Author contributions: S.M.: collection and assembly of data, data analysis and interpretation, and final approval of manuscript; K.B., H.M., and G.D.C.O.: collection and/or assembly of data and final approval of manuscript; U.L.: data analysis and interpretation, manuscript writing, financial support, and final approval of manuscript; U.J.: concept and design, data analysis and interpretation, provision of study materials, manuscript writing, financial support, and final approval of manuscript; R.S.: Concept and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS January 10, 2013.
Notch signaling is a key regulator of cell-fate decisions and is essential for proper neuroectodermal development. There, it favors the formation of ectoderm, promotes maintenance of neural stem cells, inhibits differentiation into neurons, and commits neural progenitors to a glial fate. In this report, we explore downstream effects of Notch important for astroglial differentiation. Transient activation of Notch1 during early stages of neuroectodermal differentiation of embryonic stem cells resulted in an increase of neural stem cells, a reduction in neurons, an induction of astroglial cell differentiation, and an induction of neural crest (NC) development. Transient or continuous activation of Notch1 during neuroectodermal differentiation led to upregulation of Sox9 expression. Knockdown of the Notch1-induced Sox9 expression reversed Notch1-induced astroglial cell differentiation, increase in neural stem cells, and the decrease in neurons, whereas the Notch1 effects on NC development were hardly affected by knockdown of Sox9 expression. These findings reveal a critical role for Notch-mediated upregulation of Sox9 in a select set of neural lineage determination steps controlled by Notch. STEM CELLS2013;31:741–751
During neural differentiation, neuroepithelial cells originating from the neural tube differentiate to neurons or glial cells, that is, oligodendrocytes and astrocytes . Neurons differentiate first, followed by differentiation of glial cells. How the relatively simple initial neuroepithelium can give rise to the correct numbers and types of the different subclasses of neurons and glial cells is still rather incompletely understood, but it is becoming increasingly evident that several key intracellular signaling mechanisms play important roles in this process.
One signaling mechanism that exerts profound and multifaceted effects on neural differentiation is the Notch signaling pathway . The Notch signaling pathway is a highly conserved mechanism of intercellular communication that controls developmental processes . In mammals, one of the five ligands (Delta 1, 3, 4, Jagged1, and 2) binds to one of the four Notch receptors (Notch1–4) on the surface of neighboring cells. After ligand binding, the Notch intracellular domain (NICD) is proteolytically cleaved from the transmembrane region allowing it to translocate into the nucleus. Nuclear NICD binds to RBP-J (RBPJk/CBF1/RBPSUH/SUH/CSL) and assembles into a transcriptional activation complex with coactivator proteins and chromatin remodeling enzymes, finally inducing transcription of downstream target genes .
Notch signaling plays important roles at a number of distinct steps during development and maturation of the nervous system, ranging from proliferation within the neuroepithelium to synaptic plasticity in mature neurons . During development of the nervous system, Notch signaling is required for the generation and maintenance of neural stem cells and proper control of neurogenesis in the embryonic and adult brain [5, 6]. Notch signaling is required for gliogenesis in the developing peripheral and central nervous system (CNS), as judged by loss of glial specification when the RBP-J gene was conditionally ablated in the developing nervous system . Furthermore, introduction of NICD into adult neural stem cells promotes glial differentiation . Notch functions are not restricted to the CNS. During neural crest (NC) development, Notch activity first helps to establish the NC domain within the ectoderm via lateral induction and subsequently influences the diversification of NC progeny via lateral inhibition . It is becoming increasingly clear that the Notch downstream signaling response is quite cell context dependent [3, 10–12], but how this downstream diversity relates to the fact that Notch signaling at an early stage of neural differentiation blocks differentiation and maintains the stem cell state while at later stages promoting gliogenesis is not established. While hairy and enhancer of split (HES) proteins are induced by Notch signaling and inhibit neural commitment by antagonizing the activity of proneural genes [6, 13], they do not account for the inductive roles of Notch activity during later stages of neuroectodermal development.
