TIF1β regulates the pluripotency of embryonic stem cells in a phosphorylation-dependent manner

Abstract

Transcription networks composed of various transcriptional factors specifically expressed in undifferentiated embryonic stem (ES) cells have been implicated in the regulation of pluripotency in ES cells. However, the molecular mechanisms responsible for self-renewal, maintenance of pluripotency, and lineage specification during differentiation of ES cells are still unclear. The results of this study demonstrate that a phosphorylation-dependent chromatin relaxation factor, transcriptional intermediary factor–1β (TIF1β), is a unique regulator of the pluripotency of ES cells and regulates Oct3/4–dependent transcription in a phosphorylation-dependent manner. TIF1β is specifically phosphorylated in pluripotent mouse ES cells at the C-terminal serine 824, which has been previously shown to induce chromatin relaxation. Phosphorylated TIF1β is partially colocalized at the activated chromatin markers, and forms a complex with the pluripotency-specific transcription factor Oct3/4 and subunits of the switching defective/sucrose nonfermenting, ATP-dependent chromatin remodeling complex, Smarcad1, Brg-1, and BAF155, all of which are components of an ES-specific chromatin remodeling complex, esBAF. Phosphorylated TIF1β specifically induces ES cell–specific genes and enables prolonged main-tenance of an undifferentiated state in mouse ES cells. Moreover, TIF1β regulates the reprogramming process of somatic cells in a phosphorylation-dependent manner. Our results suggest that TIF1β provides a phosphorylation-dependent, bidirectional platform for specific transcriptional factors and chromatin remodeling enzymes that regulate the cell differentiation process and the pluripotency of stem cells.

Embryonic stem (ES) cells exhibit the characteristic properties of infinite self-renewal and the ability to differentiate into various cell lineages, including the germ line. In addition to extrinsic signals such as leukemia inhibitory factor (LIF), bone morphogenetic protein (BMP), and fibroblast growth factor, the existence of an intrinsic transcription network comprising core factors such as Oct3/4, Sox2, and Nanog has been described (1, 2). However, the molecular mechanisms responsible for self-renewal, maintenance of pluripotency, and lineage specification during the differentiation of ES cells are still largely unknown.

Transcription intermediary factor–1β (TIF1β), also known as KRAB-associated protein 1 (KAP1) or tripartite motif protein 28 (TRIM28), is thought to act as a scaffold protein to recruit chromatin-modifying enzymes and transcriptional repressors such as heterochromatin protein 1 (HP1), the histone methyltransferase SETDB1, and the histone deacetylase–containing complex NuRD, which together silence transcription by triggering the formation of heterochromatin (3, 4). In fact, TIF1β is concentrated in heterochromatin after the differentiation of F9 carcinoma cells via association with HP1 (5). Moreover, the interaction of TIF1β with HP1 has been shown to be essential for progression of differentiation (6). These reports suggest that TIF1β is necessary for the proper progression of the process of differentiation from undifferentiated stem cells. Conversely, recent publications have implicated other functions of this protein. For example, TIF1β-null embryos have been shown to arrest at E5.5 (7), suggesting the functional importance of TIF1β in very early stages of development. Moreover, RNAi-based screening has suggested that TIF1β may be important for the maintenance of ES cells, although no detailed analyses of this gene in ES cells has been conducted (8).

In this study, we found that TIF1β is a critical switch that changes the pluripotent and differentiated states of ES cells. TIF1β is specifically phosphorylated in pluripotent ES cells at the C-terminal serine 824 amino acid, a modification that has been shown to induce chromatin relaxation (9). This phosphorylation is critical for the maintenance of the pluripotency and induction of iPS cells (i.e., induced pluripotent stem cells) from somatic cells. Our results suggest that TIF1β provides a phosphorylation-dependent, bidirectional platform for the transcription factors and chromatin remodeling components that regulate the pluripotency and differentiation of ES cells.

Results and Discussion

Identification of TIF1β That Promotes Pluripotent State of Mouse ES Cells.

