Characterization of the Phosphoinositide 3-Kinase-Dependent Transcriptome in Murine Embryonic Stem Cells: Identification of Novel Regulators of Pluripotency


  • Author contributions: M.P.S.: conception and design, collection and assembly of data, data analysis and interpretation and manuscript writing; B.K.: collection and assembly of data, data analysis and interpretation; B.T.: collection and assembly of data, data analysis and interpretation; R.K.: data analysis and interpretation; J.V.: data analysis and interpretation; O.H.: collection and assembly of data; H.S.: data analysis and interpretation; M.J.W.: conception and design, financial support, data analysis and interpretation, manuscript writing and final approval of manuscript.

  • First published online in STEM CELLS Express January 15, 2009.


Phosphoinositide 3-kinase (PI3K)-dependent signaling has been implicated in the regulation of embryonic stem (ES) cell fate. To gain further insight into the mechanisms regulated by PI3Ks in murine ES cells, we have performed expression profiling using Affymetrix GeneChips to characterize the transcriptional changes that arise as a result of inhibition of PI3K-dependent signaling. Using filtering of greater than 1.5-fold change in expression and an analysis of variance significance level of p < .05, we have defined a dataset comprising 646 probe sets that detect changes in transcript expression (469 down and 177 up) on inhibition of PI3Ks. Changes in expression of selected genes have been validated by quantitative reverse transcription polymerase chain reaction. Gene ontology analyses reveal significant over-representation of transcriptional regulators within our dataset. In addition, several known regulators of ES cell pluripotency, for example, Nanog, Esrrb, Tbx3, and Tcl-1, are among the downregulated genes. To evaluate the functional involvement of selected genes in regulation of ES cell self-renewal, we have used short interfering RNA-mediated knockdown. These studies identify genes not previously associated with control of ES cell fate that are involved in regulating ES cell pluripotency, including the protein tyrosine phosphatase Shp-1 and the Zscan4 family of zinc finger proteins. Further gain-of-function analyses demonstrate the importance of Zscan4c in regulation of ES cell pluripotency. STEM CELLS 2009;27:764–775


Embryonic stem (ES) cell lines exhibit the remarkable property of pluripotency, that is, the ability to differentiate into all cell lineages comprising the adult organism [1, 2]. Pluripotency is underpinned by ES cell self-renewal, widely defined as symmetrical division of ES cells to generate two identical undifferentiated daughter cells. Regulation of ES cell self-renewal is of great interest because the ability to maintain and expand pluripotent ES cells is essential if the therapeutic potential offered by ES cell-derived progeny to regenerative medicine is to be realized.

The pluripotency of ES cells is maintained by the coordinated actions of extrinsic regulators, signaling pathways, and transcription factors. Leukemia inhibitory factor (LIF), via activation of Stat3 and c-myc, was the first cytokine implicated in maintenance of mouse embryonic stem (mES) cell pluripotency [2–5], while bone morphogenetic proteins 2 and 4 [6] and glycogen synthase kinase-3/Wnt signaling [7] have been more recently reported to enhance self-renewal in the presence of LIF. The transcription factors Oct4, Sox2, and Nanog are widely recognized as playing key roles in maintenance of the undifferentiated phenotype [8]. Interestingly, overexpression of Nanog alone is sufficient to sustain ES cell self-renewal in the absence of LIF [9], with Nanog dimerization critical [10], while transient downregulation of Nanog predisposes cells to differentiation [11]. The demonstration that a set of promoter sequences are cobound by all three factors in human embryonic stem cells [12] and by Nanog and Oct4 in mES cells [13] supports the view that Oct4, Sox2, and Nanog act in concert to maintain pluripotency, although recently this transcriptional network has been further extended [14].

A number of additional genes also contribute to regulation of pluripotency, including Tcl1, Tbx3, Esrrb, Dppa4, Unigene Mm.343880 [15], Sall4 [16], Zfp206 [17], Zfx [18], Tcf3 [19], Klf 2,4, and 5 [20, 21], Zic3 [22], Pim1 and 3 [23]. These reports indicate that although Nanog, Oct4, and Sox2 play a central “permissive” role in maintaining ES cell pluripotency, other regulators further elaborate this network. Additional levels of complexity have been illustrated by the characterization of proteins that interact with Nanog in mES cells [24] and the demonstration that epigenetic changes also influence ES cell fate [25–27].

Phosphoinositide 3-kinases (PI3Ks) are a family of lipid kinases, which regulate a range of physiological processes including proliferation, development, growth, and migration [28]. The class I PI3Ks phosphorylate PI(4,5)P2 to generate PI(3,4,5)P3 and interestingly ablation of the p110β catalytic isoform of this class results in lethality at the preimplantation stage [29]. PI3Ks have been implicated in the regulation of ES cell proliferation because PTEN null ES cells (PTEN is a negative regulator of the PI3K pathway) display accelerated cell cycle progression [30]; an active form of the PI3K p110α catalytic subunit can restore proliferation of ERas null ES cells [31], and inhibition of PI3Ks with 25 μM LY294002 reduces ES cell proliferation [32]. In contrast, we have demonstrated that PI3K-dependent signaling is also required for optimal maintenance of mES cells in their undifferentiated state [33], and other groups have independently confirmed our findings in mouse, monkey, and human ES cells [34–37]. Mechanistically, we have revealed that inhibition of PI3Ks leads to elevated extracellular regulated kinase/mitogen-activated protein kinase signaling [33] and demonstrated that expression of Nanog as well as several putative Nanog target genes is regulated in a PI3K-dependent manner [38].

