Dax-1 Knockdown in Mouse Embryonic Stem Cells Induces Loss of Pluripotency and Multilineage Differentiation

Authors

  • Olfa Khalfallah,

    1. Institut de Pharmacologie Moléculaire et Cellulaire CNRS UMR 6097, Valbonne, France
    2. Université de Nice-Sophia Antipolis, Valbonne, France
    Search for more papers by this author
  • Matthieu Rouleau,

    1. INSERM U898, Faculté de Médecine, Université de Nice-Sophia Antipolis, Nice, France
    Search for more papers by this author
  • Pascal Barbry,

    1. Institut de Pharmacologie Moléculaire et Cellulaire CNRS UMR 6097, Valbonne, France
    2. Université de Nice-Sophia Antipolis, Valbonne, France
    Search for more papers by this author
  • Barbara Bardoni,

    1. Institut de Pharmacologie Moléculaire et Cellulaire CNRS UMR 6097, Valbonne, France
    2. Université de Nice-Sophia Antipolis, Valbonne, France
    Search for more papers by this author
  • Enzo Lalli

    Corresponding author
    1. Institut de Pharmacologie Moléculaire et Cellulaire CNRS UMR 6097, Valbonne, France
    2. Université de Nice-Sophia Antipolis, Valbonne, France
    • Institut de Pharmacologie Moléculaire et Cellulaire, CNRS UMR 6097, 660 route des Lucioles, Sophia Antipolis, 06,560 Valbonne, France

    Search for more papers by this author
    • Telephone: 33-(0)493957755; Fax: 33-(0)493957708


  • Author contributions: O.K.: collection and assembly of data, data analysis; M.R.: collection and assembly of data, data analysis; P.B.: financial support; B.B.: financial support, data analysis; E.L.: conception and design, financial support, data analysis, manuscript writing.

  • First published online in STEM CELLS EXPRESS March 26, 2009.

Abstract

Dax-1 (Nr0b1) is an orphan member of the nuclear hormone receptor superfamily that has a key role in adrenogonadal development and function. Recent studies have also implicated Dax-1 in the transcriptional network controlling embryonic stem (ES) cell pluripotency. Here, we show that Dax-1 expression is affected by differentiating treatments and pharmacological activation of β-catenin–dependent transcription in mouse ES cells. Furthermore, Dax-1 knockdown induced upregulation of multilineage differentiation markers, and produced enhanced differentiation and defects in ES viability and proliferation. Through RNA interference and transcriptome analysis, we have identified genes regulated by Dax-1 in mouse ES cells at 24 and 48 hours after knockdown. Strikingly, the great majority of these genes are upregulated, showing that the prevalent function of Dax-1 is to act as a transcriptional repressor in mouse ES cells, as confirmed by experiments using the Gal4 system. Genes involved in tissue differentiation and control of proliferation are significantly enriched among Dax-1–regulated transcripts. These data show that Dax-1 is an essential element in the molecular circuit involved in the maintenance of ES cell pluripotency and have implications for the understanding of stem cell function in both physiological (adrenal gland) and clinical (Ewing tumors) settings where Dax-1 plays a pivotal role in development and pathogenesis, respectively. STEM CELLS 2009;27:1529–1537

INTRODUCTION

Embryonic stem (ES) cells are capable to undergo cell division without differentiation (self-renewing) and to produce all three germ layers: ectoderm, endoderm, and mesoderm (pluripotency). The pluripotency of ES cells is maintained by the action of several signaling pathways and transcription factors. Among them, the POU domain protein Oct3/4 and the Nanog homeoprotein play a pivotal role [1]. These factors orchestrate transcriptional regulatory networks that control the expression of other transcription factors regulating target genes important for ES cells pluripotency. The definition of these transcriptional regulatory networks and the identification of the mechanisms controlling the expression and the activity of the factors involved in the network constitutes a major issue of the research in the stem cells domain.

Dax-1 (Nr0b1) is an unusual member of the nuclear hormone receptor superfamily. Its C-terminal domain has homology with the nuclear receptor ligand-binding domain, whereas three repeats of a unique cysteine-rich motif assemble to make its N-terminal domain. In the adult, Dax-1 is selectively expressed in tissues involved in steroidogenesis and reproductive function [2]. In humans, mutations in the DAX-1 gene cause adrenal hypoplasia congenita (AHC) associated with hypogonadotropic hypogonadism [3, 4], whereas Dax-1 deficient mice display no major adrenal and pituitary phenotype [5]. In steroidogenic tissues, DAX-1 functions as a negative regulator of steroid production (reviewed in [2]), working as a transcriptional repressor by interaction with a DNA stem-loop structure in the StAR promoter [6] and by repressing nuclear receptor transciptional activity by direct protein–protein interaction [7–9].

