Author contributions: K.K.-K.C.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript; J.Z.: collection and/or assembly of data, data analysis and interpretation, manuscript writing; N.-Y.C.: collection and/or assembly of data, data analysis and interpretation; Y.-S.C.: collection and/or assembly of data, data analysis and interpretation; H.S.S.: collection and/or assembly of data; T.K.S.: collection and/or assembly of data; S.K.-W.O.: financial support, administrative support; H.-H.N.: provision of study material, data analysis and interpretation; A.B.-H.C.: financial support, administrative support.
First published online in STEM CELLS EXPRESS June 11, 2009.
Insight into the regulation of core transcription factors is important for a better understanding of the molecular mechanisms that control self-renewal and pluripotency of human ESCs (hESCs). However, the transcriptional regulation of NANOG itself in hESCs has largely been elusive. We established a NANOG promoter luciferase reporter assay as a fast read-out for indicating the pluripotent status of hESCs. From the functional cDNA screens and NANOG promoter characterization, we successfully identified a zinc finger transcription factor KLF4 and a homeodomain transcription factor PBX1 as two novel transcriptional regulators that maintain the pluripotent and undifferentiated state of hESCs. We showed that both KLF4 and PBX1 mRNA and protein expression were downregulated during hESC differentiation. In addition, overexpression of KLF4 and PBX1 upregulated NANOG promoter activity and also the endogenous NANOG protein expression in hESCs. Direct binding of KLF4 on NANOG proximal promoter and PBX1 on a new upstream enhancer and proximal promoter were confirmed by chromatin immunoprecipitation and electrophoretic mobility shift assay. Knockdown of KLF4/PBX1 or mutation of KLF4/PBX1 binding motifs significantly downregulated NANOG promoter activity. We also showed that specific members of the SP/KLF and PBX family are functionally redundant at the NANOG promoter and that KLF4 and PBX1 cooperated with OCT4 and SOX2, and transactivated synergistically the NANOG promoter activity. Our results show two novel upstream transcription activators of NANOG that are functionally important for the self-renewal of hESC and provide new insights into the expanded regulatory circuitry that maintains hESC pluripotency. STEM CELLS 2009;27:2114–2125
Human ESCs (hESCs) are pluripotent cells derived from the inner cell mass (ICM) of the blastocyst, an organization of cells formed during early embryogenesis. The defining properties of ECSs include their ability to proliferate indefinitely while remaining undifferentiated in culture and their pluripotent capacity to generate every cell type in the body [1–3]. Systematic, genome-wide transcriptome profiling has identified hundreds of genes, including several transcription factors, which have expression patterns tightly correlated with ESC pluripotency [4–7]. Therefore, recent advances in functional genomics have enabled genome-wide genetic studies in stem cells for identification of important genes crucial to a specific biological pathway or sufficient to confer a particular cellular phenotype [8–10].
The maintenance of self-renewal and pluripotency in ESCs are controlled by extrinsic signaling pathways and intrinsic self-renewal factors and epigenetic modifications [11, 12]. Among several external signaling pathways that are involved in maintenance, fibroblast growth factor (FGF) family members and transforming growth factor β (TGFβ)/activin are important in promoting self-renewal of hESCs in feeder-free culture [13–15]. For intrinsic factors, the homeodomain transcription factors NANOG and OCT4, and the high motility group-box containing transcription factor SOX2, constitute a conserved core transcriptional regulatory network that is essential for specifying the undifferentiated state of both hESCs and mouse ESCs (mESCs) [16–18]. NANOG is specifically expressed in the pluripotent cells of the ICM and epiblast and is rapidly downregulated on differentiation of hESCs . Disruption of Nanog leads to mESC differentiation to the extraembryonic endoderm [20, 21]. Inhibition of NANOG gene expression also causes hESC differentiation to extraembryonic cell lineages [19, 22], and overexpression allows feeder-independent proliferation of hESCs . Recently, it has been shown that, although downregulation of Nanog predisposes ESCs toward differentiation, ESCs can still self-renew indefinitely in the total absence of Nanog . NANOG is considered a core element of the pluripotent transcriptional network and stabilizes the pluripotent state of ESCs. However, the mechanisms regulating NANOG expression in hESCs are not well characterized.
NANOG, OCT4, and SOX2 co-occupy and regulate their own promoters together with other developmental genes with diverse functions and collaborate to form an extensive regulatory circuitry including autoregulatory and feed-forward loops [16, 25, 26]. Although there is a considerable amount of information on the downstream target genes of NANOG, little is known about the upstream transcriptional regulation of NANOG in hESCs. cis-regulatory elements essential for NANOG expression in hESCs have previously been mapped to an upstream 404-base pair (bp) proximal promoter region containing the Octamer/Sox element . These suggest that OCT4-SOX2 act to promote NANOG transcription in pluripotent cells. However, Nanog mRNA is present in early mouse blastocysts with undetectable levels of Oct4. Moreover, a high level of Nanog is beneficial to ESC self-renewal, but overexpression of Oct4 induces differentiation . Conversely, Oct4 and Sox2 can be found in precursor cells of the blastocyst that are Nanog negative . In addition, overexpression of both Oct4 and Sox2 in differentiated cells cannot rescue Nanog expression . Taken together, these findings indicate the presence of other transcription factors, in addition to OCT4 and SOX2, which may be responsible for the pluripotent-specific expression of NANOG in hESCs.
