The transcription factor TFCP2L1 induces expression of distinct target genes and promotes self-renewal of mouse and human embryonic stem cells

TFCP2L1 (transcription factor CP2-like 1) is a transcriptional regulator critical for maintaining mouse and human embryonic stem cell (ESC) pluripotency. However, the direct TFCP2L1 target genes are uncharacterized. Here, using gene overexpression, immunoblotting, quantitative real-time PCR, ChIP, and reporter gene assays, we show that TFCP2L1 primarily induces estrogen-related receptor β (Esrrb) expression that supports mouse ESC identity and also selectively enhances Kruppel-like factor 4 (Klf4) expression and thereby promotes human ESC self-renewal. Specifically, we found that in mouse ESCs, TFCP2L1 binds directly to the Esrrb gene promoter and regulates its transcription. Esrrb knockdown impaired Tfcp2l1's ability to induce interleukin 6 family cytokine (leukemia inhibitory factor)–independent ESC self-renewal and to reprogram epiblast stem cells to naïve pluripotency. Conversely, Esrrb overexpression blocked differentiation induced by Tfcp2l1 down-regulation. Moreover, we identified Klf4 as a direct TFCP2L1 target in human ESCs, bypassing the requirement for activin A and basic fibroblast growth factor in short-term human ESC self-renewal. Enforced Klf4 expression recapitulated the self-renewal–promoting effect of Tfcp2l1, whereas Klf4 knockdown eliminated these effects and caused loss of colony-forming capability. These findings indicate that TFCP2L1 functions differently in naïve and primed pluripotency, insights that may help elucidate the different states of pluripotency.

TFCP2L1 (transcription factor CP2-like 1) is a transcriptional regulator critical for maintaining mouse and human embryonic stem cell (ESC) pluripotency. However, the direct TFCP2L1 target genes are uncharacterized. Here, using gene overexpression, immunoblotting, quantitative real-time PCR, ChIP, and reporter gene assays, we show that TFCP2L1 primarily induces estrogen-related receptor ␤ (Esrrb) expression that supports mouse ESC identity and also selectively enhances Kruppel-like factor 4 (Klf4) expression and thereby promotes human ESC self-renewal. Specifically, we found that in mouse ESCs, TFCP2L1 binds directly to the Esrrb gene promoter and regulates its transcription. Esrrb knockdown impaired Tfcp2l1's ability to induce interleukin 6 family cytokine (leukemia inhibitory factor)-independent ESC self-renewal and to reprogram epiblast stem cells to naïve pluripotency. Conversely, Esrrb overexpression blocked differentiation induced by Tfcp2l1 down-regulation. Moreover, we identified Klf4 as a direct TFCP2L1 target in human ESCs, bypassing the requirement for activin A and basic fibroblast growth factor in short-term human ESC self-renewal. Enforced Klf4 expression recapitulated the self-renewal-promoting effect of Tfcp2l1, whereas Klf4 knockdown eliminated these effects and caused loss of colony-forming capability. These findings indicate that TFCP2L1 functions differently in naïve and primed pluripotency, insights that may help elucidate the different states of pluripotency.
Embryonic stem cells (ESCs) 3 are derived from the inner cell mass of the preimplantation blastocyst (1)(2)(3). Under appropriate in vitro culture conditions, ESCs proliferate indefinitely without differentiation while retaining the capacity to generate cell lineages derived from all three primary germ layers (4). To date, although ESC-like cells from many species have been established, only ESCs derived from mice and rats possess the ability to generate germline-competent chimeric offspring and thus represent a "naïve" pluripotent state (1,2,5,6). Interestingly, the available human ESCs (hESCs) are more similar to mouse postimplantation epiblast-derived stem cells (EpiSCs) than to mouse ESCs (mESCs) in their self-renewal requirements and morphology and thus represent a "primed" pluripotency state (3,7,8). mESC self-renewal can be maintained in two distinct culture systems: serum-containing medium supplemented with leukemia inhibitor factor (LIF) (9, 10) and serum-free N2B27 medium supplemented with two small molecule inhibitors (2i), CHIR99021 and PD0325901 (11). LIF supports self-renewal by inducing activation of signal transducer and activator of transcription 3 (STAT3) (12). CHIR99021 and PD0325901 maintain self-renewal through inhibition of glycogen synthase kinase 3 (GSK3) and mitogen-activated protein kinase kinase (MEK) (11), respectively. However, hESCs requires the activin A and basic fibroblast growth factor (bFGF) cytokines to maintain their identity (3). The addition of Wnt/ ␤-catenin signaling inhibitors can further enable robust hESC propagation (13,14). Understanding how these growth factors mediate intracellular signaling pathways controlling the unique pluripotent state and the similarities and differences between naïve and primed pluripotency are hot spots in current stem cell research.
Despite the difference in growth factor requirements between mESCs and hESCs, the core transcription factors gov-erning pluripotency are similar, such as the master pluripotency genes Oct4, Nanog, and Sox2 (15). Recently, transcription factor CP2-like 1 (Tfcp2l1) has been identified as an important pluripotent factor and has become one of the core markers to identify ESCs generated from many species (16 -20). We and other groups reported that Tfcp2l1 expression is high in the inner cell mass and mESCs, down-regulated in primed stem cells, and further reduced in differentiated cells (16,17,21,22). Tfcp2l1 plays an essential role in maintaining ESC identity. In mESCs, it is a critical target in LIF-and 2i-mediated selfrenewal (16,17,23). To date, only knockdown of Tfcp2l1, but not other STAT3 targets, is able to weaken the LIF/STAT3mediated mESC self-renewal and EpiSC reprograming (16,17). In addition, overexpression of Tfcp2l1 can compensate for the function of 2i when combined with Klf2, another pluripotency gene (23). In contrast to naïve-type stem cells, Tfcp2l1 is expressed highly in early human embryos, while it declines in vitro in established primed hESCs (21,24). However, overexpression of Tfcp2l1 cannot reprogram hESCs into the naïve pluripotency state (18). Remarkably, enforced expression of Tfcp2l1 promotes self-renewal, whereas its suppression leads to hESC differentiation toward endoderm and mesoderm specification (22,24). Taken together, these findings suggest that the self-renewal-promoting function of Tfcp2l1 is conserved in mESCs and hESCs.
Tfcp2l1 has been proposed to act partially through repression of multiple lineage commitments (25). However, it is unclear whether Tfcp2l1 functions through direct activation of a selective pluripotent factor. To resolve this issue, we sought to identify genes directly regulated by Tfcp2l1 in mouse and human ESCs mainly based on gain-and loss-of-function analyses. These analyses identified Esrrb and Klf4 as two direct targets of Tfcp2l1 that are capable of mediating the self-renewalpromoting effects of Tfcp2l1 in mESCs and hESCs, respectively.

