Progenitors In the course of this study, we found that EGFR expression, which is initially restricted to the Olig2 pmn domain in the E12.5 spinal cord, then spreads to the dorsal progenitor domain, including Olig2 cells at E14.5 (Fig. 2D). Therefore, we analyzed Gab1-deficient mice to clarify the roles of Gab1 as a mediator of EGFR signaling in spinal cord histogenesis. Since Gab1 / embryos can hardly survive beyond E15 [24], we analyzed these embryos as far as E14.5. At E13.5, when some of the Olig2 progenitors in the pmn domain start to express Nkx2.2 and migrate away from the VZ as OPCs [28, 29, 41], the progenitor cell proliferative activity, as defined by the expression of ph3, was still lower in the Gab1 / spinal cords than in the wt spinal cords, throughout the D-V axis (Fig. 3A, 3G, 3M). Similarly, the number of Olig2 progenitors in the VZ was also significantly lower, although the difference between the wt and Gab1 / was not as significant at this stage as that at E12.5 (Fig. 3A, 3G, 3Q). The number of Olig2 /Nkx2.2 OPCs in the Gab1 / spinal cords was also significantly lower in the absence of any significant difference in the total number of Nkx2.2 cells (Fig. 3B, 3H, 3N, 3R). At E14.5, when the subpopulation of Pax7 dorsal progenitors starts to express Olig2 [9, 10], no significant difference in the number of Olig2 cells was observed in the ventral VZ, in contrast to the E13.5 spinal cord. However, the number of these cells in the ventral MZ was significantly lower in the Gab1 / mice than in the wt mice (Fig. 3C, 3I, 3Q). The difference in the number of Olig2 /Nkx2.2 OPCs, which had been detectable at E13.5, was also lost at E14.5 (Fig. 3C, 3I, 3O, 3S). These results

6 Hayakawa-Yano, Nishida, Fukami et al Figure 3. Analysis of E13.5 and E14.5 spinal cords of Gab1-deficient mice. Cross-sections of E13.5 E14.5 wt and Gab1 / embryos were analyzed by immunohistochemistry for Olig2 ([A E, G K], green), ph3 ([A, G], red), Nkx2.2 ([B, C, H, I], red), EGFR ([E, F, K, L], red), and Pax7 ([D, J], red; [F, L], green). Quantification of the cells is shown in panels (M T). Counting was conducted in five sections (12 m) from each embryo counted. The absolute numbers of immunoreactive cells are represented as mean SEM (blue bars: wt; red bars: Gab1 / ; n 3 embryos for each genotype;, p.05;, p.01 vs. wt). Scale bars: 50 m (shown in [I] for [A, C, G, I], in[h] for [B, H], and in [L] for [D F, J L]). (U X): Regional differences in the requirement of Gab1 for epidermal growth factor (EGF)-dependent proliferation of Olig2 progenitors in vitro. Cells from the dorsal and ventral segments of the spinal cord prepared separately from E14.5 Gab1-deficient mice were cultured for 2 or 72 hours in the presence of 1 M BrdU and in the presence or absence of 20 ng/ml of EGF, followed by triple immunocytochemistry for Olig2, Pax7, and BrdU. Representative micrographs show the Olig2 (red)/pax7 (green)/brdu (blue) to cells grown for 72 hours in the presence of EGF (U). The numbers of immunoreactive cells in ten randomly chosen fields ( 800 total cells) were counted. Expansion of Olig2 and/or Pax7 cells between 2 and 72 hours after plating is shown in relative terms (fold) (blue bars: wt; red bars: Gab1 / ) (V). Percentages of Olig2 and/or Pax7 cells relative to the total number of cells in each condition (W) and the time point (2 and 72 hours after plating, respectively) are shown (Olig2 cells, red bars; Olig2 /Pax7 cells, yellow bars; Pax7 cells, green bars). Percentages of BrdU and BrdU /Olig2 (including Olig2 /Pax7 ) cells relative to the total number of cells after 72 hours culture in the presence of EGF are shown (BrdU cells: blue bars; Olig2 /BrdU cells: pink bars) (X). Each value represents mean SEM (n 3 embryos;, p.05;, p.01 vs. wt). In panel (W), the asterisks for single Olig2 or Pax7 cells and Olig2 /Pax7 cells are shown in the same color as the respective bars. In panel (X), the asterisks for the total number of BrdU and Olig2 /BrdU cells are shown in blue and pink, respectively. Abbreviations: BrdU, 5-bromo-2 -deoxyuridine; C2h, 2 hours after plating without growth factor; C72h, 72 hours after plating without growth factor; D, dorsal cultures; dp1 P2, dp1 domain to P2 domain; E, embryonic day; E72h, 72 hours after plating in epidermal growth factor; EGFR, epidermal growth factor receptor; MZ, marginal zone; Olig2D, Olig2 progenitors in the dorsal segment; V, ventral cultures; VZ, ventricular zone; wt, wild-type.