To gain further insight into the molecular mechanisms under the control of PI3K-dependent signaling pathways in ES cells, we undertook global expression profiling using Affymetrix 3′ expression arrays. Statistical analyses have defined a set of genes that display altered expression on inhibition of PI3Ks in undifferentiated ES cells. Loss and gain of function of selected genes revealed a role for the Zscan4 family of zinc finger proteins as novel regulators of ES cell pluripotency.


Cell Culture, Plasmids, and Stable Transfections

E14Tg2a and E14 Nanog-green fluorescent protein (GFP) [11] murine ES cell lines were maintained in culture in Knockout (KO) Dulbecco's modified Eagle's medium (DMEM), supplemented with 15% (vol/vol) KO serum replacement (SR; Invitrogen, Scotland, and 103 units/ml ESGRO (LIF) (Chemicon, Temecula, CA, as described previously [33, 39]. Inhibitors, either 5 μM LY294002 or 2 μM BIO, were added to cells for the lengths of time indicated and dimethyl sulfoxide (DMSO) used as vehicle control. ZScan4 family members were cloned from cDNA generated from RNA isolated from undifferentiated E14Tg2a ES cells using the following primers; forward: 5′-ACAATGGCTTCA CAGCAGG; reverse: 5′-GTCAGATCTGTGGTAATTCC. Polymerase chain reaction (PCR) products were cloned into pcDNA3.1/V5-His-TOPO (Invitrogen) for expression analyses. To generate V5-His-epitope tagged Zscan4-expressing ES cells, E14Tg2a cells were electroporated (800 V, 3 μF) with linearized pcDNA3.1-Zscan4cV5-His. 5 × 105 cells were plated per 10-cm dish, cultured in the presence of LIF plus G418 (200 μg/ml) for 8 days and resistant colonies picked, expanded, and screened for expression by immunoblotting with anti-V5 antibodies.

Preparation of RNA and Affymetrix Microarrays

RNA was extracted using RNeasy kits (Qiagen, Hilden, Germany, with on-column DNase digestion according to the manufacturer's recommendations. RNA from three independent experiments per time point was labeled using the One-Cycle Target Labeling and Control Reagent package according to manufacturer protocols (Affymetrix, Santa Clara, CA, In summary, biotinylated cRNA was generated using the in vitro transcription labeling procedure. cRNA was fragmented and 15 μg hybridized to Mouse Genome 430 2.0 arrays (45,101 probe sets) for 16 hours at 45°C. Using an Affymetrix fluidics Station 450, arrays were washed and stained with streptavidin-phycoerythrin. Signals were detected by scanning with an Affymetrix Gene-Chip Scanner 3,000 7G. Image data generated were analyzed using GCOS 1.4 with default Affymetrix analysis settings. Original datasets have been deposited in the ArrayExpress database for open access. Robust multichip average (RMA) normalization was performed, and three pair-wise comparisons were performed using analysis of variance (ANOVA). An ANOVA p value < .05 in the control (DMSO) versus LY294002-treated comparison and a fold-change of greater than 1.5 anywhere in the time-course were the parameters used to identify differentially expressed genes (n = 646 probe sets). For differentiation of expression changes between the different conditions, hierarchical clustering of these 646 probe sets was performed using Cluster version 3.0 [40] applying mean-centering of genes, before average linkage clustering with uncentered correlation. According to the expression profile of the hierarchical cluster, a k-means clustering was performed using k = 12 groups, euclidean distance as similarity matrix and 100 runs [40].

Quantitative Reverse Transcription Polymerase Chain Reaction

Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was carried out as described previously using the LightCycler 1.5 system (Roche Diagnostics, Basel, Switzerland, [38]. The gene-specific primers used in this study are available in supporting information Table 1. Melt-curve analyses were performed to verify specific amplification and PCR efficiencies calculated and used to determine the relative quantification values for calibrator-normalized target gene expression using LightCycler software (v4.0). In all cases, transcript levels were normalized to β-actin and analyzed for statistical significance using two-tailed Student's t test.

Short Interfering RNA-Mediated Knockdown

Commercial short interfering RNAs (siRNAs) were purchased from Dharmacon (Lafayette, Co, Endoribonuclease prepared siRNAs (esiRNAs) were prepared following a protocol adapted from [41]. siRNA and lipofectamine 2000 (Invitrogen) were incubated separately with 50 μl KO DMEM, without supplements, for 5 minutes, mixed gently and incubated for a further 20 minutes. 1 × 105 cells were added to the siRNA/lipofectamine mix and plated into KO DMEM containing SR and 103 U/ml ESGRO (LIF). After 48 hours, cells were washed in phosphate buffered saline (PBS) and retransfected while still adhered. Following a further 24 hours, cells were typsinized and plated into self-renewal assays (see below). The remaining cells were either lysed in solubilization buffer as described previously [42] or RNA was extracted using 200 μl Trizol (Invitrogen) according to standard protocols.

Self-Renewal Assays

To determine the ability of ES cells to retain an undifferentiated phenotype, self-renewal assays were performed as previously described [33], with the exception that Fast violet was substituted for Fast red. Alkaline phosphatase positive colonies, indicative of undifferentiated, self-renewing ES cell colonies and unstained, differentiated colonies, were counted in triplicate for each treatment following 3-5 days in culture.