Esrrb is a direct target of Tfcp2l1 in mESCs
Previously, Smith and co-workers (26) developed a data-constrained, computational method and defined the simplest essential cassette for maintaining naïve pluripotency. This minimal set comprises 3 inputs (2i/LIF), 12 transcription factors (Oct4, Sox2, Nanog, Klf2, Esrrb, Tfcp2l1, Klf4, Sall4, Gbx2, STAT3, TCF3, and MEK), and 16 interactions, in which Esrrb and Sall4 are two potential direct targets of Tfcp2l1 (26). To validate this predicted relationship of Tfcp2l1, we designed five different approaches. First, we generated one mESC line that overexpressed FLAG-tagged mouse Tfcp2l1 using a PiggyBac vector (PB-mTfcp2l1) in which Tfcp2l1 expression was efficiently enhanced (Fig. 1A). Quantitative real-time PCR (qRT-PCR) analysis revealed that overexpression of Tfcp2l1 resulted in up-regulation of the Esrrb transcript but not Sall4 (Fig. 1B). Second, to further validate that Esrrb is regulated by Tfcp2l1, we examined Esrrb transcription under different Tfcp2l1 expression levels by using one mESC line that contains a doxycycline (Dox)-inducible mouse Tfcp2l1 transgene (17), in which Tfcp2l1 transcription was efficiently induced by Dox treatment (Fig. 1C). Esrrb has a similar expression pattern to Tfcp2l1 fol-lowing Dox addition (Fig. 1C). Third, to investigate whether Tfcp2l1 down-regulation attenuates Esrrb expression, we constructed lentiviral vectors expressing mouse Tfcp2l1-specific shRNA sequences (mTfcp2l1 sh#1, mTfcp2l1 sh#2, and mTfcp2l1 sh#3). Stable knockdown (70 -80%) of Tfcp2l1 transcript was observed following drug selection (Fig. 1D). As expected, the Esrrb transcript decreased in Tfcp2l1 shRNA cells when compared with scramble control cells (Fig. 1D). Fourth, to explore how Tfcp2l1 regulates the expression of Esrrb, we mapped the binding sites of Tfcp2l1 by analyzing the deep sequencing (ChIP-seq) date of Tfcp2l1 (27). As shown in Fig.  1E, the Tfcp2l1 protein has many binding sites in the Esrrb locus (Fig. 1E). Finally, to further determine whether Esrrb is a direct target of Tfcp2l1, we analyzed the Tfcp2l1-binding consensus motifs (Fig. 1F) and predicted two potential binding sites within the Esrrb promoter region (from Ϫ3000 ϳ ϩ1) from the JASPAR CORE database (motif 1, Ϫ1312 CCAGCCTTGAC-TAG Ϫ1298 ; and motif 2, Ϫ574 GCAGACTGGCCCAG Ϫ561 ). We then performed ChIP-qRT-PCR in PB-mTfcp2l1 46C mESCs and found a direct interaction between Tfcp2l1 and the Esrrb promoter region containing the motif 1 sequence (Fig. 1G). This suggested an interaction between Tfcp2l1 and the Esrrb promoter. To further determine whether Tfcp2l1 is a functional activator of the Esrrb promoter, Tfcp2l1 was co-transfected with reporter vectors that drive the expression of luciferase under the control of Esrrb promoter fragments (pGL3-Esrrb), containing motif 1 or mutated motif 1 sequences (Mut) (Fig. 1H). Under these conditions, we observed a 2.7-fold increase in the WT promoter activity relative to the mutant sequence in PB-mTfcp2l1 cells (Fig. 1I). Collectively, these results suggest that Esrrb is positively regulated and activated by Tfcp2l1 in mESCs.