The authors previously identified various chromatin-associated proteins specifically expressed in pluripotent mouse ES cells (10). To isolate factors that regulate pluripotency of ES cells, stable mouse ES cell lines expressing the chromatin-related proteins were established, and further screening was performed by using a functional assay based on the persistent alkaline phosphatase activity of mouse ES cells cultured in the absence of LIF. As a result, TIF1β was isolated as a unique factor that reproducibly showed the ability to maintain the pluripotent state of mouse ES cells. Although TIF1β is known to be a crucial factor for a chromatin remodeling complex consisting of corepressors and histone deacetylases that induces heterochromatin formation during differentiation (11), a recent RNAi-based screening raised the possibility that TIF1β may be an important gene for pluripotent ES cells (8). In our experiment, mouse ES cells stably expressing TIF1β showed prolonged alkaline phosphatase activity after withdrawal of LIF (Fig. S1A). Under the same conditions, control vector–transfected cells showed a flattened morphology and no phosphatase activity. RT-PCR, immunoblot, and immunofluorescent analyses all verified the sustained expression of Oct3/4 and ES cell–specific cell surface marker SSEA1 (Fig. S1 B and C) in the TIF1β-expressing cells. These results suggest that overexpression of TIF1β has the ability to prolong the pluripotent state in mouse ES cells.

When TIF1β knockdown ES cells were generated by stably expressing TIF1β-specific shRNA (Fig. S2A), the cells lost their tightly compacted morphology and had a stretched shape (Fig. S2B), as reported previously (8). The growth of the knockdown cells was apparently decreased (Fig. S2D). Moreover, TIF1β knockdown cells showed decreased expression of SSEA1 and Nanog (Fig. S2C). The data were also confirmed by semiquantitative analysis by counting the immunopositive cells (Fig. S2 E and F) and immunoblotting (Fig. S2G). In contrast, knockdown of the related gene TIF1α had no apparent effect on the morphology of the mouse ES cells (Fig. S2B) or on the expression of ES-specific markers (Fig. S2 C and E–G), suggesting that these effects are specific to TIF1β. Moreover, quantitative RT-PCR analysis revealed that shRNA-dependent knockdown of TIF1β strongly induced the expression of the primitive ectoderm marker gene Fgf5 and relatively weak expression of the extraembryonic ectoderm marker gene Eomes (Fig. S2H), suggesting that TIF1β preferentially inhibits the differentiation of ES cells into primitive ectoderm cells. These results verified that TIF1β is a specific and indispensable factor for the maintenance of pluripotency in ES cells.

Phosphorylation of TIF1β Regulates Pluripotency of Mouse ES Cells.

As we previously reported (10), TIF1β was highly expressed in the pluripotent mouse ES cells, and TIF1β expression decreased when ES cells spontaneously differentiated after culturing without LIF (Fig. 1 A and C). To our surprise, the phosphorylation of TIF1β serine 824 (S824), which has been shown to induce active relaxation of chromatin (9), was quite high in all of the pluripotent mouse ES cells and germline stem (GS) cells we tested (Fig. S3A), and this phosphorylation was dramatically decreased when the cells differentiated (Fig. 1 B and C and Fig. S3B). In contrast, this phosphorylation was not observed in other cells, like mouse embryonic fibroblasts (MEFs), NIH 3T3 cells (Fig. 1D), or HaCaT keratinocytes. TIF1β was not phosphorylated much in the mouse embryonic carcinoma (EC) cell lines F9 and P19 (Fig. S3A), suggesting that this modification is specific to pluripotent ES cells and GS cells. During embryogenesis, whereas TIF1β was detected homogeneously in the entire E3.5 embryo, TIF1β was strongly phosphorylated in the inner cell mass of the embryos, from which ES cells can be established (Fig. S3C).

Fig. 1.