Sox9 is a transcription factor that plays an important role in the development of the CNS and NC. Sox9 belongs, together with Sox8 and Sox10, to the SoxE group of sex-determining region-related HMG-box family. In the CNS, Sox9 is essential for gliogenesis [14, 15] and necessary and sufficient to initiate the induction of embryonic and adult neural stem cells . It is further expressed directly after specification in NC development  and favors NC at the expense of CNS development . As in the CNS, Sox9 blocks differentiation into neurons and supports glial development in the peripheral nervous system . Sox9 has been implicated in Notch-mediated gliogenesis as Sox9 expression is downregulated upon RBP-J ablation in the developing nervous system . Due to the fact that Sox9 expression is upregulated following activation of Notch in embryonic stem cells (ESCs) undergoing neural differentiation [10–12], regulation of Sox9 by Notch in the nervous system warrants further analysis.
In this study, we examined the role of Sox9 in Notch1-mediated control of neural differentiation. Temporally controlled activation of Notch1 in ESC during in vitro differentiation toward the neural lineage revealed roles for Notch signaling in increasing the number of neural stem cells, reducing the number of neurons, inducing glial cell differentiation and inducing NC development. Notch1 upregulated Sox9 expression at all these steps; however, knockdown of the Notch1-induced Sox9 expression by siRNA specifically reversed glial and dose-dependently neuronal differentiation as well as stem cell promotion to wild-type levels while having no effect on Notch1-induced NC development. These data define Sox9 as an important Notch downstream gene in controlling a subset of Notch-mediated responses during neuroectodermal development.
MATERIALS AND METHODS
Cell Culture and Differentiation
The ESC line EB5-NERT  expressing a 4-hydroxy-tamoxifen (OHT) inducible form of the Notch1 intracellular domain was grown in ES maintenance medium consisting of KO-Dulbecco's modified Eagle's medium (DMEM) (Invitrogen, Karlsruhe, Germany, http://www.lifetechnologies.com) supplemented with 10% knockout serum replacement (Invitrogen, Karlsruhe, Germany, http://www.lifetechnologies.com), 1% pretested fetal calf serum (FCS) (Cambrex/Lonza, Cologne, Germany, http://www.lonza.com/), nonessential amino acids (Invitrogen, Karlsruhe, Germany, http://www.lifetechnologies.com, stock solution diluted 1:100), 2 mM L-glutamine (Invitrogen, Karlsruhe, Germany, http://www.lifetechnologies.com), 5 × 10−5 M 2-mercaptoethanol (Sigma-Aldrich, Munich, Germany, http://www.sigmaaldrich.com), and 10 ng/ml (103 U/ml) cell culture tested leukemia inhibitory factor as described previously . We have shown recently that in vitro differentiation of control EB5 cells was not affected by OHT application . Neural differentiation of EB5-NERT cells was carried out according to the protocol of Ying et al. using a monolayer cell culture in a N2B27 supplemented medium . Briefly, trypsinized cells were washed, centrifuged, and resuspended in N2B27 medium (49% Neurobasal medium, 49% DMEM/F12 + Glutamax, 1% B27 supplement, 0.5% N2 supplement, 0.5% L-glutamine, 12.5 μg/μl insulin [all Invitrogen, Karlsruhe, Germany, http://www.lifetechnologies.com], 37.5 μg/ml bovine serum albumin [Sigma-Aldrich, Munich, Germany, http://www.sigmaaldrich.com], and 0.1 mM 2-mercaptoethanol) at a cell density adjusted to 0.125–1 × 105 cells per well of a six-well plate (Corning, Amsterdam, The Netherlands, http://www.corning.com) in 2 ml medium. For the experiments comprising 1–16 days of differentiation, the medium was changed every 1 or 2 days. Activated Notch1 was induced by addition of 5 or 50 nM OHT for 0–2 days of differentiation or the whole differentiation period as indicated. All differentiation experiments were repeated at least three times. To confirm Sox9 and Pax6 as direct target genes of Notch1 in neuroectodermal differentiation, blocking of protein synthesis was achieved by addition of 50 μg/ml cycloheximide (CHX) for 4 hours. The ESC line Notch1ΔETetOn was grown and differentiated as described previously [11, 22]. Activated Notch1 was produced in Notch1ΔETetOn cells by addition of 0.5 μg/ml doxycycline for 0–2 days of differentiation.