To address the molecular mechanisms of pluripotent cell specific expression, we investigated the regulatory elements that are involved in the control of NANOG transcription. In this study, we established a NANOG promoter luciferase reporter assay as a fast read-out for indicating the pluripotent status of hESCs. This assay system was used for functional cDNA screens to identify genes that maintain the undifferentiated hESC state. From the screen, KLF4 and PBX1 were identified as novel regulators that maintained the undifferentiated state of hESCs by regulating the expression of NANOG. Promoter characterization, chromatin immunoprecipitation (ChIP), electrophoretic mobility shift assay (EMSA), and luciferase assay confirmed that KLF4 and PBX1 directly bind to the NANOG promoter and independently regulate NANOG expression. Furthermore, NANOG promoter activity is transactivated synergistically by KLF4 and PBX1 and in cooperation with OCT4 and SOX2. Thus, our results presented here extend the understanding of the transcription regulatory network in hESCs where KLF4 and PBX1 act as direct transcriptional activators of NANOG and may be novel integral components of core regulatory circuitry in hESCs.
MATERIALS AND METHODS
Sequence Analysis and Plasmid Construction
Upstream human NANOG genomic sequences on human chromosome 12 (accession no. NT_000012.9) were retrieved from the public databases at the National Center for Biotechnology Information (www.ncbi.nlm.nih.gov). Comparative DNA analysis and sequence alignment comparisons between human, monkey, and mouse were performed using the University of California Santa Cruz (UCSC) Human Genome Browser at the UCSC Genome Bioinformatics website (http://genome.ucsc.edu) and the multiple sequence analysis tool ClustalW2. For NANOG promoter luciferase (NANOG-Rluc) reporter plasmid construction, the 5-kb NANOG promoter region (−4941/+239 relative to the transcription start site) was amplified from genomic DNA of hESCs. For details of plasmid construction, please refer to the supporting information data .
Cell Culture, Transfection, and Differentiation
hESC lines HES-2 and HES-3 from ES Cell International (Singapore, http://www.escellinternational.com) was cultured in medium conditioned by mitomycin-C-inactivated immortalized mouse embryonic fibroblast (ΔE-MEF) feeder supplemented with 4 ng/ml of FGF-2 (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) on Matrigel (BD Bioscience, San Diego, http://www.bdbiosciences.com)-coated plates as previous described . 293FT cells (Invitrogen) were cultured at 37°C/5% CO2 in complete medium containing Dulbecco's modified Eagle's medium (DMEM; high glucose) and 10% fetal bovine serum (FBS) according to the manufacturer's protocol.
For cDNA overexpression functional screens, hESCS were seeded with ΔE-MEF conditioned medium onto matrigel-coated 24-well plates for 3 days before transient transfection. NANOG-Rluc reporter vector was cotransfected along with each cDNA clone in expression vectors and also with internal control vector (CMV-Fluc, CMV promoter driven Firefly luciferase) into hESCS using Lipofectamine2000 reagent (Invitrogen), following the manufacturer's protocol. Twenty-four hours after transfection, medium was changed to either ΔE-MEF conditioned medium or embryoid body differentiation medium (KO-DMEM supplemented with 20% FBS, 1% nonessential amino acids, 1 mM L-glutamine, and 1% penicillin-streptomycin [all from Invitrogen] and 0.1 mM β-mercaptoethanol [Sigma, St. Louis, http://www.sigma.com]) and cultured for 2 days with daily medium changing. Transfections were performed in triplicate and repeated at least three times. Transfection of 293FT cells was performed in a similar way as hESCs but without changing medium after transfection. To differentiate hESCs as EBs, hESCs were dissociated into clumps and cultured in suspension as EBs in differentiation medium on low attachment plate (Corning Life Sciences, Acton, MA, http://www.corning.com/lifesciences). The culture was maintained in suspension for 7 days. The EBs were dissociated and plated for another 7 and 14 days on gelatin-coated dishes .
RNA samples were extracted from undifferentiated hESCs and day 7 and 14 EBs using Qiagen RNAeasy kit (Hilden, Germany, http://www1.qiagen.com) according to the manufacturer's instructions and reverse transcribed into cDNA using Superscript III reverse transcriptase (Invitrogen). The cDNAs were used for reverse transcriptase-polymerase chain reaction (RT-PCR) analyses with primer pairs listed in supporting information Table 1. Quantitative real-time PCR analyses were performed using an ABI PRISM 7500 Sequence Detection System and SYBR green PCR master mix (Applied Biosystems, foster City, CA, http://www.appliedbiosystems.com), as previously described . Primers used for real-time PCR amplification are listed in supporting information Table 1.
