Previous studies have demonstrated that the transcription factor Sox2 is essential during the early stages of development. Furthermore, decreasing the expression of Sox2 severely interferes with the self-renewal and pluripotency of embryonic stem (ES) cells. Other studies have shown that Sox2, in conjunction with the transcription factor Oct-3/4, stimulates its own transcription as well as the expression of a growing list of genes (Sox2:Oct-3/4 target genes) that require the cooperative action of Sox2 and Oct-3/4. Remarkably, recent studies have shown that overexpression of Sox2 decreases expression of its own gene, as well as four other Sox2:Oct-3/4 target genes (Oct-3/4, Nanog, Fgf-4, and Utf1). This finding led to the prediction that overexpression of Sox2 in ES cells would trigger their differentiation. In the current study, we initially engineered mouse ES cells for inducible overexpression of Sox2. Using this model system, we demonstrate that small increases (twofold or less) in Sox2 protein trigger the differentiation of ES cells into cells that exhibit markers for a wide range of differentiated cell types, including neuroectoderm, mesoderm, and trophectoderm but not endoderm. We also demonstrate that elevating the levels of Sox2 quickly downregulates several developmentally regulated genes, including Nanog, and a newly identified Sox2:Oct-3/4 target gene, Lefty1. Together, these data argue that the self-renewal of ES cells requires that Sox2 levels be maintained within narrow limits. Thus, Sox2 appears to function as a molecular rheostat that controls the expression of a critical set of embryonic genes, as well as the self-renewal and differentiation of ES cells.
Disclosure of potential conflicts of interest is found at the end of this article.
Recent studies have begun to identify gene regulatory networks involved in maintaining the self-renewal and pluripotency of embryonic stem (ES) cells [1, 2]. One of these networks is controlled, in part, by the transcription factors Sox2 and Oct-3/4, which behave as master regulators during mammalian embryogenesis [3, 4]. Sox2 and Oct-3/4 have been shown to physically interact and to cooperatively regulate their own expression by binding to adjacent high-mobility-group (HMG) and pituitary octamer unc (POU) motifs (collectively known as HMG/POU cassettes) within enhancers required for the transcription of each gene [5, 6]. Moreover, Sox2 and Oct-3/4 have been implicated directly in the transcription of a growing list of other essential Sox2:Oct-3/4 target genes that also possess HMG/POU cassettes [7, , , , , –13]. Although it is widely believed that Sox2 and Oct-3/4 positively regulate Sox2:Oct-3/4 target genes [5, , , , , , , –13], recent studies have shown that overexpression of Sox2 in ES cells and embryonal carcinoma (EC) cells results in decreased expression of its own gene, as well as decreased expression of several other Sox2:Oct-3/4 target genes, including the Oct-3/4, Nanog, and Fgf-4 genes . Interestingly, overexpression of Oct-3/4 inhibits its own promoter, but, unlike Sox2, Oct-3/4 does not appear to globally inhibit this set of genes.
Previous work demonstrated that small increases or decreases in Oct-3/4 expression promote the differentiation of ES cells into extraembryonic endoderm and mesoderm or trophectoderm, respectively . Other studies have shown that reducing the level of Sox2 promotes the differentiation of ES cells into trophectoderm-like cells . Thus, it appears that the self-renewal of ES cells requires the maintenance of master regulators, in particular Oct-3/4 and Sox2, within narrow limits by mechanisms that carefully titrate their expression. Nonetheless, the effect of increasing Sox2 expression on the fate of ES cells is poorly understood. Given the critical role of Sox2 in regulating genes required for the self-renewal and pluripotency of ES cells, we predicted that overexpression of Sox2 would reduce the expression of one or more essential Sox2:Oct-3/4 target genes and trigger ES cell differentiation. To test this prediction, we engineered mouse ES cells for inducible overexpression of Sox2. Our studies indicate that small increases in the level of Sox2 promote the differentiation of ES cells into cells that express markers associated with a wide range of differentiated cell types.
