SEARCH

SEARCH BY CITATION

Keywords:

  • Sox1;
  • Neuroectoderm;
  • Pax6;
  • Radial glia;
  • Embryonic stem cells

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The transcription factors Sox1 and Pax6 are expressed sequentially during early mouse embryonic neurogenesis. Sox1 expression starts upon formation of neuroectoderm, whereas Pax6 is subsequently expressed in radial glial cells, the latter giving rise to most neurons of the cerebral cortex. Here we used mouse embryonic stem (ES) cells to study the role of Sox1 and Pax6 in regulating differentiation of neural progenitors. For this purpose, we investigated the effect of overexpression and knockdown of Sox1 and Pax6, using three differentiation protocols. We show that (a) expression of Sox1 or Pax6 in uncommitted ES cells favored neuroectodermal lineage choice; (b) continuous Sox1 expression maintained cells at the neuroepithelial stage and prevented expression of Pax6 and other radial glial cell markers; (c) Sox1 knockdown facilitated exit from the progenitor stage, whereas Pax6 knockdown decreased formation of radial glia; (d) forced Pax6 expression in neuroepithelial cells triggered their differentiation into radial glia and neurons; and (e) Pax6 expression induced cell migration, a feature typical of radial glia-derived early neurons. We conclude that Sox1 enhances neuroectodermal commitment and maintenance but blocks further differentiation. In contrast, Pax6 is involved in the progression of neuroectoderm toward radial glia. STEM CELLS2009;27:49–58


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Early neurogenesis in mammals is characterized by a series of events beginning with the induction of neuroectoderm constituting the neural plate, which will later fold to give rise to the neural tube. In mouse, neuroectoderm is characterized by the expression of the SoxB1 transcription factor family members, such as Sox1, Sox2, and Sox3 [1], which play an important role in its specification and maintenance [1–3]. Neuroectodermal cells may directly give rise to neurons, but mostly they differentiate into a more advanced type of neuronal precursors, namely radial glial cells. During embryogenesis, radial glia progressively replaces neuroectoderm and thereby constitutes the major progenitor cell population during development and postnatal life [4]. These cells differ from earlier neuroectodermal cells by the expression of several markers, such as vimentin, brain-lipid binding protein (BLBP), glutamine synthase, and L-glutamate/L-aspartate transporter [5]. Radial glial cells can differentiate into all three neural lineages, but this potential differs according to developmental stage [5]. There is now also increasing evidence for subpopulations of radial glial cells giving rise to different types of neurons [6, 7].

Sox1 is a member of the SoxB1 transcription factor family and is involved in very early steps of neurogenesis. In contrast to Sox2 and Sox3, its expression is restricted to the neural plate and neural tube [8]. It has been shown that Sox1 overexpression favors neuroectodermal lineage choice from uncommitted mouse embryonic stem (ES) cells [3]. However, it remains unclear whether it maintains cells at the neuroectodermal stage or induces their further differentiation [1, 2]. Interestingly, Sox1 knockout mice have only a mild phenotype, consisting mainly of eye abnormalities and seizures [9], probably because other Sox members play a compensatory role.

Pax6 is a key transcription factor in the development of the central nervous system. In the mouse, its expression appears first in radial glial cells [5]. Pax6sey/sey mice [10] express a nonfunctional Pax6, leading to severe developmental defects, in particular a massive deficit in cortical development [6], eye development [11], axonal pathfinding [12], and neuronal positioning [13–17]. There is now good evidence for the existence of Pax6-positive and Pax6-negative radial glia. Indeed, radial glia in certain brain areas, such as the ganglionic eminence, does not express Pax6 [6], and Pax6-deficient ES cells give rise to a subtype of radial glia that allows development of GABAergic but not glutamatergic neurons [7]. Whether Pax6 is driving development of radial glia or is just necessary for the generation of cortical neurons from radial glia remains presently unknown.

The sole expression of Pax6 can confer neuronal properties to certain cell types [18, 19]; however, it can also assume different roles during eye [20] or pancreas [21] development. Therefore, the role of Pax6 during embryogenesis seems to be largely context-dependent.

In this study, we used mouse embryonic stem cells to investigate the role of Sox1 and Pax6 during early developmental steps of neurogenesis. We analyzed the effects of overexpression and knockdown of Sox1 or Pax6. We show that their overexpression results in increased neural commitment during nondirected differentiation of embryonic stem cells. However, effects on subsequent steps of differentiation distinguished the two transcription factors: Sox1 overexpression maintained cells at the neuroectodermal stage, whereas Pax6 overexpression enhanced their transition toward radial glia, enhanced generation of neurons, and increased cell migration. We also show that knockdown of Sox1 expression in neuronal progenitors resulted in the exit from the neuroepithelial state, whereas knockdown of Pax6 expression repressed commitment toward the radial glia lineage.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Reagents

Reagents and their sources were as follows: pDONR221 vector (Invitrogen, Carlsbad, CA, http://www.invitrogen.com); the murine CGR8 embryonic stem cell line (European Collection of Cell Culture); the stromal bone marrow MS5 cell line (provided by Katsuhiko Itoh [22]); cell culture media, fetal bovine serum, serum replacement, penicillin, streptomycin, N2 supplement, nonessential amino acids, sodium pyruvate (Gibco, Paisley, U.K., http://www.invitrogen.com); basic human fibroblast growth factor (Invitrogen); blasticidin (Invitrogen); tetracycline (Invitrogen); Gateway Clonase enzymes (Invitrogen); small interfering RNA (siRNA) against Sox1 (Mm_Sox1_1_HP siRNA; Mm_Sox1_3_HP siRNA; Qiagen, Hilden, Germany, http://www1.qiagen.com); siRNA against green fluorescent protein (GFP; GFP-22 siRNA; Qiagen); and siRNA against Pax6 (ON-TARGETplus SMARTpool, L-062890-00; Dharmacon, Inc., Lafayette, CO, http://www.dharmacon.com).

