Acetylation of Sox2 Induces its Nuclear Export in Embryonic Stem Cells


  • Author contributions: G.A.B.: concept and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; M.P.K.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; H.Z.: collection and/or assembly of data, data analysis and interpretation; A.V.T.: collection and/or assembly of data; D.Q.: collection and/or assembly of data; D.W.: collection and/or assembly of data, data analysis and interpretation; S.K.: concept and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript.

  • First published online in STEM CELLS EXPRESS July 09, 2009.


Embryonic stem (ES) cells require a coordinated network of transcription factors to maintain pluripotency or trigger lineage specific differentiation. Central to these processes are the proteins Oct4, Nanog, and Sox2. Although the transcriptional targets of these factors have been extensively studied, very little is known about how the proteins themselves are regulated, especially at the post-translational level. Post-translational modifications are well documented to have broad effects on protein stability, activity, and cellular distribution. Here, we identify a key lysine residue in the nuclear export signal of Sox2 that is acetylated, and demonstrate that blocking acetylation at this site retains Sox2 in the nucleus and sustains expression of its target genes under hyperacetylation or differentiation conditions. Mimicking acetylation at this site promotes association of Sox2 with the nuclear export machinery. In addition, increased cellular acetylation leads to reduction in Sox2 levels by ubiquitination and proteasomal degradation, thus abrogating its ability to drive transcription of its target genes. Acetylation-mediated nuclear export may be a commonly used regulatory mechanism for many Sox family members, as this lysine is conserved across species and in orthologous proteins. STEM CELLS 2009;27:2175–2184


Embryonic stem (ES) cells are the in vitro analog of the inner cell mass of the blastocyst and, like the blastocyst, are capable of generating all cell lineages during embryonic development. ES cells also have the ability to proliferate indefinitely in culture and maintain their pluripotent state through a network of specific transcription factors [1, 2]. One critical transcription factor within this network, Sox2, belongs to a large family of proteins characterized by their similarity to the Sry (sex-determining region of chromosome Y) protein, and having a conserved high mobility group (HMG) DNA binding domain. There are 20 Sox proteins in mammals which are widely expressed in embryonic as well as adult tissues, and they regulate diverse developmental processes [3]. Generally, Sox proteins require other transcription factors such as partner proteins to control their activities [4, 5]. For example, in ES cells, Sox2 heterodimerizes with Oct4, and together they bind to a consensus DNA sequence present in the promoter region of their transcriptional target genes [6, 7]. Given that many Sox proteins play critical roles during early embryonic development, it is not surprising that they would be regulated by several complementary mechanisms. For example, several Sox proteins (specifically Sry, Sox9 and Sox10) are regulated in part by nuclear import and export [8]. Excluding a transcription factor from the nucleus blocks its ability to bind to target DNA sequences and control target gene expression. In addition to subcellular shuttling, many Sox proteins are also regulated by post-translational modifications. One well-characterized example is protein kinase A (PKA)-mediated Sox9 phosphorylation which increases its DNA binding affinity [3].

Post-translational modifications have also been reported as a regulatory mechanism for Oct4, Nanog, and Sox2. Oct4 and Sox2 are both sumoylated [9, 10], but this modification has varying effects on the function of these two proteins. Sumoylation of Oct4 increases its stability and DNA binding activity, but hampers the ability of Sox2 to bind DNA. A C-terminal motif of Oct4 is shown to be phosphorylated in HEK293 and Hela cells [11]. Phosphorylation of Oct4 at Serine 229 and Tyrosine 327 has a differential effect on the protein's ability to transactivate target gene expression [12]. Recent analysis of the human ES cell phosphoproteome indicates that both Oct4 and Sox2 are phosphorylated [13], but the functional consequences of these modifications has not been determined. Similarly, Nanog is also reported to be phosphorylated [14] with no characterization of its effects on the protein function. Together, these reports suggest that post-translational modifications may be essential for regulating the activity of these proteins.

In addition to protein phosphorylation and sumoylation, acetylation of lysine residues acts as a regulatory signal to control protein activity and/or stability in a wide variety of cellular processes. For example, the function of the tumor suppressor protein p53 is activated by acetylation, and deacetylation modulates its effect on cell growth and apoptosis [15]. Acetylation can improve protein stability, as in the case of E2F [16] or destabilize the protein, as in the case of HIF-1α [17]. One of the best characterized enzymes that catalyzes acetylation is p300/cAMP-response-element-binding-protein-binding protein (CBP), an acetyltransferase known to modify both histone and nonhistone proteins [18]. Recent reports demonstrate the involvement of p300/CBP in the ES cell transcription network. These studies show that p300/CBP is recruited to genomic sequences bound by Oct4, Sox2, and Nanog, and that depletion of any of these factors reduces the binding of p300/CBP at such sites [19]. This suggests that p300/CBP activity in ES cells is closely linked to Oct4, Nanog, and Sox2, and plays a critical role in the ES cell transcription network. The interplay between p300/CBP and Sox2 is also supported by in vitro studies using the Fgf4 enhancer [20–22].