Esrrb mediates the self-renewal-promoting effects of Tfcp2l1 in mESCs
Tfcp2l1 overexpression can substitute for LIF or CHIR99021 to support mESC self-renewal (17,23,28). To determine whether Tfcp2l1 functions in maintaining mESC pluripotency depend on Esrrb in serum or serum-free conditions, we first decreased Esrrb expression in PB-mTfcp2l1 mESCs with two shRNAs specific to mouse Esrrb mRNA (mEsrrb sh#1 and mEsrrb sh#2). qRT-PCR analysis confirmed that the Esrrb transcript levels were decreased (40 -60%) in these cells ( Fig. 2A). Next, we cultured these cells in serum-containing medium without LIF for 8 days and found that scramble control mESCs retained normal mESC morphology, positive AP activity, and expression of the pluripotency marker OCT4 (Fig. 2, B-D), whereas knockdown of Esrrb induced differentiation in PB-mTfcp2l1 mESCs (Fig. 2, B-D). To further verify this observation, we seeded these cell lines in serum-free N2B27 medium supplemented with PD0325901. After three passages, we found that down-regulation of Esrrb extinguished the pluripotency supported by PB-mTfcp2l1, as indicated by the flat cell morphology and decreased AP activity (Fig. S1, A and B). Overall, these data suggest that Tfcp2l1 requires Esrrb to maintain the naïve pluripotency of mESCs.
Because knockdown of Tfcp2l1 impairs the undifferentiated state of mESCs under LIF conditions, we consequently wanted

Tfcp2l1 promotes ESC self-renewal via distinct targets
to test whether elevated expression of Esrrb diminishes this phenotype. First, we overexpressed HA-tagged mouse Esrrb using a PB vector (PB-mEsrrb), which caused significant increases in ESRRB (Fig. 2E). Second, we decreased Tfcp2l1 expression in PB and PB-mEsrrb mESCs with two shRNAs specific to mouse Tfcp2l1 mRNA (mTfcp2l1 sh#1 and mTfcp2l1 sh#2). qRT-PCR analysis confirmed that the Tfcp2l1 transcript levels were decreased (40 -60%) in these cells (Fig. 2F). After being cultured in serum-containing medium without LIF for 8 days, only the PB-mEsrrb mESCs infected with the scramble or Tfcp2l1 shRNA lentivirus displayed ESC morphological characteristics and expressed the pluripotency marker OCT4, whereas PB mESCs infected with the scramble or Tfcp2l1 shRNA lentivirus differentiated (Fig. 2, G and H). Notably, Tfcp2l1 protein and Esrrb protein also interact in mESCs (29). Therefore, Tfcp2l1 promotes mESC self-renewal via Esrrb at the protein and transcriptional levels.