7 1416 Role of Gab1 in Spinal Cord Development Table 2. In vitro properties of EGF-responsive Olig2 progenitors derived from E14.5 spinal cord Condition Region Nestin NG2 Sox10 Nkx2.2 O4 GF Dorsal Ventral EGF Dorsal Ventral Dorsal and ventral halves of E14 mouse spinal cords were independently dissociated and cultured in the presence or absence of EGF (20 ng/ ml) for 72 hours, followed by fixation and double immunocytochemistry of Olig2 with several progenitor markers. Values are percentages of immunopositive cells in the Olig2 cells (mean SEM, n 3)., p.05;, p.01 vs. ventral cultures. Abbreviations: EGF, epidermal growth factor; GF, without growth factor. indicate that Gab1 transiently contributes to the expansion of Olig2 progenitors in the pmn domain between E12.5 and E13.5. In the dorsal segment, the total number of Olig2 cells, which may also include some populations migrating from the ventral segment [9], Pax7 /Olig2 cells, and EGFR /Olig2 cells were significantly lower in the Gab1 / mice, in the absence of any significant differences in the total number of Pax7 cells and EGFR /Pax7 cells (Fig. 3D 3F, 3J 3L, 3P, 3T), indicating that Gab1 is required for the expansion of dorsally originating Olig2 progenitors. Similar to the observation in the E12.5 spinal cord, there was no detectable increase in the number of apoptotic cells as recognized by the presence of pyknotic nuclei in the Gab1 / spinal cord at these stages either (data not shown). To clarify whether the EGF signal for the proliferation of these Olig2 progenitor cells is mediated by Gab1, we carried out primary cultures of E14.5 dorsal and ventral spinal cords of Gab1-deficient mice separately in the presence of 1 M BrdU and in the presence or absence of EGF for 2 and 72 hours, followed by triple ICC for Olig2, Pax7, and BrdU (Fig. 3U 3X). In the case of dorsal cord cell cultures, exposure of wt cells to EGF resulted in robust expansions of the absolute numbers of Olig2 /Pax7 cells, Olig2 /Pax7 cells, and total number of Pax7 cells (5.1-fold, 20.7-fold, and 4.3-fold increase from 2 hours after plating, respectively) but a decrease in that of Olig2 /Pax7 cells (0.29-fold). The percentages of these cell populations relative to the total number of cells changed similarly. On the other hand, exposure of the Gab1 / cells to EGF resulted in some expansion, to a relatively lesser extent, of Olig2 /Pax7 cells (2.1-fold increase), similar expansion of Olig2 /Pax7 cells (25.3-fold), similar decrease of Olig2 / Pax7 cells (0.24-fold), and a decrease in the total number of Pax7 cells (0.83-fold), as compared with the wt cell cultures. The percentages of Olig2 /Pax7 cells, Olig2 /Pax7 cells, whole Pax7 cells, total BrdU cells, and BrdU /Olig2 (including Olig2 /Pax7 ) cells in the Gab1 / cultures were significantly lower than those in the wt cells at 72 hours after plating in the presence of EGF. The proportion of Olig2 cells within BrdU cells was also significantly lower (wt: % vs. Gab1 / : %; p.0088). It is noteworthy that the number of Olig2 cells, including a few Olig2 /Pax7 cells, but not the total number of Pax7 cells and Olig2 /Pax7 cells in the Gab1 / spinal cord cultures was significantly lower than that in the wt spinal cord cultures at 2 hours after plating, confirming the results of the immunohistochemical analysis described above. In the case of the ventral cord cultures, to our surprise, there was no significant difference in the EGF-dependent expansion of Olig2 cells between the wt and Gab1 / cultures. Note that most of Olig2 cells ( 95%) expanded in the presence of EGF in both dorsal and ventral cultures were BrdU-incorporated dividing or newly generated cells. These results indicate that, although Gab1 mediates EGFdependent proliferation of Olig2 progenitors, including Olig2 /Pax7 cells, derived from the E14.5 dorsal spinal cord, it is not essential for the EGF-dependent proliferation of Olig2 progenitors derived from the E14.5 ventral spinal cord, unlike those from the E12.5 ventral spinal cord in vitro, which is consistent with the transient requirement of Gab1 in vivo. Some expansion of Olig2 cells in the Gab1 / dorsal cord cultures may represent contamination by the ventral cord population. There was no significant difference in the expansion ratio of Olig2 /Pax7 cells between wt cultures and Gab1 / cultures, whereas that of the total population of Pax7 cells increased in the wt but rather decreased in the Gab1 / dorsal cord cultures, suggesting that Pax7 single-positive progenitors may become Olig2 /Pax7 progenitors in the presence of EGF, and that Gab1 is not necessary for this transition. Indeed, in the absence of EGF, the number of Olig2 /Pax7 cells did not increase at all in either the wt or Gab1 / cultures, whereas the number of Olig2 /Pax7 cells was somewhat decreased in the Gab1 / cultures 72 hours after plating. It has been suggested that Olig2 progenitor cells derived from the dorsal Pax7 progenitors generate OPCs just like the corresponding cells from the ventral cord [9, 10]. On the other hand, the requirement of Gab1 for the EGF-dependent proliferation of Olig2 progenitor cells is context-dependent (i.e., spatially and temporally regulated as described above). Since the timing of differentiation of dorsally derived OPCs is likely to be later than that of the ventrally derived OPCs [9, 10], it is possible that differences in the cellular context depending on the differentiation status cause the difference in Gab1 requirement. To clarify whether the differentiation status of EGF-responsive Olig2 progenitor cells is regionally different in the E14.5 spinal cord, we compared the cellular phenotype of EGF-responsive Olig2 progenitors derived from the dorsal segment with that of the cells derived from the ventral segment. We carried out a similar in vitro analysis to that described in Table 1 and supplemental online Figure S3. We cultured cells from the dorsal and ventral halves of E14.5 spinal cord separately, followed by double immunocytochemistry for Olig2 with Nestin, NG2, Nkx2.2, O4, or Sox10, which is another OPC marker [42] (Table 2 and supplemental online Fig. S5). Along with the expansion of Olig2 cells in the presence of EGF, significant differences between the dorsal and ventral cultures appeared in the expressions of NG2, Sox10, Nkx2.2, and O4. The proportions of cells expressing these markers within the population of Olig2 cells in the presence of EGF were significantly lower in the dorsal cultures than those in the ventral cultures, whereas most of the Olig2 cells in either of the two cultures coexpressed Nestin. In particular, the majority of Olig2 cells in the dorsal cultures showed no expression of Sox10 or NG2 (also expressed by OPCs) [43] (70.1% and 89.1%, respectively), whereas a small proportion (0.8%) was also O4, indicating that these dorsally derived EGF-responsive Olig2 progenitor cells were more undifferentiated or specified to nonoligodendrocytic lineage than the corresponding ventrally derived cells.