In this study, we report that p300/CBP can acetylate Sox2 in its DNA-binding domain, and that enhancing global acetylation in ES cells with deacetylase inhibitors promotes Sox2 ubiquitination. We identify a key lysine residue in the nuclear export signal of Sox2 that is acetylated, and demonstrate that blocking acetylation at this lysine maintains Sox2 in the nucleus and sustains expression of its target genes. Mimicking acetylation at this residue enhances the interaction of Sox2 with the nuclear export machinery. Furthermore, conservation of this lysine across species and orthologous proteins suggests acetylation-mediated nuclear export may be a regulatory mechanism for many Sox family proteins.


Embryonic Stem Cell Culture, Treatment, and Differentiation

J1 mouse ES cells were grown under typical ES cell conditions on gelatinized tissue culture plates in feeder-free conditions. For embryoid body formation, 2 × 106 cells/mL ES cells were plated in low attachment dishes in the presence of complete growth medium lacking leukemia inhibitory factor (LIF), and cultured for days as noted with a change in medium every other day. For drug treatments, adherent ES cells at 75% confluence were washed in phosphate buffered saline solution, then incubated with normal ES cell media plus 100 ng/mL trichostatin A (Millipore, Billerica, MA, for times as noted.

Plasmid Construction

For plasmid construction, 6xHis-tagged Nanog, Oct4, Sox2, and Sox2 K75A or K75Q expression vectors were generated by Gateway LR recombination (Invitrogen, Carlsbad, CA, of pDEST17 (Invitrogen) and Gateway donor vectors containing the coding sequences for Nanog, Oct4, Sox2, and Sox2 K75A or K75Q. Next, pWT GFP-Sox2 and pK75A GFP-Sox2 was generated by Gateway LR recombination of pDS-EF1-GFP-XB (ATCC) and Sox2 or K75A Sox2 Gateway donor vectors. Point mutations in K75 were generated by two-step polymerase chain reaction (PCR) mutagenesis. External primers were designed that flank the HindIII and SacII sites in Sox2 and that would amplify approximately 700bp, including the K75 codon. Forward external primer: 5′-ctggcagttccctactctcg-3′, reverse external primer: 5′- gctccatgggctctgtggtcaagtccga-3′. Mutagenic primers were designed with nucleotide changes that would alter the K75 codon to an alanine or a glutamine, and were then ordered in both the forward and reverse orientations: K75Q primer: 5′-ggagatcagccagcgcctgggcgcggag-3′. K75A primer: 5′-ggagatcagcGCgcgcctgggcgcggag-3′. Point mutant plasmids were generated by first amplifying products from a Sox2 template with a mutant-forward primer and an external-reverse primer, and separately with a mutant-reverse primer and an external-forward primer. These two PCR products were combined and used as template in the secondary PCR reaction, amplifying the whole mutant fragment with the external-forward and external-reverse primers. The mutant fragments were subcloned back into the Sox2 plasmid with HindIII and SacII.

Generation of Wild Type and K75A Green Fluorescent Protein-Sox2 Embryonic Stem Cell Lines

A quantity of 2.5 × 106 ES cells were seeded onto mitotically inactivated DR4 MEFs on 10 cm gelatin-coated plates. Cells were transfected with ClaI-linearized pDS-EF1-GFP-XB-WT Sox2 or pDS-EF1-GFP-XB-K75A Sox2 using Effectene (Qiagen, Hilden, Germany, Twenty-four hours after transfection, media was replaced and Geneticin (Invitrogen) was added at 0.5 mg/mL. Geneticin-resistant colonies were selected and expanded 10 days after transfection.