Generation of Cell Lysates and Immunoblotting

Cells, on ice, were washed three times with PBS prior to lysis in solubilization buffer, as described previously [42]. To isolate nuclear proteins, 10 μl of freshly made nuclear extraction buffer (20 mM HEPES pH7.9, 400 mM NaCl, 1 mM EGTA, 1 mM EDTA, 25% glycerol, 1 mM dithiothreitol) containing a cocktail of phosphatase and protease inhibitors (Roche Diagnostics) was added to the pelleted nuclei and mixed for 15 minutes at 4°C. Cell debris were harvested via centrifugation for 2 minutes and the supernatant retained (nuclear extract). Protein concentrations were determined using Bio-Rad protein assay kit according to the manufacturer's instructions. Twenty μg of each lysate was fractionated by SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblotted onto nitrocellulose [42]. The following primary antibodies were used: Mouse monoclonal anti-V5 epitope tag antibody (ABgene, Surrey, U.K., at 1:1,000 dilution; rabbit polyclonal anti-Shp-1, Shp-2, Stat3 (Santa Cruz Biotechnology Inc., Santa Cruz, CA,, or p85 (Upstate, Charlottesville, VA, antibodies at 1:4,000 dilution; rabbit polyclonal anti-phosphorylated p44/42MAPK (Cell Signaling Technologies, Danvers, MA,, CST 9101), phosphorylated ribosomal protein S6 (CST 2211), phosphorylated Tyr 705 Stat3 (CST 9131) at 1: 1,000 and anti-phosphotyrosine antibody, 4G10 (Upstate), at 1:10,000. Goat anti-rabbit or rabbit anti-mouse secondary antibodies, conjugated to horseradish peroxidase (DAKO, Glostrup, Denmark, were used at 1:20,000 dilution and blots developed using enhanced chemiluminescence (GE Healthcare, Life Sciences, Little Chalfont, Bucks, United Kingdom, Blots were stripped and reprobed as described previously [43].

Immunocytochemistry and Flow Cytometry

For immunocytochemical staining, ES cells were cytospun onto a coverslip and fixed with 4%(wt/vol) paraformaldehyde/1%(wt/vol) saponin in PBS at room temperature. Donkey blocking serum was added prior to incubation with anti-V5 antibody. After two PBS washes, goat α mouse IgG Cy3 (GE Healthcare) was applied for 1 hour. Samples were washed and 10 μg/ml 4′-6-diamidino-2-phenylindole in PBS added for 1 hour. Cells were mounted in fluorescence mounting medium (DAKO) and visualized using a Zeiss LSM510 confocal microscope. Flow cytometry of Nanog-GFP ES cells was performed using a fluorescence-activated cell sorter (FACS) Canto cytometer (Becton Dickenson/BD Biosciences, Erembodegem, Belgium, with FACS Diva software. Dead cells were excluded based on forward and side scatter parameters.


Affymetrix Microarray Analyses Define the PI3K-Dependent Transcriptome in Murine ES Cells

PI3K-mediated signaling has been demonstrated to play a role in regulating self-renewal of murine, monkey, and human ES cells [33–37]. To gain further insight into the molecular mechanisms regulated by PI3K signaling in ES cells that contribute to controlling ES cell fate, we have characterized the transcriptional profile regulated by PI3K-mediated pathways. Our previous studies have shown that treatment with the broad selectivity PI3K inhibitor, LY294002, at a dose of 5 μM (close to its IC50 value [44]), leads to a loss of ES cell self-renewal after 3-4 days, an effect that is partially reversible up to 48 hours of inhibition [33]. LY294002 (at 5 μM) was used to inhibit the PI3K activity profile in undifferentiated E14Tg2a ES cells cultured in the presence of LIF. ES cells were incubated in the presence of LIF, with or without LY294002, for 24, 48, or 72 hours, and for each treatment and time point, RNA was extracted from each of three independent biological replicates. Mouse Genome 430 2.0 arrays (Affymetrix) were used to assess global gene expression in each sample. robust multichip average (RMA) normalization was applied to the raw expression data and pairwise comparisons carried out between the controls and LY294002-treated samples. An ANOVA p value of < .05 was applied to the data together with a fold-change cut-off of 1.5-fold or greater. These filters generated a dataset containing 646 probe sets, corresponding to 469 probe sets detecting transcripts downregulated and 177 probe sets detecting transcripts upregulated on inhibition of PI3K signaling (complete probe set information is available in supporting information Table 2). Taking into account current gene annotations, the 646 probe sets map to 514 distinct genes, including several Riken cDNAs, predicted genes, and expressed sequence tags. Of these genes, 363 are downregulated, and 149 are upregulated. An initial hierarchical clustering of the 646 probe sets suggested that the dataset falls into 12 groups. This was used to cluster the data using k-means (k = 12), and the corresponding heatmap is shown in Figure 1A with a full probe set list for each cluster given in supporting information Table 3. The 20 genes showing the most statistically significant changes in gene expression are shown in Figure 1B.

Figure 1.

