SEARCH

SEARCH BY CITATION

Keywords:

  • Stat3;
  • Self-renewal;
  • Microarray;
  • Leukemia inhibitor factor;
  • grg5

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Mouse embryonic stem (ES) cells can be propagated in vitro while retaining their properties of pluripotency and self-renewal under the continuous presence of leukemia inhibitor factor (LIF). An essential role has been attributed to subsequent activation of the Stat3 transcription factor in mediating LIF self-renewal response. To date, however, downstream target genes of Stat3 in ES cells are still unknown. To isolate these genes, we performed a microarray-based kinetic comparison of LIF-stimulated (undifferentiated) ES cells versus ES cells induced to differentiate by shutting down Stat3 activity through either LIF deprivation or, more specifically, expression of a Stat3 dominant-negative mutant. In each case, we chose the earliest time at which ES cells lose their self-renewal properties, as illustrated by a decrease in the number of embryoid bodies and blast cell colony formation as well as germ layer marker expression. Comparison of the two independent approaches revealed similarly regulated genes that are likely to be involved in the Stat3 effects on ES cell self-renewal. For instance, upregulation of growth factors such as the transforming growth factor-β relative Lefty1 or transcriptional regulators such as Id1 and Id2 and down-regulation of the groucho-like protein Aes1 (grg5) were found. Promoter analysis of the aes1 gene revealed three functional Stat3 consensus sites, as shown by luciferase assays. Furthermore, chromatin immunoprecipitation experiment demonstrated that Stat3 is recruited to the promoter of aes1 in ES cells. These data demonstrated that the aes1 gene is a direct transcriptional target of Stat3 in ES cells.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Mouse embryonic stem (ES) cells are permanent pluripotent cells derived from the inner cell mass of the blastocyst. These cells undergo self-renewing divisions while retaining the capacity to generate all types of fetal and adult cell lineages and to participate fully in embryonic development when reintroduced into a blastocyst [1]. Mouse ES cells can be propagated in culture in the presence of leukemia inhibitor factor (LIF), which activates the gp130 signaling pathways, resulting in Jak-mediated Stat3 phosphorylation. Several studies have shown that Stat3 is one major player in this process [25]. Other transcriptional regulators, such as Oct3/4 [6], Nanog [7, 8], or Id1-3 [9], also participate in the maintenance of ES cell self-renewal, but Stat3 activation remains prominent because in the absence of LIF stimulation, endogenous expression of these factors is insufficient to sustain ES cell self-renewal. However, very little is known about the downstream target genes of activated Stat3 in ES cells. As their characterization could be of help to understand the molecular mechanisms underlying self-renewal, we used a microarray approach in this study to analyze gene profiling in ES cells with several repetitions of the experiments, allowing a robust statistical analysis of the results. To perform this approach, ES cells were first induced to differentiate upon LIF withdrawal, which led to the disruption of the signaling pathways lying downstream of gp130. To specifically target the stage when ES cells start losing their self-renewal potential without extensive engagement in differentiation, we characterized the earliest stage of appearance of mRNA-encoding germ layer markers and diminution of the ability of ES cells to generate in vitro embryoid bodies and blast cell colonies. In a more specific second strategy, we expressed a dominant-negative Stat3 mutant and identified several potential downstream Stat3 target genes. Comparison of the results obtained in the two independent microarray experiments revealed an overlap of several transcripts that are similarly regulated and are thus likely to participate in the Stat3-mediated self-renewal response.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Cell Culture

Gs2, IOUD2, and E14 cells were cultured in the absence of feeder in Glasgow's minimum essential medium (Life Technologies, Rockville, MD, http://www.lifetech.com) supplemented with 15% fetal calf serum (FCS) (Hyclone, Logan, UT, http://www.hyclone.com), 450 μM monothioglycerol(MTG)(Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com), and 1,000 U/ml LIF (Esgro, Temecula, CA, http://www.esgro-lif.com). For the Gs2 cells, the medium was supplemented with 1 μg/ml tetracycline (Sigma-Aldrich). Upon tetracycline removal, this cell line expresses a dominant-negative form of Stat3 (Stat3F), whose expression induces a 50% reduction of endogenous Stat3 activation [4, 6].