Protein Purification and Protein-Protein Interactions

WT, K75A, and K75Q His-Sox2 were acetylated using recombinant p300 (Upstate, Charlottesville, VA, in an in vitro histone acetyltransferase assay as in the “In vitro histone acetyltransferase (HAT) assay” section of methods. For Crm1 interaction studies, acetylated and nonacetylated proteins were incubated with 300 μg of whole cell lysates from ES cells in NP-40 buffer (50 mM Tris pH 8.0, 150 mM NaCl, 1% NP-40, 5 mM EDTA). After 4 hours of incubation at 4°C on a rotator, bound complexes were pulled down using nickel resin (Qiagen), washed in NP-40 buffer three times, resuspended in SDS sample buffer, and electrophoresed on an SDS-PAGE gel and blotted onto a PVDF membrane according to standard procedures. For Figure 1B, extracts were immunodepleted with antibodies to histone H3 (Millipore) before incubating with either acetyl lysine antibody (Cell Signal), Sox2 antibody (Abcam, Cambridge, U.K., or IgG control (Abcam). Blots were probed with primary antibodies: Nanog and Sox2 (Millipore), Crm1 (BD Biosciences, San Diego, CA,, acetyl lysine (Cell Signaling Technology, Beverly, MA,, tubulin (Calbiochem, San Diego, CA, and ubiquitin (Boston Biochem Inc, Cambridge, MA, Nuclear/cytoplasmic fractions in Figures 1B, 2D, 2E, and 2F were prepared using a nuclear and cytoplasmic protein extraction reagents kit (ThermoScientific/Pierce, Rockford, IL, as per the manufacturer's directions.

Figure 1.

Sox2 is acetylated in vitro and in vivo. (A): Histone acetyltransferase assay in which recombinant Oct4, Nanog and Sox2 were incubated with recombinant p300 and analyzed by SDS-PAGE followed by fluorography. (B): Endogenous acetylated Sox2 was immunoprecipitated from nuclear and cytoplasmic fractions of embryonic stem cells using acetyl lysine antibody, and detected by antibodies against Sox2. (C): Partial amino acid sequence of mouse Sox2 (SwissProt accession number P 48432) showing mapped acetylated lysines identified by mass spectrometry. Lysines acetylated by recombinant p300 are highlighted in red, and lysines acetylated by embryonic stem cell nuclear extract are marked with asterisks. (D): Conservation of the Sox2 nuclear export sequence in different species and with other Sox proteins. Lysine 75 from mouse Sox2 is conserved across species and other Sox proteins, and is highlighted in bold. Abbreviations: ES, embryonic stem; HAT, histone acetyltransferase; IP, immunoprecipitated.

Figure 2.

Acetylation promotes Sox2 nuclear export and subsequent ubiquitination. (A): Diagram of the Sox2 protein. The high mobility group box is denoted by the rectangle, with the red area representing the nuclear export signal (NES) identified by sequence homology. The mutation of lysine to alanine at lysine 75 is underlined. (B): Pull-down showing strong interaction of p300-acetylated wild-type (WT) Sox2 with Crm1 (lane 2), a weak association of p300-acetylated K75A Sox2 with Crm1 (lane 4), and a strong association of unacetylated K75Q Sox2 with Crm1 (lane 8). Acetyl lysine Western shows that wild type (WT), K75A, and K75Q proteins used in the reaction were acetylated by recombinant p300. Coomassie staining shows that equal amounts of all proteins were used in the reaction. (C): K75A GFP-Sox2 is capable of maintaining Nanog expression in the absence of endogenous Sox2. Top: Western blot for Sox2 from K75A GFP-Sox2 embryonic stem (ES) cells, 48 hours after treatment with control siRNA oligos (Lamin and Oct4), pan-Sox2 siRNA oligo, or siRNA oligos targeting only endogenous Sox2. Bottom: Western blot for Nanog from K75A GFP-Sox2 ES cells after siRNA knockdown of Sox2 shows that Nanog is maintained by K75A GFP-Sox2 in the absence of endogenous Sox2. Nanog protein levels are expressed relative to Lamin siRNA knockdown. (D): K75A GFP-Sox2 is maintained in the nucleus of ES cells. Western blot showing levels of endogenous Sox2, WT GFP-Sox2 and K75A GFP-Sox2 in nuclear and cytoplasmic fractions of ES cells in the presence of 100 ng/mL trichostatin A (TSA) for times indicated. Tubulin and histone H3 were used as loading and fractionation controls. (E): TSA treatment reduces Sox2 protein levels. Western blot showing levels of nuclear and cytoplasmic Sox2 in untreated and TSA-treated cells. H3 and Tubulin Western blots are controls for nuclear and cytoplasmic fractionation. (F): TSA treatment promotes Sox2 ubiquitination. Immunoprecipitation of Sox2 from nuclear and cytoplasmic fractions of untreated or TSA-treated ES cells, followed by Western blots for ubiquitin and Sox2. Relative ubiquitination is calculated as the ratio of image density in the ubiquitin Western blot to the image density of the Sox2 Western blot, which indicates TSA-treated ES cells have approximately three times as much ubiquitinated Sox2 as untreated ES cells. Abbreviations: TSA, trichostatin A; WB, Western blot; WT, wild type.