Clustering and gene ontology analyses of gene expression changes occurring in embryonic stem (ES) cells upon inhibition of phosphoinositide 3-kinases (PI3Ks).(A): k-means clustering of 646 probe sets into 12 clusters. (B): Probe sets with the 20 highest significance scores in the dataset are presented. The Affymetrix probe set ID is shown along with the ANOVA value and FC in expression between controls and LY294002-treated samples at each time point. (C): ES cells were cultured in the presence of leukemia inhibitory factor plus or minus 5 μM LY294002 for the times indicated. Expression of Gm397 and Shp-1 was analyzed by quantitative reverse transcription polymerase chain reaction, and target gene expression was normalized relative to β-actin levels. The averages and SEM of duplicate samples from each of three independent biological replicates are shown: *, p < .05; **, p < .005; ***, p < .0005, in a Student's t test. (D): Schematic presenting temporal changes in ES cell behavior and gene expression arising after inhibition of PI3K signaling. Abbreviations: ANOVA, analysis of variance; DMSO, dimethyl sulfoxide; FC, fold-change; h, hour; ID, identifier.

Our cluster analyses reveal some interesting trends. For example, a gradual decline in expression is observed over time in the controls in clusters 1 and 2, whereas inhibition of PI3Ks reduces expression to a much greater extent. This pattern is mirrored in clusters 9, 10, 11, and 12, and to a lesser extent in clusters 4 and 5, where there is a gradual increase in expression of transcripts in these clusters in the controls over time. Again, inhibition of PI3Ks reduces expression within these clusters.

Over-Representation of Transcriptional Regulators in the PI3K-Dependent Transcriptome in Murine ES Cells

Using the database for annotation, visualization, and integrated discovery [45] Gene Functional Classification tool, with a Mus musculus background dataset and medium stringency, the 469 downregulated probe sets generates a set of 376 gene identities, which form 14 distinct functional groups (Table 1 Part A and supporting information Table 4 for full gene list). The 232 probe sets not included are listed in supporting information Table 5. The largest functional group, with the highest enrichment score, comprises genes encoding proteins involved in transcriptional regulation and DNA binding, including Klf2, Klf4, Nanog, Esrrb, Tbx3, and Zfp42, previously associated with ES cell pluripoteny [8, 15, 20, 46]. Genes within this group are located in a range of k-clusters (supporting information Table 6). Significantly, Oct4 and Sox2 are not contained within our dataset of downregulated genes, indicating that, at least at the transcriptional level PI3K-dependent signaling targets specific pluripotency networks. A variety of transporter molecules, located in the plasma membrane and within mitochondria, are represented in functional groups 12-14. Interestingly, given the reports that PI3K signaling can regulate ES cell proliferation [30–32], classical cell cycle regulators, with the exception of cyclin D1 in functional group 10, do not appear to be over-represented in our dataset of downregulated genes.

Table 1. Functional group clustering of genes down and upregulated on inhibition of phosphoinositide 3-kinase-dependent signaling in embryonic stem cells
inline image

Part B of Table 1 summarizes the results of analyses using the upregulated dataset. Again, the largest group with highest enrichment comprises genes that encode proteins involved in transcriptional regulation and DNA binding including Nfat, Sox4, Sox11, Lef1, Zic5, Zfp36, and Wt1, as well as the DNA methyltransferases Dnmt3a and Dnmt3b. A second functional group comprises histones, suggesting that loss of pluripotency and commitment to differentiation may lead to changes in histone composition. Functional groups 4 and 5 consist of a number of kinases and phosphatases, as well as ring finger proteins, indicating significant changes in regulatory processes following inhibition of PI3Ks. The complete lists of genes that cluster in these functional groups are available in supporting information Table 7, with nonclustering genes listed in supporting information Table 8.

Analysis of Kyoto encyclopedia of genes and genomes (KEGG) pathways, summarized in the part C of Table 1, reveals that in the dataset of downregulated probe sets, genes involved in sugar metabolism (galactose, fructose, mannose) and glycolysis are over-represented, which could relate to major changes in ES cell metabolism as the cells exit the self-renewing state and commit to a program of differentiation. Interestingly, several genes involved in the TGFβ pathway are also present in the downregulated genes, including Bmp4, Inhibinβ, Id4, and Pitx2. In the upregulated dataset, KEGG pathways involving antigen processing and presentation and axon guidance are over-represented (Part D of Table 1), suggesting a possible bias in differentiation toward neurogenesis resulting from PI3K inhibition.

A Common Group of PI3K and Nanog Targets

We have previously demonstrated a role for PI3Ks in the maintenance of Nanog levels [38] and so were interested in determining whether some of our PI3K regulated genes are also direct targets of Nanog. Using the VENNY program [47], we investigated the overlap between our PI3K transcriptome and Nanog target genes identified by Loh [13]. Five hundred randomly generated probe sets corresponding to 444 genes had 38 Nanog associations (8.56%), whereas 90 PI3K-regulated genes from 481 (18.71%) were found to have direct associations with Nanog and are listed in supporting information Table 9A. Among the downregulated genes are Lefty2, MRas, Rfx4, Tbx3, Bmp4, Esrrb, Fgf4, Fgf5, Eed, and Klf4, while upregulated genes include Lef1, Myc, Dnmt3b, Zic5, and Smad1. GO analyses revealed only two functional groups, the largest containing transcriptional regulators (enrichment score 4.11) and the second kinases (genes listed in supporting information Table 9B). These data indicate that the set of genes regulated by both Nanog and PI3K-dependent signaling comprise to a large extent regulators of transcription.

