Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Jan 31.
Published in final edited form as: Stem Cells. 2015 Feb;33(2):367–377. doi: 10.1002/stem.1870

Interrogation of a Context-Specific Transcription Factor Network Identifies Novel Regulators of Pluripotency

Ritu Kushwaha 1,2, Nirmala Jagadish 1,2, Manjunath Kustagi 6,2, Mark J Tomishima 3, Geetu Mendiratta 1, Mukesh Bansal 6, Hyunjae R Kim 6, Pavel Sumazin 6, Mariano J Alvarez 6, Celine Lefebvre 6, Patricia Villagrasa-Gonzalez 6, Agnes Viale 4, James E Korkola 1, Jane Houldsworth 1, Darren R Feldman 5, George J Bosl 5, Andrea Califano 6,7,8,*, R S K Chaganti 1,5,8,*
PMCID: PMC4305010  NIHMSID: NIHMS635905  PMID: 25336442

Abstract

The predominant view of pluripotency regulation proposes a stable ground state with coordinated expression of key transcription factors (TFs) that prohibit differentiation. Another perspective suggests a more complexly regulated state involving competition between multiple lineage-specifying TFs that define pluripotency. These contrasting views were developed from extensive analyses of TFs in pluripotent cells in vitro. An experimentally-validated, genome-wide repertoire of the regulatory interactions that control pluripotency within the in vivo cellular contexts is yet to be developed. To address this limitation, we assembled a TF interactome of adult human male germ cell tumors (GCTs) using the Algorithm for the Accurate Reconstruction of Cellular Pathways (ARACNe) to analyze gene expression profiles of 141 tumors comprising pluripotent and differentiated subsets. The network (GCTNet) comprised 1305 TFs, and its Ingenuity Pathway analysis identified pluripotency and embryonal development as the top functional pathways. We experimentally validated GCTNet by functional (silencing) and biochemical (ChIP-seq) analysis of the core pluripotency regulatory TFs POU5F1, NANOG, and SOX2 in relation to their targets predicted by ARACNe. To define the extent of the in vivo pluripotency network in this system, we ranked all TFs in the GCTNet according to sharing of ARACNe-predicted targets with those of POU5F1 and NANOG using an Odds-Ratio analysis method. To validate this network, we silenced the top 10 TFs in the network in H9 ES cells. Silencing of each led to downregulation of pluripotency and induction of lineage; 7 of the 10 TFs were identified as pluripotency regulators for the first time.

INTRODUCTION

Recent studies have provided deep insights into the core pluripotency regulatory circuitry of the embryonic stem (ES) cell state involving interaction between transcription factors (TFs), noncoding RNAs and Polycomb group (PcG) proteins (Marson et al., 2008; Jaenisch and Young, 2008; Schuettengruber and Cavalli, 2009; Young, 2011). These studies have lead to the concept that pluripotency is a stable ground state maintained by the coordinated expression of TFs, such as the canonical triad POU5F1, NANOG, and SOX2, which prohibits differentiation. In contrast, others have argued for an unstable nature of the pluripotency state, with competition between lineage-specifying TFs acting to maintain a delicate regulatory balance necessary for its maintenance (Loh and Lim, 2011). Consistent with such a view, Shu et al (2013) showed that lineage specifiers that are pluripotency rivals and are also not commonly expressed in ES cells can reprogram mouse somatic cells to pluripotency. These contrasting views suggest that in-depth elucidation of the mechanisms underlying pluripotency require a far more detailed view of the regulatory programs that control it, in vitro as well as in vivo. Although we and others have shown that evaluating the overlap of regulatory programs controlled by TF-pairs is critical in assessing their epistatic, synergestic, or complementary role in regulating specific pathophysiologic processes (Carro et al., 2010; Ravassi et al., 2010; Della Gatta et al., 2012), an experimentally-validated, genome-wide repertoire of transcriptional interactions (i.e., interactome) representing pluripotency and differentiation processes in stem cells is not yet available to allow these kinds of analyses to be performed. Thus, over all, the complete regulatory control landscape of the pluripotent cell remains largely elusive.

Recently, we have successfully assembled several cell-context-specific transcriptional networks for mammalian cells using the Algorithm for the Reconstruction of Accurate Cellular Networks (ARACNe) (Basso et al., 2005, 2010; Palomero, et al., 2006; Lim et al., 2009; Margolin et al., 2009; Zhao et al., 2009; Carro et al., 2010; Lefebvre et al., 2010; Cadeiras et al., 2011). These networks withstood rigorous experimental validation and their analysis helped elucidate previously unsuspected regulatory mechanisms (Zhao et al., 2009; Carro et al., 2010; Lefebvre et al., 2010; Della Gatta et al., 2012). To apply these methods to the reconstruction and interrogation of an interactome representing the in vivo regulatory landscape of pluripotent cells, we chose the adult human male germ cell tumor (GCT) as the model system. GCTs originate from lineage-restricted germ cells and a subset, embryonal carcinoma (EC), re-acquires pluripotency comparable to that of the blastocyst. EC cells represent a biological counterpart of ES cells; including presenting very similar gene expression profiles (GEPs) (Sperger et al., 2003). GCTs comprise two main histologic subsets, seminoma (SEM) and non-seminoma (NS) (Ulbright, 1993; Chaganti and Houldsworth, 2000). SEM presents as germ cell-like, non-differentiating tumors and NS comprises EC and its developmental derivatives teratoma, which displays somatic lineage differentiation, and choriocarcinoma and yolk sac tumor, which display features of extra-embryonal differentiation. EC development in vivo involves downregulation of pluripotency and upregulation of lineage differentiation.

We used ARACNe to analyze a large GEP dataset derived from fresh-frozen GCT biopsies representing all differentiation lineages and normal testis (Korkola et al., 2005, 2006, 2009). This analysis yielded the first de novo, genome-wide transcriptional interactome (GCTNet), representing the context-specific regulatory programs that determine the in vivo behavior of these tumors, including control of pluripotency, differentiation, and tumorigenesis. We evaluated the quality of the interactome with specific focus on established pathways controlling pluripotency and developed an experimentally validated map of the transcriptional programs controlled by POU5F1, NANOG, and SOX2. Time -series GEP analysis followed by silencing of these factors, as well as genome-wide chromatin immunoprecipitation (ChIP) assays, were used to validate their ARACNe-inferred targets and provide further insight into the specific, time dependent, functions regulated by them during loss of pluripotency and differentiation in vitro in NT2/D1 EC cells. Analysis of all genetic programs co-regulated by TF pairs in vivo in the interactome revealed the first genome-wide map of cooperative regulation in a pluripotent cell and its developmental derivatives and provided critical insights to identify additional candidate pluripotency regulators. Specifically, by ranking all TFs in the network based on the statistical significance of the programs they co-regulate with POU5F1 and NANOG, we identified several novel candidate pluripotency regulators. Functional validation of the 10 most significant regulators identified in this approach by lentivirus-mediated short hairpin RNA (shRNA) silencing in H9 ES cells showed that each of them was necessary to maintain the pluripotency state. These 10 TFs included 3 that were previously identified, and 7 that were previously not identified, as pluripotency regulators. These results thus demonstrate that the GCTNet interactome represents bona fide regulatory interactions in this tumor system; they further elucidate the novelty and complexity of the pluripotency control system.