Western Blot Analysis
Protein samples were electrophoresed on a SDS polyacrylamide gel (NuPAGE 4-12% gradient gel; Invitrogen) and were transferred onto a polyvinylidene difluoride membrane (Millipore, Billerica, MA, http://www.millipore.com). The membrane was probed with rabbit anti-NANOG antibody (1:250; Abcam, Cambridge, U.K., http://www.abcam.com), goat anti-KLF4 antibodies (1:1000; Abcam), or rabbit anti-PBX1 antibody (1:1000; Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com) and followed by horseradish peroxidase-conjugated anti-rabbit or anti-goat secondary antibodies (Bio-Rad, Hercules, CA, http://www.bio-rad.com). Loading consistency was determined with mouse anti-β-actin (1:3000; Santa Cruz Biotechnology, Santa Cruz, CA, http://www.scbt.com). Then, the membrane was detected by ECL Detection Reagents (GE Healthcare, Piscataway, NJ, http://www.gehealthcare.com) and exposed onto x-ray films (Amersham Biosciences).
Luciferase Reporter Assay
A dual luciferase system (Promega, Madison, WI, http://www.promega.com) was used. Please refer to supporting information data for details .
RNA Interference Knockdown Analysis
hESCs were seeded onto matrigel-coated 24- or 96-well plates 1 day before transfection. KLF4 or PBX1 shRNA plasmid (Superarray, Frederick, MD, http://www.sabiosciences.com) or siRNA against OCT4 (Qiagen) were cotransfected together with NANOG-Rluc reporter plasmid and also with internal control vector, CMV-Fluc, into hESCs using Lipofectamine 2000 reagent (Invitrogen), following the manufacturer's protocol. Cells were harvested 72 hours after transfection and assayed for luciferase activity using the Dual-Luciferase assay system (Promega). The sequences targeted by shRNA are as follow: KLF4-1, TGGACTTTATTCTCTCCAATT; KLF4-2, CTGGACTTTATTCTCTCCAAT; PBX1-1, AGCTGTCACTGCTACCAATGT; PBX1-2, GATCCTGCGTTCCCGATTTCT.
ChIP was performed as described previously  with anti-KLF4 polyclonal antibodies (sc-12538; Santa Cruz Biotechnology) or anti-PBX1 polyclonal antibodies (sc-888; Santa Cruz Biotechnology). The details are described in the supporting information data.
Recombinant proteins of the PBX1 and KLF4 DNA binding domain (GST tagged) were used for gel shift assays. The details are described in the supporting information data.
Establishment of NANOG-Rluc Assay in hESCs
To establish a transient reporter-based assay for indicating the pluripotent status of hESC and identifying upstream transcriptional regulators of NANOG, functional cDNA screens were used to overexpress candidate genes in hESCs that were cultured in differentiation medium without FGF-2 after transfection. hESCs transiently transfected using a 5-kb human NANOG promoter luciferase (NANOG-Rluc) reporter showed high luciferase activity when cultured in medium conditioned by MEF. Conversely, quick downregulation of NANOG-Rluc activity was observed when hESCs were cultured in EB differentiation medium for 3 days (Fig. 1A), indicating that the reporter-based assay system reflects the pluripotent status of the hESCs. No change in CMV promoter luciferase activity was observed (Fig. 1A).
Identification of KLF4 and PBX1 from Functional Screens with the NANOG-Rluc Assay
To identify upstream transcriptional regulators of NANOG, this NANOG-Rluc reporter assay system was used for functional cDNA overexpression screens. Luciferase activity derived from this NANOG-Rluc vector, cotransfected into hESCs with the respective cDNA expression vector, was used as a fast and high signal-to-noise read-out for gain of activity when cultured in differentiation medium without FGF-2 after transfection. Thus, high luciferase activity read-out is indicative that candidate genes can induce the NANOG promoter activity when overexpressed in hESC cultured in differentiation medium. Candidate genes were chosen for the screen because they have similar expression pattern to NANOG during hESC differentiation. Most of them are ESC-specific transcription factors and showed downregulation during hESC differentiation as indicated from recent genome-wide profiling studies of mESCs and hESCs through transcriptome analyses [4–7]. From the screens, KLF4, which encodes a Kruppel-like zinc finger transcription factor , and PBX (pre-B cell leukemia) 1a and 1b, which encode two different isoforms of a homeodomain transcription factor [32, 33], caused a strong induction in NANOG-Rluc activity despite being in differentiation medium (three- to fourfold compared with the empty vector control), indicating either a direct or indirect effect on the activation of the NANOG promoter (Fig. 1B). Positive control OCT4/SOX2/NANOG only showed a slight but significant increase (p < .05) in this NANOG-Rluc assay (Fig. 1B). The overexpression of respective cDNA 3 days after transfection was confirmed by RT-PCR (data not shown). Transfected hESCs in differentiation medium reduced luciferase activity noticeably after 3 days of culture as shown in Figure 1A, which correlated to a downregulation of endogenous NANOG mRNA and protein as indicated by quantitative real-time PCR and Western blot analysis, respectively (Fig. 1C, 1D). Overexpression of KLF4 and PBX1 activated the NANOG-Rluc activity and also highly upregulated endogenous NANOG mRNA and protein expression. Results indicated that KLF4 had a higher activation than PBX1 on endogenous NANOG expression (Fig. 1C, 1D). These results indicate that this transient NANOG promoter luciferase reporter assay contains sufficient cis-regulatory information to recapitulate endogenous NANOG expression. Moreover, both KLF4 and PBX1 may target on NANOG promoter and act as transcriptional regulators.