Recently, it has been shown that Dax-1 is expressed at high levels in undifferentiated murine ES cells and that its expression is rapidly downregulated during their differentiation [10, 11]. Furthermore, Dax-1 has been shown to be a direct target of, and interact with, Nanog [12, 13]. These data suggest that Dax-1 is a component of the transcriptional network present in ES cells to repress differentiation genes and to preserve their self-renewal and pluripotency potential.

This work was undertook to identify the molecular function and the target genes of Dax-1 in mouse ES cells. Here, we show that Dax-1 is rapidly downregulated at the mRNA (messenger RNA) and protein level by different treatments inducing ES cell differentiation. Our findings indicate that Dax-1 also functions as a transcriptional repressor in murine ES cells and show that both its N-terminal and C-terminal domains possess a promoter-specific transcriptional repressor activity. An RNA interference approach coupled to transcriptome analysis was used to identify genes regulated by Dax-1 in ES cells. Remarkably, Dax-1–regulated transcripts were significantly enriched in mRNAs encoding structural proteins and transcription factors involved in multilineage differentiation and cell cycle regulation. Consistent with these findings, the expression of markers associated with different terminal differentiation lineages were upregulated following Dax-1 knockdown.

MATERIALS AND METHODS

Plasmids

Full-length mouse Dax-1, Dax-1 1-202, and Dax-1 203-472 were cloned in the pG4MpolyII vector [14] to express them as fusion proteins with the yeast Gal4 DNA binding domain. pTK-luc and β-globin-luc reporter plasmids containing Gal4 binding sites have been described [14]. To study β-catenin–dependent transcription, the TOPflash and FOPflash luciferase reporters (Millipore, Billerica, MA, http://www.millipore.com) were used. pCH110 [14] was cotransfected to express β-galactosidase for normalization of luciferase activities.

Cell Culture

CGR8 mouse embryonic stem cells were routinely cultured on mitomycin-treated fibroblast feeder cells in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) containing 4.5 g/L glucose, 10% fetal calf serum (StemCell Technologies, Grenoble, France, http://www.stemcell.com), 0.1 mM nonessential amino acids, 0.1 mM β-mercaptoethanol, penicillin-streptomycin (all from Invitrogen), and 1,000 U/mL leukemia inhibitory factor (LIF). Before experiments, CGR8 cells were plated on gelatin-coated flasks and cultured for at least three passages to get rid of feeder cells. Three different protocols were used to induce differentiation of ES cells: (a) culture without LIF (−LIF); (b) 1% dimethyl sulfoxide (DMSO) in the absence of LIF (+DMSO); (c) 2 μM retinoic acid (RA) in the presence of LIF (+RA). These treatments are known to differentially modulate differentiation of mouse ES cells toward both ectoderm and mesoderm (−LIF, +DMSO) or neural and endodermal lineages (+RA), respectively [10]. In some experiments, treatment with 2 μM of the glycogen synthase kinase-3 (GSK-3) inhibitor 6-bromoindirubin-3′-oxime (BIO; Calbiochem, San Diego, http://www.emdbiosciences.com) was associated with differentiation protocols. Data are expressed as the percentage of the values in cells kept in basal culture conditions and represent the average ± SEM of three experiments.

Quantitative Reverse Transcription-Polymerase Chain Reaction

Quantitative reverse transcription-polymerase chain reaction (RT-qPCR) for transcript quantification was performed using the SYBR®Green I dye assay on a LightCycler 480 instrument (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com), using TATA-binding protein as a reference transcript, as described [15]. Primer sequences were designed using the Primer Express software (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com) and are reported in supporting information Table 1. Results were calculated using the ΔΔCt method [16].

Immunofluorescence and Immunoblots

Immunofluorescence and immunoblots were performed as described [17], using the 2F4 mouse anti DAX-1 antibody [18], mouse β-tubulin antibody (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and rat anti-mouse Flk-1 (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). Dax-1 protein expression in Figure 4 and supporting information Figure 2 was quantified by the ImageJ software and corrected by the relative β-tubulin expression.

Flow Cytometry

ES cells were fixed with 70% ethanol, treated with RNAse and stained with propidium iodide (50 μg/mL). Cells were analyzed with a FACScan instrument (Becton, Dickinson).

Reporter Genes Transfection and Luciferase Assays

CGR8 cells growing in 24-well plates were transfected with TOPflash and FOPflash reporters or Gal4 constructs using Lipofectamine 2000 (Invitrogen), following the manufacturer's instructions. After transfection, some samples were treated with BIO for 24 hours. Luciferase activity was measured as previously described [17]. Data represent the average values ± SEM of four experiments.