Validation of Microarray Expression Data

We have previously reported that inhibition of PI3K-dependent signaling results in downregulation of expression of Nanog, Lefty2, Fry, Silver, and Rfx4 and upregulation of Lef1 and Wt1 [38], validating the changes in expression of these genes detected by microarray. Expression of other genes (see below for selection criteria) is shown in Figure 1C for Gm397 and Shp-1, while AF067061, 1700061G1Rik, LOC32781, and Ypel2 are shown in supporting information Figure S1. In 14 out of 16 cases, highly comparable expression patterns were measured by both microarray and qRT-PCR, while in the other two cases, expression was similar, indicating a high degree of agreement between the two approaches. Figure 1D presents a summary of the temporal events and changes in gene expression observed on inhibition of PI3K signaling in ES cells and compiles data presented here and from our previous work [33, 38].

Identification of Signaling Pathways Downstream of PI3Ks Involved in Regulation of Genes of Interest

PI3K signaling, via protein kinase B, normally suppresses activity of glycogen synthase kinase 3 (GSK-3), and we have demonstrated previously that PI3K-dependent inhibition of GSK-3 activity is involved in the regulation of Nanog gene and protein expression [38]. We were interested in determining whether other genes are regulated by similar mechanisms and so compared the effects of inhibition of PI3Ks alone, GSK-3 alone (using the inhibitor BIO [7], which in essence mimics PI3K activation), and simultaneous inhibition of both PI3Ks and GSK-3. Expression of Gm397, AF067061, LOC327811, and Baz1a (all Cluster one) decreased on inhibition of PI3Ks, whereas inhibition of GSK-3 did not reverse this effect, suggesting regulation of these genes is independent of GSK-3 (Fig. 2). In contrast, inhibition of GSK-3 was able to overcome the effects of PI3K inhibition on expression of Shp-1 and 1700061G19Rik (clusters 11 and 4, respectively), such that their expression returned to the levels seen in the presence of LIF alone, consistent with their expression being dependent on GSK-3. Expression of Ypel2, mapping to cluster 10, was inhibited by either LY294002 or BIO, whereas addition of both together had no additive effect, suggesting a more complex pattern of regulation. Overall, these data suggest that PI3K signaling regulates changes in gene expression via both GSK-3-dependent and independent processes.

Figure 2.

Glycogen synthase kinase 3 (GSK-3)-dependent and independent mechanisms are involved in the control of expression of phosphoinositide 3-kinase target genes. E14tg2a embryonic stem cells were cultured in the presence of LIF plus or minus 5 μM LY294002, plus or minus 2 μM of the GSK-3 inhibitor BIO [7, 48] for 72 hours. Expression of selected genes was analyzed by quantitative reverse transcription polymerase chain reaction, and target gene expression was normalized relative to β-actin levels. The averages and SEM of duplicate samples from each of two independent biological replicates are shown: *, p < .05; **, p < .005; ***, p < .0005, in a Student's t test. Abbreviation: LIF, leukemia inhibitory factor.

Functional Analysis of PI3K-Target Genes in Control of ES Cell Fate

Our earlier studies have demonstrated that PI3K signaling is required for maintenance of expression of Nanog [38], a critical regulator of ES cell pluripotency [9, 49]. We wanted to investigate if other genes, also regulated in a PI3K-dependent manner, contribute to the regulation of ES cell fate, and our primary aim was to identify potentially novel regulators. Genes were selected for further functional analysis based on the following criteria: (a) if expression was at least partly restricted to the early embryo; (b) if they were annotated, but had no assigned function; (c) if they were downregulated 24-48 hours following PI3K inhibition; and (d) encoded potential signaling molecules. Our first strategy was to perform a loss-of-function screen for the genes we selected for further analyses. esiRNAs were generated to our target genes, based on the method of Kittler et al., [41] and were used to knockdown expression in ES cells following transient transfection. Transfectants were assessed for their ability to generate self-renewing or differentiated ES cell colonies in a clonal assay based on alkaline phosphatase staining. Out of 21 genes initially screened, six Gm397, AF067061, Ypel2, Baz1a, Shp-1, and 1700061G19Rik, were selected for more detailed analyses. We performed loss of function studies using both in-house generated esiRNAs and commercially sourced siRNAs targeting each of these genes. Despite achieving greater than 80% reduction in expression of AF067061 and Baz1a using Smartpool siRNAs (Dharmacon), no consistent significant effects on self-renewal in the presence of LIF and serum were observed (data not shown). In contrast, transient knockdown of Gm397 with Smartpool siRNAs led to a consistent and significant reduction in the percentage of self-renewing colonies, compared to nontargetting siRNA, shown in Figure 3A. Similar results were obtained with in-house generated esiRNAs (not shown). A single siRNA targeting the 3′ untranslated region of Gm397 also led to a reduction in alkaline phosphatase positive, self-renewing colonies, although this was less marked than with the siRNA pool. Consistent with this reduction in the number of alkaline phosphatase positive colonies, knockdown of Gm397 also led to reduction in expression of markers of pluripotency, including Nanog, Oct4, and Rex1 (Fig. 3Aii). We also observed that reducing expression of the tyrosine phosphatase, Shp-1, with either esiRNAs (not shown) or Smartpool siRNAs (Fig. 3Bi) consistently led to a significant reduction in ES cell self-renewal. Consistent with these biological effects, concentrations of siRNAs of 20 nM and above led to significantly reduced expression of Shp-1 protein (Fig. 3Bii). Furthermore, using a Nanog-GFP reporter ES cell line, described in [11], knockdown of Gm397 or Shp-1, reduced the proportion of cells expressing high levels of Nanog, consistent with a reduction in maintenance of pluripotency (Fig. 3C).