RNA Isolation, Real-Time Reverse Transcriptase Polymerase Chain Reaction and Analysis of Transcript Levels

Total RNA was purified with RNeasy Miniprep kit (Qiagen), DNAse treated (RQ1, Promega) and reverse transcribed (iScript cDNA synthesis kit; Bio-Rad, Hercules, CA, PCR reactions were performed using 10 ng of cDNA in an ABI7500 Fast Real-Time PCR System as per manufacturer's instructions (Applied Biosystems, Foster City, CA, Taqman gene expression assays (all from Applied Biosystems): glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Mm99999915_g1), Trp53 (assay ID Mm00441964_g1), Rex1 (Mm01194090_g1), Oct4 (Mm00658129_gH), FoxH1 (Mm00514851_m1), REST (Mm00803268_m1), Fgf4 (Mm03053741_s1), and Klf4 (Mm00516105_g1). A custom Taqman assay was designed for Nanog as follows: forward primer sequence 5′- TCCTCGCCCTTCCTCTGAA-3′, reverse primer sequence 5′- CAGGACTTGAGAGCTTTTGTTTGG-3′, reporter sequence 5′- CAGCCCTGATTCTTCT-3′. Gene expression units are calculated by taking the difference of the three GAPDH Ct values from the three target Ct values, then raising 2 to the power of the difference and multiplying by a constant, which is 1,000 for this analysis. The equation is represented as one gene expression unit = 1,000*(2∧[TargetCt-GAPDHCt]).

Chromatin Immunoprecipitation

ChIP was performed using 2.0 × 107 ES cells as described in the online protocol provided by Upstate (Millipore). Purified DNA was amplified with promoter-specific primers and Applied Biosystems SYBR PCR mastermix (Applied Biosystems). Relative enrichment was calculated as 2∧(ΔCT(control ChIP)- ΔCT[experimental ChIP]), where ΔCT is equal to the CT(immunoprecipitated sample) - CT(input). Primers are as follows: for the Nanog promoter, forward primer 5′-ggatgtctttagatcagaggatgccc-3′; reverse primer, 5′-ccacagaaagagcaagacaccaacc-3′. For the Trp53 promoter: forward primer, 5′-ttgccctcaacccacggaag-3′; reverse primer, 5′-ccagtcttcggagaagcgtg-3′. For the Rex-1 promoter: forward primer, 5′-gcatcctctgcttgtgtaaattc-3′; reverse primer, 5′- ctcagttatgcaaatgcctcttc-3′. ChIPs were performed with anti-GFP antibody (Abcam).

RNA Interference

To allow siRNA-lipid complexes to form, siRNA oligos to the 3′UTR of Sox2 (Qiagen) and SMARTpool siRNA targeting Oct4, Sox2 and Lamin (Dharmacon, Chicago, IL, were incubated with DharmaFECT4 (Dharmacon) in OptiMEM (Invitrogen) for 30 minutes at room temperature. Next, 0.3 × 106 ES cells were added to the transfection reagents for final concentrations of 84 nM siRNA and 1.17 μL/mL of DharmaFECT4. Protein was harvested from cells after 48 hours incubation with siRNA.

In Vitro Histone Acetyltransferase Assay

Approximately 5 pmol of His-tagged p300 was incubated with either 250 ng of core histones, His-tagged Oct4, Nanog, or Sox2 in the presence of 5 μM 3H-acetyl CoA (Amersham Biosciences, Piscataway, NJ, for 1 hour at 30°C, as described in the online protocol provided by Upstate (Millipore). Similarly, His-tagged Sox2 was acetylated using 125 μg of ES nuclear extract (prepared as in reference [39]) for 1 hour at 30°C. Reactions were stopped by adding SDS sample buffer and electrophoresed on an SDS-PAGE gel. The gel was fixed, treated with fluorography enhancing solutions (PerkinElmer, Waltham, MA,, dried, and exposed to film.

Protein Sequence Analysis by Liquid Chromatography - Mass Spectrometry and Liquid Chromatography - Tandem Mass Spectrometry

To map acetylation sites, we carried out two independent experiments: 1) His-tagged Sox2 was in vitro acetylated by recombinant p300, and 2) His-tagged Sox2 was acetylated by incubation with ES cell nuclear extract (prepared as in reference [39]). The method for both protocols is described in the “In Vitro HAT assay” section of Methods. In both cases, the acetylation reaction was stopped by adding SDS sample buffer and samples were electrophoresed on an SDS-PAGE gel. The gel was stained by Coomassie blue and the Sox2 band was excised from the gel, denatured, reduced, and alkylated in 8 M GuHCl, 10 mM DTT and 20 mM iodoacetamide. Samples were then exchanged into 100 mM ammonium bicarbonate and divided into equivalent aliquots with one fraction treated with trypsin and the remaining treated with chymotrypsin overnight. Peptides were later extracted by removing the ammonium bicarbonate solution and liquid chromatography - mass spectrometry and liquid chromatography - tandem mass spectrometry was then performed via capillary LC and a quadrupole-time of flight mass spectrometer. Samples were searched using MASCOT software (Matrix Science Inc, Boston, MA, with acetylation as a variable modification.