Vector Constructions

We previously described the construction of the 2K7 lentivectors [23] and the EF1Tet system [24]. The coding sequences of murine Pax6 and Sox1 were cloned into pDONR221. The lentiviral constructs were generated as follows: to generate the 2K7bsdEF1-α/TetR, pENTR EF1-α and pENTR TetR were recombined into 2K7bsd; to generate the 2K7GFPEF1-αTetO2/Pax6, pENTR EF1-αTetO2 and pENTR Pax6 were recombined into 2K7GFP; to generate the 2K7bsdEF1-αS Sox1, pENTR EF1-αS and pENTR Sox1 were recombined into 2K7bsd; and to generate the 2K7bsdEF1-α/monomeric red fluorescent protein 1 (mRFP1), pENTR EF1-α and pENTR mRFP1 were recombined into 2K7bsd.

Cell Culture

The CGR8 ES cell lines were maintained in BHK-21 medium supplemented with 10% fetal calf serum, L-glutamine, nonessential amino acids, sodium pyruvate, penicillin and streptomycin, and leukemia inhibitory factor. CGR8 ES cells were cultured on gelatin-coated dishes.

ES Cell Differentiation

Neuronal differentiation on MS5 was carried out as described [25]. Briefly, irradiated MS5 cells (1.25 × 105 cells per well) were seeded in six-well plates. The next day, ES cells were plated on the MS5 layer in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with nonessential amino acids, 2-mercaptoethanol, and 15% knockout serum. Five days later, cells were trypsinized and seeded on polyornithine-coated six-well plates in complete DMEM supplemented with N2 supplement and human basic fibroblast growth factor (10 ng/ml). Feeder-free/serum-free neuronal differentiation was carried out as described [26]. Embryoid bodies were generated by the hanging drop method as described [27].

Lentiviral Transductions and Generation of ES Cell Lines

Lentivector production and mouse ES cell transduction were performed as previously described [23]. To generate a CGR8 ES cell line inducible for Pax6 expression, we first expressed TetR under the control of the constitutive EF1-α promoter, using the 2K7bsdEF1-αTetR lentivector. We next transduced cells to express Pax6 under the control of the inducible EF1-αTetO2 promoter using the 2K7GFPEF1-αTetO2/Pax6 lentivector, to generate a Pax6-inducible cell line. We then sorted highly GFP-positive cells by flow cytometry to allow (a) optimal Pax6 expression upon induction and (b) monitoring of the presence of the Pax6 expression cassette in ES cells. We also generated a Sox1-overexpressing CGR8 ES cell line using a 2K7bsd in which the EF1-αS promoter drives the expression of Sox1. We also generated a CGR8 ES cell line overexpressing a red fluorescent protein. For this purpose, we constructed a 2K7bsd lentivector expressing mRFP1 under the control of the EF1-α promoter, and we transduced and selected ES cells with blasticidin as described [23].

siRNA Transfection

CGR8 ES cells were cultured for 5 days on MS5 cells, subsequently replated at 8 × 105 cells per cm2 in DMEM high-glucose containing N2 supplement and bFGF (10 ng/μl), and transfected with 50 nM siRNA against GFP, Sox1, and Pax6. Cells were analyzed by immunostaining 2 or 3 days after replating.

Immunofluorescence Microscopy

Immunofluorescence was carried out according to standard techniques. In brief, ES cells were grown on glass coverslips coated with either a MS5 feeder layer or polyornithine in six-well plates. Cells were fixed with 2% paraformaldehyde for 30 minutes, washed with Hanks' balanced saline solution (HBSS), and permeabilized with 0.5% (vol/vol) Triton X-100 for 30 minutes. Cells were then exposed to primary antibodies overnight at 4°C. After two washes in HBSS containing 1% serum (blocking buffer), cells were stained with secondary antibodies at room temperature for 1 hour (1:1,000 dilution in blocking buffer). Cell nuclei were stained with 1 μg/ml 4′,6-diamidino-2-phenylindole for 10 minutes. Visualization analysis took place on a Zeiss Axioplan microscope (Carl Zeiss, Jena, Germany, http://www.zeiss.com) equipped for epifluorescence. The dilutions for the primary antibodies in blocking buffer were as follows: mouse monoclonal anti-nestin antibody (1:2,500) (Chemicon, Temecula, CA, http://www.chemicon.com), rabbit polyclonal anti-class III β-tubulin antibody (1:2,500) (Covance, Berkeley, CA, http://www.covance.com), mouse IgM anti-vimentin (1:200) (Developmental Studies Hybridoma Bank, Iowa City, IA, http://www.uiowa.edu/∼dshbwww), rabbit anti-BLBP (1:1,000) (Chemicon), mouse IgM anti-RC2 (1:200) (Developmental Studies Hybridoma Bank), goat anti-Pax6 (1:200) (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com), and goat anti-Sox1 (Santa Cruz Biotechnology). For secondary detection, Alexa Fluor 488 or 555 conjugates were used (1:1,000) (Molecular Probes, Eugene, OR, http://probes.invitrogen.com). For negative controls, immunostaining was performed without first antibody.