Tfcp2l1 promotes ESC self-renewal via distinct targets
tiated or died, whereas an average of 28 AP-positive colonies were generated from 1 ϫ 10 5 PB-mTfcp2l1/scramble transfectants in three independent replicates. In parallel experiments, an average of 13 and 9 colonies were obtained from PB-mTfcp2l1/mEsrrb sh#1 and PB-mTfcp2l1/mEsrrb sh#2 transfectants, respectively (Fig. 3, C and D). These ESC-like colonies tolerated single-cell dissociation and could be continuously expanded in LIF/serum. At the same time, they acquired the mESC-specific transcriptional program and expressed high levels of Nanog, Stella, Rex1, and Nr0b1, but a low level of the EpiSC-specific marker Fgf5 (Fig. 3E); this suggests that they had been reprogrammed into a naïve state. Collectively, these results indicate that augmented Esrrb expression is likely important for Tfcp2l1 to convert mouse EpiSCs to the ESC state.

KLF4 is critical for TFCP2L1 to maintain hESC identity
Ectopic expression of TFCP2L1 can bypass the requirement for activin A and bFGF in short-term self-renewal of hESCs, whereas suppression of TFCP2L1 abolishes hESC pluripotency (22,24). To screen the functional targets of human TFCP2L1, we performed an RNA-Seq analysis. We identified 623 genes that were up-regulated and 449 genes that were down-regulated at least 2-fold by human TFCP2L1 overexpression (PB-hTfcp2l1) compared with empty vector PB (GEO accession no GSE115075) (Fig. 4A). Among these increased genes, we focused on pluripotency genes and identified KLF4 and KLF5. qRT-PCR was then used to confirm the mRNA levels of these two candidates in PB and PB-hTfcp2l1 cells (Fig. 4B). Subsequently, to examine whether the expression of the two candidates decreased during hESC differentiation, such as TFCP2L1, we performed monolayer differentiation. As shown in Fig. 4C, KLF5 is expressed in both pluripotent hESCs and differentiated cells, whereas the Klf4 transcript is down-regulated as hESCs differentiate (Fig. 4C). In particular, the change in KLF4 expression is similar to that seen for TFCP2L1, as well as other pluripotency markers, such as OCT4 and NANOG (Fig. 4C), which implies that KLF4 contributes to hESC maintenance and that there is association between KLF4 and TFCP2L1.

Tfcp2l1 promotes ESC self-renewal via distinct targets
To evaluate the ability of KLF4 to promote hESC self-renewal, we constructed a human KLF4-overexpressing hESC line with the PB vector (PB-hKlf4) (Fig. 4D). After cultured in hESC basal medium without activin A and bFGF for 6 days, PB cells differentiated, whereas PB-hKlf4 cells maintained an undifferentiated morphology and sustained AP-positive activity (Fig. 4E). Accordingly, immunofluorescence showed high expression of the pluripotency marker OCT4 (Fig. 4F). However, similar to TFCP2L1 overexpressing hESCs (22), PB-hKlf4 ESCs exhibited some evidence of differentiation at the edge of each colony and could not been maintained for long-term cultures. Therefore, overexpression of KLF4 enables short-term self-renewal of hESCs.
To demonstrate that Klf4 is a key downstream mediator of Tfcp2l1, we knocked down the human KLF4 transcript (hKlf4 sh#1 and hKlf4 sh#2) in PB-hTfcp2l1 transfectants (Fig. 4G). Knockdown of KLF4 resulted in PB-hTfcp2l1 hESC differentiation in the absence of activin A and bFGF, as indicated by the flat cell morphology, decreased AP activity and low expression of OCT4 (Fig. 4, H and I). Conversely, they highly expressed differentiation genes, such as GATA4, GATA6, and FOXA2 (Fig. 4I), suggesting that knockdown of KLF4 eliminates the ability of TFCP2L1 to promote self-renewal. To determine whether TFCP2L1 directly binds to the KLF4 locus, we analyzed the TFCP2L1-binding consensus motifs (Fig. 1F) and investigated the role of TFCP2L1 in the regulation of KLF4 expression. To detect whether TFCP2L1 binds to the KLF4 promoter, we predicted several potential binding sites within the KLF4 gene (from Ϫ2500 to ϩ200) from the JASPAR CORE database. Then we used ChIP-qRT-PCR in PB-hTfcp2l1 hESCs and found the direct interaction of TFCP2L1 with the KLF4 promoter (Fig. 4J). To further characterize the role of TFCP2L1 in KLF4 expression, we first constructed WT and mutant pGL3-Klf4 cells (Fig. 4K) and transfected the modified mESCs with the TFCP2L1 gene. Under these conditions, we observed that TFCP2L1 overexpression enhanced the luciferase activity in pGL3-Klf4 cells. TFCP2L1 overexpression did not increase the KLF4 promoter activity in mutant pGL3-Klf4 cells (Fig. 4L), indicating that KLF4 is a direct target of TFCP2L1. To further investigate whether TFCP2L1 protein and KLF4 protein interact in hESCs, we performed co-immunoprecipitation in flag-tagged TFCP2L1 and HA-tagged KLF4 overexpressing hESCs. We found that the TFCP2L1 and KLF4 proteins also interact (Fig. S2). Taken together, these results suggest that KLF4 is a direct target of TFCP2L1 and that the function of the latter in hESC self-renewal is largely dependent on KLF4. It will be of great interest to investigate whether KLF4 regulates its own expression via TFCP2L1.
To verify that ESRRB is not essential for human TFCP2L1 to promote hESC self-renewal and that Klf4 is dispensable for