Statistical Analysis

To determine significance between WT GFP-Sox2 and K75A GFP-Sox2 ES cells, we made comparisons in SigmaStat v3.5, using two-way analysis of variance (ANOVA) followed by a post hoc Tukey's test. A p value of <0.05 is considered statistically significant; n = 3 for each condition.


Sox2 Can Be Acetylated by p300/CBP

To examine whether p300/CBP is required for the activity of Oct4, Nanog, or Sox2, we tested if any of these factors could be acetylated by p300/CBP. We used His-tagged Sox2, Oct4, and Nanog proteins as substrates in an in vitro acetyl transferase assay with recombinant p300 and 3H-labeled acetyl CoA. Sox2, but not Nanog or Oct4, was robustly acetylated by p300 (Fig. 1A). To investigate whether Sox2 was acetylated in vivo, we used antibodies against acetylated lysine to immunoprecipitate all acetylated proteins from nuclear and cytoplasmic ES cell extracts, and then performed a Western blot for Sox2 to determine the abundance of acetylated Sox2. We observe acetylated Sox2 in the nucleus of ES cells (Fig. 1B), consistent with the nuclear localization of p300/CBP [23]. In parallel, we immunoprecipitated total Sox2 from the same lysates, and performed a Western blot for Sox2 to determine the relative abundance of Sox2 in the nuclear and cytoplasmic compartments. We detected more Sox2 in the nuclear fraction than in the cytoplasm. These results suggest that Sox2 is also acetylated in vivo.

We next mapped the acetylation sites on Sox2 by two independent approaches. First, we analyzed in vitro acetylated Sox2 (Fig. 1A) using mass spectrometry, and found acetylated lysine residues at K37, K60, K67, K75, K89, K97, K105, K111, K117, K119, and K123 (Fig. 1C). With the exception of K123, all of the acetylated lysines reside within the HMG box of Sox2. To map the residues likely to be acetylated in vivo, we incubated bacterially-expressed His-Sox2 with ES cell nuclear extract in the presence of acetyl CoA and analyzed the acetylated peptides using mass spectrometry. We detected acetylation at seven lysine residues, K37, K60, K67, K75, K89, K97, and K105 (Fig. 1C). All of the lysines identified as acetylated by this approach were also identified as acetylated by recombinant p300, strongly suggesting that p300/CBP acetylates Sox2 in ES cells at lysine residues within the HMG box.

In addition to the highly conserved DNA-binding HMG box region in Sox2, there are other subdomains within Sox2 as predicted by sequence homology. Notably, we find a leucine-rich nuclear export signal (NES) within the HMG box of Sox2 which is highly homologous to the functional NES in Sox9 and Sry (Fig. 1D). Lysine 75 (K75), which was acetylated using both methods, falls within this putative Sox2 NES. We compared the NES from mouse Sox2 to Sox2 protein sequences from other species, as well as to other Sox family members. The NES region and the relative position of K75 of mouse Sox2 are identical across vertebrate Sox2 proteins (Fig. 1D), and show a high degree of conservation across related Sox family members. Strong sequence conservation in this subdomain suggests that it may be important for regulating the function of Sox proteins. The presence of an acetylated lysine residue in this region may indicate an additional means of regulation, or may confer specificity to a regulatory process.