Immunostaining Quantifications

Immunostaining quantifications were performed on an ImageXpressMicro acquisition system (Molecular Devices Corp., Union City, CA, http://www.moleculardevices.com). Automated image acquisition was performed using the 10× Plan Fluor objective, and 18 fields were analyzed for each condition. The “neurite outgrowth” analysis module was used to quantify cellular extensions positive for β3-tubulin, nestin, RC2, and vimentin. Quantifications are expressed as values for total outgrowth divided by total cell number. The “cell scoring” analysis module was used to quantify immunostainings for BLBP, and quantifications are expressed as percentage of positive cells. All values were normalized on values obtained from the ES control cell line.

Western Blotting

Cells were lysed on ice in 1% Triton X-100 in 50 mM NaCl, 10 mM MgCl2, 1 mM EGTA, and 50 mM Tris-HCl, pH 7.4, containing protease inhibitor mixture (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) and sonicated. Samples from equal cell numbers were separated on 10% SDS-polyacrylamide gel electrophoresis gels and transferred to polyvinylidene difluoride membranes for Western blotting according to standard techniques. Antibodies used and dilutions were as follows: goat anti-Pax6 (1:100) (Santa Cruz Biotechnology), goat anti-Sox1 (1:1,000) (Santa Cruz Biotechnology), and rabbit anti-histone H3 (1:5,000) (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com).

Cell Sorting

GFP-positive ES cells were sorted by flow cytometry on a FacSorter (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com).

Live Imaging Studies

Live imaging studies were performed on an ImageXpressMicro acquisition system (Molecular Devices). Cells were kept in serum replacement medium supplemented with 1 μg/ml tetracycline at 37°C with 5% CO2 supply. Fluorescence images of GFP and mRFP1-expressing cells were acquired every 5 minutes for 48 hours. Analysis of cell motion was performed using MetaXpress analysis software (Molecular Devices). Mean cell motion was calculated using the following formula: ∑ (cell surface area of picture n−cell surface area of picture n−1)/∑ cell surface area of picture n.

Real-Time Polymerase Chain Reaction

Reactions were run on an ABI Prism 7900 HT detection system (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com). The data were normalized to TBP, Rps9, and GAPDH housekeeping genes using the geNorm software (http://medgen.ugent.be/∼jvdesomp/genorm/). Sequences of the primers used were as follows: nestin forward, 5′-AGATCGCTCAGATCCTGGAA-3′; nestin reverse, 5′-GGTGCTATCATCTGCATCGTT-3′; β3-tubulin forward, 5′-TATTCAGGCCCGACAACTTT-3′; β3-tubulin reverse, 5′-GGGTGTCAACCAGAGGAAGT-3′; BLBP forward, 5′-TCCAGCTGGGAGAAGAGTTT-3′; BLBP reverse, 5′-CCAACCGAACCACAGACTTA-3′; endogenous Pax6 forward, 5′-GTTGGTGTGTTCCCTGTCCT-3′; endogenous Pax6 reverse, 5′-ACCGCCCTTGGTTAAAGTCT-3′; transduced Pax6 forward, 5′-GCCGCCAGAACACAGGTA-3′; transduced Pax6 reverse, 5′-CGCTGTGACTGTTCTGCATC-3′; TetR forward, 5′-AAAAATAAGCGGGCTTTGCT-3′; TetR reverse, 5′-CCCCTTCTAAAGGGCAAAAG-3′; endogenous Sox1 forward and reverse, Mm_Sox1_1_SG QuantiTect Primer Assay (Qiagen).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Engineering Pax6-Inducible and Sox1-Expressing ES Cell Lines

We initially intended to generate a mouse ES cell line constitutively expressing Pax6. However, this cell line was difficult to expand and progressively lost Pax6 expression (data not shown). A possible explanation is cell cycle arrest or decreased cell proliferation caused by Pax6, which has been shown to occur in different cell types upon Pax6 overexpression [19, 28]. For this reason, we generated a Pax6-inducible ES cell line that could be stably maintained and expanded in culture. We used the previously described EF1-Tet system [29], with slight modifications (described in Materials and Methods). We first checked the expression of transduced Pax6 in ES cells expressing TetR alone or ES cells inducible for Pax6 expression in the induced and noninduced state. A 14-fold induction was observed after 24 hours of incubation with 1 μg/ml tetracycline (supporting information Fig. 1A). To investigate Pax6 expression at the protein level, Western blotting was performed. Undifferentiated ES cells were cultured with different concentrations of tetracycline for 24 hours, and total proteins were extracted. We observed a dose-responsive induction of Pax6 expression (supporting information Fig. 1B). We also checked the subcellular localization of Pax6 by immunostaining to confirm its nuclear localization in tetracycline-treated cells (supporting information Fig. 1C, 1D). We next tested inducibility of Pax6 expression during neuronal differentiation on MS5 cells. ES Pax6 cells were cultured with or without tetracycline from day 3 on and maintained until day 6. Western blot analysis of Pax6 expression was performed at days 4, 5, and 6 (supporting information Fig. 1E). Interestingly, neuronal differentiation led to significant Pax6 expression even in the absence of tetracycline. This might be due to reduction of the TetR expression level during differentiation (supporting information Fig. 1F). For further experiments, we decided to compare the cell line expressing TetR alone with the Pax6-inducible cell line, both treated with 1 μg/ml tetracycline from day 3 or 4 on. Throughout the text, cells expressing TetR alone are referred to as ES control and cells inducible for Pax6 expression are referred to as ES Pax6.