In addition to this apparent activation of the endogenous NANOG promoter, cotransfections of the KLF4 and PBX1 expression vectors with a NANOG promoter EGFP reporter vector in 293FT cells led to a strong and moderate induction of EGFP expression, respectively, as detected by cellular EGFP fluorescence on the microscope and assayed quantitatively for EGFP fluorescence using a fluorometer (Fig. 1E, 1F). KLF4 was found to have the strongest induction (sevenfold relative to vector control), whereas PBX1 had lower induction but comparable to OCT4-SOX2 induction on NANOG promoter-driven EGFP expression (Fig. 1F). Together, these results provide strong evidence that the 5-kb NANOG promoter is a direct target of KLF4 and PBX1.
KLF4 and PBX1 Expression Are Associated with hESC Pluripotency
KLF4 and PBX1 were found to be expressed in undifferentiated hESCs and downregulated in differentiated cells and thus may play a role in regulating hESC pluripotency. We assayed the expression of KLF4 and PBX1 in hESCs differentiated as EBs after 7 and 14 days by RT-PCR (Fig. 2A). Similar to the trends observed for the NANOG gene, KLF4 and PBX1 transcript levels decreased in hESC-derived EBs from 7 to 14 days (Fig. 2A), relative to the undifferentiated hESC control. Quantitative real-time PCR was also performed to confirm the expression of KLF4 and PBX1 during hESC differentiated as EBs. KLF4 transcript level was markedly reduced to approximately 23% of its expression level in undifferentiated hESCs. Similarly, PBX1 expression levels showed a 40% reduction during EB differentiation (Fig. 2B). The decrease in KLF4 and PBX1 mRNA correlated with a comparable decrease in their protein expression during EB differentiation (Fig. 2C). Therefore, KLF4 and PBX1 gene expression is associated with the hESC pluripotent state, and its expression decreases as cells differentiate.
Comparative DNA Analysis Identified Conserved Regions of the NANOG Promoter
NANOG is a key marker of pluripotency in hESCs . As such, functional regulatory elements controlling its expression are expected to be conserved because of evolutionary constraints and purifying selection . To this end, a 5-kb 5′ upstream regulatory region of the human NANOG gene was submitted to the UCSC genome browser for multiple sequence alignment. Of the 5 kb of noncoding genomic sequence that we compared by comparative DNA analysis, we identified four areas of significant sequence conservation (data not shown). This comparative alignment analysis between this region and those of the monkey and mouse Nanog ortholog shows four conserved regions (CR1–CR4) of homology between these species (60–94% conservation; supporting information Fig. 1). A schematic representation of the four conserved regions in the NANOG 5-kb upstream region is shown in Figure 3A, denoting the relative size and location of each of the conserved regions. CR1 is within the proximal promoter region containing the OCT/SOX element [25, 26] (Fig. 3A). Together, these finding suggests that, in addition to this well-characterized OCT/SOX element, other functionally conserved cis-regulatory elements may exist to regulate expression of the NANOG gene.
Functional Analysis of the Conserved cis-Regulatory Regions of NANOG Promoter
To identify the conserved cis-regulatory elements that regulate the transcription of human NANOG gene, a series of deleted fragments of the 5′ upstream region were cloned into the luciferase reporter vector. The lengths of the deletion constructs from the 5′ and 3′ ends of the NANOG promoter were based on the identified CR1–CR4 containing putative enhancer sequences (Fig. 3A). Each deletion construct was transiently transfected into hESCs and 293FT cells (Fig. 3B). The entire wild-type 5-kb (−4941/+239, WT-pro) fragment showed approximately 25-fold higher promoter activity than vector control. Interestingly, deletion of CR4 (ΔCR4) or CR3 and CR4 (ΔCR3/4) increased the promoter activity by approximately 2- and 2.5-fold than the entire 5-kb fragment (WT-pro), respectively (approximately 50- and 75-fold higher than vector control, respectively; Fig. 3B). These results suggested that the upstream region (−4941/−1897) of the NANOG promoter containing CR3 and CR4 may contain negative cis-regulatory elements that regulate transcription activity. In addition, although the slight decrease in activity compared with ΔCR3/4 seen with the deletion of CR2 (ΔCR2/3/4) was statistically significant (p < .005), a much more notable reduction was observed with all the fragments containing a deletion of CR1 (ΔCR1, ΔCR1/4, ΔCR1/3/4; Fig. 3B). Taken together, the highest promoter activity was observed after deletion of the CR3 and CR4 fragment. These results suggested that important cis-regulatory pluripotent-specific elements lies within CR1 and CR2 of the NANOG promoter. This is consistent with previous studies that showed that the NANOG proximal promoter within CR1 is essential for pluripotent expression in hESCs [25, 26].