Figure 3.

siRNA-mediated knockdown of Gm397 or Shp-1 reduces the ability of embryonic stem (ES) cells to self-renew. (A): Smartpool siRNA targeting Gm397 (397pool), a single siRNA targeting the 3′ untranslated region of Gm397 (397 3′) or NTC siRNA, were transiently transfected into E14tg2a ES cells. A concentration of 50 nM for each siRNA was used. Three days post-transfection cells were replated at low density and self-renewal measured after a further 4 days. (i): Photographic images of representative colonies are shown (scale bar = 200 μm) and the percentage of self-renewing, alkaline phosphatase positive colonies presented, with the values corresponding to the average ± SEM of three independent experiments with significance determined using a Student's t test. (ii): Knockdown of Gm397, Rex1, Oct4, and Nanog transcripts was monitored by quantitative reverse transcription polymerase chain reaction and target gene expression was normalized relative to β-actin levels. Values correspond to the average ± SEM from three independent biological experiments, and significance was determined using a Student's t test. *, p < .05; **, p < .01. (B): As for (A) using siRNAs targeting Shp-1 (gray bars), or nontargetting controls (white bars). (i): Photographic images of representative colonies transfected with 50 nM siRNAs are shown (scale bar = 200 μm) and the percentage of self-renewing, alkaline phosphatase positive colonies presented, with the values corresponding to the average ± SEM of three independent experiments with significance determined using a Student's t test. (ii): Knockdown of Shp-1 protein was monitored by immunoblotting extracts with a Shp-1 specific antibody. The immunoblot was reprobed with an anti-GAPDH antibody to assess equivalence of loading. A representative experiment is shown. (C): Effect of siRNA knockdown on expression of GFP in the Nanog-GFP reporter ES cell line was assessed by flow cytometry 5 days after transfection with siRNAs. Values represent average percentage ± SD. of Nanog-GFP-expressing cells (percentage of Nanog-GFP+). Abbreviations: GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GFP, green fluorescent protein; NTC, nontargetting control; siRNA, short interfering RNA.

We further investigated the mechanism of action of Shp-1. Following knockdown of Shp-1, patterns of phosphotyrosyl proteins in basal and LIF stimulated conditions did not noticeably alter (Fig. 4A) and neither did LIF-stimulated activation of MAPK signaling or phosphorylation of S6 ribosomal protein (a read-out of PI3K signaling). However, levels of phosphorylation of Stat3 at Y705 were elevated following Shp-1 knockdown compared with controls (Fig. 4A). We investigated the possibility that Stat3 is a substrate for Shp-1 using ES cells expressing either wild-type Shp-1 or a substrate-trapping form of Shp-1 (C453S Shp-1) under the control of Tet-off regulated expression [39]. Consistent with the results of Shp-1 knockdown, when induced to express C453S Shp-1, phosphorylation of Stat3 at Y705 was maintained at elevated levels following LIF-stimulation and wash-out, compared to uninduced cells (Fig. 4Bi), whereas wild type Shp-1 had little effect (Fig. 4Bii).

Figure 4.

Perturbation of Shp-1 action in embryonic stem (ES) cells alters Stat3 phosphorylation.(A): siRNA-mediated knockdown of Shp-1 was carried out as described in Figure 3. About 72 hours after transfection, cells were washed free of LIF, incubated for a further for 4 hours, and then treated with 104 U/ml LIF for 15 minutes. Cytosolic cell extracts were generated, triplicate blots prepared, and immunoblotting carried out using the antibodies indicated to detect tyrosine-phosphorylated proteins (α-pY), pY705 Stat3 (α-pStat3), phosphorylated and active Erk1/2 (α-pErk) or phosphorylated S6 protein (α-pS6). Blots were stripped and reprobed to show loading with anti-p85 phosphoinositide 3-kinase antibodies (α-p85) and expression of Shp-1 (α-Shp-1). CON, ES cells treated only with Lipofectamine; NTC, ES cells transfected with nontargetting siRNAs and Shp-1, ES cells transfected with Shp-1 siRNAs. (B): ES cell transfectants expressing either (i) a substrate-trapping mutant of Shp-1 (C435S Shp-1) or (ii) wild-type Shp-1, under the regulation of the Tet-off expession system, were induced to express exogenous Shp-1 by removal of Tet for 24 hours. Cells were washed free of LIF, incubated for 4 hours (CON) and then treated for 10 minutes with 104 U/ml LIF. LIF was then washed out and cell extracts prepared at the time intervals indicated after LIF washout. Immunoblotting was performed to detect phosphorylation of Y705 of Stat3 (α-pStat3). Blots were stripped and reprobed to show loading for Stat3 (α-Stat3) and expression of exogenous Shp-1 (α-Shp-1). Abbreviations: LIF, leukemia inhibitory factor; NTC, nontargetting control; siRNA, short interfering RNA.