MATERIALS AND METHODS

Additional details about the experimental procedures can be found in the Supplemental Information.

Gene expression profiles (GEPs) of GCT

The GEP data from 141GCT samples previously reported by us was obtained using Affymetrix U133A+B microarrays (data available in GEO database, accession No GSE 3218, Korkola et al., 2006, 2009).

Cell culture

NT2/D1 cells were maintained and propagated as described previously (Houldsworth et al., 2002). The H9 human ES cells were obtained from the MSKCC Stem Cell Core Facility.

Chip-seq and ChIP PCR

ChIP assay of NT2/D1 and H9 cells was performed according to described methods (Schmidt et al., 2009).

Lentivirus-mediated shRNA silencing of POU5F1, NANOG and SOX2

NT2/D1 ES cells stably transduced by lentiviral vectors were collected on days 2, 3, 6, and 7 from POU5F1, NANOG and SOX2 silenced cells for RNA isolation and protein estimation. RNA was converted into double-stranded cDNA and cRNA and checked on an Agilent bioanalyzer, fragmented, and hybridized to Affymetrix HG-U133Plus microarrays.

ARACNe analysis

The raw Affymetrix CEL files obtained after the imaging protocol were further processed to obtain probe-set level consensus using the Affymetrix MAS5 protocol (affy package from Bioconductor; http://www.bioconductor.org, (Gautier, L et al., 2004). This was followed by a log2 transformation and then quantile normalization as implemented in geWorkbench (http://www.geworkbench.org). The probe-set level expression values were then profiled (Margolin et al., 2006) to obtain a MI vs p-value curve. This curve described MI values to be used as a threshold corresponding to a user specified p-value for ascribing significance to ‘edges’ in the resulting ARACNe network. A p-value cutoff threshold of 0.05 (Bonferroni corrected for multiple testing) was used when running ARACNe (threshold corresponded to a Mutual Information value of 0.489). A Data Processing Inequality (DPI) value of 0 was used in the analysis. ARACNe was run in bootstrapping mode with a 100 separate runs of sample with replacement modifications to the original normalized expression dataset. The consensus network from the 100 separate runs was obtained considering all edges occurring in at least 95 of the 100 separate bootstrapped networks, and represents the GCT Net interactome.

RESULTS

Assembly of a core transcriptional network for GCTs

The ARACNe algorithm was used to analyze 135 GEPs derived from GCT biopsies, representing pluripotent and differentiation-associated histologic subsets of this malignancy, and from 6 normal testes as controls, to produce a comprehensive map of transcriptional interactions. Out of 3,099 probe sets annotated as TFs in the Affymetrix gene expression platform (HG-U133A+B) by Gene Ontology (GO) (Liu et al., 2003), ARACNe predicted targets for 2,612 probe sets, corresponding to 1,305 distinct TFs that virtually recapitulate all TFs expressed in this cellular context. A mutual information threshold (MI) ≥ 0.489 was selected, corresponding to a p≤ 0.05, the Bonferroni-corrected p-value, for the inference of any false positive interactions in this dataset. ARACNe inferred ~260,000 unique regulatory interactions between the 1,305 TFs and 17,025 target genes, which together comprise the complete GCTNet. To determine whether the GCTNet provided an adequate regulatory context to study pluripotency and lineage differentiation, we first selected the 130 TFs (top 10%) with the largest regulon size (i.e., the number of their ARACNe-inferred targets). Analysis of GO biological categories revealed that these regulators were highly enriched in those controlling pluripotency, cell fate, development and differentiation (Table S1A). Similarly, the most enriched pathway in Ingenuity analysis of the top 130 TFs was “Role of Oct4 in Mammalian Stem Cell Pluripotency” (Table S1B). Taken together, these data confirm that known pluripotency and differentiation-related regulators are especially well represented in the GCTNet suggesting that this interactome provides a valuable environment to study these processes.

To perform a systematic, albeit non-comprehensive quality control analysis for the inferred interactome, we undertook a detailed experimental validation of the ARACNe-inferred targets of the 3 most relevant TFs in the pluripotency pathway, POU5F1, NANOG, and SOX2. For this, we first analyzed the regulatory programs represented by their targets in the GCTNet and then proceeded with their experimental validation by ChIP-seq and profiling assays, following their shRNA-mediated silencing using NT2/D1 EC and H9 ES cells, which were selected as the closest in vitro counterparts of the in vivo GCT cells for studying programs controlling pluripotency and differentiation.

ARACNe-inferred targets of POU5F1, NANOG, and SOX2 direct distinct regulatory pathways in the GCTNet

Among the 17,025 regulatory targets represented in the GCTNet, 835 (5%) were found to be either individual or joint targets of POU5F1, NANOG, and SOX2; individually, the regulons of these 3 TFs included 338, 376, and 307 genes, respectively. Interestingly, while POU5F1 and NANOG shared 127 common targets (max overlap ω ≈ 38%, p = 0, i.e., -value below machine precision by Chi-square test), POU5F1 and SOX2 as well as NANOG and SOX2 shared only 40 (ω ≈ 13%, p < 5.1E-56) and 41 (ω ≈ 13%, p < 2.7E-62) common targets, respectively, suggesting a smaller albeit still highly significant regulatory program overlap; finally, the intersection of all 3 regulons included 21 genes (Figure 1A, Table S2A and S2B). Taken together, these results suggest that, in addition to a small core shared program regulated by all 3 TFs, SOX2 controls a partially orthogonal transcriptional program, while NANOG and POU5F1 display virtually overlapping programs in this interactome. Considering that both experimental and computational methods to dissect molecular interactions have relatively high false negative rates (FNRs), the observed overlap between POU5F1 and NANOG programs is extremely high, suggesting that they control almost identical programs. Assuming an FNR of 30% for ARACNe, the true overlap would be Ω = 78%, while at 37% FNR, it would be Ω =100%.

Figure 1.