Phosphorylation of TIF1β promotes pluripotency-specific marker expression. (A) Immunoblotting of endogenous TIF1β in ES D3 cells cultured with or without LIF for 8 d. (B) Immunoblotting of phosphorylated TIF1β (S824) using the same samples as A. (C) Densitometric semiquantification of the protein bands of A and B. (D) Specific phosphorylation of TIF1β in pluripotent mouse ES cells. Phosphorylation of TIF1β in undifferentiated ES D3 was compared with that in other types of cells, such as MEFs or NIH 3T3 cells. (E) Increased expression of pluripotency-specific marker proteins in ES cells stably expressing phosphorylation-mimicked TIF1β. WT, SD, and SA indicate WT TIF1β and two mutated TIF1β, S824D and S824A, respectively. Cntl indicates empty vector–transfected cells. These cells were cultured with or without LIF for 8 d. Bottom graph shows the semiquantified protein amounts of this blot. (F) Morphology of TIF1β-stably expressing ES cells cultured in the absence of LIF. TIF1β-expressing ES D3 cells established in E were cultured without LIF for 8 d. (G) Immunofluorescent analysis of Oct3/4 and SSEA1 proteins in ES cells stably expressing the SD or SA type of TIF1β. ES cells were cultured with or without LIF for 8 d. Cntl indicates empty vector–transfected cells. (H) Quantitative methylation analyses of pluripotency-specific genes by combined bisulfite restriction analysis: DNA methylation status of the indicated promoters in ES D3 cells stably expressing TIF1β-SD, TIF1β-SA, or empty vector (Cntl) cultured as in E was analyzed.

The effect of phosphorylation of TIF1β (S824) on the regulation of pluripotency in mouse ES cells was then evaluated. Serine 824 of TIF1β was mutated into aspartic acid (SD) or alanine (SA) to mimic phosphorylated or nonphosphorylated versions of TIF1β, respectively. Stable expression of the TIF1β-SD mutant has been shown to induce constitutive chromatin relaxation (9). Mouse ES cells stably expressing TIF1β-SD protein maintained high levels of Oct3/4 and Nanog protein expression after an 8-d culture without LIF, comparable to expression levels in the control vector–transfected ES cells cultured with LIF (Fig. 1E). In contrast, TIF1β-SA–expressing ES cells apparently lost the ability to express these marker proteins at a level similar to the control cells cultured without LIF. TIF1β-SD–expressing cells showed relatively higher levels of expression of these markers than WT TIF1β-expressing cells. This phosphorylation-dependent maintenance of ES cell pluripotency was also observed in a neomycin-resistant cell pool generated by retrovirus-based gene transfer, suggesting that maintenance of ES cell pluripotency by TIF1β-SD is not a result of the selection of particular clones but rather stems from the phosphorylation-dependent effects of TIF1β. Consistent with these observations, the morphology of TIF1β-SD- or TIF1β-SA–expressing ES cells looked like the appearance of the ES cells cultured with or without LIF, respectively (Fig. 1F). WT TIF1β-expressing cells showed partial activity for the maintenance of pluripotency. Immunofluorescent staining with Oct3/4 and SSEA1 antibodies (Fig. 1G) further supports the importance of this modification. Moreover, DNA methylation analysis of pluripotency-specific genes confirmed lower methylation level of these promoters in the ES cells stably expressing TIF1β-SD (Fig. 1H). Summarizing these findings, TIF1β has the ability to promote the pluripotency of mouse ES cells in a C-terminal phosphorylation–dependent manner.

Phosphorylation of TIF1β Regulates Differentiation of ES Cells and Induction of iPS Cells.

The effects of TIF1β on the differentiation of mouse ES cells were then tested. For differentiation to neuronal cells, embryoid bodies prepared from dissociated ES cells were cultured by floating culture in a serum-free medium for 8 d as described previously (12). After 5 d of adhesion culture, the differentiated cells showed significant outgrowth of neurofilament-200 (NF200) and TuJ1-positive neurites from the embryoid bodies (Fig. 2A). However, ES cells expressing TIF1β-SD did not show apparent outgrowth of neurites. In contrast, TIF1β-SA–expressing ES cells efficiently differentiated to neuronal cells. This phosphorylation-dependent effect of TIF1β on neuronal differentiation was not a result of the toxic effect of TIF1β-SD on the differentiated neurons, as transduction of the TIF1β-SD virus did not have a significant damaging effect on the neurons fully differentiated from mouse ES cells. These results indicate that TIF1β inhibits the differentiation of ES cells in a phosphorylation-dependent manner.

Fig. 2.