KH2 ES cells and inducible Flag-Sox2 KH2 ES cell clones were maintained in Dulbecco's modified Eagle's medium containing 15% FBS supplemented with 100 μM β-mercaptoethanol and 10 ng/ml leukemia inhibitory factor (LIF) (Chemicon, Temecula, CA, http://www.chemicon.com) on gelatin-coated tissue culture plastic. Stock cultures and all experimental cultures were maintained at 37°C in a moist atmosphere of 95% air and 5% CO2.
The pBS31 and pCAGGS-FLPe-puro vectors were obtained from Dr. R. Jaenisch . The cytomegalovirus (CMV)-Sox2F vector was described previously . The Flag-Sox2-EcoRI vector was created by inserting an EcoRI site downstream of the Flag-Sox2 coding region in the CMV-Sox2F vector by site-directed mutagenesis using the primer pair 5′ GGGCTGGACTGCGAATTCGAGAAGGGGAG 3′ and its complement (the mutation is indicated by boldface type). The pBS31-Flag-Sox2 vector (also referred to as pgkATGfrt-Flag-Sox2) was created by inserting an EcoRI fragment, which contained the coding region for Flag-Sox2, from the Flag-Sox2-EcoRI vector into the unique EcoRI site of the pBS31 vector . This placed the coding sequence of Flag-Sox2 downstream of the minimal CMV tetracycline-inducible promoter and upstream of the rabbit β-globin polyadenylation signal sequence.
Generation of ES Cells with Inducible Flag-Sox2 Transgene Expression
KH2 ES cells obtained from Open Biosystems (Huntsville, AL, http://www.openbiosystems.com) were used to insert a single copy of the Flag-Sox2 transgene by Flipase (FLP) recombination into the collagen 1A1 locus under the control of a minimal CMV tetracycline-inducible promoter using a method described previously . Treatment and selection of Flag-Sox2 ES cells were as follows: approximately 1.25 × 107 KH2 ES cells were electroporated with 25 μg of pBS31-Flag-Sox2 and 40 μg of an expression vector for the FLP recombinase (pCAGGS-FLPe-puro ). Properly integrated genes were inserted via the FLP recognition target (FRT) sites in the collagen 1A1 locus and replaced the neomycin-resistance cassette with the pBS31-Flag-Sox vector, which contains a 3-phosphoglycerate kinase (PGK) promoter and ATG start codon to initiate the expression of the promoterless and ATG-less hygromycin-resistance cassette present in the collagen 1A1 locus of KH2 ES cells. Thus, positive clones are hygromycin-resistant and neomycin-sensitive. Twenty-four hours after electroporation and plating, colonies were selected for hygromycin resistance by the addition of 50 μg/ml hygromycin B. Ten days later, colonies were picked and expanded in the absence of hygromycin and genotyped positively and negatively by polymerase chain reaction (PCR). Several pure Flag-Sox2 ES clones were frozen and analyzed, and they behaved in a similar manner in multiple experimental conditions.
Genomic DNA was isolated using a high-salt method (adapted from [19, 20]) from hygromycin-selected Flag-Sox2 ES cell colonies. Primer pairs were designed to amplify parental sequences, as well as correctly targeted sequences. The primer pair that detects parental sequences is forward coll1A1 primer 5′ TTTTCGATCAGAAACTTCTCG 3′ and reverse PGK promoter primer 5′ GTTCTCCTCTTCCTCATCTCC 3′. These primers will amplify an ∼250-base pair PCR product. To detect the correctly recombined sequences, we used the same forward coll1A1′ primer described above, as well as a reverse primer specific to the minimal CMV tetracycline-responsive promoter, 5′ CGAGCTCTGCTTATATAGGCC 3′. This primer pair will amplify a 1.1-kilobase pair product. We combined the usage of these two primer sets to detect positive Flag-Sox2 ES cell clones.