ES Cell Differentiation

Gs2 ES cells were first maintained for 24 or 48 hours in the absence of LIF or tetracycline. Then they were induced to differentiate in nonadherent conditions as described below in the absence of LIF but in the presence of 1 μg/ml tetracycline to avoid expression of Stat3F. The number of day-3 embryoid bodies (EB3) was determined by plating for 3 days 500 ES cells in a 35-mm Petri dish in 1% methylcellulose (Fluka Methocell MC, Sigma-Alrich) in Iscove's modified Dulbecco's medium (IMDM) supplemented with 10% FCS, 300 μM MTG, and 80 μg/ml insulin (Sigma-Aldrich). Blast cell colonies were formed by the two-step method described by Kennedy et al. [10]. EB3 was first formed by plating 4.5 × 104 ES cells in a 100-mm Petri dish in 10 ml IMDM supplemented with 15% FCS, 450 μM MTG, 200 μg/ml iron-saturated transferrin (Sigma-Aldrich), and 50 μg/ml L-ascorbic acid (Sigma,-Aldrich) for 3 days. Cells were then dissociated by trypsinization, and for each condition, the same number (3 × 104) of living cells was plated in 1% methylcellulose (Methocult, Fluka, Buchs, Switzerland, http://www.buchs-sg.ch) in 35-mm Petri dishes in 5 ml IMDM supplemented with 10% FCS, 450 μM MTG, 25 μg/ml L-ascorbic acid (Sigma-Aldrich), 200 μ/ml iron-saturated transferrin (Sigma-Aldrich), and 5 ng/ml vascular endothelial growth factor. After 4 days, blast colonies (BC4) were counted.

Sample Preparation

A total of 4 × 105 cells were seeded in a 100-mm cell-culture plate under standard conditions (+LIF, Tet ON) in the absence of either tetracycline (Tet OFF) or LIF (−LIF) for 16, 24, and 48 hours. Total RNA was extracted with Trizol (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) and purified using Qiagen RNeasy kit (Qia-gen, Hilden, Germany, http://www1.qiagen.com). Concentration and quality were determined using an Agilent labchip Bioanalyzer (Agilent Technologies, Palo Alto, CA, http://www.agilent.com).

Microarray

We used an in-house c-DNA chip from the CEA (Evry, France, http://www.cea.fr). Approximately 9,000 clones were amplified and spotted onto GAPS II-coated slides (Corning, Corning, NY, http://www.corning.com). A complete description of the chips is available on the National Center for Biotechnology Information database (GEO, http://www.ncbi.nih.gov/geo/; platform #GPL1091). Total RNA 20 μg was reverse-transcribed at 42°C for 1 hour 30 minutes in a final volume of 30 μl with 400 U of Superscript II RNAseH reverse transcription (Invitrogen), oligodT (2 μg), dNTP (0.5 mM), and a 4:1 ratio of aminoallyl-dUTP (Sigma-Aldrich)/dTTP. RNA was then degraded with 2 U RNAseH (Invitrogen) at 37°C for 15 minutes. The reaction mixture was then filtered and concentrated using a Microcon 30 (Millipore, Billerica, MA, http://www.millipore.com), and indirect incorporation of Cy3 and Cy5 fluorescent labels (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) was performed in 10 μl of 50 mM NaCOOH for 1 hour at room temperature. Unincorporated dyes were quenched with 4.5 μl of 4 M hydroxylamine (Sigma-Aldrich) for 15 minutes. Labeled cDNA were pooled and purified on a nucleospin column (Macherey Nagel, NY, http://www.macherey-nagel.com), dried under vacuum, and resuspended in 30 μl of hybridization buffer (× 6 SSPE, × 2.5 Denhardt, × 0.5 SDS, 50% formamide) supplemented with 10 μg mouse cot1 DNA, 10 μg polydA, and 10 μg yeast tRNA (Invitrogen). Probes were heat-denatured and hybridized overnight at 42°C on the chip previously saturated by incubation in 20 × standard saline citrate (SSC), 10% SDS, 0.01% bovine serum albumin for 1 hour at 50°C. The chips were then washed twice in 500 ml of × 0.1 SSC and 0.1% SDS and once in 500 ml of × 0.1 SSC for 15 minutes at room temperature.