8 Hayakawa-Yano, Nishida, Fukami et al Figure 4. Reduction in the number of Olig2 progenitors in the epidermal growth factor receptor-deficient mouse. Cross-sections of E13.5 E14.5 wt and Egfr / embryos were analyzed by double immunohistochemistry for Olig2 ([A, B, E, F, I, J, M, N], green) and ph3 ([A, B], red), Nkx2.2 ([E, F, I, J], red), or Pax7 ([M, N], red). Quantification of the cells is shown in panels (C, D, G, H, K, L, O). The absolute numbers of immunoreactive cells in the five sections (12 m) counted are represented as mean SEM (blue bars: wt; red bars: Egfr / ; n 3 embryos for each genotype;, p.05;, p.01 vs. wt). Scale bars: 50 m (shown in [B] for [A, B], in[j] for [I, J], and in [N] for [E, F, M, N]). Abbreviations: dp1 P2, dp1 domain to P2 domain; E, embryonic day; MZ, marginal zone; Olig2D, Olig2 progenitors in the dorsal segment; VZ, ventricular zone. Reduction in the Number of Olig2 Progenitors in the Developing Spinal Cord of EGFR-Deficient Mice To confirm the requirement of the EGF signal for the expansion of Olig2 progenitors in vivo, we analyzed the developing spinal cord of EGFR-deficient (Egfr / ) mice employing the same methods as those for the analysis of Gab1-deficient mice. Unlike the Gab1 / spinal cord, the Egfr / spinal cord showed no significant difference in the number of Olig2 progenitors at E12.5 compared to wt spinal cord (data not shown). After E13.5, however, a similar reduction in the number of Olig2 progenitors to that in the Gab1 / spinal cord was observed in the Egfr / mice (Fig. 4). At E13.5, although a significant reduction in the number of Olig2 cells was observed in both the VZ and MZ, no significant reduction in the overall cellular proliferation as determined based on ph3 expression was observed in the VZ (Fig. 4A 4D). The number of Olig2 /Nkx2.2 OPCs was also reduced, whereas there was no significant change in the total number of Nkx2.2 cells (Fig. 4E 4H). In the ventral segment, at E14.5, a significant reduction in the number of Olig2 cells was observed only in the MZ, whereas no change in the numbers of Olig2 /Nkx2.2 OPCs and total number of Nkx2.2 cells was observed (Fig. 4I 4L). In the dorsal segment, the number of Olig2 cells, including Olig2 /Pax7 cells, was significantly reduced, whereas the total number of Pax7 cells did not change (Fig. 4M 4O). We did not find any change in the number of apoptotic cells as recognized by the presence of pyknotic nuclei (data not shown). These results indicate that the EGF signal certainly contributes to the expansion of Olig2 progenitor cells, at least to some extent, in the developing spinal cord. Moreover, together with the in vitro results from the analysis of Gab1 / mice, Gab1 may mediate this action of EGF. The milder phenotype of the Egfr / mice than the Gab1 / mice with respect to the reduction in the number of ventrally derived Olig2 progenitors may be caused by the difference in the genetic background of the mice (ICR and C57BL6, respectively). In fact, the number of total and ratio of migrating (localized in MZ) Olig2 /Nkx2.2 OPCs in ICR mice at E13.5 were less than that in the C57BL6 mice (Figs. 3, 4), indicating a delay of spinal cord development of ICR mice at this time point. Alternatively, Gab1 may mediate other unidentified signals, inducing the expansion of Olig2 progenitors, especially until E13.5. In this case, PDGF signal, which is a major factor for OPC proliferation in vivo [44], may be excluded from the candidates. Gab1-deficiency did not affect PDGF-AA-dependent proliferation of Olig2 progenitors derived from E12.5 or E14.5 spinal cord, at least in vitro (Fig. 2 and supplemental online Fig. S6).

9 1418 Role of Gab1 in Spinal Cord Development Involvement of Akt Downstream of Gab1 in the EGF-Dependent Proliferation of Olig2 Progenitors Finally, in order to determine which pathway, the PI3K/Akt pathway and/or the Ras/MAPK pathway, downstream of Gab1 is activated by the EGF signaling in the developing spinal cord, we first performed immunoblotting of the phosphorylated forms of various signaling molecules in the wt and Gab1 / cells (Fig. 5A). Since E12.5 and E14.5 dorsal Gab1 / spinal cord cells cannot proliferate in response to EGF, these cells were first grown for 72 hours in the presence of 20 ng/ml FGF2 and lysed after EGF or FGF2 stimulation for 15 minutes. The average percentages of the expanded Olig2 cells in the E12.5 spinal cords in the presence of FGF2 were 55% and 50.9% in the wt and Gab1 / cells, respectively, and those in the E14.5 dorsal spinal cords were 41.2% and 