Figure 1

Open in a new tab

(A). Unique and interacting targets of POU5F1, NANOG, and SOX2 identified by ARACNe (see Table S2A-B). (B), (C) and (D) GO analysis of POU5F1, NANOG and SOX2 target genes. Blue bars represent observed percentage of the genes for the given TF in a given GO category. Red bars represent the percentage of genes in the whole genome in a given GO category. Significance (P-value) of enrichment is based on a Fischer Exact Test (also see Table S2 C-E).

Interestingly, while GO categories enriched in targets individually controlled by either POU5F1 or NANOG included regulation of cell and nucleic acid metabolism (Figure 1B and 1C, Tables S2C and S2D), common targets of both TFs were highly enriched in categories representing control of stem cell and/or developmental related functions. In sharp contrast, SOX2 targets were mainly involved in developmental processes (Figure 1D, Table S2E). Finally, the 21 target genes that were inferred as targets of all 3 TFs (RHOBTB3, RIF1, NONO, DLG3, TTC3, MRS2L, TBC1D23, MED12, JARID2, PBX3, USP28, A4GALT, FGF4, ATP12A, CSH1, HLA-DPB2, ZNF589, PDPN, DPPA4, DND1, TERF1) preside over important functions associated with regulation of stem cell maintenance and development, differentiation and morphogenesis, compartment specification, cell cycle and mitosis, and regulation of transcription, suggesting a core stem cell regulatory program controlled by all 3 TFs in this biological system.

ARACNe-inferred targets of POU5F1, NANOG, and SOX2 are validated by ChIP-seq and ChIP-PCR

Analysis of DNA-binding sites predicted by ChIP-Seq assays in both NT2/D1 EC and H9 ES cells for each the of 3 TFs revealed their significant enrichment in proximal promoter regions (10kb, 5kb, and 2kb) of ARACNe-inferred targets, despite the differences between the in vivo context from which these were inferred and the in vitro context in which they were validated (Figure S1 A-H). Finally, we performed individual ChIP-PCR assays for the top 10 ARACNe-inferred targets of each TF, in both NT2/D1 EC and in H9 ES cells in a very small region [−1,000bp,+500bp] surrounding the transcription start site (TSS) (Figure S1G and S1H and Table S2F-K). This analysis confirmed specific binding of POU5F1 and SOX2 to the promoter of each of their top 10 targets, in both cell lines. NANOG was found to bind specifically to the promoter of 9 out of 10 of its ARACNe-inferred targets in both cell lines and to the promoter of DND1, in H9 ES but not NT2/D1 EC cells. These data suggest that actual binding specificity of these TFs to their ARACNe-inferred targets is significantly higher than that revealed by ChIP-related analysis, due to the previously reported false negative rates of this genome-wide technique (Margolin et al., 2009).

Functional pathways mediated by POU5F1, NANOG, and SOX2 are identified by target analysis following their shRNA mediated silencing

To functionally characterize the ARACNe-inferred targets of the 3 TFs, we achieved their stable silencing by lentivirus-mediated shRNA in NT2/D1 EC cells, followed by GEP. Cells were harvested on day 6, following puromycin selection of clones that integrated the construct. Downregulation of the silenced TFs was assessed by Western blot (data not shown). Enrichment of ARACNe-inferred targets in the differentially expressed genes following silencing of each TF was highly significant by gene set enrichment analysis (GSEA) (Subramanian et al., 2005) (PPOU5F1 =3.8E-18, PNANOG = 5.4E-19, PSOX2 = 1.72-12, respectively) (Figures 2 A-C and Table S3A). Based on GO enrichment analysis, categories of downregulated POU5F1 and NANOG targets were primarily enriched in cellular organization and DNA metabolism categories, whereas upregulated targets were enriched in cellular metabolism categories (Figures 2 D-G and Table S3A). Conversely, downregulated SOX2 targets were enriched in regulation of histone methylation and modification and stem cell differentiation and chromosome organization categories, whereas upregulated targets were enriched in control of nervous system development, neurological system process, regulation of axon extension, and chondrocyte differentiation categories, in addition to other developmental processes (Figures 2H and I, Figure S2 and Table S3A).

Figure 2. Silencing of POU5F1, NANOG and SOX2 in NT2/D1 EC cells.

Figure 2

Open in a new tab

(A), (B) and (C) GSEA of POU5F1, NANOG and SOX2 target genes on the differential expression of genes in the POU5F1, NANOG and SOX2 knockdown cells respectively when compared to controls. The bar code plot indicates the position of target genes. Genes with largest -log10 (p-value) are on the left and those with a value of 0 are on the right). The y-axis is the GSEA enrichment score. The curve represents a running weighted (weight = 1) cumulative enrichment score. The red vertical bars show position of genes in set S1. The small black bars above the red bars show positions of ChIP-PCR validated ARACNe targets. The grey scale bar at the top shows deeper shades of grey for larger local change in enrichment score. The title contains the GSEA Normalized Enrichment Score and the GSEA p-value for the enrichment of set S1 in set S2. (D) and (E) GO enrichment of downregulated and upregulated targets of POU5F1. (F) and (G). GO enrichment of downregulated and upregulated targets of NANOG. (H) and (I). GO enrichment of downregulated and upregulated targets of SOX2. The blue bars represent observed percentage of targets genes in a particular GO category. The red bars represent the percentage of genes in the whole genome in a particular GO category (as annotated by GO). The significance (p-value) of this enrichment is based on a Fischer Exact Test (also see Table S3 and Figure S2).

In addition to analysis of these genes following stable silencing, we also analyzed the 7-day time-course following silencing. Cells were harvested on days 2, 3, 6, and 7 following transfection to produce GEPs. Controls included non-transfected cells and cells transfected with non-target RNA. At each time point, the corresponding TF protein levels were also assessed by Western blot (data not shown). We first examined the overall pattern of expression change of target genes during the time-course for all 3 TFs. Most genes showed progressive up- or downregulation starting on day 2. A smaller proportion of genes however showed a more complex oscillatory pattern, showing up- or downregulation at day 2 but reversing this trend at day 7. These patterns are illustrated in an animation (Figures S3). Intriguingly, despite the fact that silencing of each TF induced loss of pluripotency and differentiation, the specific time-dependent programs as well as their enrichment in specific functional categories were TF- specific. Targets of POU5F1 and NANOG that were downregulated by day 2 were predominantly enriched in regulators of pluripotency, cell cycle, chromatin modification, and metabolism, whereas upregulated targets were enriched in regulation of metabolism and signal transduction. For SOX2, instead,development related and signal transduction genes were highly enriched in the up- as well as in the downregulated targets.