Microarray Analysis

All of the conditions were compared with Gs2 cells continuously maintained in the presence of LIF and tetracycline. Experiments were performed in both labeling orientations, allowing compensation for different dye incorporation efficiencies. For each time point, eight and six slides were performed independently for the Tet OFF and −LIF experiments, respectively. Slides were scanned in both Cy3 and Cy5 channels with AxonGenePix scanner. Grid alignments and spot detections were performed with the GenePix-Pro software (Axon Instruments/Molecular Devices Corp., Union City, CA, http://www.moleculardevices.com). Data are available at the NCBI gene expression and hybridization array data repository (GEO, http://www.ncbi.nlm.nih.gov/geo/; accession no. GSE1150, −LIF; and accession no. GSE 1151, Tet OFF). The data were thereafter processed with the GeneSpring software (Silicon Genetics, Palo Alto, CA, http://www.agilent.com/chem/genespring). After normalization by the lowess method, the data were filtered for nonreliable signals: using the GeneSpring software, the intensity in the control channel (+LIF/ON) for each spot was plotted against the standard deviation of normalized values for the totality of the spots to determine a threshold value for the mean of intensities in the control channel. The genes for which this value was under the threshold displayed high standard deviation and were thus excluded from the analysis. The mean of ratio (−LIF or Tet OFF vs. +LIF/ON) was calculated for each gene and asked for statistical variation from 1 by a Student's t-test at a 1% significant level (p < .01). Hierarchical clustering was also done with the GeneSpring software using the Pearson's correlation.

Real-Time Reverse Transcription–Polymerase Chain Reaction

Quantitative reverse transcription–polymerase chain reaction (RT-PCR) was performed on total RNA purified with the High Pure RNA extraction kit (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com). cDNA was synthesized from 2 μg of RNA with Superscript II reverse transcription (Invitrogen) using random hexamer primers. Quantitative PCR was performed on an ABI Prism 5700 sequence detection system (Applied Biosystems, Foster City, CA, http://www.appliedbiosystems.com) using SYBR green chemistry. Primers were designed with the Primer Express software (Applied Biosystems) and are available on request.

Chromatin Immunoprecipitation

Chromatin immunoprecipitation (ChIP) of STAT3 binding regions was performed on 2 × 107 cells, as described before [11], with protein A sepharose beads instead of Staph A cells. Ten microliters of the chromatin supernatant was saved before the IP as INPUT. IP was performed with anti-Stat3 antibody (2 μl; Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com). Precipitated DNA samples were resuspended in 30 μl. PCR was performed on 1 μl of pure extracts or a 100-fold dilution (INPUT sample) with the Taq polymerase from Takara, using primers 5′-agt gga agc cag gca tag tg-3′ and 5′-agg tga ctc cag gac att gg-3′, amplifying part (187 bp) of the aes1 promoter (nucleotides −3419 to −3606) containing the putative Stat3 binding sites. Thirty-six cycles were performed at an annealing temperature of 66°C.

Transfection and Luciferase Measurements

pRcCMV and pRcCMVStat3 expression vectors were a gift from J. Bromberg [12]. An annealed oligonucleotide corresponding to the −3490 to −3419 region of aes1 promoter containing the three putative (agctcttctgggtaatagctcaactcctaattcccagaatg-cagttcactccaggatc) or mutated (agctcacctgggtcatagctcaactccta-aaccccagtctgcagaccactcgaggatc) stat3 binding sites was cloned between the SstI and BglII sites of the pGL2-promoter luciferase reporter plasmid (Promega, Madison, WI, http://www.promega.com) to generate the pAES1 or pAES1mut plasmids, respectively.

293T cells were grown in Dulbecco's modified Eagle's medium containing 10% FCS. A total of 105 cells was seeded in a 35-mm dish 24 hours before transfection using Fugene6 reagent (Roche Diagnostics). Total DNA for transfection was 1.6 μg, including 0.75 μg of empty pRcCMV or pRc CMV-Stat3 expression vector, 0.75 μg of luciferase reporter construct (pAES1 or pAES1mut), and 0.1 μg of β-galactosidase internal control vector to correct for variation in transfection efficiency.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Earliest Stages of ES Cell Differentiation