39.2%, respectively. Moreover, the proportions of expanded Olig2 cells within the EGFR cell population in the presence of FGF2 derived from E12.5 whole and E14.5 dorsal spinal cords of the wt mice were 60.3% and 75.2%, respectively. Therefore, we should have been able to pick up the signal response in the Olig2 progenitors to FGF2 or EGF stimulation at least to some extent. In the E12.5 cultures, although significant increases in the phosphorylation of Akt1 at Thr-308 (2.8-fold relative to that in GF ) and ERK1/2 (p42/ p44) (3.8- and 2.7-fold relative to that in GF, respectively) were detected upon EGF stimulation in the wt cells, virtually no such increase in the phosphorylation of Akt1 at Thr-308 (1.2- fold relative to that in GF ; p.05 compared with wt) was detected in the Gab1 / cells, although phosphorylation of ERK1/2 (3.3- and 3.0-fold relative to that in GF, respectively) was detected. This result indicates that Gab1 is required for the activation of the PI3K/Akt but not the Ras/MAPK cascade by EGF signaling in the E12.5 spinal cord progenitor cells. In both the wt and Gab1 / cells, FGF2 also induced significant activation of ERK1/2, whereas only a small increase in the phosphorylation of Akt1 at Thr-308 was detected. Thus, no involvement of Gab1 in FGF2 signaling could be confirmed biochemically. Slight phosphorylation of Akt1 at Ser-473 was detected, but without any difference between the pre- and poststimulation for any growth factor (data not shown). Finally, neither EGF nor FGF2 activated STAT3 recognized by the phosphorylation on Tyr-705 in E12.5 spinal cord cultures, indicating that STAT3 is not essential for EGF signaling in this context. In the E14.5 dorsal spinal cord cultures, phosphorylation of Akt1 at Thr-308 was significantly increased in the wt cells (4.6-fold relative to that in GF ) following EGF stimulation, as seen in the E12.5 cultures, whereas only a slight increase was seen in the Gab1 / cells (1.8-fold relative to that in GF ; p.05 compared with wt). The increase in the phosphorylation of ERK1/2 was almost to the same extent (4.3- and 3.5-fold relative to that in GF in the wt and Gab1 / cells, respectively) as that in the E12.5 cultures. On the other hand, in contrast to the E12.5 cultures, EGF stimulation induced a significant increase in the phosphorylation of Akt1 at Ser-473 in the wt cells and a slight increase in the Gab1 / cells (4.5- and 2.3-fold relative to that in GF, respectively). Similar to the E12.5 cultures, no STAT3 activation was detected. Thus, Gab1 is required for the activation of the PI3K/Akt cascade, but not the Ras/MAPK cascade in EGF signaling in the E14.5 dorsal spinal cord progenitor cells. In the E14.5 ventral cord cultures, in which Gab1 deficiency did not affect the EGF-dependent proliferation of Olig2 progenitors, to our surprise, EGF stimulation induced phosphorylation of Akt1 at Thr-308 and Ser-473 in the wt cells (3.9- and 3.6-fold relative to that in GF, respectively) and to a lesser extent in the Gab1 / cells (1.8- and 1.7-fold relative to that in GF, respectively; p.05 compared with wt), whereas there was no change in the phosphorylation of ERK1/2 (3.6- and 4.0-fold relative to that in GF in the wt and Gab1 / cells, respectively), suggesting the existence of alternative pathways or a lower threshold level of Akt1 and/or ERK1/2 activation for inducing the proliferation of Olig2 progenitors. Next, to clarify whether Akt activation mediated by Gab1 is essential for the EGF-dependent proliferation of Olig2 progenitors derived from the E12.5 spinal cord and E14.5 dorsal spinal cord, we tested if expression of constitutively active Akt1 can rescue the defect in the Gab1 / spinal cord in vitro. Toward this end, cells from E12.5 spinal cords and dorsal halves of E14.5 spinal cords of Gab1 / mice were transuded by lentiviruses expressing m Akt1, which is an active form of Akt1 [43] with Venus (a modified GFP [45]) through an internal ribosomal entry site (supplemental online Materials and Methods) and cultured in the presence of EGF for 72 hours. The virus-mediated expression and function of m Akt1 were confirmed by immunoblot analysis of E12.5 spinal cord cultures with antibodies against phosphorylated Akt1 (Thr-308) and Akt substrates (Fig. 5D). In the E12.5 cultures, the proportion of Olig2 cells within the cells infected with the m Akt1-expressing viruses was significantly higher than that within the cells infected with control viruses expressing Venus only in Gab1 / cultures and reached the level observed in the wt cells infected