Small Interfering RNA Selection and Transfection

Synthetic double-stranded Stealth small interfering RNA (siRNA) oligonucleotides (Invitrogen) were selected using the online BLOCK-iT RNAi Designer tool (http://rnaidesigner.invitrogen.com/rnaiexpress/) according to their highest probability to produce efficient knockdown. The sequences of the three Dax-1 oligonucleotides used were (sense strand):

siDax-1 #1 ACCUGCACUUCGAGAUGAUGGAGAU

(Starting at nucleotide 940 of Dax-1 mRNA—reference sequence NM_007430)

siDax-1 #2 ACCAGAUCAGAUCCGCUGAACUGAA

(Starting at nucleotide 1,297 of Dax-1 mRNA—reference sequence NM_007430)

siDax-1 #3 ACCAACACGACGCAGGAAAUGCUUA

(Starting at nucleotide 972 of Dax-1 mRNA—reference sequence NM_007430)

The medium GC Stealth oligonucleotide (Invitrogen) was used as a control. siRNAs were transfected into CGR8 cells growing in 24-plates using Lipofectamine 2000 (Invitrogen), following the manufacturer's instructions. Cells were either lysed 24 or 48 hours after transfection and analyzed for RNA and protein expression or transfected again 48 hours after the first transfection using the same protocol and analyzed for RNA and protein expression at 72 and 96 hours.

Expression Microarray and Data Analysis

Samples and Labeling

Total RNA was extracted by the Trizol (Invitrogen) method and purified on RNeasy columns (Qiagen, Hilden, Germany, http://www1.qiagen.com). RNA concentration was measured by spectrophotometry and its integrity checked using an Agilent Bioanalyzer instrument. One microgram total RNA was amplified and labeled with Cy3 and Cy5 fluorochromes using the Amino Allyl MessageAmp aRNA kit according to the manufacturer's (Ambion, Austin, TX, http://www.ambion.com) protocol. Cy3-and Cy5-labeled cRNAs were fragmented using fragmentation buffer (Agilent Technologies, Palo Alto, CA, http://www.agilent.com), dissolved in hybridization buffer (Agilent), and hybridized to pan-genomic mouse microarrays of the RNG/MRC resource (http://www.microarray.fr). These microarrays harbor 24,109 50mer oligonucleotides. Probe sequences are available on the MEDIANTE web site (http://www.microarray.fr:8080/merge/index). They are archived at GEO (http://www.ncbi.nlm.nih.gov/projects/geo/query/acc.cgi?acc = GPL1476) as platform GPL1476.

Hybridization Procedures

Cy3-and Cy5-labeled cRNAs dissolved in hybridization buffer were hybridized on previously processed slides (treated with 50 mM ethanolamine in 50 mM borate buffer, pH 9.0 at 20°C for 1 hour) in a total volume of 500 μL using the Microarray Hybridization Chamber (Agilent) at 62°C for 17 hours using a four rpm agitation. Slides were washed with washing buffer #1 (6× SSC, 0.005% Triton X-102) at 20°C for 10 minutes and with washing buffer #2 (0.1× SSC, 0.005% Triton X-102) at 4°C for 5 minutes, dried, and stocked under vacuum. Fluorescence data were acquired from slides using a GenePix 4000B instrument (Molecular Devices Corp., Union City, CA, http://www.moleculardevices.com).

Data Analysis

TIF images containing the data from each fluorescence channel were quantified with the GenePix Pro 6.0 program (Axon Instruments, Union City, CA, http://www.moleculardevices.com). Data were analyzed using programs of the TM4 software suite (http://www.tm4.org) [19]. Data were imported from the MEDIANTE database as .gpr files and converted into .mev files using the ExpressConverter software. They were subsequently normalized using the lowess method and the flip dye routine on the MIDAS software. Analysis was performed using the Significance Analysis of Microarrays (SAM) method [20] on MeV 4.0 after percentage cutoff filtering (= 100%). Transcripts SAM parameters were set as follows:

24 hours FDR = 11.9%

48 hours FDR = 13.8%

Experiment Design

Gene expression profiles were analyzed by two-color microarrays in CGR8 cells transfected with siDax-1 #1 and #3 and compared with cells transfected with the control siRNA in two replicate experiments at both 24 and 48 hours. For each experimental point, two technical replicates (dye-swaps) were examined.

Statistics

The GEMS Launcher tool in the Genomatix suite (http://www.genomatix.de) was used to calculate the probability of common elements existence in the lists of Dax-1–regulated transcripts at 24 and 48 hours and Dax-1 bound promoters identified in the study by Kim et al. [21]. The MatInspector tool was used to search for the presence of Sf-1/Lrh-1 binding sites in the promoters of genes regulated by Dax-1.