Gm397 is a Member of the Zscan4 Family of Zinc Finger Proteins

We were particularly interested in the role of Gm397 in control of ES cell fate because it contains a SCAN domain, which mediates protein-protein interactions [50], along with four conserved zinc finger motifs (Fig. 5A). Database interrogation determined that Gm397 belongs to a family of nine predicted genes (supporting information Table 10), data largely confirmed by a paper published simultaneously to our analyses [51]. This family of genes is collectively referred to as the ZScan4 family. For comparison, we have adopted the nomenclature used by Falco et al., [51] thus Gm397 corresponds to Zscan4c. From ES cells, we cloned cDNAs corresponding to Zscan4c and ZScan4f, ZScan4-ps2, and ZScan4-ps3, the latter two classified as pseudogenes by Falco et al., [51], although our data suggest otherwise. Given the high levels of homology, the Affymetrix probe sets that hybridize to Zscan4 sequences are likely to detect all gene members, with the exception of Zscan4b and Zscan4e, while the primers we used for quantitative PCR (supporting information Fig. S2) are located in regions of very high homology, again likely to detect most, if not all, Zscan4 family RNA species. Furthermore, our siRNA knockdown strategies used primers that we predict will target most Zscan4 transcripts in ES cells.

Figure 5.

Gm397 is a member of the Zscan4 family of SCAN-domain-containing Zinc finger proteins.(A): Schematic showing the key structural features of Gm397/Zscan4c. The SCAN-B domain is shown in turquoise, the three highly conserved zinc finger motifs in red and the zinc finger motif of lower homology in purple. (B): Predicted protein sequences of Zscan4 family members cloned from embryonic stem cell cDNAs. In this study, the sequence in black represents Zscan4c/Gm397. Changes in the predicted amino acid sequences encoded by the other Zscan4 genes cloned are shown in the appropriate positions. The SCAN-B homology domain (aa 9-163) is underlined, with core residues (aa 53-87) indicated by bold typeface. Residues that share >51% identity with the SCAN consensus sequence are shaded in turquoise, while conserved amino acid differences are shaded pink. The three zinc-finger motifs (aa 396-418; 453-475 and 480-503, with pfam E-values of 9.8 × 10−6, 5.7 × 10−6, and 1.4 × 10−4) are highlighted in bold italic script, with conserved residues in red. The zinc finger motif with lower homology (aa 425-447) is in italics with conserved residues in blue.

Expression of Zscan4c Enhances ES Cell Self-Renewal

To complement our loss of function studies, we examined the effect of overexpression of a V5-His-epitope-tagged version of Zscan4c on ES cell behavior. A number of independent clones expressing Zscan4c-His-V5, as well as vector only controls, were generated and expression of Zscan4c-His-V5 verified by immunoblotting and immunocytochemistry (Fig. 6A). In the presence of optimal and suboptimal levels of LIF, the clone expressing the highest levels of Zscan4c-His-V5 (clone 17) generated a higher proportion of undifferentiated colonies than did the control clones (Fig. 6Bi). When self-renewal of all Zscan4c clones was examined, we found there was a good relationship between the level of Zscan4c expression and enhancement of the proportion of alkaline positive colonies at suboptimal doses of LIF (Fig. 6Bii). These results are consistent with Zscan4c playing a role in maintenance of ES cell self-renewal, which appears dependent on Zscan4c expression levels.

Figure 6.

Overexpression of Zscan4c enhances embryonic stem (ES) cell self-renewal. E14 ES cells were transfected with pcDNA3.1neo encoding a V5-His epitope-tagged version of Zscan4c, or empty vector as a control. (A): Independent clones were selected and expression of Zscan4c in nuclear extracts monitored by immunoblotting with anti-V5 epitope tag antibodies ([i], upper panel) and reprobed with an anti-SHP-2 antibody ([i], lower panel). Immunohistochemical staining with anti-V5 antibodies detects Zscan4c in cytosol and nuclei only in Zscan4-expressing ES cell clones ([ii], Zscan4c C17 and control C7 shown). (B): The effect of Zscan4c expression on ES cell self-renewal was examined for all clones, as well as control clones, at different doses of LIF, as indicated. Alkaline positive (self-renewing) colonies were scored. (i): Data for the highest expressing Zscan4 clone C17 show mean ± SEM (n = 6). (ii): Data for all Zscan4c clones in comparison to control clone seven and nonexpressing clone 18 are shown at the suboptimal dose of 10 U/ml. Abbreviations: C, clone; LIF, leukemia inhibitory factor.


Here, we report the characterization of the PI3K-dependent transcriptome in murine ES cells. We [33] and others [34–37] have previously defined a role for PI3K-dependent signaling in optimal maintenance of ES cell self-renewal; hence, definition of the gene expression patterns that are regulated via PI3K pathways provides new insight into the molecular mechanisms underlying maintenance of ES cell pluripotency.