An analysis of variance (ANOVA) (Anscombe, 1948) of the GEPs at the 4 time points versus controls for each TF yielded a set of genes, whose expression was most variable at these time points (Figure 3A). Pair-wise overlap of affected genes following silencing was very significant (p-value below machine precision by Chi-square test). A set of experimentally consistent targets (core program) controlled by each TF was then obtained by selecting their ARACNe-inferred targets whose expression was affected by ANOVA analysis (Figures 3C-E). Enrichment of TF-specific core programs in GO categories was also computed and compared confirming the previous results, (Tables S3B-D). The overlap of enriched GO categories for NANOG and POU5F1 core programs (p-value below machine precision by Chi-square test) was much more extensive than for the overlap of SOX2 and NANOG core programs (p ≤ 4.66E-239) and for SOX2 and POU5F1 core programs (p ≤ 6.05E 134, Figure 3B; Table S3E). These results suggest that experimental validation of ARACNe-inferred targets by TF silencing substantially confirms their functional organization.

Figure 3. ANOVA and PC analysis of silencing time series of POU5F1, NANOG and SOX2 in NT2/D1 EC cells.

Figure 3

Open in a new tab

(A). ANOVA genes (with 0.05 p-value cut-off) in silencing time series POU5F1, NANOG and SOX2 in NT2/D1 EC cells. (C), (D) and (E). GSEA of POU5F1, NANOG and SOX2 ARACNetarget genes on the all the genes in the genome ranked by their ANOVA score in their respective silencing time series in NT2/D1 EC cells. The bar code plot indicates the position of respectiveARACNe target genes. (B). Overlapping categories from GO of the intersection of ARACNe targets and ANOVA genes of POU5F1, NANOG and SOX2 silencing time series. (F-G). Principal component contribution in silencing time series of POU5F1, NANOG, and SOX2. The ANOVA F-test p-value is also shown. (H). Scatter-plot of the 2 main principal components in the PCA. (see Table S3B-E).

Finally, we analyzed the time series GEPs by Principal Component Analysis (PCA) (Pearson, 1901). On average, PCA identified 3 statistically significant components capturing the gene expression trend across the 4 discrete time-points, representing the 7 days following the shRNA-mediated TF silencing, with 2 components capturing most of the data variance. Figures 3F and G project the cell-state trajectory over the most significant components (PC1 and PC2) across the four time-points following silencing of the individual TFs. As shown, the time-dependent trajectory following SOX2 silencing is highly distinct from those following silencing of NANOG and POU5F1, which instead are strikingly similar. Thus, PCA analysis further confirmed that NANOG and POU5F1 have largely overlapping programs, while SOX2 controls a relatively independent program across the time-course, which induces loss of pluripotency while resulting in a different cell endpoint state, as shown in Figure 3H.

Taken together, these assays support the accuracy of the ARACNe-inferred transcriptional programs and suggest that the GCTNet interactome constitutes a reasonably accurate and comprehensive map of the transcriptional interactions that regulate GCT behavior, including control of their pluripotent state.

GCTNet interrogation identifies novel pluripotency regulators

We proceeded to test whether GCTNet interrogation could help identify novel pluripotency regulators based on the statistical significance of the overlap of their transcriptional targets with those of NANOG, POU5F1, and SOX2. To assess whether such an analysis could be performed for any TF of interest without requiring additional experimental data, we designed it to depend only on the availability of the GCTNet, without requiring any experimental signature following shRNA mediated silencing of a TF of interest.

Specifically, for any pair of TFs represented in the GCTNet we computed their Transcriptional Overlap Odds Ratio, TOOR, i.e., the ratio between the number of overlapping targets of two TFs divided by the number of overlapping targets expected by chance, as well as the associated p-value. From more than 800,000 possible combinations of the 1,305 TFs in the interactome, we considered only TF-pairs with at least 20 overlapping targets and selected the 10% most significant ones (TOOR ≥ 8.1), resulting in 12,120 significant pairs, covering 910 of the 1,305 TFs. These pairs were displayed in a proximity map, where the distance between two TFs is inversely proportional to the statistical significance of their TOOR (i.e., the larger the overlap the closer the TFs). Interestingly, most of the previously reported ES cell state regulators formed distinct clusters in this map (Figure 4). A conclusion of this analysis is that many TFs in the GCTNet have significantly overlapping programs with established ES cell regulators, including POU5F1, NANOG, and SOX2, suggesting that if these are indeed involved in pluripotency control, then regulation of the ES cell state may indeed be much more complex than currently appreciated.

Figure 4.

Figure 4

Open in a new tab

Layout of all the TFs in the genome based on the overlap of their targets (quantified by an odds-ratio of overlap) . TFs placed closer have a larger odds-ratio of overlap of their targets. The ES cell state regulator TFs are represented by large circles colored in green. ( see Table S5)

To test this hypothesis, we proceeded to identify and experimentally validate the 10 TFs with the most significant regulatory program overlap with POU5F1 and NANOG. We selected this specific pair given their established relevance in pluripotency control and the fact that their regulatory programs are virtually overlapping. However, the same analysis could have been easily performed for any desired TF or TF cluster in the GCTNet.

To compute the statistical significance of the pooled odd ratios for each TF with two distinct TFs (i.e., NANOG and POU5F1) we used the Mantel–Haenszel statistics (Mantel and Haenszel, 1959). (Table S4). Based on this analysis, the ten TFs with the most significant pooled TOOR included: PRDM14, UTF, ZNF296, ZNF618, ZNF208, ZNF589, ETV4, TEAD4, VENTX, and TFCP2L1.

Analysis of POU5F1/NANOG program overlap identifies novel pluripotency regulators

To investigate their role as pluripotency regulators, we performed lentivirus-mediated shRNA silencing of these 10 TFs in H9 ES cells and evaluated lineage initiation by monitoring RNA and protein levels of a panel of relevant marker genes as a read out for loss of pluripotency. For each TF, a minimum of four distinct shRNA hairpins were first evaluated for their silencing efficiency. The 2 hairpins inducing the most efficient TF silencing were then used in the loss of pluripotency assay, which was performed in triplicate. Cells were harvested on day 5 following puromycin selections of clones that integrated the construct and TF silencing was confirmed by qRT-PCR. Cell cycle analysis of cells silenced for the 4 zinc finger TFs showed decrease in S-phase cells (Figure S4). RNA from the silenced cells was assayed for expression of multiple lineage markers by qRT-PCR, including those for neuronal (NES, NEFM, NEUROD, PAX6, PROM1, and GFAP), ectodermal (OLIG2, KTR7, KTR19, and CDH3), mesodermal (T, TWIST1, MSX1, and FOXF2), endodermal (AFP, GATA4, DAB2, NODAL, SOX7, SOX17, and GATA6), and trophoectodermal (HAND1, EOMES, and CDX) lineages. As shown in Figure S5, each lineage was expressed in the silenced cells of each TF as indicated by expression of one or more of these markers, thus phenocopying the pattern observed in cells silenced for POU5F1 and NANOG (controls). Such multilineage marker expression of ES cells silenced for POU5F1 and NANOG was previously shown in several studies (Niwa et al., 2000; Mitsui et al., 2003; Chazaud et al., 2006; Ivanova et al., 2006; Wang et al., 2012).