RESULTS

Dax-1 Is Rapidly Downregulated at the Transcript and Protein Levels During Differentiation of Mouse ES Cells

Previously published transcriptome data have shown that the Dax-1 transcript is rapidly downregulated during mouse ES cell differentiation [10, 11]. To confirm these data at both the transcript and protein level, we studied the modulation of Dax-1 mRNA and protein in the CGR8 mouse ES cell line cultured in the absence of feeder cells, following three different differentiation protocols: LIF withdrawal (−LIF), LIF withdrawal associated to +1% DMSO (+DMSO), and 2 μM RA (+RA). All differentiation schemes induced a rapid downregulation of Dax-1 mRNA, with the +RA condition inducing the fastest and most dramatic effect (Fig. 1A). Remarkably, the Dax-1 transcript was transiently induced by RA treatment at early times (6-12 hours) before the occurrence of a dramatic fall in its levels at 24-72 hours. Expression of other pluripotency-associated transcripts (Rex-1, Oct3/4, Nanog) was also modulated by all three differentiation protocols, with Rex-1 being most rapidly downregulated. At the protein level, Dax-1 modulation by the three differentiation treatments followed the pattern of its mRNA, being most dramatically downregulated starting from 36 hours after beginning of the +DMSO and +RA treatments and at a reduced extent by simple removal of LIF from the culture medium (Fig. 1B). −LIF and +DMSO treatments equally powerfully induced expression of the primitive ectoderm marker Fgf5, whereas RA treatment preferentially induced expression of the neuroectoderm marker Isl1 (Fig. 1C).

Figure 1.

Dax-1 expression is rapidly downregulated at the messenger RNA (mRNA) and protein levels by differentiation-inducing treatments in mouse embryonic stem (ES) cells. (A): Time course of modulation of Dax-1 (red), Rex-1 (green), Oct3/4 (yellow), and Nanog (blue) mRNAs in CGR8 mouse ES cells by LIF withdrawal (left panel), −LIF +1% DMSO (center panel), and 2 μM RA (right panel) treatments. RNA was extracted from cells cultured in basal conditions (white histograms; 0) or subjected to the three differentiation-inducing treatments for 6, 12, 24, 36, 48, and 72 hours, respectively. (B): Immunoblot showing the time course of Dax-1 protein modulation by the same differentiating treatments in CGR8 cells. β-tubulin expression in the same samples is shown as a loading control. (C): Induction of Fgf5 (upper panel) and Isl1 transcripts (lower panel) by LIF withdrawal (blue histograms), −LIF +1% DMSO (white histograms), and 2 μM RA (red histograms) at various time points. Data are expressed as the percentage of the values in cells kept in basal culture conditions at the same time points. Abbreviations: DMSO, dimethyl sulfoxide; LIF, leukemia inhibitory factor; RA, retinoic acid.

Dax-1 is a nuclear protein in undifferentiated murine ES cells, and its levels are dramatically reduced in ES cells cultured in the absence of LIF, treated with DMSO or with RA for 48 hours (Fig. 2). Remarkably, in differentiated cells, Dax-1 appeared to be absent or to be expressed at very low levels with a cytoplasmic localization, especially after RA treatment.

Figure 2.

Immunofluorescence showing Dax-1 expression in mouse embryonic stem cells grown in the presence of LIF (+LIF) or cultured in the absence of LIF (−LIF), without LIF +1% DMSO (+DMSO) or after treatment with 2 μM RA (+RA). Treatments lasted for 48 hours. Nuclei were counterstained with DAPI dye and images were acquired by confocal microscopy. Scale bar = 50 μm. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; DMSO, dimethyl sulfoxide; LIF, leukemia inhibitory factor; RA, retinoic acid.

β-Catenin–Dependent Transcription Affects Dax-1 Expression in Mouse ES Cells

Dax-1 expression has been shown to be controlled by Wnt signaling in the gonad through interaction of β-catenin with Tcf/Lef transcription factors bound to the Dax-1 promoter [22]. Since Wnt signaling has an essential role in human and mouse ES cells for the maintenance of their pluripotency [23], we used BIO, a pharmacological inhibitor of GSK-3, which regulates negatively the Wnt effector β-catenin [23], to investigate the role of Wnt signaling in the control of Dax-1 expression in mouse ES cells. BIO treatment of CGR8 cells activates the β-catenin responsive promoter TOPflash, whereas it has no effect on the expression of the FOPflash reporter, which contains mutated Tcf binding sites (Fig. 3A). In mouse ES cells subject to differentiation by LIF withdrawal and DMSO treatment, BIO substantially rescued the expression of the pluripotency-associated gene Rex-1 (Fig. 3C, 3E), as also previously described [23]. Dax-1 expression was rescued by BIO treatment up to the 72 hours time point in cells cultured in the absence of LIF, whereas it was upregulated by BIO only to a reduced extent in samples where DMSO treatment was associated with LIF withdrawal (Fig. 3B, 3D). Conversely, BIO further stimulated Dax-1 expression at 6 hours after start of RA treatment, whereas it was unable to rescue its expression at later times (Fig. 3F). Interestingly, early stimulation by BIO under RA treatment was neither evident for Rex-1 nor was its expression rescued at later time points (Fig. 3G). BIO treatment did not significantly modulate Dax-1 expression in ES cells cultured in the presence of LIF (supporting information Fig. 1).

Figure 3.