We also generated a cell line with constitutive expression of Sox1 using the 2K7bsdEF1-αS Sox1 (described in Materials and Methods). This cell line grew normally and Sox1 expression was well maintained over time. We checked Sox1 overexpression by Western blotting (supporting information Fig. 1G) and immunofluorescence (supporting information Fig. 1H, 1I), confirming constitutive Sox1 expression restricted to the cell nucleus. Throughout the text, this cell line is referred to as ES Sox1.

We next aimed at the characterization of the cell lines we generated in two differentiation models: (a) embryoid bodies and (b) directed neuronal differentiation (stromal cell-dependent [25] and stromal cell-independent [26]). We observed that transgene expression was lost over time (supporting information Fig. 3). This loss of transgene expression was strong in 13-day-old embryoid bodies, where many cells lost transgene expression (supporting information Fig. 3A, 3H), whereas the loss was less pronounced in cells differentiated for 6 days on MS5 stromal cells (supporting information Fig. 3C, 3J). However, when cells were dissociated after culture on MS5 stromal cells, transgene silencing became more significant (supporting information Fig. 3E, 3L). Thus, there is transgene silencing; the determining factors are time and possibly also culture conditions. The impact of transgene silencing is discussed below. The three differentiation systems are illustrated in supporting information Figure 2.

Overexpression of Either Sox1 or Pax6 Induces Massive Neuronal Differentiation in Embryoid Bodies

To investigate the role of Sox1 and Pax6 in neuroectodermal lineage choice, we generated embryoid bodies (EBs) as previously described [27] with the different cell lines mentioned above. In the ES Pax6 cell line, Pax6 expression was induced with tetracycline throughout differentiation. EBs generated from control cells generated few neurons (Fig. 1A–1D), whereas EBs from the ES Sox1 (Fig. 1E-H) and ES Pax6 (Fig. 1I–1L) cell lines were characterized by greatly enhanced neurogenesis. Sox1 EBs tended to be round and multilayered, whereas Pax6 EBs tended to be smaller and more flat (data not shown). However, neuron-rich areas in Sox1 EBs and in Pax6 EBs were similar. At day 13 of differentiation, EBs generated from ES Sox1 contained many Sox1-positive/β3-tubulin-negative cells (Fig. 1M), often arranged in rosette structures positive for nestin and surrounded by β3-tubulin-positive neurons (Fig. 1M, 1N). Pax6-positive cells were present in rosette areas and at their periphery and were associated with RC2-rich areas (Fig. 1O).

thumbnail image

Figure 1. Neuronal differentiation in embryoid bodies derived from different ES cell lines. (AL): Development of embryoid bodies derived from wt, Pax6-overexpressing, and Sox1-overexpressing cells. (AD): Embryoid bodies generated from ES control cells generated few neurons. (EL): Increased neurogenesis occurred in embryoid bodies overexpressing either Sox1 (EH) or Pax6 (IL). Red: β3-tubulin immunostaining; blue: DAPI staining. (MO): Analysis of 13-day-old embryoid bodies generated from ES Sox1 cells. Most cells were Sox1-positive and were often arranged in rosettes surrounded by Sox1-negative/β3-tubulin-positive cells (M). These rosettes were positive for nestin and were surrounded by mature neurons (N); Pax6 and RC2-rich areas were present at the periphery of rosettes (O). (PX): Analysis of embryoid bodies generated from ES Pax6 cells. (P): In 9-day-old embryoid bodies, most cells were Pax6-positive. (Q): in 13-day-old embryoid bodies, a minority of cells remained Pax6-positive. (R): Same field as (Q), showing correlation of Pax6 expression with GFP expression. (S): Rare Pax6-positive and GFP-negative cells present in rosette structures, suggesting endogenous Pax6 expression. There was inverse correlation between GFP-positive cells and staining for Sox1 (T), nestin (U), RC2 (V), BLBP (W), and β3-tubulin (X). Scale bars = 20 μm (M) and 100 μm (all other panels). Abbreviations: BLBP, brain-lipid binding protein; DAPI, 4′,6-diamidino-2-phenylindole; ES, embryonic stem; GFP, green fluorescent protein; wt, wild-type.

Download figure to PowerPoint

As mentioned before, prolonged differentiation in the embryoid body model resulted in transgene silencing (supporting information Fig. 2A, 2H). However, silencing of the Sox1 and the Pax6 transgene was observed at different steps of differentiation. Sox1 was highly expressed in uncommitted and neuroepithelial cells but virtually undetectable in radial glial cells (i.e., cells stained for BLBP and Pax6) (supporting information Table 1). The most interesting pattern was observed for Pax6. As the Pax6 expression vector contained a GFP cassette, silencing of the transgene could be monitored as loss of GFP fluorescence (compare Fig. 1Q with 1R), and the emergence of Pax6-positive, GFP-negative cells indicated endogenous Pax6 expression. The Pax6 transgene was uniformly expressed in early EBs (i.e., uncommitted cells; Fig. 1P). In more developed EBs (day 13), Pax6 expression was generally decreased. Sox1-positive cells were invariably GFP-negative (Fig. 1T), suggesting that silencing of the Pax6 transgene was necessary to reach the neuroectodermal stage (supporting information Table 1). However, in more developed EBs, there was also an emergence of Pax6-positive, GFP-negative areas (Fig. 1S), most likely corresponding to expression of endogenous Pax6. These endogenous Pax6-positive cells were generally associated with rosettes (Fig. 1S), similar to those seen for the radial glia marker BLBP.