Specific Transactivation of the NANOG Promoter by KLF4 and PBX1
We next studied the specific location within this 5-kb NANOG promoter that was mediated by the transcriptional activation of KLF4 and PBX1. A series of NANOG promoter deletion reporter vector were cotransfected with the KLF4 expression vector into 293FT cells and assayed for luciferase activity. Through deletion constructs driving the luciferase reporter, overexpressed KLF4-mediated transactivation was localized to CR1. This region, when deleted either by itself or along with CR4 or with CR3 and CR4 (ΔCR1, ΔCR1/4, ΔCR1/3/4), significantly reduced luciferase activity to approximately 30% of the entire 5-kb wild-type NANOG promoter (WT-pro) activity, whereas deletion of CR4 itself or along with CR3 or with CR2 and CR3 (ΔCR4, ΔCR3/4, ΔCR2/3/4) had no significant effect (Fig. 3C). For overexpressed PBX1-mediated transactivation, similarly, we cotransfected PBX1 expression vector with a series of NANOG promoter deletion reporter vector into 293FT cells. Significant reduction of luciferase activity was observed when either CR2, CR3, and CR4 (ΔCR2/3/4) or only CR1 (ΔCR1) was deleted. Results showed that these fragments retained approximately 70% of wild-type promoter (WT-pro) activity and further deletion of CR1 along with potential negative regulatory element CR4 or CR3 and CR4 (ΔCR1/4, ΔCR1/3/4) resulted in a slight increase in luciferase activity. The highest PBX1-mediated transactivation was obtained from the fragment containing CR1 and CR2 (ΔCR3/4; Fig. 3D). Taken together, these results suggested that the important element(s) for PBX1 transactivation was localized to CR1 and CR2 of NANOG promoter and for KLF4 transactivation was localized to CR1 of NANOG promoter.
KLF4 and PBX1 Bind to and Regulate NANOG Expression
Because the NANOG promoter contained important elements (CR1 and CR2) for KLF4 and PBX1 transactivation, we asked whether NANOG is a direct target of KLF4 and PBX1. To identify whether there was a specific protein–DNA interaction between KLF4/PBX1 and the NANOG promoter, an EMSA was performed. Twelve different probes within CR1 and CR2 of the NANOG promoter were used for EMSA (Fig. 4A). Results showed that the recombinant DNA binding domain KLF4 binds to the CR1 of the NANOG proximal promoter, whereas PBX1 binds to two regions (CR1 and CR2) on the NANOG promoter (Fig. 4B). Supershift assay using hESC nuclear extracts also showed that the hESC protein complex mobility was reduced after incubation with anti-KLF4 or anti-PBX1 antibodies (supporting information Fig. 2). Using the ChIP assay, we further confirmed these interactions and showed that it was occurring in vivo in undifferentiated hESCs. The same region as in EMSA showed significant enrichment with KLF4 antibodies, and the same region within CR1 and another within CR2 showed significant enrichment with PBX1 antibodies (Fig. 4C). The same results were also obtained with another hESC line H1 as shown in supporting information Figure 3. The neighboring regions on both sides of these KLF4 and PBX1 elements were not significantly enriched, showing the specificity of the binding. Irrelevant mock ChIP did not show any significant enrichment in these corresponding regions (Fig. 4C; supporting information Fig. 3). In summary, these data clearly indicate that KLF4 and PBX1 bind to the NANOG promoter in vivo. Two KLF4 and three PBX1 consensus binding sites can be found at the peak region shown by KLF4 and PBX1 ChIP, respectively (Fig. 4C, 4D; supporting information Fig. 1) [35, 36]. Furthermore, mutagenesis experiments showed that both GGGGGTGTG and GGGGTGGG motif sequences were required for in vitro KLF4 protein–DNA interactions (Fig. 4D). Similarly, mutagenesis experiments also showed that all three motif (TCATGTT and ATTTCTT motifs on CR1 and AATCATC motif on CR2) sequences were required for in vitro PBX1 protein–DNA interaction (Fig. 4D). In summary, our results showed that KLF4 and PBX1 directly bind to the NANOG promoter through conserved binding motifs.