To identify downstream Stat3 target genes specifically involved in self-renewal regulation, we first determined the time window when a consistent decrease in ES cell potential occurred without extensive differentiation. To this end, we used the previously described Gs2 ES cell line expressing the tetracycline regulatable dominant-negative form of Stat3, which induces the differentiation within 6 days [4]. This Stat3F mutant differs from the wild-type protein by a mutation in the Y705 residue (Y705-F) and is believed to act by competitive inhibition of Stat3 phosphorylation at gp130 docking sites and, possibly, by sequestrating Stat3 partners in the cytoplasm. As shown in Figure 1A, Gs2 ES cells cultured under standard condition (+LIF/ON) formed characteristic round-shaped colonies, whereas morphological changes occurred 96 hours after LIF (−LIF) or tetracycline withdrawal (Tet OFF), with the presence of most large flattened differentiated colonies. To determine the time of engagement into differentiation in the two conditions, we also examined the expression of marker genes for the three germ layers [6, 13, 14]. The ectoderm (FGF-5) and parietal and visceral endoderm (GATA-4) markers were detected 48 hours after Stat3F expression and increased during differentiation, whereas the mesoderm marker brachyury appeared later, at 72 hours (Fig. 1B). We also tested the ability of ES cells maintained for 24 or 48 hours without LIF or tetracycline to subsequently form EB3 and BC4 in the absence of LIF and the presence of tetracycline to avoid Stat3F expression during the differentiation step (Table 1). A decrease in BC4 formation was observed when the cells were precultured for 48 hours without tetracycline (55% of the control). More pronounced effects were seen when the cells were LIF-deprived for 48 hours because only 54% (EB3) and 35% (BC4) of the colonies were obtained. Overall, for a given duration of preculture, LIF deprivation is more potent than Stat3F expression for abrogating pluripotency. This feature, consistent with the morphology (Fig. 1A) and the gene expression level of differentiation markers and microarray data (Fig. 1B and see below), may reflect the already described partial inhibition of endogenous Stat3 by Stat3F ([4] and data not shown).

thumbnail image

Figure Figure 1.. Morphological changes and germ layer marker expression on Gs2 embryonic stem cell differentiation. (A): Phase-contrast images of Gs2 embryonic stem cells plated at 5 × 103 cells per cm2 and cultured under standard conditions (leukemia inhibitor factor [LIF] 100 U/ml and tetracycline [Tet] 1 μg/ml; control), without Tet (Tet OFF), or without LIF (−LIF) for 96 hours. All photographs are at the same magnification, × 4. (B): Reverse transcription–polymerase chain reaction kinetic analysis of lineage-specific marker expression after Tet removal. As a positive control, cells were cultured for 96 hours without LIF.

Download figure to PowerPoint

Table Table 1.. Decrease in embryonic stem cell–derived EB3 and BC4 formation after a first-step culture in the absence of leukemia inhibitor factor (LIF) or tetracycline (Tet)
  1. a

    Embryonic stem cells were maintained under standard conditions (+LIF, Tet ON) or without LIF or tetracycline for 24 or 48 hours, as indicated, and subsequently induced to differentiate in the absence of LIF and in the presence of tetracycline as described in Materials and Methods. The number of day-3 embryoid bodies (EB3) and day-4 blast colonies (BC4) was measured. Shown is the number ± SD of colonies obtained from a typical experiment (n = 3). The percentages (mean ± SD) from three independent experiments appear between brackets.

Preculture condition    
LIFTetTime (h)EB3BC4
+ON 50 ± 1.5(100)115 ± 2.12(100)
+OFF2444 ± 7.5(88 ± 1)93 ± 13.4(85 ± 6)
  4845 ± 9(91 ± 2)60 ± 10.6(55 ± 3)
ON2443 ± 4(93 ± 10)87 ± 5.6(87 ± 15)
  4824 ± 5(54 ± 8)35 ± 4.9(35 ± 7)

Taken together, these data demonstrate that pluripotency is lost by a significant proportion of ES cells after Stat3F expression or, to a greater extend, upon a 48-hour LIF removal. We thus chose to perform gene profiling within the first 48 hours of LIF deprivation or Stat3 activity inhibition.