These results suggest that (a) Sox1 expression in noncommitted cells favors neuroectodermal differentiation, but transgene silencing has to occur to allow further progression toward radial glia, and (b) Pax6 overexpression in noncommitted cells favors neural lineage commitment, but transgene silencing has to occur to allow progression through the neuroepithelial stage, and expression of endogenous Pax6 is associated with further progression toward a radial glia stage.

Characterization of Early Neuronal Differentiation of Mouse ES Cells on MS5 Cells

We continued our studies in a different experimental system, namely neuronal differentiation of mouse ES cells on MS5 cells [25]. We first investigated the emergence of markers and morphological features characterizing neuroepithelial and radial glial cells and compared it with normal embryogenesis.

During mouse development, Sox1 starts to be expressed in neuroepithelial cells forming the neural plate around embryonic day (E) 8 [30], whereas Pax6 expression starts later, around E9, in radial glial cells [31]. Nestin and RC2 are expressed both in neuroepithelial cells and radial glial cells, whereas Pax6, BLBP, and vimentin are restricted to radial glial cells [4]. Neuroepithelial cells have relatively short cytoplasmic extensions, whereas radial glial cells have a pronounced bipolar shape with long cytoplasmic extensions [5]. In wild-type ES cells cultured on MS5 cells, endogenous Sox1 expression started at day 3 of differentiation and extended progressively to the majority of cells around day 5. Endogenous Pax6 started to be expressed at day 4 of differentiation and remained restricted to a subset of cells over the following days (Fig. 2A–2D). Interestingly, cells expressing high levels of Pax6 expressed no or only low levels of Sox1 (Fig. 2D). Sox1-positive cells were characterized by nestin and RC2-positive extensions that were relatively short (Fig. 2E, 2F). They were also either BLBP-negative (Fig. 2G), or weakly BLBP-positive. Sox1 expression was not associated with β3-tubulin expression (Fig. 2H). In contrast, Pax6-positive cells were characterized by long nestin and RC2-positive extensions (Fig. 2I, 2J), and expressed generally high levels of BLBP (Fig. 2K). Many β3-tubulin-positive mature neurons retained Pax6 expression (Fig. 2L). Therefore, this culture system provides a mixture of neural cells at diverse stages of differentiation resembling stages found during early embryonic neurogenesis.

thumbnail image

Figure 2. Expression patterns of Sox1, Pax6, and radial glial cell markers in differentiating wild-type embryonic stem (ES) cells. In ES cells cultured on the MS5 stromal cell line, Sox1 expression started around day 3 (A), and its expression progressively extended at days 4 (B) and 5 (C). Pax6 expression started around day 4 (B), and its expression extended over the following days but remained restricted to a minority of cells (C, D). Note that high Pax6-expression was correlated to low Sox1 expression and vice versa. Sox1-positive cells had short nestin-positive (E) and RC2-positive extensions (F) and generally expressed low levels of BLBP or no BLBP (G) and no β3-tubulin (H). Pax6-positive cells had long nestin-positive (I) and RC2-positive extensions (J), expressed generally high levels of BLBP (K), and were often associated with β3-tubulin expression. Scale bars = 50 μm. Abbreviation: BLBP, brain-lipid binding protein.

Download figure to PowerPoint

Sox1 Maintains Cells at the Neuroepithelial Stage, Whereas Pax6 Induces Their Differentiation into Radial Glial Cells