TIF1β regulates both differentiation of ES cells and induction of iPS cells. (A) TIF1β inhibits the neuronal differentiation of ES cells in a phosphorylation-dependent manner. Aggregated ES cells were cultured in GMEM/KSR–based medium in a suspension culture for 8 d followed by adhesion culture on a poly-L-ornithine/laminin–coated dish for 5 d. The differentiated cells were immunostained with TuJ1 (green) and neurofilament 200 (NF200, red) antibodies. Nuclei were stained with DAPI. (B) Phosphorylation-dependent regulation of iPS cell induction by TIF1β. MEFs were infected with four retroviruses that express Oct3/4, Sox2, Klf4, and c-Myc with a combination of TIF1β-expressing retrovirus or control virus (Cntl). Induction of iPS cells was monitored by their morphology and alkaline phosphatase activity. The alkaline phosphatase staining was conducted 12 d after infection. The assay was performed in triplicate. (C) Quantitative analysis of induction of iPS cells. The number of alkaline phosphatase–positive and morphologically ES cell–like cells in B were counted. (D) Expression of endogenous pluripotency-specific marker genes in the established ES-like cells when cultured in the presence of LIF (1,000 U/mL). Expression of various pluripotency-specific marker genes was analyzed by RT-PCR with the specific primers designed for the endogenous mRNAs. The number indicates the clone name of stable cells.

This study further investigated whether the phosphorylation of TIF1β affects induction of iPS cells from somatic cells such as fibroblasts. With retroviral gene transfer using four factors—Oct3/4, Sox2, Klf4, and c-Myc, as reported previously (13)—alkaline phosphatase–positive ES-like colonies appeared 12 d after infection. As shown in Fig. 2 B and C, introduction of the TIF1β-SD gene, in addition to the four factors, to MEFs promoted the induction of ES-like cells by 33%. In contrast, addition of the nonphosphorylatable TIF1β-SA strongly inhibited this process. These results indicate that phosphorylation of TIF1β is critical for iPS induction. The timing of the induction of iPS cells with TIF1β-SD retrovirus was 2 to 3 d faster than the control virus–infected cells in multiple experiments. In contrast, infection of ES cells with TIF1β-SA retrovirus generated iPS colonies with similar timing to that of control iPS cells. When these ES-like colonies were picked up and established as clones and their expression of pluripotency-specific marker genes were analyzed with RT-PCR, the ES-like clones generated by infection with the TIF1β-SD virus showed more complete expression of endogenous pluripotency-specific markers than the control clones induced with the four factors (Fig. 2D). In contrast, ES-like clones established with the TIF1β-SA virus clearly showed incomplete expression of these markers even in the presence of high doses of LIF. Most of the TIF1β-SA–expressing colonies were difficult to establish as clones in the presence of LIF, and could not be expanded for further analysis.

The effects of TIF1β mutants were not caused by their toxicity on the MEFs or established iPS cells, but were rather a result of the iPS induction process, because ES cells stably expressing TIF1β-SD or TIF1β-SA were morphologically indistinguishable from the control ES cells and the expression of various ES markers were mostly comparable to those of the control vector–transfected cells when they were cultured in the presence of LIF (Fig. S4C). Moreover, MEFs infected with high titer TIF1β-SD- or TIF1β-SA–expressing retrovirus looked healthy and no apparent toxic effect or growth inhibition was observed (Fig. S4 D and E). These results preclude the possibility that the phosphorylation status of TIF1β affects the viability of the established iPS cells cultured with LIF or MEFs infected with the viruses. Thus, our data suggest that the phosphorylation of TIF1β, which induces active relaxation of chromatin, affects the efficiency of iPS induction and the quality of established clones of iPS cells.

Euchromatin-Specific Localization of Phosphorylated TIF1β in Mouse ES Cells.