Regulation of Dax-1 expression by β-catenin–dependent transcription in mouse embryonic stem cells. (A): CGR8 cells were transiently transfected with TOPflash (white histograms) and its negative control FOPflash (black histograms) luciferase reporters to monitor Tcf/β-catenin transcriptional activity. The pCH110 β-galactosidase expression plasmid was cotransfected with the luciferase reporter vectors. The glycogen synthase kinase-3 inhibitor BIO (2 μM) induced TOPflash activity, whereas it had no significant effect on FOPflash. Luciferase activities were normalized to β-galactosidase activities. *, p < .05; **, p < .01 (analysis of variance [ANOVA]) BIO-treated samples compared with samples in control conditions. (B, D, F) Time course of Dax-1 and (C, E, G) Rex-1 mRNAs expression in CGR8 cells cultured in the absence of LIF (B, C), without LIF +1% DMSO (D, E), or with 2 μM RA (F, G) and treated (black histograms) or not (white histograms) with 2 μM BIO. Data are expressed as the percentage of the values in cells kept in basal culture conditions at the same time points. *, p < .05; **, p < .01 (ANOVA) BIO-treated samples compared with samples in basal conditions. Abbreviation: BIO, 6-bromoindirubin-3′-oxime.

Dax-1 Knockdown Induces Upregulation of Early Differentiation Markers of the Three Embryonic Germ Layers

With the aim to determine the function of Dax-1 in mouse ES cells, we specifically extinguished its expression using synthetic RNA oligonucleotides (siRNAs) selected for their highest probability to induce efficient gene silencing and monitored the effects of Dax-1 knockdown on the expression of marker genes for embryonic germ layers at different time points. Dax-1 protein expression can be efficiently downregulated by two out of the three different siRNAs tested (supporting information Fig. 2). For further experiments, we selected the siRNA producing the highest repression of both Dax-1 mRNA and protein expression (siDAX-1 #1 in supporting information Fig. 2). This siRNA, used in a double-transfection protocol, efficiently repressed Dax-1 protein expression in mouse ES cells starting from 24 up to 96 hours after transfection (Fig. 4A). To investigate the functional effects of Dax-1 knockdown on the differentiation program of murine ES cells cultured in the presence of LIF, we monitored the expression of different marker genes of the pluripotency state (Oct3/4, Nanog, Rex-1) and of early markers of the three embryonic germ layers (Fgf5 for primitive ectoderm; Brachyury for mesoderm; HNF4α for endoderm) by RT-qPCR in murine ES cells silenced for Dax-1 expression at different times after siRNA transfection. Efficient downregulation of Dax-1 mRNA by the specific siRNA at all times was confirmed by these experiments (Fig. 4B). Upon Dax-1 knockdown, no significant variation was detected for the transcripts encoding Rex-1 and Oct3/4, whereas Nanog was downregulated at 96 hours. Expression of differentiation marker genes was induced by Dax-1 silencing, with Fgf5 and HNF4α being upregulated at higher levels than Brachyury. These data show that Dax-1 knockdown in murine ES cells induces the activation of multiple differentiation pathways even in the presence of LIF in the culture medium and implicates this transcription factor in the maintenance of the pluripotency state in mouse ES cells.

Figure 4.

Dax-1 knockdown induces the expression of early differentiation marker genes in mouse embryonic stem cells. (A): A representative experiment showing efficient Dax-1 protein repression by a specific small interfering RNA (siRNA) 24-96 hours after transfection. β-tubulin expression in the same samples is shown as a loading control. Quantification of Dax-1 expression in cells transfected with the specific siRNA as a percentage of the expression in cells transfected with sicontrol, after correction for β-tubulin expression, is shown for each time point. (B): Expression of Dax-1, pluripotency genes (Rex-1, Oct3/4, and Nanog), and early differentiation marker genes (Fgf5, Brachyury, and HNF4α) in cells with silenced Dax-1 expression at 24 (yellow histograms), 48 (blue histograms), 72 (red histograms), and 96 (green histograms) hours. Data are expressed as the percentage of the values at the same time points in cells transfected with the control siRNA. *, p < .05; **, p < .01, (analysis of variance) siDax-1 transfected samples compared with sicontrol transfected samples.

Differentiation Is Engaged in Mouse ES Cells Following Silencing of Dax-1 Expression

To assess whether prolonged Dax-1 knockdown in mouse ES cells not only triggered the expression of early embryonic germ layers markers but also induced cell differentiation, we probed the expression of marker genes specific of different cell lineages: ectoderm (keratinocyte K5 and K14, neural nestin), mesoderm [heart- and muscle-specific troponin T (TnT) and α-actinin] and endodermal Gata4. The expression of these transcripts was measured in mouse ES cells transfected with the control siRNA or the Dax-1 – specific siRNA at 96 hours after the start of the procedure. The most highly induced transcripts were K5 and nestin, with α-actinin and Gata4 being induced at intermediate levels and K14 and TnT showing little or no induction (Fig. 5A). These findings are consistent with the preferential induction of early ectodermal and endodermal markers after Dax-1 knockdown (Fig. 4B). We also investigated the expression of Flk-1, an early mesoderm marker expressed in endothelial cell precursors, in Dax-1–silenced ES cells [24]. Although Flk-1 was not expressed by ES cells transfected with a control siRNA, cell membrane staining for this antigen was readily detectable in ES cell cultures silenced for Dax-1 expression (Fig. 5B).