We then further confirmed the RNA expression results by immunofluorescence analysis of H9 ES cells following silencing of each TF, using a representative antibody for each lineage, including PAX6 (neuronal), HAND1 (trophoectodermal), NODAL (endodermal), KRT7 (ectodermal), and MSX1 (mesodermal). As shown in Figure 5, silencing of ZNF208 resulted in lineage marker expression patterns very similar to those following POU5F1 silencing. Silencing of other TFs, including TEAD4, UTF1, ZNF296, TFCP2L1, ZNF589 and PRDM14, also displayed similar lineage expression patterns, whereas silencing of ETV4 and VENTX induced a pattern dominated by expression of endodermal lineage markers (FigureS5 and S6). We further quantitated the protein expression of selected lineage markers by Western blotting (Figure S7). We also noted that lineage markers were expressed in clonal clusters, within silenced cell colonies.

Figure 5.

Figure 5

Open in a new tab

Immunofluorescence staining showing lineage development in (A) POU5F1 and (B) ZNF208 silenced H9 ES cells. Staining with DAPI, specific lineage marker primary antibodies PAX6 (neuronal), HAND1 (Trophoectodermal), NODAL( Endodermal), KRT7(Ectodermal), and MSX1( mesodermal) and the merged staining respectively, is shown. Scale bar for panel a and b in A and B is 100μm and for panel c, d and e in A and B is 50μm. (see Figure S5)

Overall, these results clearly establish that each of the 10 computationally inferred TFs play an individual role in pluripotency maintenance. It is noteworthy that 4 of these TFs are zinc finger genes of poorly annotated function (ZNF296, ZNF618, ZNF208, ZNF589) and 3 of them are developmental regulators (ETV4, TEAD4, and VENTX) with no previous indication of a role in pluripotency regulation. The remaining 3, UTF1, PRDM14, and TFC2PL1, have previously been established to play roles in pluripotency regulation in the POU5F1-NANOG-SOX2 pathway (Okuda et al., 1998; Loh et al., 2006; Nishimoto et al., 2013; Chen et al., 2008, van den Berg et al., 2010). These results thus strongly support the hypothesis that pluripotency regulation is more complex than recognized and involves a broad spectrum of partners of the canonical regulators, and possibly non-partners as well, as was indicated in the studies of Shu et al. 2013.

DISCUSSION

In this study, we used our previously published GEP data from adult human male GCTs to reverse-engineer and experimentally validate a genome-wide transcriptional regulatory network in a cellular context that is relevant to the study of pluripotency, differentiation, and tumorigenesis in vivo. This interactome provides the first de novo genome-wide resource representing the in vivo regulatory landscape of pluripotent cells for the study of processes relevant to pluripotency maintenance and differentiation. As shown in prior related efforts in cancer, including glioma (Zhao et al., 2009; Carro et al., 2010; Sumazin et al., 2011), T cell acute lymphoblastic leukemia (T-ALL) (Palomero et al., 2006; Della Gatta et al., 2012), and B cell lymphoma (Mani et al., 2008; Lefebvre et al., 2010), we expect that the availability of comprehensive and accurate models of regulation will be transformational in the elucidation of a variety of in vivo pluripotent cell related mechanisms.

Analysis of the GCTNet revealed multiple pathways of pluripotency, differentiation, and tumorigenesis simultaneously operating in this complex biological system. Notably, “Role of OCT4 in Mammalian Embryonic Stem Cell Pluripotency” emerged as the most significantly enriched functional pathway in the GCTNet, confirming that this interactome is uniquely suited for the study of pluripotency and lineage differentiation in vivo, thereby establishing a strong biological rationale for the validation of the canonical TF triad POU5F1, NANOG, and SOX2 in this system. The ARACNe-inferred interactions of the 3 TFs in the network, validated by rigorous functional assays in vitro by silencing experiments offer, for the first time, a glimpse into their core regulatory programs, with insights into specific time-dependent functions that they regulate during loss of pluripotency and lineage specification.

Our experiments validated the accuracy of ARACNe inferences. Indeed, single gene ChIP-PCR assay results showed that all 3 canonical pluripotency TFs bind to virtually all of their top ARACNe-inferred targets. This result highlights an advantage of the regulatory models developed by genome-wide ARACNe-based target identification compared to those assembled by highthroughput ChIP approaches for a handful of genes. The latter identify, for a given TF, an enormous number of binding targets, often in the thousands. Yet, among these, only a handful is functionally regulated by the TF in a specific context. The ARACNe-inferred targets, on the other hand, represent expressed functional targets compared to binding targets identified by ChIP-Chip, as is previously highlighted in the study of BCL6 regulation in human B cells (Basso et al., 2010).

To avoid the issue that TF binding does not imply regulation, we performed functional validation of the ARACNe-inferred targets by 7-day time course GEP following shRNA-mediated silencing of the 3 TFs in NT2/D1 cells. High enrichment of ARACNe-inferred targets in differentially expressed genes was observed following shRNA-mediated silencing of each TF. These assays confirmed the very high enrichment of ARACNe-inferred targets in differentially regulated genes following silencing of the corresponding TF, and revealed enrichment of several functional categories among the inferred targets. This included downregulation of cell cycle and stem cell differentiation and upregulation of metabolic gene categories. Developmental gene categories showed a mixed effect with both up- and downregulation, as expected, since these are related to multiple developmental pathways, only some of which are represented among the differentiated cells. Silencing of SOX2 showed a rather distinct pattern of expression, compared to NANOG and POU5F1, including upregulation of genes predominantly related to developmental process categories. PCA also showed that NANOG and POU5F1 programs are substantially overlapping, while SOX2 controlled an orthogonal program across the time-series and induced a distinct phenotypic endpoint at terminal differentiation by day 14.