Microarray Analysis After LIF Deprivation

We first analyzed the modulation of gene expression upon LIF removal for 16, 24, and 48 hours in the presence of tetracycline (Tet ON). For each time point, six independent hybridizations were performed. The data were globally normalized and filtered as described in the experimental procedure section. Among a total of 5,891 genes, 277 were identified as differentially expressed at least at one time point according to a Student's t-test with a 99% confidence level (p < .01, supplemental online data 1). Kinetic analysis revealed that most of the statistically significant modulations occurred at 48 hours (277 vs. 43 and 46 genes at 16 and 24 hours, respectively). These differentially expressed genes were then grouped according to their expression profile by hierarchical clustering with the Pearson's correlation. Gene clusters with related expression pattern were easily distinguishable (Fig. 2). mRNAs increasing from 16 to 48 hours were grouped in cluster A, whereas those showing a decrease during this period were grouped in cluster B. Examples of known genes from these two clusters, showing the highest variations (mean of ratio > 1.5), are given in Figure 2. Among those, several genes were expected to vary upon ES cells differentiation. We indeed found a decrease in mRNA levels of osteopontin (spp1) [15], activin beta-B (Inhbb) [14, 16], and zinc finger protein 57 (zfp57) [17] and an increase in mesoderm-specific transcript (mest) mRNA. Other genes previously reported to vary (socs3 [18], zfp42/rex1 [6], and CD9 [15]) were also found but are not shown in Figure 2 because of a t-test p value above .01 (.02, .03, and .07, respectively). Interestingly, a decrease in Stat3 and socs3 expression, both described as Stat3 targets, was also found [19, 20]. These observations support evidence for a differentiation of ES cells and validate our approach. Several other genes whose expression has not previously been reported to vary were identified, some of which play an important role in the regulation of cell cycle or differentiation. For instance, mRNA levels of the helix-loop-helix (HLH) transcriptional regulators Id1 and Id2 were found to increase. This was also the case for mRNA-encoding cell-cycle regulators, such as the cyclin-dependent kinase inhibitor p57kip2 and cyclin D3. In addition, this study also pointed to several genes encoding proteins that can interfere with signaling pathways such as transforming growth factor (TGF)-β or Wnt pathway, known to be important for the control of stem cells properties [21, 22]. This is the case for lefty1 (ebaf), a member of the TGF-β superfamily, whose expression increased after LIF removal, and for aes1 (grg5), encoding a groucho-like partner of Tcf/LEF transcription factors, the nuclear mediators of Wnt signaling [23], whose expression is diminished.

thumbnail image

Figure Figure 2.. Modulation of gene expression upon LIF withdrawal. Gene expression was analyzed after LIF removal for 16, 24, and 48 hours. Statistically significant modulations in gene expression (Student's t-test, p < .01) were grouped by hierarchical clustering with the Pearson's correlation. Shown are examples of known genes with variation above 1.5-fold. Color code: yellow, no variation; red, upregulation; green, downregulation.

Download figure to PowerPoint

Microarray Analysis After Expression of a Stat3 Dominant-Negative Form

To analyze variations in gene expression induced by Stat3F (Tet OFF/+LIF), we followed the same time course and performed eight independent hybridizations for each time point (16, 24, and 48 hours). The data were processed as for the first experiment. Six hundred thirty-nine genes were found to be differentially expressed at least at one time point according to a Student's t-test with a 99% confidence level (p < .01, supplemental online data 2). Kinetic analysis revealed that most of the modulations took place at 48 hours (417 vs. 195 and 116 genes at 16 and 24 hours, respectively). Hierarchical clustering of these genes revealed two clusters of upregulated (C and C') and one of downregulated (D) genes (Fig. 3). Examples of genes included in these two clusters and displaying a high variation (mean of ratio > 1.5-fold) are shown in Figure 3. With these criteria, we identified 12 downregulated genes, and among them, only 3 are known, namely fibronectin (fn1), asparagine synthetase (asns), and apolipoprotein E (apoE) (Fig. 3).

thumbnail image

Figure Figure 3.. Modulation of gene expression after Stat3F induction. Expression of Stat3F was induced by tetracycline removal for 16, 24, or 48 hours, and gene expression was analyzed. Statistically significant modulations in gene expression (t-test, p < .01) were grouped by hierarchical clustering with the Pearson's correlation. Shown are examples of known genes with variation above 1.5-fold. Color code: yellow, no variation; red, upregulation; green, downregulation.