Extract Preparation and Western Blotting
Nuclear extracts from untreated and doxycycline-treated Flag-Sox2 ES cells were prepared using the Pierce NE-PER nuclear and cytoplasmic extraction kit, as described in the manufacturer's protocol (Pierce, Rockford, IL, http://www.piercenet.com). Extracts were supplemented with protease and phosphatase inhibitors and stored at −80°C. Bicinchoninic acid analysis (Pierce) was performed to quantitate the level of protein in each sample, and equal amounts of protein were separated on a 4%–20% Tris-HEPES SDS-polyacrylamide gel electrophoresis (PAGE) gel. Western blot analysis was performed with an anti-Sox2 antibody (AB5603; Chemicon) at a 1 μg/ml concentration with an alkaline phosphatase-conjugated goat α-rabbit secondary antibody (1:1,000). Endogenous Sox2 migrated as a ∼35-kDa protein (as described previously ), and Flag-Sox2 migrated at a slower rate because of the highly acidic Flag epitope. The alkaline phosphatase-conjugated antibodies were detected using the enhanced chemifluorescence kit (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) and scanned on a Storm PhosphorImager (Molecular Dynamics, Sunnyvale, CA, http://www.mdyn.com). Quantitation was performed using the ImageQuant 5.0 analysis software (Molecular Dynamics).
Colony Forming Assay
Flag-Sox2 ES cells or parental KH2 ES cells were plated in the presence or absence of doxycycline at low density (950 cells per 60-mm dish). After 108 hours, cells were fixed and stained with Coomassie Blue reagent (6:3:1 solution of water, isopropanol, and acetic acid, respectively, with 0.5 mg/ml Coomassie Blue). The ability to form colonies was assessed by estimating the number of total colonies and scoring them according to morphology (ES-like, mixed, or differentiated). Cells were photographed using an Olympus light microscope (Olympus, Tokyo, http://www.olympus-global.com) with a PA1–10A adapter attached to a 10-megapixel Canon EOS Rebel XTi digital camera (Canon, Tokyo, http://www.canon.com). The field of view at ×40 magnification was 1 mm. The size of the photographs was scaled accordingly.
Cell Cycle Analysis
H2 Flag-Sox 2 ES cells were plated at 2 × 105 cells per 60-mm dish in 3 ml of medium. After 24 hours, cells were refed, and, where indicated, 4 μg/ml doxycycline was added to the medium. After the appropriate time point, cells were prepared for cell cycle analysis. Cells were trypsinized with 0.1% trypsin and allowed to dislodge from the dish. The cells were counted, centrifuged at 2,800 rpm for 5 minutes, and resuspended in Vindelov's reagent  at a density of 1 × 106 cells per milliliter. Cells were placed at 4°C in darkness for 1–2 hours prior to cell cycle analysis. Cells were analyzed via flow cytometry by the University of Nebraska Medical Center Flow Analysis Core facility (Omaha, NE) using a FACSCalibur (Becton, Dickinson and Company, San Jose, CA, http://www.bd.com) instrument to detect propidium iodide.
RNA Isolation and cDNA Synthesis
Flag-Sox2 ES cells were grown for 9 and 72 hours in the presence or absence of 4 μg/ml doxycycline. RNA was isolated by Tri-Reagent (Molecular Research Center, Inc., Cincinnati, http://www.mrcgene.com) and 1-bromo-3-chloropropane phase separation according to the manufacturer's protocol with the exception of increasing the precipitation step to overnight at −20°C and performing a second precipitation with ethanol-sodium acetate before washing with ethanol and drying the RNA pellet. RNA pellets were resuspended in 200 μl of HPLC H2O. The concentration of RNA was determined by UV spectrophotometry. One-half microgram of RNA was treated with amplification-grade DNase I (Invitrogen). Next, cDNA was synthesized using the SuperScript III First-Strand Synthesis SuperMix (Invitrogen).