Analysis of overlap of ARACNe-inferred targets between TFs in the interactome, implying target sharing, led to several important insights and discoveries. First, it showed that 910 of the 1305 TFs in the interactome (69%) presented highly significant regulatory program overlap, thus emphasizing the importance of considering transcriptional programs jointly regulated by multiple TFs as well as the context-specific complexity of combinatorial TF regulation in the dissection of biological process of interest. Indeed, focusing on regulatory program overlap between the NANOG/POU5F1 pair and other TFs, we identified a large number of TFs as high probability candidate pluripotency regulation co-factors. This suggests a much broader pluripotency regulatory network than currently recognized involving both previously identified co-regulators as well as a large number of novel regulators. As a proof of principle, as well as to show that results stemming from the analysis of the proposed pluripotent tumor system regulatory network could be effectively replicated in non-cancer human pluripotent stem cells, we successfully investigated the role of the top 10 of these TFs in pluripotency regulation in H9 ES cells.

Indeed, lentivirus mediated shRNA silencing of each of these candidate regulators elicited loss of pluripotency markers and gain of multilineage differentiation markers, recapitulating the behavior of ES cells in which POU5F1 and NANOG were silenced, as controls. The top 2 of these TFs, PRDM14 and UTF1, have previously been shown to be important ES cell state regulators that interact with the core regulators (Okuda et al., 1998; Zhao et al., 2008; Chia et al., 2010; Ma et al., 2011; Jia et al., 2012; Chan et al., 2012; Grabole et al., 2013; Yamaji et al., 2013). TFPC2L1 has been shown to have significant target sharing with POU5F1 and NANOG and play a role in pluripotency regulation (Chen et al., 2008; van den Berg et al., 2010; Ye et al., 2013). However, the 4 human zinc finger TFs (ZNF296, ZNF618, ZNF208, and ZNF589) have not been previously functionally characterized, although the murine Zfp296 has recently been shown to induce reprogramming of murine fibroblasts earlier and more efficiently in combination with Oct4, Sox2, Klf4 and C-myc suggesting a role in pluripotency regulation for this gene (Fischedick et al., 2012). Interestingly, the KRAB- containing ZNF proteins have recently been shown to activate the endogenous POU5F1 promoter thereby suggesting a role for these genes in regulating pluripotency (Juarez-Moreno et al., 2013; Ji et al., 2014). The remaining 3 TFs (ETV4, TEAD4, and VENTX)are functionally divergent and have not been previously reported as ES cell state regulators. ETV4 is an oncogene that is rearranged and/or overexpressed in prostrate and other neoplasms (Oh et al., 2012), TEAD4 specifies trophoderm in preimplantation mouse embryos (Nishioka et al., 2008), and VENTX is a homeobox TF that promotes expansion of human hematopoietic stem cells (Gao et al., 2012). In addition, human as well as Xenopus orthologs of VENTX have been shown to be targets of POU5F1 and NANOG suggesting that this gene may play a role in pluripotency regulation (Jung et al., 2010; Scerbo et al., 2012). Thus, 7 of the top 10 predicted pluripotency regulators, based on the overlap of their regulatory programs with those of the 3 core regulators, are shown to be novel, bona fide pluripotency regulators. The specific mechanism(s) by which these novel TFs play a role in lineage induction and whether they can by themselves or in combination with the core regulators reprogram somatic cells to pluripotency are issues that need to be addressed. Our results presented here nevertheless vividly emphasize the complexity of the pluripotency network in the in vivo context.

CONCLUSION

This study represents the development of the first experimentally validated genome-wide map of transcriptional interactions in a pluripotent tumor system in vivo. We show here the feasibility of systematically unraveling the complexity of context-specific gene interaction in a biological system that can be used to study processes of interest. The extensive validation of the functional targets of the classical triad of core pluripotency regulatory TFs and the discovery of an extensive network of novel pluripotency regulators that interact with them suggest that a systems biology approach as used here has merit and can help further elucidate in vivo regulation of the stem cell state and lineage differentiation.

Supplementary Material

Supp FigureS1-S7
Supp FigureVideoS3
Download video file (32.4MB, mp4)
Supp Material
Supp TableS1
Supp TableS2
Supp TableS3
Supp TableS4

ACKNOWLEDGEMENTS

This study was supported by the NIH grants CA008748, CA121852, CA109755, GM091240, NYSTEM grant CO29153, and the Byrne Fund. The MSKCC Genomics Core Facility performed the microarray and ChIP-seq. We thank the MSKCC Molecular Cytology Core Facility for help with IF analysis and imaging. The authors declare no conflicts of interest.