Acetylation at K75 Promotes Sox2 Nuclear Export

Dynamic regulation of nuclear and cytoplasmic localization has been shown to be important in the regulation of other Sox family members. For example, the subcellular distribution of Sox10 modulates transactivation of its target genes [24], and the Sox9 NES is critical for triggering male specific sexual differentiation [25]. Interestingly, acetylation of Sry in the nuclear localization signal retains Sry in the nucleus [26]. Sox2 itself has been shown to translocate from the cytoplasm into the nucleus during very early stages of embryogenesis [27], suggesting that subcellular shuttling might be a means of regulating its activity. Since we find that one of the acetylated lysines, K75, is located within the Sox2 NES, and that this lysine is conserved across Sox2 orthologs as well as through other Sox family members, we hypothesized that acetylation of K75 could regulate the activity of Sox2 by affecting its subcellular distribution. Proteins containing NES sequences regulate their nuclear export through interaction with Crm1, a nuclear export receptor [28]. To test whether Crm1 regulates nuclear export of Sox2 and whether acetylation at K75 affects this process, we generated a mutant Sox2 protein with an alanine substitution at this residue (K75A), preventing it from being acetylated, but not affecting acetylation at the other lysines in the protein. Wild-type and K75A recombinant His-tagged Sox2 proteins (Fig. 2A) were used as substrates for in vitro acetylation by recombinant p300. After confirming their acetylation status by Western blot analysis using a pan-acetyl lysine antibody (Fig. 2B), the Sox2 proteins were bound to nickel resin, incubated with ES cell extract, and assayed for interaction with Crm1 by Western blot. Both unacetylated wild-type and K75A Sox2 proteins show a low level of interaction with Crm1. However, acetylation of wild-type Sox2, unlike the K75A mutant, strongly enhanced its interaction with Crm1 (Fig. 2B), suggesting that acetylation of K75 promotes association of Sox2 with Crm1. To test whether acetylation at K75 is sufficient to promote association with Crm1, we mutated K75 to a glutamine (K75Q) which functions as an acetyl-mimetic [29]. When used in the same Crm1-binding assay, we see a robust interaction with Crm1, even in the absence of acetylation. This interaction was enhanced by acetylation, indicating acetylation of additional residues may be required for optimal Crm1 interaction. These data indicate that the K75Q acetyl-mimetic mutation is sufficient to promote Sox2 interaction with the nuclear export machinery, and supports our hypothesis that acetylation at K75 promotes Sox2 nuclear export.

To analyze the function of the K75A substitution in vivo, we generated two stable ES cell lines expressing either wild-type (WT GFP-Sox2) or K75A Sox2 (K75A GFP-Sox2) fused to green fluorescent protein (GFP). Because overexpression of Sox2 has been shown to differentiate ES cells [30], we verified that both WT GFP-Sox2 and K75A GFP-Sox2 ES cell lines have not committed to differentiation by assaying the expression of a variety of lineage markers (Bmp4, Hand1, GATA6, Sox17, Cdx2, and Pax6) and the expression of pluripotency markers Nanog and Oct4 (supporting information, Fig. S2). Having confirmed that this level of exogenous Sox2 expression did not lead to ES cell differentiation and that levels of pluripotency markers were similar to unmodified ES cells, we tested whether K75A GFP-Sox2 was able to functionally replace endogenous Sox2. We used siRNA oligos designed to target the 3′UTR of Sox2. These siRNAs do not affect expression of the GFP-tagged protein, and they allow us to analyze the effect of the K75A Sox2 mutation on ES cells in the absence of endogenous Sox2. All three 3′UTR-specific siRNA oligos significantly reduced endogenous Sox2 protein levels within 48 hours of treatment, without affecting K75A GFP-Sox2 expression (Fig. 2C; lanes 5-7). As a positive control, we used siRNA oligos against the coding sequence of Sox2, which efficiently knocked down both the endogenous and GFP-tagged Sox2 protein (Fig. 2C; lane 4). Positive control siRNAs against Oct4 decreased the levels of endogenous Sox2, and negative control siRNAs against Lamin showed no effect on Sox2 protein levels (Fig. 2C; lanes 2 and 3). To test whether K75A Sox2 was able to rescue Sox2 target gene expression, we analyzed Nanog protein levels after knockdown of endogenous Sox2 alone (using siRNA #3), or endogenous and K75A GFP-Sox2 (pan-Sox2 siRNA). As expected, we saw an approximately 40% decrease in Nanog protein levels with the pan-Sox2 siRNA (Fig. 2C, bottom panel) and no change with knockdown by siRNA #3. This demonstrates that K75A GFP-Sox2 is capable of replacing endogenous Sox2 to maintain Nanog expression and that mutation of K75 does not inherently disrupt the transcriptional function of Sox2.

Since blocking acetylation at K75 impairs the association of Sox2 with the nuclear export machinery, we wanted to examine the effects of global acetylation on the subcellular distribution of Sox2. To hyperacetylate cellular proteins, we treated WT GFP-Sox2 and K75A GFP-Sox2 ES cells with the deacetylase inhibitor trichostatin A (TSA), then prepared nuclear and cytoplasmic fractions at various time points and analyzed relative Sox2 protein levels. We found that endogenous and WT GFP-Sox2 was present in both the nucleus and cytoplasm of ES cells, and that all Sox2 protein levels decreased after 6 hours of TSA treatment. In stark contrast, K75A GFP-Sox2 protein was detected exclusively in the nuclear fraction under both TSA and untreated conditions (Fig. 2D), indicating that blocking acetylation at K75 is sufficient to prevent nuclear export of Sox2 in vivo.