Quantitative Polymerase Chain Reaction
The cDNA generated from Flag-Sox2 ES cells grown in the absence or presence of doxycycline were subjected to SYBR Green quantitative polymerase chain reaction (qPCR) on the Cepheid SmartCycler using software version 2.0c (Sunnyvale, CA, http://www.cepheid.com). Gene expression was assayed using RT2 Real-Time SYBR Green PCR master mix from SuperArray Bioscience Corporation (Frederick, MD, http://www.superarray.com) according to the manufacturer's protocol using previously described gene-specific primers for Fgf-4, Utf1, Nanog, and Oct-3/4 . Gene-specific primer sequences were also obtained from PrimerBank  or manually designed for the genes shown in supplemental online Table 1. Relative gene expression, for each gene in the untreated and doxycycline-treated Flag-Sox2 ES cells, was normalized to glyceraldehyde-3-phosphate dehydrogenase. Gene expression for doxycycline-treated cells is reported as the average difference in CT value (cycle number at which significant increase in the fluorescence is detected above the background value) relative to expression of the gene in the control (untreated) cells.
Engineering Mouse ES Cells for Inducible Expression of Sox2
To test the prediction that small increases in the levels of Sox2 would trigger the differentiation of ES cells, we generated mouse ES cells in which Sox2 levels could be elevated over a relatively short time frame with the aid of an inducible, Tet-On promoter. We initiated these studies with KH2 ES cells, which can be readily engineered for stable integration of a single transgene into a predetermined site . KH2 ES cells express the M2 tetracycline transactivator (M2-rtTA) under the control of the ROSA26 locus and possess a transgene inserted into the collagen 1A1 locus. The transgene in the collagen 1A1 locus contains a PGK-driven neomycin resistance gene cassette flanked by FLP recombinase sites, plus a hygromycin resistance gene, which lacks an ATG start codon and promoter. We transfected KH2 ES cells by electroporation with an expression vector for the FLP recombinase and a vector containing the Flag-Sox2 transgene under the control of a tetracycline minimal CMV promoter flanked by FLP recombinase sites (Fig. 1A). Under these conditions, the FLP recombinase excises the parental KH2 neomycin-resistance cassette flanked by FLP recombination sites through the facilitation of recombination with the electroporated Flag-Sox2 plasmid via its FLP recombination sites. Successful replacement of the neomycin resistance gene cassette with the Flag-Sox2 gene cassette also inserts the PGK promoter and an ATG start codon in front of the promoterless, ATG-less hygromycin-resistance cassette. This creates a hygromycin-resistant, G418-sensitive cell. We isolated three positive Flag-Sox2 ES cell clones (H2, F6, and G3), which were identified by PCR genotyping (Fig. 1A). All Flag-Sox2 ES cell clones exhibited a morphology indistinguishable from that of the parental KH2 ES cells when cultured on gelatin-coated plates in the presence of LIF (described below) or when plated on irradiated mouse embryonic fibroblasts (data not shown). Furthermore, they grew at rates typical of mouse ES cells.
Increasing Sox2 Levels Triggers Differentiation and Loss of ES Cell Characteristics
We initially determined that Flag-Sox2 was expressed when the F6, G3, and H2 Flag-Sox2 ES cell clones were exposed to doxycycline (Fig. 1B). Flag-Sox2 protein migrates at a slightly slower rate because of the highly acidic nature of the Flag peptide. Hence, we were able to identify both the exogenous and endogenous forms of Sox2 protein with a Sox2 antibody. We also determined that the expression of Flag-Sox2 increased in response to doxycycline in a dose-dependent manner (Fig. 1C). Importantly, exposure of Flag-Sox2 ES cells to 4 μg/ml doxycycline for 24 hours resulted in moderate increases (approximately twofold) in Sox2 expression in multiple experiments. Higher levels of Flag-Sox2 protein were induced when the cells were exposed to 8 and 16 μg/ml of doxycycline; however, higher levels of Flag-Sox2 were detrimental to the survival of ES cells (data not shown). Therefore, in subsequent experiments we used 4 μg/ml doxycycline for induction of Flag-Sox2 transgene expression. At this concentration of doxycycline, the induction of Flag-Sox2 remained below a level of threefold (relative to Sox2 in the uninduced cells) even after 3 days, which was the longest time used in the vast majority of our studies. Moreover, the level at which Flag-Sox2 could be induced by 4 μg/ml of doxycycline did not vary significantly (typically twofold) with increasing passage of the Flag-Sox2 ES cells (e.g., passages 5 and 18) (data not shown).