As high expression levels of Sox1 or Pax6 were present in morphologically and immunocytochemically distinct cellular types, we investigated whether their overexpression alters the phenotype of ES cell-derived neuronal precursors. ES Sox1 and ES Pax6 cells were cultured on MS5. In ES Pax6 cells, Pax6 expression was induced at day 4 of differentiation. Cells were analyzed by immunostaining at day 6. In the ES Sox1 cell line, virtually all cells were Sox1-positive (Fig. 3 A) and Pax6-negative (Fig. 3B). Cells had short nestin-positive extensions (Fig. 3C), whereas RC2 (Fig. 3D) and BLBP expression remained low (Fig. 3E) and almost no β3-tubulin cells were found (Fig. 3F). In ES Pax6 cells, virtually all cells were Pax6-positive (Fig. 3H). However, the levels of Pax6 expression varied. Interestingly, only cells with very low Pax6 expression (corroborated by low GFP expression) expressed Sox1 (Fig. 3G), suggesting that Pax6 gene inactivation is required for Sox1 expression, similar to the case described above for EB experiments (supporting information Table 1). There were also many cells highly positive for nestin (Fig. 3I), RC2 (Fig. 3J), and BLBP (Fig. 3K), and many β3-tubulin-positive cells were present around the colonies (Fig. 3L). We then aimed at quantifying the extent of neuronal differentiation by investigating specific neuronal markers at the mRNA level. The MS5 differentiation system did not allow us to perform these experiments, as MS5 cells are a source of contaminating mRNA. We therefore used a feeder-free neuronal differentiation system [26], allowing us to monitor neuronal differentiation at the mRNA level in the whole cell population. After 3 days of differentiation tetracycline was added to the medium, and cDNA was prepared from cells after 5 days of differentiation. In these conditions, the majority of cells differentiated toward the neuronal lineage. However, in contrast to MS5-driven differentiation [25], non-neuronal cells persisted in this culture system, as previously described [26]. Radial glial cell markers, such as BLBP and vimentin, can also be expressed by non-neuronal cell types, such as astrocytes [32] and epithelial cells [33]; therefore, we chose to investigate the levels of β3-tubulin, as well as the levels of endogenous Sox1 and Pax6, using primers specific for their 3′ untranslated regions by real-time polymerase chain reaction to investigate the extent of differentiation of the different cell lines. In these conditions, β3-tubulin (Fig. 3M) and Pax6 (Fig. 3N) remained low in Sox1-overexpressing cells, whereas endogenous Sox1 levels were unchanged (Fig. 3O). This suggests that cells were retained at the neuroepithelial stage. In contrast, Pax6 ES cells expressed high levels of β3-tubulin (Fig. 3M). Endogenous Sox1 was only modestly decreased (Fig. 3O, p < .05), and endogenous Pax6 levels were also decreased (Fig. 3N, p = .06), suggesting negative autoregulation of the Pax6 promoter, as previously described [34].

thumbnail image

Figure 3. Forced Sox1 expression prevents radial glial cell differentiation, whereas Pax6 promotes it. ES Sox1 cells were cultured for 6 days on MS5 and were immunostained against Sox1 and radial glial cell markers. Virtually all cells expressed Sox1 (A) but not Pax6 (B); they also expressed low levels of nestin (C), RC2 (D), and BLBP (E) and were β3-tubulin negative (F). ES Pax6 cells were cultured for 6 days on MS5, with tetracycline from day 4 on. They were immunostained against Sox1, radial glial cell markers, and β3-tubulin. Cells positive for Sox1 were GFP-low (G), whereas virtually all cells were positive for Pax6 (H). Many cells had long nestin (I) and RC2-positive (J) cytoplasmic extensions and were positive for BLBP (K), and many cells were β3-tubulin-positive (L). Scale bars = 50 μm. (MO): Real-time polymerase chain reaction on neuronal markers after 5 days of differentiation in feeder-free neuronal differentiation. Abbreviations: A.V., arbitrary values, BLBP, brain-lipid binding protein; DAPI, 4′,6-diamidino-2-phenylindole; ES, embryonic stem; GFP, green fluorescent protein; wt, wild-type.

Download figure to PowerPoint

Next, radial glial cell markers and β3-tubulin were investigated at the protein level by immunofluorescence after cell dissociation. ES control, ES Pax6, and ES Sox1 cells were cultured for 5 days on MS5, and subsequently replated and cultured for 24 or 96 hours. Quantification of radial glial cell markers was performed using an automated image analysis system (described in Materials and Methods). Supporting information Figure 4 shows representative pictures of immunostainings at days 1 and 4 on polyornithine. Quantification of the different markers showed increased levels of nestin, vimentin, BLBP, and RC2 at day 1 on polyornithine in ES Pax6 compared with ES control and ES Sox1 cells (Fig. 4 A). At day 4 (Fig. 4B), β3-tubulin was highly upregulated in ES Pax6 compared with ES control and ES Sox1 cells. Upregulation of radial glial cell markers was less pronounced, suggesting that most cells were already differentiated into neurons. Taken together, these results suggest that (a) Sox1 overexpression maintains cells at the neuroepithelial stage and prevents their differentiation into radial glial cells, and (b) Pax6 expression in neuroepithelial cells enhances their differentiation into radial glial cells and ultimately into neurons.

thumbnail image

Figure 4. Quantification of radial glial cell markers at the protein level. Embryonic stem (ES) control, ES Pax6, and ES Sox1 cells were cultured on MS5, and tetracycline was added to the cell culture medium at day 4. At day 5, cells were replated on polyornithine-coated coverslips, and tetracycline treatment was maintained for the next 72 hours. Immunostainings were performed after 24 or 96 hours to investigate expression of radial glial cell markers and of β3-tubulin to identify mature neurons. Automated quantification of immunostainings was performed at days 1 (A) and 4 (B) after replating. As MS5 cells were immunoreactive for vimentin, pictures were cropped to analyze only vimentin expressed from ES-derived cells. Blue bars: ES control cells. Yellow bars: ES Sox1 cells. Purple bars: ES Pax6 cells. Error bars indicate SEM. ∗, p < .05; ∗∗, p < .01. Abbreviation: A.V., arbitrary values.