Figure 5.

Terminal differentiation is triggered by Dax-1 silencing in mouse embryonic stem cells. (A): Expression of keratinocyte markers K14 and K5, cardiac muscle TnT, and α-actinin, neural nestin, and endodermal Gata4 in CGR8 cells transfected with siDax-1 at 96 hours after transfection. Data are expressed as the percentage of the values at the same time points in cells transfected with the control small interfering RNA (siRNA). *, p < .05; **, p < .01 (analysis of variance) siDax-1 transfected samples compared with sicontrol transfected samples. (B): Expression of the early mesoderm marker Flk-1 is triggered by Dax-1 knockdown in CGR8 cells (lower panels) at 96 hours after transfection. No induction of Flk-1 can be detected at the same time after transfection of the control siRNA (upper panels). Nuclei were counterstained with DAPI dye and images were acquired by confocal microscopy. Scale bar: 50 μm. (C) Morphology of CGR8 cells transfected with sicontrol (upper panels) or siDax-1 (lower panels) at 96 hours after transfection. Left panels, phase contrast pictures. Right panels, DAPI staining of nuclei. Scale bar = 200 μm. Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; TnT, troponin T.

Effect of Dax-1 Knockdown on ES Cell Viability and Proliferation

Associated to the effect on the expression of differentiation marker genes, Dax-1 knockdown also induced a drastic reduction of the size of ES cell clusters (Fig. 5C). Flow cytometric analysis of cell cycle distribution revealed a higher percentage of cells in the sub-G1 peak in cells transfected with siDax-1, compared with cells transfected with sicontrol (supporting information Table 2). In addition, the proliferative fraction of cells with silenced Dax-1 expression was substantially decreased (54.5 in sicontrol-transfected cells vs. 43 in siDax-1-transfected cells). These data indicate that Dax-1 is a critical factor in sustaining viability and proliferation of mouse ES cells.

Dax-1 Is a Transcriptional Repressor in Mouse ES Cells

At the transcriptional level, in the adrenogonadal-pituitary axis DAX-1 functions as a repressor of the expression of genes mostly involved in steroidogenic function and reproduction (reviewed in [2]). A powerful transcriptional repression domain has been identified in the C-terminal portion of the protein [7, 14]. Significantly, DAX-1 mutants found in AHC patients have an impaired transcriptional repressor activity which, in the case of missense mutants, is caused by abnormal nuclear localization of the protein [14, 25, 26]. To investigate the transcriptional properties of Dax-1 in murine ES cells, we fused the full-length protein (1-472) and its N- (1-202) and C-terminal (203-472) domains to the yeast Gal4 DNA-binding domain and measured their effect on basal transcription of the thymidine kinase (TK) and β-globin promoters-luciferase containing upstream Gal4 binding sites. Interestingly, while both Gal4 - Dax-1 and Gal4 - Dax-1 (203-472) worked as efficient repressors of both the TK and the β-globin promoters, we observed a differential effect of the Gal4 - Dax-1 (1-202) construct, which repressed efficiently the TK but not the β-globin promoter (Fig. 6). These data show that Dax-1 works as a transcriptional repressor also in mouse ES cells and that both its N- and C-terminal domains have a repressor activity, which is promoter-specific in the case of the N-terminal domain of the protein.

Figure 6.

Dax-1 is a transcriptional repressor in mouse embryonic stem cells. β-globin and TK promoters luciferase reporters were transfected into mouse CGR8 cells along with plasmids encoding yeast Gal4 DBD (black histograms), Gal4 DBD fused to full-length Dax-1 (1-472; red histograms), Gal4 DBD - Dax-1 1-202 (green histograms), Gal4 DBD - Dax-1 203-472 (purple histograms), and pCH110 to express β-galactosidase. Corrected luciferase activities are expressed as the percentage of the values in cells transfected with Gal4 DBD. *, p < .05; **, p < .01 (analysis of variance) Gal4 DBD Dax-1 fusions compared with Gal4 DBD. Abbreviation: TK, thymidine kinase.