While the K75A GFP-Sox2 protein was retained in the nucleus more efficiently than the WT GFP-Sox2 or endogenous Sox2 proteins, all forms of the protein are reduced after 6 hours incubation with TSA. We surmised that the decrease in protein level may result from proteasomal degradation. To explore this further, we prepared nuclear and cytoplasmic fractions from ES cells treated for 6 hours with TSA or untreated (Fig. 2E). From these fractions, we immunoprecipitated (IP) total Sox2, then performed Western blots for ubiquitin and Sox2. As seen previously, TSA treatment decreases Sox2 levels in ES cells (Fig. 2E, top panel). When we examine ubiquitination of Sox2 from the Sox2 IPs, we see that Sox2 ubiquitination is elevated in the TSA-treated cells in both the nuclear and cytoplasmic compartments, despite the relative reduction in total Sox2 protein levels (Fig. 2F). We calculated the image density of the ubiquitinated bands, and divided that value by the image density of the immunoprecipitated Sox2. This “relative ubiquitination” measurement indicates that there is approximately three times as much ubiquitinated Sox2 in the TSA-treated cells as is in the untreated cells (20.63 versus Six.6 nuclear, 26.74 versus Eight.78 cytoplasmic, Fig. 2F). These results suggest that there is an increase in Sox2 ubiquitination downstream of cellular acetylation, and this could account for the protein loss observed in TSA treatments. This may also explain the absence of acetylated Sox2 in the cytoplasmic fractions (Fig. 1B).

Blocking Acetylation at K75 Sustains Sox2 Target Gene Expression

Based on our prior observations, we expected that sustained nuclear K75A GFP-Sox2 in the presence of TSA should correlate with sustained Sox2 target gene expression. We treated WT GFP-Sox2 and K75A GFP-Sox2 ES cells with TSA for 0.5, 1, 2, 3, and 4 hours, then harvested RNA from the cells to analyze expression of several Sox2 target genes. As shown in Figure 3A through 3C, K75A GFP-Sox2 ES cells have significantly sustained Sox2 target gene expression (Nanog, Trp53, and Rex1) compared to WT GFP-Sox2 ES cells during the time course of TSA treatment. We confirmed these observations using ChIP analysis to examine the binding of WT GFP-Sox2 and K75A GFP-Sox2 to the promoters of these target genes. Figures 3D through 3F show that K75A GFP-Sox2 maintains and even increases its binding to target promoters during TSA treatment, whereas WT GFP-Sox2 is rapidly lost, consistent with the observed decrease of Sox2 target gene expression (Fig. 3A-3C).

Figure 3.

Blocking acetylation at K75 sustains Sox2 target gene expression and promoter occupancy during trichostatin A (TSA)-induced differentiation. (A–C): Gene expression analysis of Sox2 target genes Nanog(A), Trp53(B), and Rex1(C) by quantitative polymerase chain reaction (qPCR) in wild type (WT) GFP-Sox2 and K75A GFP-Sox2 embryonic stem (ES) cells in presence of 100 ng/mL trichostatin A (TSA) for times indicated. Data points represent gene expression normalized against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (control) levels. Results shown are the average of three replicates. *p < .05. (D–F): ChIP assays were performed against GFP using WT GFP-Sox2 and K75A GFP-Sox2 ES cells that had been treated with 100 ng/mL TSA for the times indicated. Quantitative polymerase chain reaction amplified DNA using primers that span the Oct4-Sox2 binding site of the Nanog promoter (D), the transcription start and part of exon one of Trp53(E), and the Oct4-Sox2 binding site of the Rex1 promoter (F). Values are expressed as fold enrichment relative to IgG control ChIP. Results shown are the average of three independent polymerase chain reactions. Abbreviations: ChIP, chromatin immunoprecipitation; TSA, trichostatin A; WT, wild type.

We also examined expression of Sox2 target genes during differentiation using an embryoid body (EB) assay. We differentiated WT and K75A GFP-Sox2 ES cells into EBs and harvested them for RNA after 1, 3, 5, and 7 days. This revealed that Nanog and several other Sox2 target genes (Oct4, Rex1, Trp53, FoxH1, Fgf4 and REST) were all expressed at significantly higher levels in K75A GFP-Sox2 EBs than in the WT GFP-Sox2 EBs (Fig. 4A-4G; p < .05), similar to the results observed after TSA treatment. Klf4, a non-Sox2 target, showed little change in gene expression in both WT and K75A GFP-Sox2 EBs (Fig. 4H). Collectively, these results indicate that acetylation of Sox2 at K75 located within the Sox2 nuclear export sequence plays a significant role in regulating Sox2 protein activity and ES cell pluripotency.

Figure 4.