To obtain insights into the mechanisms responsible for phosphorylated TIF1β–dependent regulation of stem cells and induction of iPS cells, subcellular localization of TIF1β was analyzed in pluripotent mouse ES cells by immunofluorescent staining. Endogenous TIF1β was diffusely localized in the nucleoplasm of mouse ES cells (Fig. 3A Upper), as described previously (5). In contrast, staining of phosphorylated TIF1β with an antibody specific for phosphorylated TIF1β at S824 revealed a punctate staining pattern in the nucleus (Fig. 3A Lower). Various anti-bodies that show localization similar to that of phosphorylated TIF1β were surveyed using immunofluorescent staining. The results showed that transcriptionally activated euchromatin markers such as histone H3K4me³ and H3K9Ac partially colocalized with phosphorylated TIF1β. In contrast, this punctate staining of phosphorylated TIF1β had a completely different localization from that of the heterochromatin markers histone H3K9me³ and HP1α (Fig. 3B). TIF1β has been reported to be involved in heterochromatin formation with HP1 (11) and localized in the heterochromatin foci in differentiated cells (Fig. 3C) (5). However, this characteristic localization of TIF1β in the differentiated cells was not observed in pluripotent ES cells (Fig. 3A Upper), and ES-specific phosphorylation of TIF1β was not detected in a fibroblast-like cell line, NIH 3T3 cells (Fig. 1D). These results suggest that phosphorylated TIF1β plays a unique role in pluripotent ES cells, distinct from the role of heterochromatin formation in various differentiated cell lines. When TIF1β was transiently knocked down by siRNAs, ES cells have started to change their morphology with decreased expression of Nanog gene in 4 to 6 d after transfection (Fig. S5A), which is similar timing as culture in the absence of LIF (10). Under these conditions, TIF1β knockdown cells showed increased number of H3K9me³ foci and HP1α (Fig. 3D and Fig. S5 B and C). In contrast, there was no apparent change in distribution of H3K4me³ foci in the nucleus (Fig. S5B). These results suggest that TIF1β is an important factor for the inhibition of H3K9me³ and HP1α foci formation in pluripotent mouse ES cells.

Fig. 3.

Subcellular localization of phosphorylated TIF1β in mouse ES cells. (A) Immunofluorescent staining of pluripotent ES D3 cells cultured in the presence of LIF with TIF1β or a phosphorylated TIF1β (S824)–specific antibody. (B) Localization of phosphorylated TIF1β in transcriptionally activated histones. Mouse ES cells cultured in the presence of LIF were immunostained with antibodies against phosphorylated TIF1β (green) and modified histones (red). (C) Subcellular localization of TIF1β in mouse NIH 3T3 cells. The cells were immunostained with TIF1β (green) and HP1α (red) antibodies. (D) Quantification of H3K9me³ foci in TIF1β-knockdown cell nuclei. Mouse ES D3 cells were transfected with control or TIF1β siRNA and immunostained with TIF1β and H3K9me³ antibodies. The number of immunopositive H3K9me³ foci in the nuclei was counted. (E) ATM is one of the responsible kinases for TIF1β phosphorylation in pluripotent mouse ES cells. Knockdown of ATM in ES D3 cells with the shRNA caused a decrease in TIF1β phosphorylation and loss of Oct3/4 and Nanog expression. (F) qRT-PCR analysis of pluripotency-specific ES cell marker genes. Total RNA extracted from ES cells cultured without LIF for 8 d was used for the analysis. The bar graph plots average expression values and SDs of mRNA levels normalized to control Gapdh mRNA levels. Values were calculated relative to the empty vector–transfected control samples. All experiments were performed in duplicate. C, control; W, WT TIF1β; D, TIF1β-SD; A, TIF1β-SA.

Ataxia telangiectasia mutated (ATM), a serine/threonine protein kinase that is activated by DNA double-strand breaks, has been reported as a responsible kinase for C-terminal phosphorylation of TIF1β upon DNA double-strand break (9). We asked whether ATM also phosphorylates TIF1β in mouse ES cells. As shown in Fig. 3E (Upper), the shRNA specifically targeted against ATM significantly decreased the phosphorylation of TIF1β and protein level of Oct3/4 and Nanog (Fig. 3E Lower), suggesting that ATM is one of the responsible kinases for the C-terminal phosphorylation of TIF1β in mouse ES cells. Interestingly, ATM was shown to be important for reprogramming of somatic cells into iPS cells (14). TIF1β is not directly phosphorylated by JAK in mouse ES cells (Fig. S6A).