Download figure to PowerPoint

siRNA-Mediated Knockdown of Sox1 and Pax6

We described above that Sox1 overexpression resulted in a decreased expression of Pax6 mRNA during neuronal differentiation and vice versa (Fig. 3). We next investigated how Sox1 or Pax6 knockdown at the neuroepithelial stage would affect further differentiation. For this purpose, we differentiated CGR8 cells 5 days on MS5 (i.e., approximately 80% of Sox1-positive neuroepithelial cells and 20% of Pax6-positive radial glial cells), dissociated them and transfected them with siRNA against Sox1 or Pax6, and used siRNA against GFP as a control. Interestingly, Sox1 downregulation was not sufficient to induce Pax6 expression (Fig. 5 B). However, siRNA against Sox1 provoked a premature differentiation of cells toward β3-tubulin-positive neurons (Fig. 5D). siRNA against Pax6 did not alter the proportion of Sox1-positive cells (Fig. 5C). However, it decreased the proportion of BLBP-positive cells (Fig. 5E). These results suggest that (a) knockdown of Sox1 allows further neuronal differentiation but is not sufficient to induce Pax6 expression, and (b) knockdown of Pax6 represses commitment toward radial glial cells.

thumbnail image

Figure 5. Effect of acute knockdown of Sox1 and Pax6 in neuroepithelial cells. Wild-type CGR8 cells were differentiated for 5 days on MS5 stromal cells, subsequently replated, and transfected with siGFP, siSox1, or siPax6. (AC): Immunostaining against Sox1 (red) and Pax6 (green) 48 hours after transfection with siGFP (A), siSox1 (B), and siPax6 (C). (D): quantification of the percentage of β3-tubulin-positive cells 48 hours after transfection with siGFP and siSox1. (E): Quantification of the percentage of BLBP-positive cells 72 hours after transfection with siGFP or siPax6. Error bars indicate SEM. Scale bars = 20 μm. ∗, p < .05, **, p < .07. Abbreviations: BLBP, brain-lipid binding protein; siGFP, short interfering RNA against green fluorescent protein; siPax6, short interfering RNA against Pax6; siSox1, short interfering RNA against Sox1.

Download figure to PowerPoint

Pax6 Overexpression Increases Migration of Neuronal Precursors

Pax6 is known to play an important role in the migration of differentiating radial glial cells [13–16]. Interestingly, during neuronal differentiation of ES cells, most neuroectodermal cells were found within tightly packed colonies, whereas mature neurons were generally found at the periphery as isolated cells. Thus, Sox1 is found mainly within colonies, Pax6 is found both within colonies and in the periphery, and β3-tubulin is found mostly in the periphery (data not shown; Fig. 4). This suggests that cells undergoing differentiation migrate away from the expanding colonies, similar to young neurons migrating away from germinal centers. Interestingly, Sox1-overexpressing cells stayed within the colonies (Fig. 3A), whereas ES Pax6-induced cells had a tendency to be more dispersed (an example is shown in Fig. 3G). We therefore investigated whether Pax6 expression was increasing the motility of neuroectodermal cells during their transition toward radial glia and neurons. For this purpose, we performed time-lapse imaging analysis of ES Pax6 cells and ES control cells undergoing neuronal differentiation on MS5 cells in tetracycline-containing medium. In addition, to investigate cell-autonomous or non-cell-autonomous alteration of cell motility, we cocultured ES control or ES Pax6 cells with ES cells expressing a mRFP1 [35], (described in Materials and Methods). ES Pax6 cells (Fig. 6 A, green) were more motile than control cells (Fig. 6B, green), with colonies often being disrupted by spontaneous migration of cells. In contrast, mRFP1-expressing cells (red) cocultured with ES Pax6 (Fig. 6A) or ES control cells (Fig. 6B) behaved as ES control cells, growing as compact colonies. We next performed quantitative analysis of cell migration over 24 hours (described in Materials and Methods). Figure 6C shows that there is an increase in cell motility in ES Pax6 cells but not in mRFP1-positive cells cocultured with ES Pax6, suggesting that the increase in cell motility mediated by Pax6 is cell-autonomous. Thus, Pax6 expression appears to be involved not only in the progression of differentiation toward mature neurons, but also in the migratory behavior of differentiating cells.

thumbnail image

Figure 6. Pax6 overexpression increases cell motility in a cell-autonomous manner. Shown are representative pictures from time-lapse imaging of ES Pax6 cells (green [A]) or ES control cells (green [B]) cocultured with monomeric red fluorescent protein 1 (mRFP1)-positive ES cells (red [A, B]). (C): Quantification of cell motion. Values represent total cell motion divided by total cell surface, normalized on control (TR) cells for GFP-positive cells or on mRFP1-positive cells cocultured with TR cells for mRFP1-positive cells. Time points are indicated as hours: minutes. Scale bar = 50 μm. ∗∗, p < .01. Abbreviation: ES, embryonic stem.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

In the present study we dissected functions of Sox1 and Pax6 during neuronal differentiation of mouse embryonic stem cells. We showed that Sox1 favors neuroectodermal lineage choice and maintenance but prevents Pax6 expression and further differentiation into radial glial cells. In contrast, Pax6 expression in neuroectodermal cells favors their further differentiation into radial glial cells and subsequently into neurons (Fig. 7).

thumbnail image

Figure 7. Model for the roles of Sox1 and Pax6 during early neurogenesis. Sox1 expression favors neuroepithelial fate choice and maintenance. Pax6 prevents the maintenance of the neuroepithelial state and triggers further differentiation of neuroepithelial cells into radial glial cells. Transitory Pax6 expression in uncommitted cells favors neural lineage commitment. Abbreviations: ES, uncommitted embryonic stem cells; N, neuron; NE, neuroepithelial cells; RG, radial glial cells.