Genes Regulated by Dax-1 in Mouse ES Cells

To identify the genes regulated by Dax-1, we studied gene expression profiles in murine CGR8 ES cells at 24 and 48 hours after Dax-1 knockdown (Fig. 7A). To exclude off-target effects, a stringent analysis was performed using the SAM method to identify the transcripts that are consistently upregulated and downregulated by both effective Dax-1 siRNAs (siDax-1 #1 and #3) in two independent experiments (supporting information Tables 3 and 5). The expression of the great majority (90%) of all Dax-1–regulated transcripts was upregulated by Dax-1 knockdown at 24 hours, consistently with its transcriptional repressor activity. Conversely, the percentage of transcripts upregulated by Dax-1 knockdown dropped at about 70% at 48 hours, probably because of secondary effects induced by modulation of direct Dax-1–target genes (Fig. 7A). Dax-1–regulated transcripts were annotated and clustered according to their Gene Ontology classification using the DAVID software (http://niaid.abcc.ncifcrf.gov/). Gene expression analysis confirmed the activation of multiple differentiation pathways both at 24 (supporting information Table 4) and 48 hours (supporting information Table 6) after Dax-1 knockdown in murine ES cells. In particular, genes related to muscle, heart (mesoderm), and neural (ectoderm) differentiation were significantly regulated following Dax-1 knockdown. It was remarkable that Dax-1 also regulated the expression of genes involved in cell proliferation and apoptosis.

Figure 7.

Genes regulated by Dax-1 knockdown in mouse embryonic stem cells. (A): Genes regulated by Dax-1 knockdown in two different experiments using each time two different Dax-1 specific siRNAs (siDax-1 #1 and #3 in supporting information Fig. 2) identified by SAM. In red, genes upregulated by Dax-1 knockdown; in green, genes downregulated by Dax-1 knockdown. Left panel, genes regulated at 24 hours after siRNA transfection; right panel, genes regulated at 48 hours after siRNA transfection. (B): Comparison of genes regulated by Dax-1 knockdown at 24 (left) and 48 (right) hours with genes bound by Dax-1 in the study by Kim et al. [21].

26.8% of the genes significantly modulated by Dax-1 at 24 hours were also found in the study by Kim et al. [21] as bound by Dax-1 in their promoter region (p = .34 E−9). Conversely, Dax-1 was found bound to the promoter of 22.4% of the genes modulated at 48 hours after its knockdown (p = .59 E−6) (Fig. 7B and supporting information Tables 3 and 5). These sets of genes can thus be considered as bona fide direct Dax-1 target genes. Kim et al. [21] described that promoter sets bound by Dax-1 and Nanog were more closely related between themselves than it was each group to promoters bound by the other pluripotency-associated Myc transcription factor. For comparison, there was also a significant enrichment of Dax-1–regulated genes in the group of Nanog bound promoters at 24 (p = .14 E−5) but not 48 hours after Dax-1 knockdown and no significant enrichment in the group of Myc bound promoters at both time points.

DISCUSSION

The atypical orphan nuclear receptor Dax-1 has been suggested to have a function in ES cells, which is independent from its known function as a repressor of steroidogenesis [2, 6, 11]. Here, we have shown that Dax-1 is downregulated at both the transcript and protein level by treatments that induce differentiation of murine ES cells. Dax-1 knockdown rapidly induced the upregulation of molecular markers of the three embryonic germ layers and finally produced enhanced differentiation at the cellular level (Fig. 5A, 5B). Dax-1 silencing also severely affected ES cell viability and proliferation (Fig. 5C, supporting information Table 2). These results show that sustained Dax-1 expression in mouse ES cells is required to maintain their pluripotency, confirming and extending previous studies [11, 13], even in the absence of early downregulation of the pluripotency genes Oct3/4 and Nanog. These results suggest that Dax-1 lies downstream of the major pluripotency genes in mouse ES cells and is one of their critical effectors to repress a set of differentiation-related genes.

To identify genes regulated by Dax-1 using the RNAi technique in murine ES cells, we have focused our analysis on early times after Dax-1 knockdown, with the purpose to avoid interference from secondary events triggered by cell differentiation. Remarkably, Dax-1–regulated transcripts include genes specifically expressed in differentiated tissues, confirming that Dax-1 function is important in preserving pluripotency of murine ES cells. Strikingly, at the earliest time point studied (24 hours after transfection), 90% of the genes significantly regulated by Dax-1 were induced by its knockdown. This is consistent with the transcriptional repressor function of Dax-1 in ES cells (Fig. 6), similar to its known activity in steroidogenic cells [6, 15]. These data suggest that no endogenous activating ligand for Dax-1 is present in murine ES cells and that the N-terminal domain of Dax-1 can synergize with the C-terminal domain to produce promoter-specific transcriptional repression. These results are consistent with recent structural data showing that the Dax-1 C-terminal domain lacks a ligand-binding pocket that may accommodate ligands able to convert it into a transcriptional activator. Furthermore, in the Dax-1 C-terminal domain 3D structure, its helix H12 is docked into its own coactivator-binding groove [27]. In addition, our results confirm and reinforce recent studies showing that certain DAX-1 AHC mutants that leave the N-terminal domain of the protein intact are endowed with a partial transcriptional repressor activity [28, 29]. The activity of the DAX-1 N-terminal domain may thus be important in the specification of the AHC phenotype.