Gene Expression Analysis by RT-qPCR
RNA was prepared using the RNeasy Plus Kit according to manufacturer's protocol (Qiagen, Hilden, Germany, http://www.qiagen.com), integrity checked on an agarose gel, and concentrations determined photometrically using the NanoDrop ND1000 (Thermo Scientific, Wilmington, DE, http://www.thermoscientific.com). Total RNA (0.5 or 1 μg) was reversely transcribed with the “First Strand cDNA Synthesis Kit” (Fermentas/Thermo Scientific, St. Leon-Rot, Germany, http://www.thermoscientific.com) using an oligo(dT)18 primer at 37°C for 60 minutes according to manufacturer's protocol. In the case of Notch1ΔETetOn cells, 1 μg total RNA was reversely transcribed with Superscript III Reverse Transcriptase (Invitrogen, Karlsruhe, Germany, http://www.lifetechnologies.com) using an oligo(dT)18 primer at 50°C for 60 minutes according to manufacturer's protocol. mRNA expression levels of Fgf5, Gapdh, Id3, Nestin, Pax6, Oct4, and Sox9 were measured by qPCR using the TaqMan Assays-on-Demand system (Applied Biosystems/Life Technologies, Darmstadt, Germany, http://www.lifetechnologies.com, Supporting Information Table S1 for Assay number and amplicon sizes). The expression of the genes Brachyury(T), glial fibrillary acidic protein (GFAP), Hprt, Foxd3, Snai1, Snai2, Sox2, Sox17, and Tubb3 was measured by SYBR green detection and S100b by Universal Probe Library (Roche Diagnostics, Rotkreuz, Switzerland, http://www.roche-applied-science.com) detection (Supporting Information Table S2). All reactions were measured on a 7900HT Fast Real-Time PCR System (Applied Biosystems, Darmstadt, Germany, http://www.lifetechnologies.com) or on a LightCycler 480 (Roche Diagnostics, Rotkreuz, Switzerland, http://www.roche-applied-science.com) in 384-well PCR-plates in a 10 μl reaction volume using 0.5 μl cDNA. For TaqMan detection, the “TaqMan 2× Universal Master Mix (No AmpErase UNG)” (Applied Biosystems, Darmstadt, Germany, http://www.lifetechnologies.com) was used, whereas SYBR green detection was performed with “Power SYBR green Master Mix” (Applied Biosystems, Darmstadt, Germany, http://www.lifetechnologies.com). All assays were tested in initial experiments for amplification efficiency and these efficiencies were used for later calculation of gene expression relative to Gapdh or Hprt . Induction by activated Notch1 was calculated by the formula: I = Expressioninduced/Expressionuninduced.
Flow Cytometry Analysis
Expression of different marker proteins was monitored by intracellular flow cytometry analysis using the “IntraStain Kit” (Dako, Glostrup, Denmark, http://www.dako.com) using antibodies against Nestin, Sox2, Tubb3, and GFAP according to the manual of the manufacturer with minor changes. Briefly, 50 μl IntraStain Reagent A (Fixative) was added to up to 106 cells in 50 μl medium and incubated for 15 minutes at RT. Cells were washed by addition of 2 ml flow cytometry wash buffer (5% FCS and 0.1% sodium azide in phosphate buffered saline), centrifuged, and the supernatant was discarded. Remaining cells were resuspended in about 10 μl buffer and blocked for 20 minutes by addition of 50 μl IntraStain Reagent B +0.1% NP40 + 3% normal goat serum. Subsequently, the respective primary antibody was added and incubated for 20 minutes at 4°C. Since antibodies for the GFAP and Pax6 proteins were not directly labeled, the protocol was extended for an additional washing step with 2 ml flow cytometry wash buffer and staining with a secondary antibody in 50 μl IntraStain Reagent B +0.1% NP40 for 20 minutes at 4°C in the dark, and washed again. Used antibodies and isotype controls are itemized in Supporting Information Table S3. Filtered samples were analyzed on a FACSCanto (Becton Dickinson, Heidelberg, Germany, http://www.bd.com) using the FACS Diva software (Version 5.03).