To establish a functional role between the binding sites and KLF4/PBX1-meditated NANOG promoter activity, mutagenesis was performed to generate mutated luciferase reporter constructs driven by NANOG promoter with disruptions within the KLF4/PBX1 binding motifs (Fig. 5A). Each mutant construct or wild-type construct was transfected into hESCs and assayed for luciferase activities to test the effect of these mutations on NANOG promoter activity. As shown in Figure 5B, disruption of KLF4 binding site 2 (KLF4-M1) or site 1 (KLF4-M2) resulted in a dramatic reduction (p < .005) in luciferase activities compared with wild-type NANOG promoter activity (−1879/+239, −795/+239). Simultaneous disruption of both sites (KLF4-M3) did not cause a further decline in luciferase activity, suggesting that both sites contribute equally to NANOG promoter activity and that each site on its own is insufficient for activity. For PBX1 motifs, mutations in PBX binding site 2 (PBX1-M1) or site 3 (PBX1-M2) significantly reduced up to 60% of NANOG promoter activity. A further reduction to 75% of wild-type promoter activity was observed when both sites were disrupted simultaneously (PBX1-M3). In addition, different multiple point mutations of PBX binding site 1 (PBX-M4, -M5, -M6), which are further upstream of the NANOG promoter, resulted in a loss of promoter activity (Fig. 5B). Similar reduction of the promoter activity was seen with the vector with mutated OCT/SOX elements in the NANOG promoter (OCT/SOX-M; Fig. 5B). Taken together, these results indicate that specific binding of KLF4 and PBX1 to their respective motifs on the NANOG promoter is necessary and important in regulating the expression of NANOG in hESCs.
To further address the functional roles of KLF4 and PBX1, KLF4 and PBX1 RNA interference (RNAi) constructs were cotransfected with the NANOG promoter driving luciferase reporter into hESCs. OCT4 depletion by RNAi was used as a positive control. The depletion of KLF4 and PBX1 reduced NANOG promoter activity to the same extent as mutating the KLF4 or PBX1 binding motifs (Fig. 5C), and differentiated cell morphology was observed in transfected hESCs (unpublished data). Reduction of endogenous NANOG expression in KLF4 or PBX1 knockdown hESCs was also confirmed by quantitative real-time PCR (Fig. 5D). Western blot analysis was performed to confirm the efficiency of KLF4 or PBX1 knockdown. KLF4 or PBX1 protein was markedly decreased in hESCs transfected with KLF4 or PBX1 RNAi constructs, respectively (Fig. 5E). In summary, all these data highly demonstrate that KLF4 and PBX1 directly bind to the NANOG promoter and independently regulate NANOG expression.
Functional Redundancy of the SP/KLF Family in NANOG Promoter Regulation
Previous transcriptome analyses showed that several Klf family and PBX protein family members are expressed in mESCs and hESCs, respectively, with some of them showing stem cell-specific expression patterns [37, 38]. We also confirmed their expression by quantitative real-time PCR (data not shown). To determine whether SP/KLF and PBX family members are functionally redundant in regulating NANOG promoter, we tested the ability of each to activate NANOG promoter in 293FT cells. As shown in Figure 6A, in addition to KLF4, KLF1, KLF2, and SP3 showed significant ability to transactivate the NANOG promoter, although their activities were less than that of KLF4, suggesting that the function of these genes overlap in hESCs. No significant activation of promoter activity was observed for other SP/KLF family members tested. On the other hand, in Figure 6B, other members of the PBX family tested were capable of activating NANOG promoter activity. Therefore, the data suggest that PBX1, PBX2, and PBX3 also shared redundant function in hESCs.
Combinatorial Effects of KLF4 and PBX1 in Regulation of NANOG
The fact that KLF4, PBX1, OCT4, and SOX2 can all activate the NANOG promoter raises the possibility that these proteins may cooperate and synergize with each other. To test for these combinatorial interactions between transcription factors at the NANOG promoter, we tested the ability of various combinations of these four gene products to transactivate the NANOG promoter. KLF4 and PBX1 were cotransfected alone or in different combinations with OCT4 and SOX2 with the NANOG promoter luciferase reporter in 293FT cells. As shown in Figure 6C, PBX1 alone was able to transactivate the NANOG promoter to around threefold compared with vector control. Addition of OCT4 and SOX2 together to PBX1 slightly enhanced this activation to about fourfold. Although OCT4 and SOX2 together can only slightly activate the NANOG promoter in 293FT cells, KLF4 alone significantly activated NANOG promoter to an approximate ninefold increase. Addition of OCT4 and SOX2 together with KLF4 resulted in a dramatic increase to 13-fold in transactivation of NANOG promoter activity. In addition, concurrent expression of both KLF4 and PBX1 resulted in a significant increase in transactivation of NANOG promoter compared with KLF4 or PBX1 alone and was comparable to KLF4 together with OCT4 and SOX2 transactivation. In a combination of all four factors, it showed the highest activation of NANOG promoter activity (∼16-fold increase; Fig. 6C). Taken together, these data imply that KLF4 and PBX1 synergistically transactivated the NANOG promoter and cooperated with OCT4 and SOX2 in regulating NANOG promoter activity.
Here we described the establishment of a transient NANOG promoter reporter-based assay for indicating the pluripotent status of hESCs. Since hESC lines were derived, much attention has been paid to improve the culture conditions for hESCs, with particular emphasis on defining the factors that are sufficient for maintaining hESC self-renewal and pluripotency, so that no animal feeder, matrix, or conditioned medium is required . Our results suggested that this transient NANOG promoter reporter-based assay (Fig. 1) provides an effective means of identifying direct regulators of NANOG or indirectly identifying important factors in maintenance of stem cell pluripotency. This system could also be extended to study the regulation of other pluripotent marker promoters in hESCs and could be easily applied for future drug screening or chemical screening to improve our current xeno-free culture conditions for hESC self-renewal and pluripotent maintenance.