Tfcp2l1 promotes ESC self-renewal via distinct targets
mouse Tfcp2l1 to support mESC stemness, we repressed human ESRRB in PB-hTfcp2l1 hESCs (Fig. S3A) and decreased mouse Klf4 in PB-mTfcp2l1 mESCs (Fig. S4, A and B), respectively. As expected, all of these ESCs remained undifferentiated (Figs. S3, B and C, and S4, C and D). These observations further demonstrate that the molecular mechanism by which Tfcp2l1 promotes self-renewal is different in naïve and primed pluripotency states.

Discussion
Tfcp2l1 plays an important role in maintaining mouse and human ESC identity. Although many modulators that regulate Tfcp2l1 expression have been explored, such as the LIF/STAT3 signaling pathway (16,17), Wnt/␤-catenin signaling pathway (17,23,28), JMJD1A (30) and Oct4 (31), it remains unclear how Tfcp2l1 safeguards ESC pluripotency. It may act as either an activator or as a repressor of transcription in ESCs. To resolve these issues, we sought to determine the direct targets of Tfcp2l1 and identified the Esrrb and Klf4 transcription factors. Surprisingly, our findings suggest that Tfcp2l1 preferentially activates Esrrb to reinforce the pluripotent status of mESCs, whereas Tfcp2l1 largely depends on Klf4 to promote hESC self-renewal (Fig. 5). To the best of our knowledge, this may be the first report demonstrating that one transcription factor (Tfcp2l1) maintains the undifferentiated state of naïve and primed pluripotency through induction of different downstream targets.
Although we previously reported that pluripotency gene Nanog is important for Tfcp2l1 to promote mESC self-renewal, it is unclear whether Nanog is directly controlled by Tfcp2l1 (17). In this study, we determined that Esrrb is a direct target of Tfcp2l1 and could mediate the influence of Tfcp2l1 on maintaining and inducing naïve pluripotency (Figs. 1-3). This is unsurprising, given that both Tfcp2l1 and Esrrb share many similar and associated features in mESCs. First, both genes are specific markers of pluripotency that are negatively regulated by the GSK3/TCF3 axis, and overexpression of either phenocopies the GSK3 inhibitor effect if supporting mESC pluripotency (17,23,28). Second, their expression levels both are dramatically reduced upon mESC differentiation into EpiSCs (16,17,32). Thus, overexpression of either factor not only renders mESCs independent of LIF but also reprograms EpiSCs to enter a naïve pluripotency state (16,17,32,33). Third, LIF/STAT3 signaling is necessary for the up-regulation of Esrrb expression induced by Nanog and the GSK3 inhibitor in partially reprogrammed cells (pre-iPSCs) (34), although Esrrb was shown to be a direct downstream target of Nanog and Wnt/␤-catenin signaling in mESCs (28,32). Importantly, overexpression of Esrrb can resume the reprogramming halted by inhibition of LIF/STAT3 signaling (34). Esrrb thus serves as an LIF activitydependent downstream effector for the establishment of complete pluripotency establishment (34). Nevertheless, Esrrb is not a direct target of STAT3 (16,17,33). Tfcp2l1, as one of the most important direct targets of STAT3 (16,17), may bridge the indirect interactions between the LIF/STAT3 pathway and Esrrb. Of note, LIF/serum conditions allow self-renewal of Esrrb-null mESCs (28,32), whereas other groups found that knockdown of Esrrb impairs LIF-mediated mESC self-renewal (35, 36). These inconsistencies may be due to the cross-compensatory functions between Esrrb and other genes and may be triggered by complete loss but not down-regulation of Esrrb. For example, Esrrg can replace Esrrb to induce the generation of iPSCs (37). It will be of great interest to identify the gene redundant with Esrrb in Esrrb-null mESCs.
In disagreement with the role in mESCs, elevated expression of TFCP2L1 does not induce ESRRB expression in human ESCs (Fig. 4A). We further explored this and demonstrated that KLF4 is a direct target of TFCP2L1, and it can mediate TFCP2L1-induced self-renewal in hESCs (Fig. 4, E-I). In fact, great progress has been made around the roles of Klf4 and Tfcp2l1 in mESCs, but little was known about their roles in hESCs until now. In mESCs, both genes are direct targets of the LIF/STAT3 signaling pathway and up-regulate Nanog expression (16,17,38). Meanwhile, they maintain and induce naïve