As predicted by our earlier studies , overexpression of Sox2 in ES cells triggered the differentiation of ES cells, even in the presence of LIF. In fact, exposure to doxycycline for 72 hours, which elevated Sox2 protein levels just under twofold, induced the formation of cells that exhibited several different cell morphologies typical of differentiated cells (Fig. 2). More specifically, the Flag-Sox2 ES cell clones treated with doxycycline displayed an increase in cytoplasmic to nuclear ratio, flattened morphology, and visually reduced cell numbers (described below), whereas only approximately 1% of the untreated Flag-Sox2 ES cells displayed these morphological properties. No such change was observed when the parental KH2 cells were exposed to doxycycline (Fig. 2). To determine whether the morphological changes were reversible, the Flag-Sox2 ES cells were exposed to doxycycline for 2 days, and then individual cellular clusters were followed for an additional 2 days in the absence (supplemental online Fig. 1, top) or presence (supplemental online Fig. 1, bottom) of doxycycline. Importantly, removal of the doxycycline did not cause the cells to revert to the morphology of ES cells, and they continued to exhibit the morphology of cells continuously exposed to doxycycline (supplemental online Fig. 1, bottom). These and other data discussed below argue that ES cells permanently differentiate in response to increasing Sox2 protein levels.
To determine the extent to which elevating Sox2 induces ES cells to differentiate, we compared the clonal growth of ES cells in the presence and absence of doxycycline. Clonal growth of ES cells provides a relatively simple means for determining the self-renewal capacity of ES cells. When plated at clonal densities, ES cells give rise primarily to colonies that exhibit an ES cell-like morphology (Fig. 3A, left). However, at these densities, ES cells also produce a smaller number of mixed colonies, which consist of ES cells and differentiated cells (Fig. 3A, center), and colonies consisting only of differentiated cells (Fig. 3A, right). Thus, we compared H2 Flag-Sox2 ES cells and the parental KH2 ES cells at clonal density in the presence or absence of doxycycline. After 5 days, the colonies that formed were fixed and stained with Coomassie Blue. The total number of colonies formed was counted, and the number of colonies formed by the untreated parental KH2 ES cell line was arbitrarily set to 1. We determined that increased expression of Flag-Sox2 in H2 ES cells reduced the number of colonies formed (Fig. 3B). More importantly, H2 Flag-Sox2 ES cells treated with doxycycline formed colonies containing differentiated cells far more frequently than the uninduced H2 Flag-Sox2 ES cells or the parental KH2 ES cells in the presence or absence of doxycycline (Fig. 3C). Furthermore, the percentage of colonies exhibiting the ES cell morphology was reduced from greater than 80% to less than 15% when the H2 Flag-Sox2 ES cells were cultured in the presence of doxycycline. This indicates that ES cell self-renewal is dramatically reduced when Sox2 protein levels are increased.
To confirm our earlier visual observation that increasing the level of Sox2 reduces the number of cells formed over time, we determined the number of cells produced by ES cells cultured in the presence or absence of doxycycline for 24, 48, and 72 hours. At 24 and 48 hours, little difference was observed between doxycycline-treated and control populations. However, after 72 hours, the cell number in the doxycycline-treated population was reduced more than twofold (supplemental online Fig. 2A). Furthermore, although ES cells typically have a small percentage of cells in the G0-G1 compartment of the cell cycle, cells exposed to doxycycline for 72 hours exhibited an increase in the number of cells in the G0-G1 compartment (supplemental online Fig. 2B). Taken together, our data indicate that increasing the concentration of Sox2 induces ES cells to differentiate and to lose their ability to self-renew.