Download figure to PowerPoint

Overlap of the Two Microarray Experiments

LIF-activated gp130 not only delivers a self-renewal signal mediated by Stat3 but also activates other pathways, such as Erk/mitogen-activated protein (MAP) kinase, which, on the contrary, seems to favor ES cell differentiation [24], implying that genes whose expression is modulated upon LIF withdrawal are not necessarily Stat3-regulated genes involved in the control of ES totipotency. To identify such genes, we crossed the data obtained by the two independent approaches. As mentioned earlier, Stat3F effect on ES differentiation and gene expression is weaker than that observed upon LIF withdrawal; we thus chose to compare the data of the microarray experiments on the basis of the statistical confidence in the modulation of gene expression rather than on the intensity of the observed variations. Comparison of the genes from clusters of upregulated (A, C, C') and downregulated (B and D) genes obtained in the two experiments (Table 2) revealed an overlap of several similarly regulated genes (hereafter referred to as common genes), including, among others, both upregulated (Id1, Id2, and Lefty1) and downregulated (socs3, Inhbb, and aes1) genes.

Table Table 2.. Common differentially expressed genes after leukemia inhibitor factor (LIF) removal or STAT3F expression
  1. a

    Differentially expressed genes identified in −LIF and Tet OFF experiments were compared. Shown are the common genes of known function. Fold changes and t-test p values for each gene are shown in the supplemental online data.

UpregulatedDownregulated
GeneDescriptionGeneDescription
Idb1Inhibitor of DNA binding 1AesAmino-terminal enhancer of split
Idb2Inhibitor of DNA binding 2InhbbInhibin beta-B/activin
Cdkn1cCyclin-dependent kinase inhibitor 1C/P57kip2PsapProsaposin
EbafLefty1MlpMARCKS-like protein
Ccng1Cyclin G1Lrrfip1Leucine-rich repeat–interacting protein 1
Rbm16RNA binding motif protein 16SparcSecreted acidic cystein-rich glycoprotein
Atp1a2ATPase, Na+K+ transporting, alpha 2Gjb5Gap junction membrane channel protein beta 5
Tacstd1Tumor-associated calcium signal transducer 1Socs3Cytokine-inducible SH2 protein 3/socs3
Ldh1Lactate dehydrogenase 1, A chainGabarapl2GABA(A) receptor-associated protein-like 2
Lamr1Laminin receptor 1MylpfMyosin light chain, phosphorylatable
CsdaCold shock domain protein AMylpfMyosin light chain, phosphorylatable
ApoeApolipoprotein EVdac1Voltage-dependent anion channel 1
ankProgressive ankylosisEmdEmerin
Ca3Carbonic anhydrase 3CryabCrystallin, alpha B

Validation of Transcript Variations

We further validated the microarray-based findings in Gs2 cell line by real-time RT-PCR with transcripts encoding several common genes. As shown in Figure 4A, Lefty1, p57kip2, Id1, and Id2 mRNA levels increased whereas the expression of aes1 and inhbb was diminished, thus confirming the gene expression evolution drawn from micro array analysis, even though for most of the genes tested, the ratios obtained with RT-PCR assays were higher than those obtained with the chips. We also showed that our results are not restricted to a single ES cell line because aes1 and Id2 mRNAs undergo downregulation and upregulation, respectively, upon LIF withdrawal in IOUD2 and E14 ES cell lines (Fig. 4B).

thumbnail image

Figure Figure 4.. Validation through real-time reverse transcription–polymerase chain reaction (RT-PCR) of the variation of several common genes. (A): Real-time RT-PCR measurement of mRNA level after Stat3F expression (Tet OFF) or leukemia inhibitor factor (LIF) removal (−LIF) for 48 hours in Gs2 cells. (B): Upregulation and downregulation of, respectively, Id2 and aes1 mRNA level in IOUD2 and E14 ES cells after LIF removal for 48 hours. Data were normalized with gapdh as standard.