Tfcp2l1 promotes ESC self-renewal via distinct targets
pluripotency (16,17,39). In hESCs, the KLF4 transcript level was markedly reduced upon hESC differentiation, which is similar to TFCP2L1 (22,40). Elevated expression of KLF4 transcription inhibits neural differentiation of hESCs and enhances pluripotency marker expression (41). On the other hand, KLF4 is a Yamanaka factor that can induce iPSCs from human fibroblasts (42). All of these data highlight an important characteristic of KLF4 in hESCs and imply that the interaction between TFCP2L1 and KLF4 is likely a key contribution of Tfcp2l1 to promote hESC self-renewal. Interestingly, in mESCs, Klf4 has overlapping functions with Klf2 and Klf5 in LIF/serum medium (43,44). However, in hESCs, we found that knockdown of KLF4 is enough to trigger differentiation in TFCP2L1-overexpressing hESCs (Fig. 4, G-I). These may be caused by the different culture conditions. In addition, KLF proteins exhibit unequal potency (44). For example, depletion of Klf2 is sufficient to abolish mESC self-renewal in 2i conditions (45). Furthermore, no single Klf is sufficient to restore self-renewal of triple-knockout mESCs (44). Further studies are necessary to address the detailed mechanism of TFCP2L1 and KLF4 in maintaining hESC pluripotency.
In summary, we found two distinct transcription factors downstream of Tfcp2l1 that are functionally important for the self-renewal of mouse and human ESCs. These results provide new insights into the fully understanding of the regulatory circuitry that maintains naïve and primed pluripotency and will help to derive ESCs from different species and ultimately to safely use ESCs for future translational applications.

Plasmid construction
The coding regions of the Tfcp2l1, Esrrb, and Klf4 genes were inserted into the BglII and XhoI sites of the PiggyBac trans-poson vectors carrying a Flag or HA tag. The shRNA-expressing plasmids were constructed according to the pLKO.1-TRC protocol (Addgene plasmid no. 10878). The target-specific shRNA sequences for Tfcp2l1, Esrrb, and Klf4 were inserted into the AgeI and EcoRI sites. The shRNA sequences are as follows: mEsrrb sh#1, CGATTCATGAAATGCCTCAAA; mEsrrb sh#2, GCCGAGGACTATATCATGGAT; hKlf4 sh#1: GCCAGAATTGGACCCGGTGTA; and hKlf4 sh#2, GCCT-TACACATGAAGAGGCAT. Details of the mouse and human Tfcp2l1 shRNA vectors have been reported in our previous studies (17,22). The target specific shRNA sequences for Tfcp2l1, Esrrb, and Klf4 used in this study are listed in Table S1.