Increasing Sox2 Levels Directs Cell Fate
Previous studies with another pluripotency-associated transcription factor, Oct-3/4, demonstrated that increasing its expression in ES cells triggers differentiation into extraembryonic endoderm and mesoderm . To help determine how cell fate is affected by increased Sox2 expression, we initially performed DNA microarray analysis to provide a preliminary view of how gene expression is altered when Sox2 is overexpressed in H2 Flag-Sox2 ES cells. Using a DNA microarray that contained somewhat more than 10,000 genes, we determined that approximately 50% of the genes represented on the microarray were expressed in H2 Flag-Sox2 ES cells grown in the absence of doxycycline. Interestingly, approximately 2% of the expressed genes appeared to exhibit a change in expression of twofold or greater when the cells were treated with doxycycline for 72 hours. More specifically, approximately 100 genes appeared to exhibit an increase of twofold or greater, whereas approximately 25 genes exhibited a reduction of twofold or greater (supplemental online Table 2). To verify the validity of changes observed by DNA microarray analysis, we performed qPCR for more than 25 genes. With the exception of the Oct-3/4 and Utf1 genes, qPCR was performed on genes that appeared to be differentially expressed twofold or greater in our microarray analysis and are considered markers for ES cells or various embryonic lineages. Importantly, qPCR data demonstrated that a wide-range of genes that are elevated 72 hours after the exposure to doxycycline are associated with cells of the ectodermal (Pax6, Otx2, Sox21, Neuropilin, Six6, and crystallin mu), mesodermal (brachyury, vimentin, and Flk-1), and extraembryonic lineages (Cdx2, Cdh3, and Esx1) (Fig. 4). Interestingly, DNA microarray analysis did not identify changes in the expression of genes associated with embryonic or extraembryonic endodermal lineages, and this was consistent with measurement of Gata-6, Gata-4, Sox17, and Tcf2 by qPCR (Fig. 4). In contrast, the expression of genes associated with ES cells, such as Utf1, Fgf-4, Nanog, and endogenous Sox2 gene, decreased with increased expression of Flag-Sox2, as expected (Fig. 4). Surprisingly, Oct-3/4 mRNA exhibited little or no change in expression. The lack of change in Oct-3/4 expression at this time point may be due, in part, to the residual ES cell population and/or elevated expression of Oct-3/4 in some of the differentiated cells . Taken together, these results suggest that elevating Sox2 levels directs ES cell differentiation into several embryonic lineages.
Increasing Sox2 Levels in ES Cells Downregulates Sox2:Oct-3/4 Target Genes
Many of the morphological and cellular changes induced by the increased expression of Flag-Sox2 are readily apparent at 48 hours (supplemental online Fig. 1). However, changes in gene expression that trigger these effects are likely to occur at an earlier time point. In this regard, we previously demonstrated that overexpression of Sox2 in mouse ES cells decreases the expression of several Sox2:Oct-3/4 target genes within 24 hours . This suggested that increasing the levels of Sox2, and thus limiting the expression of several Sox2:Oct-3/4 target genes that are critical for ES cell pluripotency, may trigger the differentiation of ES cells. This led us to examine the expression of several known Sox2:Oct-3/4 target genes at 9 hours by qPCR, which is approximately 3 hours after Flag-Sox2 was first detected (data not shown). At 9 hours after the addition of doxycycline, the expressions of Nanog, FGF-4, Utf1, and the endogenous Sox2 gene were substantially reduced (Fig. 5). Although a small decrease in Oct-3/4 mRNA was evident at this early time point, little or no change in Oct-3/4 mRNA was observed at 16 or 24 hours by qPCR, and this was verified by Western blot analysis of nuclear extracts at 24 hours using an antibody that recognizes the first 134 amino acids of Oct-3/4 (data not shown). Remarkably, DNA microarray analysis indicated that the recently identified Sox2:Oct-3/4 target gene, Lefty1 , was also reduced when Sox2 levels are increased in ES cells; this was confirmed by qPCR at both the 72-hour (Fig. 4) and 9-hour (Fig. 5) time points. Finally, we assessed at the 9-hour time point the effect of increasing Sox2 on the expression of genes associated with ectoderm, endoderm, mesoderm, and extraembryonic lineages (Fig. 5). The expression of genes associated with trophectoderm was not detected at this early time point. However, we observed increases in the expression of several genes associated with ectoderm, in particular Sox21, and mesoderm, but not for endoderm.