REFERENCES

  1. Anscombe FJ. The Validity of Comparative Experiments. Journal of the Royal Statistical Society Series A. (General) 1948;111:181–211. [Google Scholar]
  2. Basso K, Margolin AA, Stolovitzky G, et al. Reverse engineering of regulatory networks in human B cells. Nat Genet. 2005;37:382–390. doi: 10.1038/ng1532. [DOI] [PubMed] [Google Scholar]
  3. Basso K, Saito M, Sumazin P, et al. Integrated biochemical and computational approach identifies BCL6 direct target genes controlling multiple pathways in normal germinal center B cells. Blood. 2010;115:975–984. doi: 10.1182/blood-2009-06-227017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cadeiras M, von Bayern M, Sinha A, et al. Drawing networks of rejection - a systems biological approach to the identification of candidate genes in heart transplantation. J Cell Mol Med. 2011;15:949–956. doi: 10.1111/j.1582-4934.2010.01092.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carro MS, Lim WK, Alvarez MJ, et al. The transcriptional network for mesenchymal transformation of brain tumors. Nature. 2010;463:318–325. doi: 10.1038/nature08712. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Chaganti RS, Houldsworth J. Genetics and biology of adult human male germ cell tumors. Cancer Res. 2000;60:1475–1482. [PubMed] [Google Scholar]
  7. Chan YS, Göke J, Lu X, et al. A PRC2-dependent repressive role of PRDM14 in human embryonic stem cells and induced pluripotent stem cell reprogramming. Stem Cells. 2012;31:682–692. doi: 10.1002/stem.1307. [DOI] [PubMed] [Google Scholar]
  8. Chazaud C, Yamanaka Y, Pawson T, et al. Early lineage segregation between epiblast and primitive endoderm in mouse blastocysts through the Grb2-MAPK pathway. Dev Cell. 2006;10:615–624. doi: 10.1016/j.devcel.2006.02.020. [DOI] [PubMed] [Google Scholar]
  9. Chen X, Xu H, Yuan P, et al. Integration of external signaling pathways with the core transcriptional network in embryonic stem cells. Cell. 2008;33:1106–1117. doi: 10.1016/j.cell.2008.04.043. [DOI] [PubMed] [Google Scholar]
  10. Chia NY, Chan YS, Feng B, et al. A genome-wide RNAi screen reveals determinants of human embryonic stem cell identity. Nature. 2010;468:316–320. doi: 10.1038/nature09531. [DOI] [PubMed] [Google Scholar]
  11. Della Gatta G, Palomero T, Perez-Garcia A, et al. Reverse engineering of TLX oncogenic transcriptional networks identifies RUNX1 as tumor suppressor in T-ALL. Nat Med. 2012;18:436–440. doi: 10.1038/nm.2610. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fischedick G, Klein DC, Wu G, et al. Zfp296 is a novel, pluripotent-specific reprogramming factor. PLoS One. 2012;7:e34645. doi: 10.1371/journal.pone.0034645. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Gao H, Wu X, Sun, Y, et al. Suppression of homeobox transcription factor VentX promotes expansion of human hematopoietic stem/multipotent progenitor cells. J Biol Chem. 2012;287:29979–29987. doi: 10.1074/jbc.M112.383018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Grabole N, Tischler J, Hackett JA, et al. Prdm14 promotes germline fate and naive pluripotency by repressing FGF signaling and DNA methylation. EMBO Rep. 2013;14:629–637. doi: 10.1038/embor.2013.67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Gautier L, Cope L, Bolstad BM, et al. affy-analysis of Affymetrix GeneChip data at the probe level. Bioinformatics. 2004;3:307–315. doi: 10.1093/bioinformatics/btg405. [DOI] [PubMed] [Google Scholar]
  16. Houldsworth J, Heath SC, Bosl GJ, et al. Expression profiling of lineage differentiation in pluripotential human embryonal carcinoma cells. Cell Growth Differ. 2002;13:257–264. [PubMed] [Google Scholar]
  17. Ivanova N, Dobrin R, Lu R, et al. Dissecting self-renewal in stem cells with RNA interference. Nature. 2006;442:533–538. doi: 10.1038/nature04915. [DOI] [PubMed] [Google Scholar]
  18. Jaenisch R, Young R. Stem cells, the molecular circuitry of pluripotency and nuclear reprogramming. Cell. 2008;132:567–582. doi: 10.1016/j.cell.2008.01.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Ji Q, Fischer LA, Brown RC, et al. Engineered zinc-finger transcription factors activate OCT4 (POU5F1), SOX2, KLF4, c-MYC (MYC) and miR302/367. Nucleic Acids Res. 2014;42:6158–6167. doi: 10.1093/nar/gku243. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jia J, Zheng X, Hu G, et al. Regulation of pluripotency and self- renewal of ESCs through epigenetic-threshold modulation and mRNA pruning. Cell. 2012;151:576–589. doi: 10.1016/j.cell.2012.09.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Juárez-Moreno K, Erices R, Beltran AS, et al. Breaking through an epigenetic wall: re-activation of Oct4 by KRAB-containing designer zinc finger transcription factors. Epigenetics. 2013;8:164–76. doi: 10.4161/epi.23503. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Jung M, Peterson H, Chavez L, et al. A data integration approach to mapping OCT4 gene regulatory networks operative in embryonic stem cells and embryonal carcinoma cells. PLoS One. 2010;5:e10709. doi: 10.1371/journal.pone.0010709. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Korkola JE, Houldsworth J, Dobrzynski D, et al. Gene expression-based classification of nonseminomatous male germ cell tumors. Oncogene. 2005;24:5101–5107. doi: 10.1038/sj.onc.1208694. [DOI] [PubMed] [Google Scholar]
  24. Korkola JE, Houldsworth J, Chadalavada RS, et al. Down-regulation of stem cell genes, including those in a 200-kb gene cluster at 12p1331, is associated with in vivo differentiation of human male germ cell tumors. Cancer Res. 2006;66:820–827. doi: 10.1158/0008-5472.CAN-05-2445. [DOI] [PubMed] [Google Scholar]
  25. Korkola JE, Houldsworth J, Feldman DR, et al. Identification and validation of a gene expression signature that predicts outcome in adult men with germ cell tumors. J Clin Oncol. 2009;27:5240–5247. doi: 10.1200/JCO.2008.20.0386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lefebvre C, Rajbhandari P, Alvarez MJ, et al. A human B-cell interactome identifies MYB and FOXM1 as master regulators of proliferation in germinal centers. Mol Syst Biol. 2010;6:377–386. doi: 10.1038/msb.2010.31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lim WK, Lyashenko E, Califano A. Master regulators used as breast cancer metastasis classifier. Pac Symp Biocomp. 2009;14:504–515. [PMC free article] [PubMed] [Google Scholar]
  28. Liu G, Loraine AE, Shigeta R, et al. NetAffx: Affymetrix probesets and annotations. Nucleic Acids Res. 2003;31:82–86. doi: 10.1093/nar/gkg121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Loh KM, Lim B. A precarious balance: pluripotency factors as lineage specifiers. Cell Stem Cell. 2011;8:363–369. doi: 10.1016/j.stem.2011.03.013. [DOI] [PubMed] [Google Scholar]
  30. Loh YH, Wu Q, Chew JL, et al. The Oct4 and Nanog transcription network regulates pluripotency in mouse embryonic stem cells. Nat Genet. 2006;4:431–440. doi: 10.1038/ng1760. [DOI] [PubMed] [Google Scholar]