Our data have revealed a function of β-catenin-regulated transcription in the regulation of Dax-1 expression in mouse ES cells. GSK-3 inhibition partially restored Dax-1 expression downregulated by −LIF and DMSO treatment and potentiated early activation of Dax-1 by RA, whereas it was insufficient to rescue Dax-1 expression at later times of RA treatment (Fig. 3). These data suggest that other factors in addition to Wnt signaling are involved in the regulation of Dax-1 expression in mouse ES cells. One important regulator of Dax-1 is represented by the homeodomain protein Nanog, a factor required for safeguarding ES cell pluripotency, which binds to the Dax-1 gene in an intronic position and positively regulates its expression [13]. Other potential regulators of Dax-1 expression in mouse ES cells include the other pluripotency factors Stat3 and Oct3/4 [30].

The important role of nuclear receptors in the regulation of ES cells pluripotency and differentiation has been revealed by both molecularly focused [31] and genomic-scale studies [32]. In particular, one nuclear receptor playing a pivotal role in maintaining Oct3/4 expression is Lrh-1 (Nr5a2) [31]. Nuclear receptor Sf-1 (Nr5a1), which is closely related to Lrh-1, has been shown to be an important regulator of Dax-1 expression in the developing gonads and hypothalamic ventro-medial nucleus [33]. We speculate that Lrh-1 may play a similar role in maintaining Dax-1 expression in undifferentiated mouse ES cells, which lack Sf-1 expression [31]. In turn, Dax-1, that works as a negative regulator of Lrh-1 transcriptional activity [9, 27], is likely to function by limiting Lrh-1 action in ES cells. Dosage/level of activity of this factor may in fact be critical to maintain the undifferentiated state of ES cells, similarly to the other pluripotency regulators Oct3/4 [34] and Nanog [35]. Similarly, Dax-1 may function in undifferentiated ES cells to restrict the activity of the other interacting nuclear receptor Esrrb [9], which is also important to maintain their pluripotency [13, 36]. However, our data suggest that Dax-1 can regulate a great majority of its target genes independently from interaction with Lrh-1, as revealed by the finding that most of the Dax-1 bona fide direct target promoters (27 out of 37) lack Lrh-1 binding sites (supporting information Table 7). Furthermore, Dax-1 function is not mediated through the modulation of the expression of the essential pluripotency factors Oct3/4, Nanog, and Rex-1 (Fig. 4), underlining the importance of the autonomous action of Dax-1 in the maintenance of ES cell pluripotency.

Recent studies have shown that Dax-1 is part of the core protein network implicated in the regulation of murine ES cells pluripotency, interacting with other essential factors and binding to a common set of gene promoters [13, 21]. These results have put forward an “additive” model for gene regulation in ES cells, with promoters bound by few factors tending to be inactive or repressed and promoters bound by multiple factors being active in the pluripotent state [21]. Our results indicate that this simple model of gene regulation should be modified to take into account the prevalent transcriptional repressor activity of Dax-1, which is required to prevent differentiation of murine ES cells. Given its activity in the maintenance of the undifferentiated state of ES cells, Dax-1 represents an interesting factor to be tested for reprogramming differentiated cell types to generate induced pluripotent stem cells [37].

In Ewing tumor cells, DAX-1 expression is under the direct control of the EWS/FLI oncogene and is required for tumorigenesis, controlling genes involved in cell cycle progression [38–40]. Remarkably, also in mouse ES cells, Dax-1 regulates genes controlling cell proliferation and apoptosis (supporting information Table 4 and 6). Even if the identity of the Dax-1–regulated genes varies between Ewing and ES cells (J. Alonso, personal communication), it is noteworthy that in both cellular systems Dax-1 appears to regulate similar pathways involved in the control of cell replication. These findings are even more intriguing since in the human fetal adrenal DAX-1 is mainly expressed in the outer, definitive zone of the cortex where precursor cells are localized [41] and its mutations are responsible for AHC, which is due to a failure of proliferation and differentiation of definitive zone cells after the postnatal involution of fetal zone cells [42]. It is then possible to speculate that also in the developing adrenal gland Dax-1 is involved in the suppression of the differentiated, steroid-secreting phenotype and in the maintenance of the proliferative capacities of adrenocortical precursor cells [2, 41]. On the other hand, DAX-1 mainly regulates genes involved in the steroidogenic pathway in differentiated adrenocortical cells [6, 43]. This indicates that DAX-1 may differentially regulate gene expression based on the cellular context. Future studies will be needed to elucidate the molecular basis of this differential transcriptional regulation in distinct cell types.

Acknowledgements

We thank Frédéric Brau and Julie Cazareth for assistance with confocal microscopy and flow cytometry and J. Alonso for communicating results before publication. This study was supported by grants from Association Française contre les Myopathies, Conseil Général 06, and Institut National du Cancer.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.

Ancillary