The studies described in this report were initiated by engineering ES cells for inducible overexpression of Sox2. Using this model system, we determined that moderate increases in the levels of Sox2 (less than twofold) reduce the self-renewal of ES cells and promote their differentiation. Moreover, our studies argue that elevating the levels of Sox2 in ES cells influences the direction of their differentiation. A major question posed by these findings is the mechanism(s) by which elevated levels of Sox2 triggers the differentiation of ES cells. Recent studies in this laboratory demonstrated that overexpression of Sox2 in EC cells and ES cells reduces the promoter activity of five Sox2:Oct-3/4 target genes . Moreover, transient increases in the expression of Sox2 in F9 EC cells were shown to reduce the endogenous expression of Sox2 and Oct-3/4 mRNA, plus the mRNA for Nanog and two other Sox2:Oct-3/4 target genes, Fgf-4 and Utf1. In view of the requirement for Sox2, Oct-3/4, and Nanog in the self-renewal of ES cells, we suspect that even transient decreases in expression of one or more of these genes may be sufficient to trigger differentiation. However, given recent reports that Sox2 is associated with more than 1,200 genes in ES cells [1, 2], it will be a major challenge to identify the genes responsible for Sox2-induced differentiation. Nonetheless, Nanog and/or other pluripotency-associated genes appear to be likely targets. In this regard, we observed a reduction of more than 50% in Nanog mRNA at both 9 and 72 hours after doxycycline-induced overexpression of Sox2. Surprisingly, at most time points examined, there was little or no change in Oct-3/4 mRNA levels. The reason for this remains to be determined. It is possible that elevated expression of Sox2 and continued expression of Oct-3/4 may be sufficient to sustain the transcription of the Oct-3/4 gene [5, 6]. However, given that the Oct-3/4 protein present in nuclear extracts in our cultures 24 hours after the addition of doxycycline migrated at the proper size on SDS-PAGE, it is unlikely that we are detecting the expression of an Oct-3/4 pseudogene or an Oct-3/4-related gene as described recently for somatic cells [25, 26]. Interestingly, expression of the Oct-3/4 gene has been reported in neural stem cells [27, 28]. Thus, expression of the Oct-3/4 gene in our differentiated cultures, in particular by the neuroectoderm-like cells, may be due to the use of a different enhancer of the Oct-3/4 gene, which is not dependent on Sox2 or Oct-3/4 .
Previous studies demonstrated that reducing the levels of either Sox2 or Oct-3/4 promotes the differentiation in ES cells into trophectoderm-like cells [15, 16, 30], whereas increasing the levels of Oct-3/4 in ES cells induces their differentiation to mesoderm-like and endoderm-like cells . In this study, we show that elevating the levels of Sox2 by twofold or less is also sufficient to induce ES cell differentiation. Thus, Sox2 and Oct-3/4 must each be maintained within narrow limits for ES cells to retain the ability to self-renew. However, overexpression of Sox2 and Oct-3/4 exerts different effects on the fate of ES cells. Elevating the levels of Oct-3/4 in ES cells induces their differentiation into cells that exhibit markers associated with extraembryonic endoderm or mesoderm . In contrast, overexpression of Sox2 promotes ES cells to differentiate into cells that express a wide range of markers associated with neuroectoderm and mesoderm, as well as markers for trophectoderm cells. However, we did not detect significant expression of genes associated with either embryonic or extraembryonic endoderm. At this time it is unclear why markers for endoderm were not detected. It is possible that our culture conditions did not favor the expansion and/or survival of endoderm-like cells. However, this is unlikely because treatment of H2 Flag-Sox2 ES cells for 72 hours with retinoic acid in essentially the same culture medium, but without doxycycline, increased the expression of EndoA, a marker of endodermal lineages (data not shown). Thus, we suspect that Sox2, in particular elevated levels of Sox2, directs differentiation of ES cells into cell types other than endoderm cells and/or actively blocks the expression of genes needed for the formation of endoderm. Future studies will be needed to clarify this issue and to determine whether removal of doxycycline at different time points after differentiation has been induced alters the pattern of differentiated cells that forms.