In this study, we identified the regions involved in regulating human NANOG transcription and showed that multiple cis-acting elements, including putative enhancer regions, act positively in NANOG promoter activity. In addition, we showed for the first time that the region between −4941 and −1897 (CR3 and CR4) may act as a putative negative regulatory element for NANOG transcription and that the promoter region between nucleotides −1897 and +293 containing CR1 and CR2 is sufficient to provide minimal promoter activity in hESC (Fig. 3B). These results further extended the previous studies to suggest CR2 is another important cis-acting pluripotent-specific enhancer of NANOG promoter.
Our study also uncovered that KLF4 and PBX1 were identified as novel regulators that maintain the undifferentiated state of hESCs by regulating the expression of NANOG. KLF4 (also known as gut-enriched Kruppel-like factor) belongs to the KLF family of evolutionarily conserved zinc finger transcription factors that regulate numerous biological processes, including proliferation, differentiation, development, and apoptosis [40–42]. KLF4 functions as a transcriptional activator or repressor . Klf4 was found to cooperate with Oct4 and Sox2 to activate Lefty1 core promoter in mESC and results suggested that Klf4 has a physiological significance in activating a subset of Oct4 target genes . Overexpression of Klf4 in mESC was found to prevent differentiation in EBs formed in suspension culture, suggesting that Klf4 contributes to ES self renewal . KLF4 in ES cells corresponds closely with that of known regulators of pluripotency OCT4, NANOG, SOX2, which have high levels of expression in the undifferentiated state and decrease rapidly upon differentiation (Fig. 2). Our findings in hESC are consistent with results from mESC . The differences we observed in KLF4 expression levels between hESC and differentiated EB imply a potentially significant role for KLF4 in stem cell pluripotency. Recently many reports showed that generation of induced pluripotent stem (iPS) cells by reprogramming somatic cells requires overexpression of KLF4 together with OCT4, SOX2, and MYC [44–46]. All these reports highlight an important role of KLF4 and clearly showed a direct correlation between KLF4 and hESC pluripotency. Another important transcription factor PBX1, a homeobox gene, was originally discovered by their involvement in chromosomal translocation in acute pre-B cell leukemia [32, 33]. PBX1 belongs to the PBC subclass of TALE homeodomain protein, which is important for vertebrate development. Expression of PBX1 was found in a wide range of fetal and adult tissues of higher organisms. PBX1 functions as a cofactor and couples with other transcription factors such as Hox and MEIS/PREP [47, 48]. PBX1, together with MEIS1, was found cooperatively to bind to the Xenopus laevis Zic3 promoter and activate its expression . Recently, Zic3 has been shown to play an important role in maintenance of pluripotency in mESCs . In our study, PBX1 was highly expressed in hESCs and downregulated during differentiation (Fig. 2), suggesting PBX1 could be important in maintaining hESC pluripotency. Genome scale location analysis (chromatin immunoprecipitation coupled with DNA microarray) results indicated that NANOG is bound to the PBX1 promoter , whereas in HCT116 colon carcinoma cell lines, PBX1 is upregulated and shown to be a direct target of NANOG . More recently, Pbx1 was identified to regulate self-renewal of long-term hematopoietic stem cells by maintaining their quiescence . Collectively, these studies also suggest an important role of PBX1 in maintaining self-renewal and pluripotency of hESCs. Moreover, genes encoding PBX family proteins were modified by trimethylations of histone H3 lysine 4 (H3K4me3) and actively expressed in hESCs, but downregulated during EB differentiation . Taken together, these results suggest that PBX family proteins could be important in maintaining hESC pluripotency.