Western blotting analysis
Western blotting was performed according to a standard protocol. Briefly, the cells were lysed in radioimmune precipitation assay buffer (P0013B, Beyotime Biotechnology, China). The protein concentrations of the samples were estimated with a BCA kit (PA115, Tiangen Biotech), and then the samples were separated on 10% PAGE gels and electrotransferred onto polyvinylidene difluoride membranes. Probing was performed with specific primary antibodies and horseradish peroxidase-conjugated secondary antibodies. We used the following primary antibodies: Flag (F1804, Sigma, 1:2000), HA (3724S, Cell Signaling Technology, 1:2000), and ␣-tubulin (32-2500, Invitrogen, 1:5000). Signals were detected using a Pro-Light horseradish peroxidase kit (PA112, Tiangen Biotech).

qRT-PCR
Total RNA was extracted using the TRIzol Up Plus RNA kit (ER501-01, TransGen Biotech). cDNA was synthesized from 1 g of total RNA using the TransScript all-in-one first-strand cDNA synthesis SuperMix for qPCR (One-Step gDNA removal, AT341-02, TransGen Biotech) according to the manufacturer's instructions. qRT-PCR was carried out with Top Green qPCR SuperMix (AQ131-04, TransGen Biotech) in a PikoReal real-time PCR machine (Thermo Scientific). The relative expression level was determined by the 2-⌬Cq method and normalized to ␤-actin expression. The primers used are listed in Table S2.

Immunostaining
Immunostaining was performed via standard protocols. Briefly, the cells were fixed in 4% paraformaldehyde for 20 min and incubated at 37°C in blocking buffer (PBS containing 5% BSA and 0.2% Triton X-100). The cells were then incubated with the Oct4 primary antibody at 4°C overnight. After three times washes with PBS, the cells were then incubated with secondary antibody for 1 h at 37°C. The nuclei were stained with Hoechst 33342 (H3570, Invitrogen, 1:10,000) for 30 s.

AP activity assay
The cells were fixed in 4% paraformaldehyde for 2 min at room temperature, washed in PBS, and incubated in AP staining reagent (85L2-1KT, Sigma) for 30 min at room temperature in the darkroom. After two washes with PBS, the cells were visualized under a Leica DMI8 microscope.

Tfcp2l1 promotes ESC self-renewal via distinct targets EpiSC reprogramming
A total of 1ϫ10 5 transfectants were plated in 0.1% gelatincoated 6-well plates and cultured in mESC medium supplemented with LIF and 2i for 12 days. Then the number of APpositive clones generated from the CD1 EpiSCs was counted under a Leica DMI8 microscope.

ChIP assay
ChIP was performed using the ChIP assay kit (P2078; Beyotime Biotechnology) according to the manufacturer's protocol. Anti-FLAG antibody (F1804, Sigma) was used for immunoprecipitation, and IgG was used as a control antibody. ChIP enrichment was determined by qRT-PCR. The primer sequences and locations within the Esrrb and Klf4 genes are listed in Tables S3 and S4, respectively.

Luciferase assay
The promoter regions of Esrrb (from Ϫ1450 to Ϫ1201) and Klf4 (from Ϫ1250 to Ϫ1051) were cloned into pGL3-base plasmid; the new constructs were named pGL3-Esrrb and pGL3-Klf4, respectively. The primers sequences are listed in Tables S5 and S6. They were co-transfected into 46C mESCs or hESCs, along with the Tfcp2l1-overexpressed construct, or empty vector, with a Renilla luciferase plasmid. After 48 h, luciferase activities were measured with the dual luciferase assay kit. (FR201, TransGen Biotech).

Co-immunoprecipitation
Cell extracts were prepared using Nonidet P-40 lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 0.5% Nonidet P-40, and protease inhibitors). The supernatant was collected and incubated with 10 l of anti-FLAG affinity gel (SG4110-16, GNI, Tokyo, Japan) for 2 h at 4°C. The beads were then washed six times with lysis buffer and resuspended in 100 l of 1ϫ SDS sample buffer for Western blotting analysis.

Accession number
RNA-seq data have been deposited to the NCBI Gene Expression Omnibus under the accession number GSE115075.

Statistical analysis
All data are reported as the means Ϯ S.D. Student's t test was used to determine the significance of differences using GraphPad Prism 6. p Ͻ 0.05 was considered statistically significant. Sample size and p values have been included in the figure legends.