Download figure to PowerPoint

Functional Stat3 Sites in the Promoter Region of aes1

Because Stat3 acts mainly as a transactivator, direct Stat3 targets are presumably found among the downregulated group of mRNA. Interestingly, we found three potential STAT consensus sites upstream of the aes1 transcription initiation site: one GAS (TTCCCAGAA; −3485) and two degenerated sites (TTCT-GGGTA, −3455 and TTCACTCCA, −3446) [25]. We assessed the functionality of these Stat3 sites by measuring the transcriptional activation of the luciferase gene under the control of the −3490 to −3419 region from aes1 promoter. Transient transfection of Stat3-overexpressing 293T cells led to twofold to threefold induction of luciferase activity from pAES1 plasmid, whereas no induction was observed when the three Stat3 sites were mutated (pAES1mut) (Fig. 5A). We then further evaluated whether Stat3 specifically binds to the region of the aes1 promoter containing these three sites in ES cells. Measurement of activated Stat3 by Western blot on nuclear extracts have shown that the nucleus of ES cells was cleared from activated Stat3 after 48 hours of LIF deprivation (data not shown). Analysis of the in vivo binding of Stat3 by ChIP assay revealed that Stat3 binds to the promoter of aes1 in LIF-stimulated ES cells (Fig. 5B; lane 1) whereas, as expected, deprivation of LIF for 48 hours abrogated the signal (Fig. 5B; lane 2). The specificity of this signal was confirmed by the absence of amplification when ChIP was performed with an irrelevant antibody (Fig. 5B, lanes 3 and 4). Taken together, these results show that Stat3 is able to bind the promoter of aes1 in ES cells and activates transcription from the Stat sites.

thumbnail image

Figure Figure 5.. Functionality of the STAT binding sites in the promoter of aes1. (A): 293T cells were cotransfected with aes1 promoter-driven luciferase expression vectors (wild-type, AES1, or mutated, AES1mut) and empty (pRc) or Stat3 expression vector (pRc S3) as indicated. Luciferase activity was measured 48 hours after transfection and normalized with β-galactosidase activity as an internal control. Values are means ± SE from three independent transfections. (B): Embryonic stem cells were LIF-starved for 48 hours and then restimulated for 15 minutes (lanes 1, 3, and 5). Polymerase chain reaction amplification with primers toward part of the aes1 promoter containing the putative STAT3 binding sites was performed on chromatin sample (INPUT, positive control) or on Stat3-immunoprecipitated or irrelevant immunoglobulin G-immunoprecipitated (negative control) chromatin samples (ChIP) as indicated.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Gene profiling analysis of stem cells supposes that their properties are controlled at the level of gene expression. The candidate genes underlying these properties can act either by their relative overexpression (or specific expression) as stem cell genes or, on the contrary, by their underexpression (or absence of expression) as loss of stemness. Ideally, to identify both sets of genes by means of microarray analysis, the two compared populations should differ only in their self-renewing properties. To date, however, for ES cells, comparisons have been performed with their derivatives largely engaged in a differentiation program. For instance, ES cell transcriptome has been compared with that from various embryonic tissues or from ES cells treated for 4 days with retinoic acid [17, 26, 27]. This precluded the isolation of the postulated loss-of-stemness genes because they were not distinguishable from differentiation markers. In our study, however, kinetic analysis revealed that we have chosen the earliest time point showing both noticeable changes in gene expression and decrease in ES cell self-renewal. Thus, the upregulated transcripts we isolated are candidate loss-of-stemness genes. Our strategy, using repetition of the experiments, gives good confidence in the results, and, importantly, cross-validation of our two independent approaches ruled out variations that may result from tetracycline removal and allowed us to identify a set of common genes as targets of Stat3 mediating the LIF/gp130 effect on ES cell self-renewal.

Some of these genes can act as potential links between LIF/gp130/Stat3 and multiple signaling pathways previously involved in the maintenance of stem cell properties such as bone morphogenic protein (BMP) and Wnt [9, 21, 22]. This is the case for the secreted Lefty1 protein of the TGF-β superfamily, an antagonist of Nodal signaling. By negatively interacting with the EGF-CFC coreceptors, Lefty antagonizes signaling of the Nodal/Vg1/GDF TGF-β subset of ligands [28, 29]. Lefty has also been implicated in the antagonism of TGF-β1 [30], BMP signaling [3032], as well as the Wnt pathway [33] and has been shown to activate the MAP kinase pathway [34]. In mouse ES cells, Lefty1 expression increases upon retinoic acid–induced differentiation [35], whereas, in contrast, activin expression falls dramatically when the cells are induced to differentiate [16]. Interestingly, Nodal has a negative effect on neural differentiation from ES cells [36], and activation of the TGF-β/activin/nodal signaling through SMAD2/3 has been described to be associated with pluripotency and required for the maintenance of the undifferentiated state in human ES cells and in ex vivo mouse blastocyst outgrowth [37]. Our results of an increase in expression of an antagonist of these ligands and a concomitant decrease of the inhibin β-B subunit of activin upon LIF withdrawal and Stat3F expression suggest that Stat3 contributes to maintaining an optimal TGF-β/activin/nodal signaling in undifferentiated ES cells.