Knockdown of Sox9 and Pax6
EB5 NERT or Notch1ΔETetOn cells were trypsinized, centrifuged, counted, and then resuspended and seeded in ES maintenance medium into six-well plates (Corning, Amsterdam, The Netherlands, http://www.corning.com). After 6–8 hours, the medium was changed to N2B27, and Notch1 induction as well as the transfection for knockdown was started. Two micromolar (Pax6) or 4 μM (Sox9) siRNA stock solutions of Sox9 or Pax6 pool of siRNAs (both Dharmacon/Thermo Scientific, Wilmington, DE, http://www.thermoscientific.com) were mixed with N2B27-medium in a 200 μl volume and incubated for 5 minutes at RT. In parallel, 4 μl DharmaFECT1 (Dharmacon/Thermo Scientific, Wilmington, DE, http://www.thermoscientific.com) was mixed with 196 μl serum-free medium and also incubated for 5 minutes at RT, and then the diluted siRNA was added to this mix and incubated for an additional 20 minutes at RT. This transfection mix was added to the cells in 2 ml N2B27 medium in the six-well plates. The final concentration of Sox9 and Pax6 siRNA was 100 nM and 50 nM, respectively. As a control, a 100 nM pool of nontargeting siRNA (negative control siRNA, Dharmacon/Thermo Scientific, Wilmington, DE, http://www.thermoscientific.com) or transfection reagent only was used.
Data Analysis and Statistics
Statistics were analyzed as paired data for the Student's t test or, in the case of uneven sample numbers, as heteroscedastic data. The qPCR data were calculated by comparison of the ΔCt values of the different sample populations, whereas flow cytometric results were calculated on basis of the resulted induction values. Different levels of significance according to the p values are indicated by *, p < .05; **, p < .01; and ***, p < .005.
Notch1 Signaling During Neuroectodermal Commitment Promotes Neural Stem Cells, Inhibits Neuronal Differentiation, and Induces Glial and NC Differentiation
To analyze the role of Notch1 signaling during neuroectodermal differentiation of ESCs in vitro, we took advantage of a previously established ESC system, EB5-NERT, in which Notch1 signaling can be activated by OHT [19, 23]. EB5-NERT cells were differentiated in a monolayer culture favoring the generation of neuroectodermal cells. Notch1 signaling was either continuously activated by the addition of OHT throughout culturing or transient during early stages of differentiation (0–2 days). In the absence of OHT, the EB5-NERT cells differentiated into neural stem cells forming rosettes by days 4–6 of neuroectodermal differentiation and further from day 8 onward into mature neuronal cells expressing the neuronal marker protein Tubb3 and from day 10 onward into glial cells expressing the glial cell marker protein GFAP (Fig. 1A, 1C, 1E, 1G). Continuous as well as transient activation of Notch1 by the application of OHT profoundly altered differentiation (Fig. 1B, 1D, 1F, 1H): by day 5, large dispersed cells with a flattened morphology appeared surrounding neuronal cluster-like colonies. These cells then further differentiated into small rosettes consisting of bipolar cells with small processes resembling neural stem cells. In the presence of OHT, only few bipolar or multipolar neuronal precursor cells expressing Tubb3 were obtained. In contrast, a considerably higher number of GFAP-positive glial cells were formed in the presence of OHT.