We reported here that KLF4 and PBX1 are both directly binding and regulating NANOG from the functional cDNA screens using the NANOG promoter reporter-based assay and characterization of the 5-kb NANOG promoter. These suggest that both KLF4 and PBX1 act as novel transcriptional regulators that maintain the undifferentiated state of hESCs. Our data are the first to directly link these transcription factors within the pluripotent transcriptional regulatory network of hESCs. Furthermore, because NANOG is known to be essential for maintaining pluripotency, our data further emphasize the position of KLF4 and PBX1 at the top of the regulatory network hierarchy. Our results clearly indicate that KLF4 binds to two consensus binding motifs in the CR1 of the NANOG promoter and PBX1 binds to three consensus binding motifs within CR1 and CR2 (Figs. 4 and 5). Interestingly, the location of these KLF4 binding motifs is very different from those in mouse Nanog promoter, which is located at the distal enhancer . In addition, three PBX binding motifs in the human NANOG promoter showed low homology to known consensus PBX1, PBX1/MEIS1, or PBX1-PREP1 binding sequences (supporting information Fig. 1) [36, 48, 53]. Nevertheless, PBX1 can still bind to these sites and regulate NANOG promoter activity (Figs. 4 and 5). Pbx proteins function as partners for nonhomeodomain proteins or independent of Hox proteins and may act even more broadly to modulate the activities of nonhomeodomain transcription factors as well . Thus, one possibility is that PBX1 may function in cooperation with other undefined factors in regulating NANOG expression. Additional experiments are needed to test this hypothesis
In addition, our study showed that KLF1, KLF2, and SP3 (members of the SP/KLF family), and PBX2/PBX3 are functionally redundant on the NANOG promoter (Fig. 6A, 6B), and they may exhibit similar binding properties in vitro and in vivo. KLF1, KLF2, and KLF4 are more closely related to one another than to other members of the KLF protein family . It is consistent with a recent study that showed that Klf family members were found to be functionally redundant and to exhibit highly similar DNA binding profiles in vitro and in vivo . More recently, studies showed that Klf4 could be replaced by Klf2 together with other three factors for reprogramming somatic cells to iPS cells. Klf1 also generated iPS cells but with a lower efficiency. Their results also showed the redundant function of Klf family protein . SP-like proteins and KLFs are highly related zinc-finger proteins that are important components of the eukaryotic cellular transcriptional machinery [53–55]. The observations here that SP3 are activators of NANOG transcription are largely consistent with previous studies involving the mouse Nanog promoter [56, 57]. Similarly, PBX family members also have redundant functions and may exhibit similar binding properties on the NANOG promoter. Studies showed that a feature of the Pbxl, Pbx2, and Pbx3 proteins is their extensive sequence identity both within and outside of their homeodomains. The similarities in their structures and patterns of expression suggest that Pbx proteins bind to a common DNA sequence and have overlapping function in most cell types . Pbx1-null mice result in embryonic lethality [59, 60]; however, Pbx3-deficient mice survive to term, but die within a few hours after birth from central respiratory failure . Pbx1 and Pbx3 have extensively overlapping patterns of embryonic expression, and these two related proteins could exhibit redundant functions. Indeed, in contrast with precocious deaths caused by the loss of Pbx1 or Pbx3, Pbx2-deficient mice are viable and display no obvious phenotype, despite widespread expression of the gene during embryogenesis . This suggests that the loss of Pbx2 is compensated by another member of the PBX family. Taken together, they are developmental important factors showing redundant function. Collectively, further in vitro and in vivo binding assays are needed to confirm that KLF1, KLF2, SP3, PBX2, and PBX3 co-occupy the NANOG promoter in a manner similar to KLF4 and PBX1 and integrate them into the KLF4/PBX1 and NANOG regulatory circuitry.
The central role of NANOG as a key factor of ESC pluripotency places strict requirements on the regulation of its expression, both spatially and temporally, to balance the number of undifferentiated cells with the ability for lineage commitment . Several studies have characterized the cis-regulatory elements in the mouse Nanog upstream region. Both the 5-kb upstream enhancer region and proximal promoter are reported to be bound and positively regulated by various transcription factors including Klfs and the Oct4-Sox2 complex [25, 26, 37]. Several transcriptional repressors have also been characterized to downregulate Nanog expression [64–66]. More recently, both TGFβ/activin- and BMP-responsive SMADs can bind and regulate the NANOG proximal promoter . Our data showed that KLF4 and PBX1 directly bind to the NANOG promoter and regulate its expression in hESCs and that NANOG promoter activity is synergistically transactivated by KLF4 and PBX1 and in cooperation with OCT4 and SOX2 (Fig. 7); both observations in mESCs and hESCs suggest a direct binding and regulation of KLF4 on NANOG. Our results presented here have expanded the understanding of the core transcriptional network in hESCs and also suggested that some of the mechanisms regulating human NANOG expression differ from that controlling mouse Nanog expression. A model for KLF4, PBX1, OCT4, and SOX2 on NANOG regulation that incorporates our findings is shown in Figure 7. The important conserved regulatory region (CR1) within the proximal promoter is bound and regulated by the OCT4-SOX2 heterodimer. Our results propose that SP/KLF and PBX family members might function redundantly though the same genomic locations on the human NANOG promoter; however, the existence of additional regulatory motifs for other related transcription factors cannot be excluded. Our study investigated the contribution of KLF4 and PBX1 proteins to NANOG promoter activity; it suggests that the in vivo expression of NANOG may involve a highly interconnected circuitry requiring cooperation between KLF4, PBX1, and OCT4-SOX2 to control endogenous NANOG expression specifically in hESCs (Fig. 7). Further studies are necessary to understand the roles of these KLF and PBX family proteins in pluripotent cells in vivo. Nevertheless, our study significantly identified novel transcriptional regulator important for hESC pluripotency and highlighted the expanded regulatory network in hESC maintenance.
This work was supported by the Biochemical Research Council of A∗STAR (Agency for Science Technology and Research), Singapore. We thank Selena Wu for help in hESC culture and technical assistance and Paul Robson for critical review of the manuscript.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.