We also demonstrated an upregulation of Id1 and Id2 mRNA level. Id proteins are inhibitors of basic HLH transcription factors, which often positively regulate differentiation. In ES cells, Id1–3 proteins were identified as direct targets of BMP signaling [9, 38] and display opposite functions depending on the presence or absence of LIF. In the presence of LIF, they mediate the BMP4 inhibitory effect on neuroectodermal differentiation, thus participating in self-renewal. However, in the absence of LIF, they rather promote differentiation into non-neural fates [9], suggesting that LIF/Stat3 signal neutralizes the BMP promesodermal effect. Our observation of an increase in Id1 and Id2 mRNAs after Stat3F expression suggests that LIF/gp130/Stat3 antagonizes the BMP signaling pathway at the level of Id1 and Id2 transcription.

Because Stat3 acts mainly as a transactivator, direct Stat3 targets are expected to be found among the downregulated group of mRNA, whereas upregulated genes are more likely to be indirect targets. We identified aes1 (grg5) as a downregulated gene and could further show that three Stat-binding sites were present in the promoter region of aes1. We established through ChIP experiments that endogenous Stat3 binds to this promoter and can stimulate transcription from these sites, strongly suggesting that Stat3 directly transactivates aes1 in ES cells. Aes1 belongs to the TLE/grg groucho-like family of transcriptional corepres-sors, which bind and repress various transcription factors, including Tcf proteins, the nuclear mediators of the canonical Wnt signaling [23]. Activation of the Wnt/β-catenin pathway leads to the maintenance of pluripotency in both ES and hematopoietic adult stem cells [21, 22, 3941]. Interestingly, Aes1 lacks the C-terminal region of TLE/grg proteins and is a naturally occurring trans–dominant-negative molecule alleviating transcriptional repression by TLE/grg proteins [23, 42, 43]. Thus, in contrast to its full-length relatives, Aes1 acts by derepressing, rather than by inhibiting, Tcf-mediated transactivation and may therefore increase the canonical Wnt signaling in ES cells. Hence, Aes1 could both contribute to the positive effect of Stat3 on self-renewal and mediate a crosstalk between LIF/gp130/Stat3 and Wnt signaling. Aes1-null mice are viable, and approximately half of the mice exhibit a growth defect [44, 45]. However, the lack of abnormal embryonic or perinatal lethality, as seen in Stat3 knockout mice [46], may imply a functional redundancy with, for instance, other splice forms of TLE/grg proteins featuring only the amino terminal domains typical of aes1 [47, 48]. Interestingly, certain functions of aes1 are revealed when the level of one of its transcription factor partners is twofold lowered, as recently reported in skeletal development [49]. Accordingly, the putative role of aes1 in maintaining the activity of various transcription factors in ES cells above a critical threshold may be revealed in a sensitized background, for instance, when the Wnt signaling is weakened. Further functional analysis on aes1 and other identified genes is required to better understand the molecular mechanisms involved in ES cell self-renewal; however, our work sheds new light on the mechanisms underlying ES cell self-renewal.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Prof. A. Smith for the gift of the Gs2 cells, J. Bromberg for pRcCMV and pRcCMV-Stat3 expression vector, and Edwige Leclercq and Marie-Hélène Courtier for helpful technical assistance with the RT-PCR experiments. D.S. is a recipient of a fellowship from Institut Gustave-Roussy. This work was supported in part by a grant from Association Contre le Cancer (ARC 4728).

Disclosures

The authors indicate no potential conflicts of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
Supplementary_data_1.pdf173KSupplemental Data 1
Supplementary_data_2.pdf230KSupplemental Data 2

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.