  31. Ma Z, Swigut T, Valouev A, et al. Sequence-specific regulator Prdm14 safeguards mouse ESCs from entering extraembryonic endoderm fates. Nat Struct Mol Biol. 2011;18:120–127. doi: 10.1038/nsmb.2000. [DOI] [PubMed] [Google Scholar]
  32. Mani KM, Lefebvre C, Wang K, et al. A systems biology approach to prediction of oncogenes and molecular perturbation targets in B-cell lymphomas. MolSyst Biol. 2008;4:169–177. doi: 10.1038/msb.2008.2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst. 1959;22:719–748. [PubMed] [Google Scholar]
  34. Margolin AA, Wang K, Lim WK, et al. Reverse engineering cellular networks. Nat Protoc. 2006;1:662–671. doi: 10.1038/nprot.2006.106. [DOI] [PubMed] [Google Scholar]
  35. Margolin AA, Palomero T, Sumazin P, et al. ChIP-on-chip significance analysis reveals large-scale binding and regulation by human transcription factor oncogenes. Proc Natl Acad Sci USA. 2009;106:244–249. doi: 10.1073/pnas.0806445106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Marson A, Levine SS, Cole MF, et al. Connecting micro RNA genes to the core transcriptional regulatory circuitry of embryonic stem cells. Cell. 2008;134:521–533. doi: 10.1016/j.cell.2008.07.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Mitsui K, Tokuzawa Y, Itoh H, et al. The homeoprotein Nanog is required for maintenance of pluripotency in mouse epiblast and ES cells. Cell. 2003;113:631–642. doi: 10.1016/s0092-8674(03)00393-3. [DOI] [PubMed] [Google Scholar]
  38. Nishimoto M, Katano M, Yamagishi T, et al. In vivo function and evolution of the eutherian-specific pluripotency marker UTF1. PLoS One. 2013;8:e68119. doi: 10.1371/journal.pone.0068119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Nishioka N, Yamamoto S, Kiyonari H, et al. Tead4 is required for specification of trophectoderm in pre-implantation mouse embryos. Mech Dev. 2008;125:270–283. doi: 10.1016/j.mod.2007.11.002. [DOI] [PubMed] [Google Scholar]
  40. Niwa H, Miyazaki J, Smith AG. Quantitative expression of Oct-3/4 defines differentiation, dedifferentiation or self-renewal of ES cells. Nat Genet. 2000;24:372–376. doi: 10.1038/74199. [DOI] [PubMed] [Google Scholar]
  41. Oh S, Shin S, Janknecht R. ETV1, 4 and 5: an oncogenic subfamily of ETS transcription factors. Biochim Biophys Acta. 2012;1826:1–12. doi: 10.1016/j.bbcan.2012.02.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Okuda A, Fukushima A, Nishimoto M, et al. UTF1, a novel transcriptional coactivator expressed in pluripotent embryonic stem cells and extra-embryonic cells. EMBO J. 1998;17:2019–2032. doi: 10.1093/emboj/17.7.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Palomero T, Lim WK, Odom DT, et al. NOTCH1 directly regulates c-MYC and activates a feed-forward-loop transcriptional network promoting leukemic cell growth. Proc Natl Acad Sci USA. 2006;103:18261–18266. doi: 10.1073/pnas.0606108103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Ravasi T, Suzuki H, Cannistraci CV, et al. An atlas of combinatorial transcriptional regulation in mouse and man. Cell. 2010;140:744–752. doi: 10.1016/j.cell.2010.01.044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Pearson K. "On Lines and Planes of Closest Fit to Systems of Points in Space" (PDF. Philosophical Magazine. 1901;2:559–572. [Google Scholar]
  46. Scerbo P, Girardot F, Vivien C, et al. Ventx factors function as Nanog-like guardians of developmental potential in Xenopus. PLoS One. 2012;7:e36855. doi: 10.1371/journal.pone.0036855. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Schmidt D, Wilson MD, Spyrou C, et al. ChIP-seq: Using high-throughput sequencing to discover protein–DNA interactions. Methods. 2009;48:240–248. doi: 10.1016/j.ymeth.2009.03.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Schuettengruber B, Cavalli G. Recruitment of polycomb group complexes and their role in the dynamic regulation of cell fate choice. Development. 2009;136:3531–3542. doi: 10.1242/dev.033902. [DOI] [PubMed] [Google Scholar]
  49. Shu J, Wu C, Wu Y, et al. Induction of pluripotency in mouse somatic cells with lineage specifiers. Cell. 2013;153:963–975. doi: 10.1016/j.cell.2013.05.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Sperger JM, Chen X, Draper JS, et al. Gene expression patterns in human embryonic stem cells and human pluripotent germ cell tumors. Proc Natl Acad Sci USA. 2003;100:13350–13355. doi: 10.1073/pnas.2235735100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Subramanian A, Tamayo P, Mootha V, et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA. 2005;102:15545–15550. doi: 10.1073/pnas.0506580102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sumazin P, Yang X, Chiu HS, et al. An extensive microRNA-mediated network of RNA-RNA interactions regulates established oncogenic pathways in glioblastoma. Cell. 2011;147:370–381. doi: 10.1016/j.cell.2011.09.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Ulbright TM. Germ cell neoplasms of the testis. Am J Surg Pathol. 1993:11075–11091. doi: 10.1097/00000478-199311000-00001. [DOI] [PubMed] [Google Scholar]
  54. van den Berg DL, Snoek T, Mullin NP, et al. An Oct4-centered protein interaction network in embryonic stem cells. Cell Stem Cell. 2010;6:369–381. doi: 10.1016/j.stem.2010.02.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Wang Z, Oron E, Nelson B, et al. Distinct lineage specification roles for NANOG, OCT4, and SOX2 in human embryonic stem cells. Cell Stem Cell. 2012;10:440–454. doi: 10.1016/j.stem.2012.02.016. [DOI] [PubMed] [Google Scholar]
  56. Yamaji M, Ueda J, Hayashi K, et al. PRDM14 ensures naive pluripotency through dual regulation of signaling and epigenetic pathways in mouse embryonic stem cells. Cell Stem Cell. 2013;12:368–382. doi: 10.1016/j.stem.2012.12.012. [DOI] [PubMed] [Google Scholar]
  57. Ye S, Li P, Tong C, et al. Embryonic stem cell self-renewal pathways converge on the transcription factor Tfcp2l1. EMBO J. 2013;32:2548–2560. doi: 10.1038/emboj.2013.175. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Young RA. Control of embryonic stem cell state. Cell. 2011;144:940–954. doi: 10.1016/j.cell.2011.01.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Zhao X, D'Arca D, Lim WK, et al. The N-Myc-DLL3 cascade is suppressed by the ubiquitin ligase Huwe1 to inhibit proliferation and promote neurogenesis in the developing brain. Dev Cell. 2009;17:210–221. doi: 10.1016/j.devcel.2009.07.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Zhao Y, Yin X, Qin H, et al. Two supporting factors greatly improve the efficiency of human iPSC generation Cell Stem Cell. 2008;3:475–479. doi: 10.1016/j.stem.2008.10.002. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp FigureS1-S7
NIHMS635905-supplement-Supp_FigureS1-S7.pdf (1.4MB, pdf)
Supp FigureVideoS3
Download video file (32.4MB, mp4)
Supp Material
NIHMS635905-supplement-Supp_Material.docx (31.9KB, docx)
Supp TableS1
NIHMS635905-supplement-Supp_TableS1.xlsx (48.3KB, xlsx)
Supp TableS2
NIHMS635905-supplement-Supp_TableS2.xls (280.5KB, xls)
Supp TableS3
NIHMS635905-supplement-Supp_TableS3.xlsx (150.8KB, xlsx)
Supp TableS4
NIHMS635905-supplement-Supp_TableS4.xlsx (105.7KB, xlsx)

RESOURCES