To further characterize the alterations in neuroectodermal differentiation induced by activated Notch1, we compared RNA expression levels of Sox9, Nestin, GFAP, and S100b (Fig. 2) as well as Oct4, Sox2, Tubb3, Pax6, Id3, Foxd3, Snai1, and Snai2 (inductions by activated Notch1 shown in Fig. 3 and normalized expression levels shown in Supporting Information Fig. S1) throughout neural differentiation in the presence of OHT, either continuously or only for the first 2 days of differentiation, or in absence of OHT. Continuous as well as transient Notch1 signaling induced expression of Sox9, a key transcription factor of neural stem cells, glial precursors, and NC cells . Activation of Notch1 during the first 2 days was sufficient to induce an upregulation of Sox9 that lasted also after Notch1 was downregulated (Fig. 2A, 2B). A similar upregulation of expression was seen also for the stem cell marker Nestin (Fig. 2C, 2D). The glial cell marker GFAP and the astrocytic marker S100b were upregulated by Notch1 signaling at late stages of differentiation consistent with an induction of glial progenitor cells and mature astrocytes by activated Notch1 (Fig. 2E, 2G, 2F, 2H). Interestingly, transient Notch1 activation favored the induction of GFAP expression, whereas continuous activation of Notch1 results in an enhanced S100b expression. Continuous as well as transient activation of Notch1 signaling during neuroectodermal induction resulted in decreased levels of the ESC marker Oct4 (Fig. 3A and Supporting Information Fig. S1A), a transient decrease of embryonic and neural stem cell marker Sox2 during the first 2 days of neuroectodermal differentiation with a later increase in Sox2, indicating neural differentiation was increased by OHT from day 10 onward (Fig. 3B and Supporting Information Fig. S1B). OHT treatments induced a transient increase of the neural and neural stem cell marker Pax6 during the first 2 days of differentiation with a subsequent decrease thereafter and for transient Notch1 activation a delayed increase from day 12 onward (Fig. 3C and Supporting Information Fig. S1C). The neuronal marker Tubb3 was strongly decreased from day 8 onward after continuous as well as transient activation of Notch1 signaling (Fig. 3D and Supporting Information Fig. S1D). Continuous as well as temporal activation of Notch1 during neuroectodermal induction upregulated expression of Id3 (Fig. 3E and Supporting Information Fig. S1E), which is involved in self-renewal and proliferation of neural stem and NC progenitor cells . Id3 expression remained high in presence of continuous Notch1 activation but decreased from day 12 onward after transient activation of Notch1 signaling, indicating that the cells resume differentiation upon termination of Notch1 activation (Fig. 3E and Supporting Information Fig. S1E).
We next assessed the expression of genes involved in NC specification, that is, Foxd3, Snai1, and Snai2. Activation of Notch1 increased expression of Foxd3, Snai1, and Snai2 from day 4 onward, with a maximum around days 6–8 of differentiation when Notch1 signaling was transiently activated (Fig. 3F–3H and Supporting Information Fig. S1F–S1H), suggesting a strong induction of NC by Notch1 signaling. We further analyzed the expression of markers representative of other lineages, that is, the epiblast and primitive ectoderm marker Fgf5, the mesoderm marker Brachyury(T), and the endoderm marker Sox17. In line with differentiation into primitive ectoderm, Fgf5 expression was upregulated in differentiating cultures between day 2 and 4 without activation of Notch1 signaling, while continuous as well as temporal activation of Notch1 signaling strongly downregulated the expression of Fgf5 (Supporting Information Fig. S2A). Downregulation of Fgf5 correlated with an early upregulation of the neural stem cell markers Pax6, Sox9, and Nestin, indicative of an accelerated differentiation into the neuroectoderm. Irrespective of Notch1 activation, the neuroectodermally differentiating cultures contained hardly any endodermal or mesodermal cells, as indicated by the absence of Sox17 expression and the low levels of Brachyury(T) in presence or absence of OHT (Supporting Information Fig. S2B). Since Notch1 signaling may influence cell death, we next determined the number of apoptotic cells in our differentiating cultures in the presence or absence of activated Notch1. Irrespective of transient or continuous activation of Notch1 signaling, the number of apoptotic cells was generally less than 5% throughout the cultures up to 15 days (Supporting Information Fig. S3). Thus, cell death is not playing a role for the outcome of Notch induced cell fate decisions in our neural differentiating cultures.
To further quantify the observed changes in neuroectodermal differentiation, the developing cell populations were characterized by flow cytometry at days 12–14 of neuroectodermal differentiation in presence or absence of Notch1 signaling using expression of Tubb3 for neuronal precursor cells, of GFAP for glial precursor cells and astrocytes, of Nestin for neural precursor cells, and of Sox2 plus Nestin for neural stem cells. As shown in Figure 4A, 4B, continuous activation of Notch1 increased the number of stem cells, neural precursor cells, and glial cells, whereas it decreased the number of neuronal precursor cells. Similarly, transient activation of Notch1 during the first 2 days of neuroectodermal differentiation increased the number of stem cells, neural precursors, and glial cells while decreasing the number of neuronal precursors (Fig. 4C, 4D). The number of stem cells however was increased to a much higher extent when Notch1 signaling was constantly active, suggesting that Notch signaling keeps the cells temporarily in an undifferentiated state but alters lineage decisions permanently (Fig. 4).