Strikingly, the most over-represented functional GO groups in both down and upregulated datasets were transcriptional regulators (Table 1). It has become widely accepted that a core circuit of transcription factors is important for maintenance of ES cell pluripotency [8, 12–14]. Interestingly, of the three central regulators, Oct4, Sox2, and Nanog, only Nanog appears to be sensitive to PI3K signaling, consistent with our previous report [38], indicating that PI3K signaling is involved in regulating specific components of the pluripotency network. It is interesting that inhibition of PI3K signaling also leads to decreased expression of a number of other transcription factors reported to be involved in control of pluripotency, including Klf4 [20, 21], Tbx-3, and Esrrb [15], which are also Nanog targets [13]. We also find it notable that Tcl1 expression is PI3K sensitive, although not a Nanog target [13], because Tcl1 functions as a coactivator of protein kinase B [52], one of the major downstream regulators of PI3K signaling. This finding suggests the presence of a regulatory feedback loop and provides a rationale to explain why loss of Tcl1 reduces ES cell self-renewal [15].

k-means nonhierarchical clusters defined subsets of genes showing similar patterns of expression. Genes in cluster 1 appear to be regulated independently of GSK-3, while expression of others, including Nanog [38], Shp-1, and 1700061G19Rik is dependent on GSK-3. This is of particular interest given the discovery that inhibition of GSK-3 is required for the ground state of ES cell pluripotency [53], along with work from our group and others demonstrating enhancement in self-renewal following inhibition of GSK-3 activity [54]. These observations indicate that PI3Ks influence self-renewal of ES cells via at least two complimentary mechanisms.

We have demonstrated that loss of expression of the PI3K-target genes Shp-1 and Gm397 reduced the ability of mESC to self-renew. Shp-1 is a protein tyrosine phosphatase, highly expressed in hemopoietic cells, which we have reported to be recruited to phosphotyrosyl proteins in ES cells [39]. Shp-1 is largely regarded as a negative regulator of cell signaling processes [55, 56]; however, the mechanism by which Shp-1 regulates ES cell pluripotency is unknown and while substrates of Shp-1 have been identified in specific cell types, none have been defined in ES cells. Following knockdown of Shp-1, or upon expression, the C453S Shp-1 substrate-trapping mutant, we were somewhat surprised to observe an increase in phosphorylation of Y705 of Stat3, consistent with Stat3 being an Shp-1 substrate. We were unable to detect consistent alterations in other signaling pathways implicated in the control of pluripotency, including MAPK and PI3K signaling. Low endogenous levels of leukaemia inhibitory factor receptor and gp130 on ES cells precluded detailed analysis of effects on receptor phosphorylation, although overall patterns of tyrosine phosphorylation did not appreciably alter. These results are somewhat paradoxical considering that knockdown of Shp-1 decreases ES cell self-renewal. Shp-1 may be regulating alternate pathways involved in control of pluripotency, and clearly, further analyses are required to determine the mechanism underlying Shp-1-dependent control of ES cell self-renewal.

We were particularly interested in Gm397 because (a) it was among a small cluster of probe sets that detect a decrease in expression within 24 hours of PI3K inhibition (Fig. 2) and (b) its pattern of expression was restricted to the early preimplantation embryo and ES cells [17, 51]. We performed extensive bioinformatics analyses that defined Gm397 as a member of the closely related Zscan4 family of genes (supporting information Fig. 1 and Table 10), consistent with a recent report [51]. Gm397/Zscan4c has many of the hallmarks of a transcriptional regulator, four zinc finger motifs and a SCAN domain, which mediates dimerization [57–60]. Our loss and gain of function studies support a role for Zscan4 members in regulation of ES cell pluripotency and the fact that Zscan4 genes do not show any genomic associations with Nanog [13] suggests this as an alternative mechanism by which PI3Ks influence pluripotency. Interestingly, Zfp206, another SCAN-domain Zfp, regulates ES cell pluripotency and controls expression of a select group of genes which includes the Zscan4 family, LOC327811 and AF067061 [17], all present in cluster one of our dataset (supporting information Table 3). Falco et al., reported high levels of expression of Zscan4d in late two-cell stage embryos, while Zscan4c and ZScan4f were expressed most highly in ES cells [51]. Interestingly, knockdown of ZScan4 transcripts during preimplantation development delayed progression from the two-cell to the four-cell stage and produced blastocysts that could not implant or proliferate in outgrowth culture [51]. This latter finding implicates Zscan4 function in the derivation of ES cells from the inner cell mass, consistent with our data supporting a role for Zscan4 in self-renewal. An interesting additional feature is the heterogeneity of Zcan4 expression reported in ES cells, shown by RNA in situ hybridization [51, 61], but not yet confirmed at the protein level. While at face value expression of Zscan4 by only a proportion of ES cells may seem at odds with our functional data supporting a role in ES cell pluripotency, given recent reports that a number of regulators, including Nanog [11], show heterogeneity of expression in cultures of ES cells, it is interesting to speculate that Zscan4 expression may specify an ES cell population with particular properties or is regulated in a cell cycle-dependent manner. Further studies are clearly required to reveal the mechanism of action of Zscan4 in ES cells.


In conclusion, we have characterized the profile of genes regulated by PI3K-dependent signaling in undifferentiated murine ES cells. Our analyses reveal over-representation of transcriptional regulators in both up and downregulated genes datasets and suggest PI3Ks regulate self-renewal via at least two complimentary mechanisms. Functional analyses define a role for the tyrosine phosphatase Shp-1 and the Zscan4 family member Gm397 in regulation of ES cell self-renewal.


This work was supported by grants from the European Community sixth Framework program, contract no. FunGenES LSHG-CT-2003-503494, Marie-Curie Actions and Biotechnology and Biological Sciences Research Council. We thank Dr. Ian Chambers, Institute of Stem Cell Research, University of Edinburgh, for the Nanog-GFP reporter embryonic stem cells.


The authors indicate no potential conflicts of interest.