Next, we speculated that phosphorylated TIF1β might be involved in transcriptional activation of pluripotency-specific genes in the activated chromatin foci of the pluripotent stem cells. To test this hypothesis, the expression of various ES cell–specific marker genes was analyzed with quantitative RT-PCR (qRT-PCR) using TIF1β-SD- and TIF1β-SA–expressing ES cells after an 8-d culture in the absence of LIF. As shown in Fig. 3F and Fig. S4A, TIF1β-SD cells had apparently higher expression of various pluripotency-specific genes than the control vector–transfected cells. In contrast, TIF1β-SA–expressing ES cells did not induce these genes. More interestingly, ectopic expression of TIF1β did not cause characteristic regulation of cancer-related genes, such as c-Myc and E-Ras (Fig. 3F and Fig. S4B). These results suggest that TIF1β can selectively activate the expression of various pluripotency-specific genes, but not cancer-related genes, of mouse ES cells in a phosphorylation-dependent manner.

Phosphorylated TIF1β Cooperates with Oct3/4 and Regulates Expression.

Next, the possibility that TIF1β could function as a transcriptional activator at the promoter regions of these specific genes by forming a complex with some pluripotency-specific transcriptional factors was further tested. As shown in Fig. 4A, TIF1β formed a complex with the pluripotency-specific transcriptional factor Oct3/4, and this interaction was dependent on C-terminal phosphorylation of TIF1β. A coactivator CBP was also associated with TIF1β in a phosphorylation-dependent manner. In contrast, Sox2 was not detected in this complex. Coimmunoprecipitation of endogenous Oct3/4 verified the interaction with phosphorylated TIF1β in pluripotent ES cells (Fig. 4B). Using deletion mutants of TIF1β, the interaction domain of TIF1β for Oct3/4 was located as the N-terminal half of TIF1β (Fig. 4C). CBP also interacts with the N-terminal of TIF1β (Fig. 4D). The interaction domain of Oct3/4 seems to be the internal domain of Oct3/4 (15), because both N-terminal and C-terminal deletion mutants of Oct3/4 interact with TIF1β-SD (Fig. S7A).

Fig. 4.

Phosphorylated TIF1β forms a functional complex on a pluripotency-specific gene promoter. (A) The phosphorylated form of TIF1β interacts with Oct3/4 and CBP, but not Sox2, in mouse ES cells. TIF1β was immunoprecipitated with Flag M2 antibody from mouse ES cells stably expressing Flag-TIF1β proteins or empty vector (–)–transfected cells. The resolved sample was immunoblotted with the indicated antibodies. (B) Endogenous Oct3/4 associates with phosphorylated TIF1β in mouse ES cells. Endogenous Oct3/4 was immunoprecipitated from ES D3 cells and the resolved sample was immunoblotted with phosphorylated TIF1β antibody. Ig indicates coimmunoprecipitation with normal IgG antibody. (C and D) N-terminal TIF1β interacts with Oct3/4 and CBP. Mouse ES cells transiently transfected with the deletion mutant of TIF1β were used for the coimmunoprecipitation assay. (E) Endogenous TIF1β associates with Smarcad1, Brg-1, and BAF155. Ig indicates normal IgG antibody. (F) Phosphorylation-dependent activation of the Nanog proximal promoter (−332 ∼ +50) by TIF1β. The indicated form of TIF1β was transiently expressed in NIH 3T3 cells, and the Nanog promoter activity was measured by luciferase assay. (G) ChIP assay of the Nanog proximal promoter. ChIP assay was performed using the Flag-TIF1β stable expressing ES cells or empty vector-transfected ES cells (Cntl). Immunoprecipitation was conducted with the Flag antibody or Oct3/4 antibody. (H) A diagram of phosphorylated TIF1β-regulated genes. Mouse ES cells overexpressing TIF1β-SD or TIF1β-SA cultured in the absence of LIF for 1 week were subjected to microarray analysis.