To corroborate the data in an alternative ESC line, where Notch activation can be temporally controlled, we used the Notch1ΔETetOn ESC line, in which Notch1ΔE is induced by doxycycline via the TetOn system [11, 22]. Remarkably similar effects were found after induction of Notch signaling by Doxycyclin: Sox9 expression is induced early during differentiation, correlating then with expression of the stem cell marker Nestin, and is further induced later during differentiation, correlating then with the expression of glial marker GFAP and inversely with expression of neuronal marker Tubb3 (Supporting Information Fig. S4).
The Notch1-Mediated Lineage Decision Between Neuronal and Glial Differentiation Is Mediated by Sox9
To explore whether Sox9 expression was directly regulated by Notch1, we transiently activated Notch1 by addition of OHT for 4 hours in the presence of the protein synthesis inhibitor CHX. Induction of Notch1 in neuroectodermally differentiating cells resulted in an increase of Sox9 expression despite the presence of CHX (Supporting Information Fig. S5), indicating a transcriptional regulation. Next, we sought to dissect the role of Notch1-induced upregulation of Sox9 expression for neuroectodermal commitment of ESCs. To this end, we used a siRNA-mediated Sox9 knockdown approach to specifically decrease Notch1-induced Sox9 levels back to control levels. The Sox9 siRNA knockdown efficiently decreased the Sox9 RNA level and resulted in a Sox9 protein level comparable to the control (Fig. 5A and Supporting Information Fig. S6A–S6C). Reducing Sox9 expression levels down to control levels during Notch1 activation during the first 5 days of neuroectodermal differentiation had a strong influence on the Notch1-induced lineage decision between glial cells, in particular astrocytes, and neuronal cells as analyzed by GFAP, S100b, and Tubb3 expression. Notch1 activation, at normal Sox9 levels, increased the number of GFAP-positive cells, S100b-positive cells, and GFAP/S100b double-positive cells and decreased the number of Tubb3-positive neuronal cells (Fig. 5B–5F). While GFAP-positive cells were increased with transient and continuous as well as high (50 nM OHT) and low (5 nM OHT) induction of Notch1 signaling, the suppression of Tubb3-positive cells required a stronger Notch1 activation by either a higher level temporarily (50 nM OHT) or by applying OHT continuously. Knockdown of Sox9 expression reversed the Notch1-induced increase in GFAP-positive glial cells and GFAP/S100b-positive astrocytes back to the levels seen without Notch1 induction (Fig. 5B–5F). While the suppression of Tubb3-positive cells by transient Notch1 signaling could be reverted with Sox9 knockdown (Fig. 5D), continuous Notch1 signaling inhibited Tubb3 induction in the presence and absence of Notch1-induced Sox9 induction. This suggests that Notch effects independent of early Sox9 induction also contribute to neuronal suppression, possibly by Notch-Hes inhibition of neuronal differentiation.
To further corroborate the significance of Sox9 for the Notch-mediated lineage decision between glial and neuronal cells, we knocked down Sox9 during neuroectodermal differentiation of Notch1ΔETetOn cells. Activation of Notch1 signaling by a low-dose of Dox upregulated Sox9 expression early (day 1) and late (between day 5 and 8) during neuroectodermal differentiation (Supporting Information Figs. S4, S7). These two upregulations of Sox9 expression corresponded well with the inductions seen with the OHT-inducible system in EB5-NERT cells (Fig. 2). While in both systems Notch1 induced Sox9 expression during early and late neuroectodermal differentiation, differences were seen in the kinetics of differentiation. EB5-NERT cells differentiated slower that the Ainv15-Notch1ΔETetOn cells (compare Supporting Information Fig. S4 with Fig. 2). Despite the different kinetics, knockdown of Notch1-induced Sox9 expression (Supporting Information Fig. S7) efficiently reverted Notch1-induced glial differentiation also in the Tet-inducible Ainv15 ESC (Fig. 5G). In line with our results with NERT cells (Fig. 5E), a low level of Notch1 activation by Dox in the Notch1ΔETetOn cells was not sufficient to decrease the number of Tubb3-positive cells by day 14, although Tubb3 expression was significantly downregulated by activated Notch1. These results further underline the importance of dose and timing of Notch signaling for lineage decision events.