with the control viruses; on the other hand, infection of the wt cells with the m Akt1-expressing virus did not result in any significant increase in the number of Olig2 cells (Fig. 5B, 5C). In contrast, in the E14.5 dorsal cultures, expression of m Akt1 could not rescue the Gab1 / phenotype. The number of Olig2 cells did not increase in either the wt or the Gab1 / cultures (Fig. 5B, 5C). These results indicate that, downstream of Gab1, activation of Akt1 is sufficient for the EGF-dependent proliferation of Olig2 progenitors derived from the E12.5 spinal cord, but not for that of the cells from the E14.5 dorsal spinal cord. Taken together, despite Gab1 mediating Akt1 but not ERK1/2 activation by EGF signaling in progenitor cells, including Olig2 progenitors derived from the mouse embryonic spinal cord, the Akt dependence of the proliferation of Olig2 progenitors varies depending on the origin of the cells. DISCUSSION Regulation of Progenitor Proliferation by Growth Factor Signalings in the Developing Spinal Cord In the developing spinal cord, after the cessation of motoneuron generation, Gab1 deficiency resulted in a reduction in the number of Olig2 progenitors in the pmn domain, followed subsequently by a reduction in the subpopulation of Pax7 dorsal progenitors expressing EGFR, without any detectable increase of apoptosis (Figs. 1, 3; data not shown); the cells also showed failure of EGF-dependent proliferation in vitro (Figs. 2, 5). In contrast, the proliferation of not only Olig2 progenitors, but also that of Olig2 progenitors in response to FGF2 stimulation in vitro, was maintained to the same extent in the Gab1-deficient cultures as that in the wild-type cultures (Fig. 2; data not shown), despite the fact that Gab1 mediates signals to the Akt pathway in FGF signaling in mouse embryonic fibroblasts (MEFs) [46]. It has been shown that FGF receptor substrate 2 (FRS2), another adaptor protein belonging to the common insulin receptor substrate family, functions as a key mediator in FGF signaling in other types of cells, including cortical progenitor cells [47, 48]. Moreover, Gab2 mediates Akt activation by FGF2 during retinoic acid-induced neural differentiation of P19 embryonal carcinoma cells [23]. In our study, however, FGF2 stimulation scarcely activated Akt1 in spinal cord progenitors

10 Hayakawa-Yano, Nishida, Fukami et al Figure 5. Involvement of Akt downstream of Gab1. (A): Immunoblot analysis of activation of known intracellular signaling molecules in EGF and FGF2 signalings. Wt and Gab1 / cells from E12.5 whole, E14.5 dorsal, or E14.5 ventral spinal cords were grown for 72 hours in the presence of 20 ng/ml FGF2 (E12.5 whole and E14.5 dorsal) or 20 ng/ml EGF (E14.5 ventral) and then treated with 20 ng/ml of EGF or FGF2 for 15 minutes. The cells were lysed and subjected to immunoblot analysis using the indicated antibodies. (B D): In vitro rescue of the Gab1 / phenotype in relation to the EGF-dependent proliferation of Olig2 progenitors by the introduction of constitutively active Akt1. Wt and Gab1 / cells from E12.5 whole and E14.5 dorsal spinal cord were infected with lentiviruses expressing Venus (control) or the active form of Akt1 and Venus and cultured for 72 hours in the presence of EGF followed by double immunocytochemistry for Olig2 with Venus (using GFP antibody). The infection efficiency as recognized by the expression of Venus was 40% 60% throughout the experiment. Counting was conducted in ten randomly chosen visual fields ( 300 Venus cells). In E12.5 cultures, expression of m Akt1 in the Gab1 / cells significantly increased the number of Olig2 cells within the Venus cell population as compared with that in cells infected with control viruses, and to the same level as that in the wt cells; on the other hand, no such increase was observed in the wt cells (B, C) (, p.05, n 4 embryos). In E14.5 dorsal spinal cord cell cultures, unlike E12.5 cultures, m Akt1 expression did not rescue the Gab1 / phenotype (B, C) (n 5 embryos). Each value (blue bars: wt; red bars: Gab1 / ) represents the mean SEM. Expression and function of m Akt1 were confirmed by immunoblot analysis of E12.5 spinal cord cells (wt) proliferating in the presence of EGF with antibodies against pakt1 (Thr-308), which recognized truncated m Akt1 (47 kda) and phosphorylated Akt substrates (D). Cells were lysed 30 minutes after washing out of the EGF. Abbreviations: E, embryonic day; EGF, epidermal growth factor; FGF, fibroblast growth factor; GF, without growth factor; IB, immunoblot; m Akt1, active form of Akt1.