Download figure to PowerPoint

The precise role of Sox1 in neurogenesis is subject to controversy: a role in maintenance of neuroectoderm has been suggested by some [1], whereas others have proposed a role in triggering differentiation steps downstream of neuroectoderm [2]. Forced Sox1 expression increases generation of both neuroectoderm and more mature neuronal cells [3, 8]; however, these results do not allow the temporal window of Sox1 action to be pinpointed. Our results elucidate this point. During neuronal differentiation of wild-type mouse embryonic stem cells, a switch from Sox1 to Pax6 expression occurs at the transition from neuroectoderm to radial glia. In embryoid bodies, forced Sox1 overexpression leads to a massive increase in neuroepithelial cells, radial glia, and mature neurons; however, this occurs concomitantly with extensive transgene silencing in later stages of neuronal differentiation (supporting information Fig. 2A). In contrast, in targeted neuronal differentiation, where little transgene silencing occurs (supporting information Fig. 2C), forced Sox1 overexpression retains cells at the neuroectodermal stage. These cells were also positive for nestin and Sox2 and negative for Oct-4 and Nanog (data not shown), confirming their neuroepithelial identity. Finally, siRNA knockdown of Sox1 at the neuroepithelial stage caused premature neuronal differentiation. Taken together, this evidence suggests a model where Sox1 favors development and maintenance of neuroectoderm and needs to be downregulated for further neurogenesis to proceed.

Pax6 is known to play an important role in neurogenesis [6, 36, 37]. In the developing mouse brain, Pax6 is expressed after the formation of the neural plate and therefore after Sox1 [5]. It is crucial for the generation of functional neurogenic radial glial cells that give rise to most cortical neurons [6], whereas other types of radial glia are generated in a Pax6-independent manner [6, 7]. In the present study, we show that Pax6 overexpression enhances expression of several proteins that characterize radial glia, such as BLBP and vimentin. It remains to be studied whether Pax6 regulates transcription of these genes directly or indirectly.

There are arguments suggesting a role of Pax6 in neurogenesis steps downstream of the generation of neurogenic radial glia, such as the migration of cells out of the ventricular zone [14]. In accordance with this study, we showed that Pax6 overexpression results in increased migration of young neurons out of neural progenitor colonies. These effects were restricted to cells overexpressing Pax6; therefore, this is the first demonstration that Pax6 overexpression induces cell-autonomous migration during the early steps of neuronal differentiation.

We also show that high levels of Sox1 prevent the expression of Pax6, and conversely, high Pax6 levels reduce Sox1 expression. Sox1 is known to act as a transcriptional activator during early neurogenesis [1] and is therefore likely to repress Pax6 expression indirectly. In contrast, Pax6 can act both as an activator or a repressor [38]. There are putative Pax6 binding sites in the Sox1 promoter region; however, we were not able to detect any binding of Pax6 to these sites by chromatin immunoprecipitation (data not shown). In addition, the effect of Pax6 overexpression on Sox1 mRNA levels was modest (Fig. 3O). Therefore, it is unlikely that Pax6 directly represses Sox1 expression at the transcriptional level.

An intriguing finding of our study is the proneural effect of Pax6 expression in uncommitted mouse ES cells, suggesting the intrinsic ability to give very early neural differentiation cues. This is unexpected in rodents, where during neurogenesis, Pax6 appears only in already neuralized tissue. Interestingly, however, in primate ES cells undergoing neuronal differentiation, Pax6 is expressed before Sox1 [36, 37], and Pax6-positive/Sox1-negative cells define a population of early neuroectoderm with broad potential [37]. Thus, our observations suggest a potential of Pax6 in very early neurogenesis, which already exists in rodents but is to be of biological relevance only in primates.

During evolution, Pax6 appears to be a widely used but versatile player in neural development. It is already important for neurogenesis in animals with a simple central nervous system, such as Drosophila [39]. In mammals, Pax6 appears to acquire a novel role in the newly emerging neocortex: it becomes a key molecule involved in the generation and specification of cortical neurons. In primates, compared with rodents, there are a broadening of Pax6 expression [6, 39, 40] and important differences in the positions of putative Pax6-binding sites [41].

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Our study has clearly defined distinct functions of Sox1 and Pax6 during in vitro neurogenesis. Extrapolated to the in vivo situation, our data suggest that Sox1 is a key regulator of the size of the neuroectodermal progenitor pool, whereas Pax6 limits the size of this pool and drives differentiation toward radial glia.

ACKNOWLEDGMENTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We thank Olivier Plastre for excellent technical assistance, Didier Chollet and Patrick Descombes for help in real-time polymerase chain reaction analysis, and Sergei Startchik for help in imaging studies. This work was supported by grants from the Swiss National Science Foundation (NFP46; Grant 404640-101109) and the Louis Jeantet Foundation. D.M.S. is a recipient of a M.D./Ph.D. scholarship from the Swiss National Science Foundation.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. ACKNOWLEDGMENTS
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional supporting information available online.

FilenameFormatSizeDescription
SC_1074_sm_suppfigure1.pdf346KSupporting Information Figure 1
SC_1074_sm_suppfigure2.tif1348KSupporting Information Figure 2
SC_1074_sm_suppfigure3.pdf497KSupporting Information Figure 3
SC_1074_sm_suppfigure4.pdf417KSupporting Information Figure 4
SC_1074_sm_supptable.pdf14KSupporting Information Table 1

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.