Surprisingly, only when Notch1 signaling was activated by Dox in the Notch1ΔETetOn cells, the induction of neural stem cells by activated Notch1 could be counteracted by reducing Sox9 levels with siRNA (Fig. 5B–5E, 5G). In the Notch1ΔETetOn cells, the treatment with Sox9 siRNA decreased the early and late induction of Sox9 by activated Notch1 (Supporting Information Fig. S7). The duration of Sox9 induction in response to 2 days of Notch1 activation differed between the EB5-NERT and the Notch1ΔETetOn cells: in EB5-NERT cells, Sox9 was increased for 12 days, whereas it was increased for 6 days in the Notch1ΔETetOn cells (compare Fig. 2 with Supporting Information Fig. S4). When Notch1 signaling was switched on continuously or transiently during neural differentiation by OHT in the NERT system, the number of neural stem cells remained at a high level, despite downregulation of the early Notch1-induced Sox9 expression. It is likely that the remaining Sox9 induction at later time points (Fig. 2) prevented stem cell differentiation in EB5 NERT cells treated with OHT and Sox9 siRNA. Taken together, our results further emphasize that the intensity and duration of the Notch-induced Sox9 upregulation are critical for Notch-mediated stem cell arrest.
To analyze whether preventing Notch1-mediated upregulation of Sox9 during early neuroectodermal differentiation of ESC would also affect NC development, we knocked down the Notch1-induced Sox9 expression and analyzed the expression of Snai1, Snai2, and Id3. The upregulation observed by addition of OHT in the EB5-NERT cells from day 0 to 2 was only partially affected by knockdown of Sox9 expression: there was a partial downregulation of Id3, but not of Snai1 and Snai2 (Supporting Information Fig. S8), suggesting that the early Sox9 induction contributes to only some extent to the Notch-induced specification and/or self-renewal of NC stem cells.
Downregulation of Pax6 Expression by siRNA Is Not Sufficient to Revert Notch1-Induced Changes During Neuroectodermal Differentiation
In addition to Sox9, we have recently identified Pax6 as a Notch1 target gene in ESC . Pax6 plays a crucial role in neuroectodermal development  and we therefore assessed the importance of Pax6 upregulation during neuroectodermal differentiation of ESC. First, we confirmed that activated Notch1 upregulates Pax6 in ESC and during early neuroectodermal differentiation (Fig. 3C and Supporting Information Fig. S1C). Upregulation was restricted to the first 2 days of neuroectodermal differentiation, and during days 4–10 of neural differentiation, activation of Notch1 downregulated Pax6 expression (Fig. 3C and Supporting Information Fig. S1C). Knockdown of Pax6 expression by siRNA, as expected, decreased Notch1-induced upregulation of Pax6 protein and RNA levels (Fig. 6A and Supporting Information Fig. S9). Manipulating Pax6 expression levels back to control levels during Notch1 activation during the first days of neuroectodermal differentiation had no significant influence on the Notch1-induced lineage decision between glial and neuronal cells or on the Notch1-induced increase in neural stem and progenitor cells as analyzed by GFAP, Tubb3, Nestin, and Sox2 expression (Fig. 6), suggesting that Pax6 upregulation is not sufficient to mediate these effects of Notch.