In addition to these proteins, two other nuclear proteins, MSH2 and Smarcad1, were found to interact with TIF1β, as has been reported in a colon cancer cell line (16). However, the interactions between these proteins and TIF1β were constitutive (Fig. S7B). Interestingly, we had previously identified Smarcad1 and MSH2 as highly expressed proteins in undifferentiated ES cells by quantitative proteomic analyses (10). MSH2 has been extensively analyzed for its function in DNA mismatch repair. Recently, MSH2 has been reported to function as a potent transcription activator (17). Smarcad1 (also called BAF60a) is an ATP-dependent switching defective/sucrose nonfermenting chromatin remodeling factor that modulates chromatin structure. Smarcad1 is highly expressed in the inner cell mass of blastocysts during development (18). Moreover, a recent publication (19) reported that esBAF, an ES-specific BAF (Brg/Brahma–associated factors) ATP-dependent chromatin remodeling complex, which is essential for ES cell self-renewal and pluripotency, contains Smarcad1 as a component of the complex. Coimmunoprecipitation experiment revealed that endogenous TIF1β forms a complex with Smarcad1 (Fig. 4E). Moreover, we noticed that this complex contains other esBAF components, Brg-1 and BAF155 (Fig. 4E). The N-terminal half of TIF1β seems to be important for these interactions because antibodies against C-terminal TIF1β (536-835 aa) but not N-terminal TIF1β (60-383 aa) can successfully coimmunoprecipitate Smarcad1, even though both antibodies immunoprecipitated enough amount of endogenous TIF1β protein. In our preliminary experiment, however, overexpression of Smarcad1 alone did not support the pluripotency of ES cells cultured without LIF. These results suggest that TIF1β could recruit a chromatin remodeling complex, presumably esBAF, via Smarcad1 to specific target promoters and allow efficient transcription of the downstream genes. Overexpression of Smarcad1 alone may not be sufficient to promote functional complex formation on the promoters of pluripotency-specific genes.

Phosphorylated TIF1β specifically induces transcriptional activation with Oct3/4 (Fig. 4F) when a reporter assay was done with a proximal Nanog promoter reporter construct (−332 ∼ +50), which contained an Oct3/4 binding site (20). This assay was performed with NIH 3T3 because endogenous TIF1β is not phosphorylated in this cell line (Fig. 1D). However, both the N-terminal and C-terminal fragments of TIF1β failed to induce luciferase activity, suggesting that phosphorylated full-length TIF1β is essential for the activation of the Nanog promoter. The ES cells stably expressing the N-terminal half or C-terminal half of TIF1β showed scattered or flattened shapes in the feeder-less culture, even in the presence of LIF and decreased level of Oct3/4 (Fig. S8A). In the absence of LIF, these cells failed to maintain the pluripotency and the expression of Oct3/4 completely disappeared.

A ChIP assay using endogenous Nanog promoter–specific primers also confirmed that TIF1β forms a complex on the endogenous Nanog promoter in a phosphorylation-dependent manner (Fig. 4G). Specific binding of TIF1β to the endogenous Nanog proximal promoter was confirmed using the control primers for the 1-kb upstream site and 1-kb downstream site of the Nanog transcriptional initiation site. The results also showed that endogenous Oct3/4 is efficiently recruited to the Nanog promoter in the presence of TIF1β-SD, suggesting that phosphorylation of TIF1β enables the recruitment of Oct3/4 to form an active complex on the Nanog promoter. When the target genes specifically regulated by phosphorylated TIF1β were surveyed by microarray analysis, one third of the Oct3/4 target genes (21) were identified (Table S1 and Fig. 4H). Various chromatin remodeling factors, such as Suz12, CHD9, Pcaf, and Smarcad1, were also found to be phosphorylated TIF1β-responsive genes. These results support the idea that TIF1β forms a specific complex with Oct3/4 on the promoters of pluripotency-specific genes, and promotes not only the gene expression of pluripotency-specific transcriptional factors such as Nanog, Sox2, and Oct3/4, but also the expression of various chromatin remodeling proteins for efficient control of pluripotent ES cell in a phosphorylation-dependent manner.