Blocking acetylation at K75 sustains Sox2 target gene expression during embryoid body differentiation. (A-G): Gene expression analysis of Sox2 target genes Nanog(A), Trp53(B), Oct4(C), FoxH1(D), Rex1(E), REST(F), Fgf4(G), and non-Sox2 target Klf4(H) by quantitative polymerase chain reaction (qPCR) in wild type GFP-Sox2 and K75A GFP-Sox2 embryonic stem cells that have been differentiated in an embryoid body assay for the times indicated. Data points represent gene expression normalized against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (control) levels. Results shown are the average of three replicates. *p < .05. Abbreviations: EB, embryoid body; GFP, green fluorescent protein; WT, wild type.


Our understanding of the core transcription factors that control ES cell regulatory processes as well as their involvement in generating induced pluripotent cells (iPS) from somatic cells has expanded greatly in the past few years. Sox2 is one of the four critical factors involved in generating iPS cells [31], and is also a vital component of ES cell pluripotency [32]. In recent years, much of the focus of investigation has been on the gene targets of these factors, with less emphasis on the upstream proteins that regulate Oct4, Nanog and Sox2 or cofactors that modify their activity. We have shown that p300/CBP, a transcriptional coactivator with diverse developmental functions, likely acetylates Sox2 and promotes its export from the ES cell nucleus by enhancing its association with Crm1, part of the nuclear export machinery.

Acetylation by p300/CBP has diverse effects on the downstream functions of its target proteins. While several reports describe increased nuclear retention of proteins upon acetylation [33], other studies suggest that acetylation can promote nuclear export [34]. Our findings provide evidence for negative Sox2 regulation by acetylation in ES cells through nuclear export followed by ubiquitination, and demonstrate that enhanced acetylation contributes to increased Sox2 ubiquitination. These studies add to the known functions of p300/CBP in ES cell proliferation and differentiation, and link its regulatory capacity to roles other than histone acetylation.

Export of proteins from the nucleus has been implicated in the control of several nuclear processes, primarily gene transcription [35]. Sox2 itself can shift its localization pattern early in mouse development from a cytoplasmic to a more nuclear distribution [26]. Clearly, sequestering a transcription factor in the cytoplasm represents a potent mechanism to negatively affect its activity. We propose that this mechanism of regulation also applies to Sox2 in ES cells, where we find that Sox2 moves from the nucleus to the cytoplasm, likely in response to differentiation signals. We have identified K75 as an acetylation site in the nuclear export signal of Sox2 and demonstrated its role in Sox2 cellular localization and biological activity during ES cell differentiation. Significantly, this lysine residue is conserved among Sox family proteins, suggesting a possible conservation of this regulatory mechanism. Our data also suggest a broader role for acetylation in the regulation of Sox2, via ubiquitination and proteasomal degradation. Increasing the amount of total acetylated protein in ES cells by deacetylase inhibitor treatment drives a threefold increase in the amount of ubiquitinated Sox2. The degradation of Sox2 appears to be independent of the acetylation at K75, as the K75-GFP Sox2 does eventually degrade with deacetylase inhibitor treatment. Further studies will be required to determine the link between Sox2 acetylation and ubiquitination and how these modifications impact the function of Sox2 in ES cells.

Besides having a nuclear export signal, Sox2 also has two nuclear localization signals, and mutating these sequences impedes nuclear import of Sox2 [36] and drives trophoectoderm differentiation [37]. It is interesting to note the common phenotype between nuclear depletion and knockout of Sox2 in ES cells [32], suggesting that a critical Sox2 protein level must be maintained in the nucleus to sustain pluripotency. Shuttling of other Sox family members and the presence of both nuclear export and import signals in Sox2 suggests that nuclear import/export may be an effective means of controlling the activity of these transcription factors.

Several of the other acetylated lysines identified in our studies show sequence conservation across Sox family proteins (data not shown). We are currently determining what role these modified lysines may play in the diverse biological functions of Sox2. These studies also suggest further interpretations of our previous work [38], where we found that the mSin3A/HDAC deacetylase complex interacts with Sox2 and promotes Nanog activation. An interplay between acetylation by p300/CBP and deacetylation by the mSin3A/HDAC complex of Sox2 at the promoters of its targets could act as a rapid switch to regulate gene expression. Information gained from deciphering the mechanisms that control Sox2 activity in ES cells are also germane to clinical applications of iPS cells. Since Sox2 is one of several critical factors required to generate iPS cells, identification of post-translational modifications that affect Sox2 function may enable iPS cells to be produced with greater efficiency, perhaps by increasing their proliferative capacity or lowering the Sox2 expression levels needed to induce reprogramming. Thus, our studies provide insight into how specific Sox2 post-translational modifications influence ES cell regulation, which can potentially be modulated to enhance iPS technology.


We thank Dr. B. Emerson at the Salk Institute for Biological Studies for helpful comments and colleagues at Novartis Institute for Biomedical Research, particularly Drs. E. Li, T. Chen, and A. Huang for insightful discussion.


The authors indicate no potential conflicts of interest.