Notch signaling controls cell fate decisions in many different cellular contexts and often acts at different steps during cell lineage progression. This is indeed the case in neural differentiation, where Notch signaling supports the generation and maintenance of neural stem cells and is required for gliogenesis. In this report, we have addressed how Notch signaling regulates neuroectodermal cell fate decisions, and we provide evidence for a critical role of Sox9 in promotion of neural stem cell maintenance and astroglial differentiation. Using inducible systems to activate Notch1 during in vitro differentiation of ESC along the neuroectodermal lineage, we identified Sox9 as a direct Notch target gene and observed several changes in neuroectodermal development (summarized in Fig. 7). Briefly, our data show that continuous or transient activation of Notch1 during neural differentiation of ESC resulted in an increase in neural stem cells, a block in generation of neurons, and an increase in immature glial cells and astrocytes. This extends previous studies, where ligand-specific Notch activation during embryonic neurogenesis promotes and accelerates differentiation of ESC, supports development of neuroectodermal cells, and inhibits differentiation along mesodermal lineages [19, 26, 27]. Subsequently, Notch signaling is essential for maintenance of neural stem and progenitor cells and blocks differentiation into neurons [6, 28], whereas it induces gliogenic differentiation [5, 7, 29–31]. Interestingly, transient activation of Notch1 during early neuroectodermal ESC differentiation was sufficient to irreversibly switch neurogenesis to gliogenesis, suggesting that Notch functions at this stage as an instructive rather than a permissive signal. In line with this, temporal activation of Notch by its ligand Jagged in neural stem cells for 5 hours is sufficient to irreversibly inhibit differentiation into neurons . Furthermore, an instructive glial determination and a block of neuronal differentiation were observed in NC stem cells after activation of Notch by ligand stimulation for 24 hours .
Interestingly, Sox9 played an important role in mediating the Notch response in promotion of astroglial differentiation and neural stem cell maintenance, while it was considerably less important for maintaining the block of neurogenesis and for inducing NC differentiation. The Notch-mediated block of neurogenesis may instead be controlled by the Hes/Hey family of transcription factors that inhibit proneural gene expression [13, 33]. The finding that Sox9 was a critical mediator of Notch1-induced glial promotion corroborates a previous report, where selective ablation of Notch in the developing nervous system reduced glial differentiation, and was accompanied by reduced Sox9 expression, although the importance of Sox9 was never directly tested . In the specialized setting of eye development, there is evidence that group E proteins Sox8 and 9 are involved in mediating Notch effects for generation of Müller glia cells, which are part of the vertebrate retina . However, it was not shown to be a general mechanism for glia cell development and group E Sox proteins were not confirmed as direct target genes of the Notch signaling. Sox9 induced by activated Notch1 is an essential mediator also for chondrogenic lineage determination , which lends further support to the model that Notch signaling regulates lineage decisions via direct induction of lineage-determining transcription factors . In accordance with this notion, in Drosophila, Notch signaling is required for expression of the transcription factor gcm, which directs glial specification .
Several studies suggest that Notch signaling plays a role in establishing the NC domain within the ectoderm and subsequently in diversifying the fates of cells that arise from NC [9, 37]. In this study, we observed the appearance of a population of large cells with a high migratory potential after activation of Notch1 signaling during neuroectodermal differentiation of ESC and an increase in Snai1, Snai2, and Foxd3 expression from day 4 onward, indicating induction of differentiation into immature NC cells. Promotion of NC lineage progression is in contrast to a previous report, which, using the same ESC differentiation protocol, did not observe NC or glial cells when Notch1 was constitutively expressed from the ROSA26-locus . The reason for this discrepancy is not clear, but may at least in part reflect different cell densities in the two studies, as cell density is known to influence the effects of ESC differentiation . Alternatively, the constitutive activation of Notch prior to neural differentiation  may have led to selection of cells that had lost the capacity to differentiate into NC and glial cells, whereas they retained neuronal differentiation capacity. In conclusion, the data presented here shed new light onto how Notch signaling can execute cell fate decisions at various steps during a cell lineage progression, and the observation that Sox9 is critical for a subset of fate choices during neural differentiation provides new insights into how Notch promotes glial differentiation.
In conclusion, this is the first study to demonstrate that Notch1 signaling induces glial cell differentiation of ESCs and neural stem cell self-renewal via the direct upregulation of the transcription factor Sox9. Furthermore, we show that Sox9 is only partially involved in the Notch1-induced decrease in neurons and does not play a major role for Notch1 effects on NC development. Our results identify Sox9 as a critical and specific mediator of Notch controlled neural lineage commitment.
We are very grateful to S. Horn, M. Boss, and K. Macha for their skillful technical assistance, Dr. Michael Wegner for providing the Sox9 antibody, and Dr. Claudia Lange for helpful suggestions. This work was supported by funding from the DFG SFB 415 project B8 to U.J., and by funding from the Swedish Research Council (project grant and DBRM), Knut och Alice Wallenbergs Stiftelse (WIRM), Cancerfonden, and EU (NotchIT) to U.L.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.