Ivanov et al. showed that the PHD domain of TIF1β functions as an intramolecular E3 SUMO ligase and that autosumoylation of TIF1β recruits SETDB1 histone methyltransferase and the CHD3/Mi2 component of the NuRD complex via SUMO-interacting motifs (11). Moreover, phosphorylation of TIF1β (S824) has been reported to repress autosumoylation of TIF1β and TIF1β-mediated transcriptional repression (22). Based on these findings, the possibility was considered that phosphorylated TIF1β might enable maintenance of a pluripotent state in ES cells through phosphorylation-dependent inhibition of SUMO-mediated heterochromatin formation. However, in mouse ES cells, phosphorylation of TIF1β did not cause significant changes in complex formation between TIF1β and the other proteins involved in transcriptional repression, such as HP1, SETDB1, and CHD3/Mi2. Moreover, significant dif-ferences in sumoylation were not found between TIF1β-SD and TIF1β-SA. On the contrary, our microarray analysis of TIF1β-SD- and TIF1β-SA–expressing ES cells suggested another mechanism: that phosphorylated TIF1β regulates gene expression of chromatin remodeling factors such as Suz12, CHD9, Pcaf, and Smarcad1. It appeared that, in mouse ES cells, C-terminal phosphorylation of TIF1β does not significantly modulate recruitment of the repressor machinery, but rather regulates transcriptional activation of specific chromatin remodeling factors and pluripotency-specific transcription factors by forming a complex with Oct3/4, coactivators, and chromatin remodeling enzymes in ES cells.

Recently, Hu et al. reported that a genome-wide siRNA screening in mouse ES cells identified TIF1β as one of the 148 factors essential for the maintenance of mouse ES cells (23). Their computational analysis suggested that TIF1β forms a unique module in the self-renewal transcription network separate from Nanog, Oct3/4, and Sox2. However, our study indicated that the phosphorylated form of TIF1β directly activates Nanog transcription by forming a complex with Oct3/4. A recent ChIP–ChIP analysis combined with a unique bioinformatic approach using an embryonic carcinoma cell line also revealed that half of the promoters bound by Oct3/4 and Sox2 were occupied by TIF1β (24). These results suggest that TIF1β functions as an important regulator of Oct3/4-dependent transcription in ES cells. We also observed that ectopic expression of phosphorylated version of TIF1β failed to maintain the pluripotency of ES cells in the absence of Oct3/4, Nanog, or Sox2 (Fig. S8B), supporting the fact that Oct3/4 is required for TIF1β-dependent regulation of pluripotency in ES cells. Interestingly, as shown in Fig. S8C, TIF1β-knockdown ES cells could not survive in 3i medium (25), implicating that TIF1β is essential for the intrinsic self-maintenance but not for the inhibition of neuronal differentiation of ES cells. Moreover, complete blocking of LIF signal with high concentration of a JAK inhibitor abrogated the effect of TIF1β (Fig. S6B) without affecting phosphorylation of TIF1β (Fig. S6A), suggesting that TIF1β cannot completely substitute for LIF signaling but promotes the pluripotency of mouse ES cells. Niwa et al. recently reported that the LIF signal is integrated into the core regulatory circuitry of pluripotency via two parallel pathways: the Jak-Stat3-Klf4-Sox2 pathway and the PI3K-Akt-Tbx3-Nanog pathway (26), both of which activate the expression of Oct3/4. Our data suggest that phosphorylated TIF1β at least in part cooperates with Oct3/4 and contributes the maintenance of pluripotency in ES cells, by regulating the expression of Oct3/4 target genes.

This study has demonstrated that phosphorylation of TIF1β at S824, which has been previously shown to actively induce chromatin relaxation, is a critical modification for regulation of mouse ES cells by forming a complex with Oct3/4 and the chromatin remodeling factors Smarcad1, Brg-1, and BAF155. Phosphorylation of TIF1β is not only essential for the maintenance of the pluripotent state of ES cells but also for the induc-tion of iPS cells from somatic cells. In contrast, nonphospho-rylated TIF1β is indispensable for the proper differentiation of stem cells (Fig. 2A). Therefore, a model is proposed in which TIF1β functions as a bidirectional platform that regulates the pluripotency and the differentiation of stem cells in a phosphorylation-dependent manner.

Methods

Cell Culture.

All mouse ES cells and GS cells were cultured on MMC-treated MEFs as described previously (10). See SI Methods for further details.

Alkaline Phosphatase Staining.

Mouse ES cells or iPS cells cultured on gelatin-coated dishes were washed twice with PBS solution and fixed in 3.7% formaldehyde/PBS solution for 5 min at room temperature. The cells were washed twice with PBS solution and incubated with BM purple AP substrate (Roche) for 30 min at room temperature.

Additional Details.

The remaining experimental details can be found in SI Methods.