In two important respects, the findings described in this report differ from those of earlier studies that examined the overexpression of Sox2 in ES cells. In one report, it was noted (data not shown) that elevating Sox2 in ES cells, by means of a retrovirus, caused massive cell death . In our studies, we did not observe significant cell death, provided that Sox2 levels were not elevated more than twofold higher than its endogenous level. However, when Sox2 levels were elevated fourfold or higher above its endogenous level, by using higher concentrations of doxycycline, we observed significant cell death. Thus, there appear to be narrow limits of Sox2 protein that can be tolerated by ES cells. In a second report, Zhao et al. claimed that overexpression of Sox2 did not alter the ability of ES cells to self-renew . In that study, ES cells were stably transfected with a Sox2 expression vector. Although Sox2 mRNA levels were reported to be elevated 2–20-fold above that of endogenous Sox2 mRNA, Sox2 protein levels were not determined. These results contrast with our finding that moderate overexpression of Sox2 reduces the capacity of ES cells to self-renew. In view of the toxic effects noted above when Sox2 is overexpressed at high levels in ES cells, we suspect that the stable clones described in the study by Zhao et al. were generated from variant cells in the transfected ES cell population, which could tolerate high levels of Sox2 . Nonetheless, there is one notable similarity between our findings when Sox2 was elevated approximately twofold and the stably transfected Sox2 cells. When the stably transfected Sox2 cells were cultured under differentiation-inducing conditions, these cells primarily formed neuroectoderm, and little if any endoderm or mesoderm was detected . Although we observed elevated expression of markers for mesoderm (e.g., Brachyury) when Sox2 was elevated in our cultures, we observed a pronounced increase in the expression of genes associated with neuroectoderm. The increased expression of genes involved in eye and brain development is particularly prominent, both in the DNA microarray analysis and after validation by qPCR. These genes include, but are not limited to, crystallin mu, Pax6, Six6, Otx2, and neuropilin. On the basis of these findings, we propose that high levels of Sox2 are highly toxic to the vast majority of mouse ES cells, whereas moderate increases in Sox2 induce ES cells to differentiate with a bias toward neuroectoderm. The latter proposal is consistent with the report that artificially sustaining the expression of Oct-3/4 in ES cells, when cultured in LIF-deficient, serum-free medium, promotes their differentiation into neuroectoderm-like cells that express Sox2 .
The findings in this study demonstrate that moderate increases in the level of Sox2 protein in ES cells reduces their self-renewal and promotes their differentiation. Importantly, our studies demonstrate that Sox2 levels, like those of Oct-3/4, must be maintained within narrow limits. As such, Sox2 and Oct-3/4 both function as a molecular rheostat to control the growth and differentiation of ES cells. This has important implications for efforts to reprogram somatic cells to a pluripotent stem cell state. Successful reprogramming will require restoring the expression levels of critical genes, such as Sox2 and Oct-3/4, to appropriate levels . In fact, recent reports have shown that improvements in the initial protocols for reprogramming fibroblasts  led to the production of cells that are remarkably similar to ES cells, which express Sox2 and Oct-3/4 at levels comparable to those of bona fide ES cells [36, –38]. Finally, although it is widely believed that reductions in the expression of Sox2 and Oct-3/4 during early development are key events during mammalian embryogenesis, the findings reported here raise the intriguing possibility that increases in the expression of Sox2 or Oct-3/4 are mechanisms used by nature to direct cell fate decisions during development. In support of this possibility, Oct-3/4 protein levels have been reported to be lower in the cells of the inner cell mass than one of their early derivatives, primitive endoderm .
Disclosure of Potential Conflicts of Interest
The authors indicate no potential conflicts of interest.
This work was funded by a grant from the National Institutes of General Medical Sciences (GM 080751). J.L.K. was supported in part by a training grant from the National Cancer Institute (CA 009476). Core facilities were supported in part by a Cancer Center Support Grant from the National Cancer Institute (CA 36727).