Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2009 Oct 2;5(4):420-33.
doi: 10.1016/j.stem.2009.07.012.

Uncovering Early Response of Gene Regulatory Networks in ESCs by Systematic Induction of Transcription Factors

Affiliations
Free PMC article

Uncovering Early Response of Gene Regulatory Networks in ESCs by Systematic Induction of Transcription Factors

Akira Nishiyama et al. Cell Stem Cell. .
Free PMC article

Abstract

To examine transcription factor (TF) network(s), we created mouse ESC lines, in each of which 1 of 50 TFs tagged with a FLAG moiety is inserted into a ubiquitously controllable tetracycline-repressible locus. Of the 50 TFs, Cdx2 provoked the most extensive transcriptome perturbation in ESCs, followed by Esx1, Sox9, Tcf3, Klf4, and Gata3. ChIP-Seq revealed that CDX2 binds to promoters of upregulated target genes. By contrast, genes downregulated by CDX2 did not show CDX2 binding but were enriched with binding sites for POU5F1, SOX2, and NANOG. Genes with binding sites for these core TFs were also downregulated by the induction of at least 15 other TFs, suggesting a common initial step for ESC differentiation mediated by interference with the binding of core TFs to their target genes. These ESC lines provide a fundamental resource to study biological networks in ESCs and mice.

Figures

Figure 1
Figure 1. Strategy to establish and quality-control ES cell lines
(A) List of ES cell lines generated and analyzed in this study. (B) Schematic diagram for the strategy. A parental ES cell line was named ES[MC1R(20)], which stands for MC1 ES cells, ROSA-TET locus [R], and clone 20. Each ES cell line was named by adding the name of a transgene and the clone number. For example, the ES cell line that was generated by integrating the Aes gene was named ES[MC1R(20):tetAes(24)]. For brevity, ES cell lines are simply referred by the name of a transgene (e.g., Aes) throughput the text and figures. (C–G) Representative results for quality control of the ES cell line: ES[MC1R(20):tetNr5a2(7)]. (C) qRT-PCR analysis of transcript levels of the ES cells cultured for 48 hours in the presence (+) or absence (−) of doxycycline: (Left) transcripts measured by a primer pair for ORF (both endogenous and transgene Nr5a2); (Middle) transcripts measured by endogenous Nr5a2-specific primer pair: (Right) transcripts measured by a primer pair for Venus (representing a transgene). Values are shown as fold-induction compared with Dox+ condition. Data are presented as means ± SEM. (D) Time-course analysis of exogenous (i.e., a transgene-derived) NR5A2 protein expression by Western blotting using an antibody against FLAG (upper panel) and β-actin (lower panel). (E) Localization of the exogenous NR5A2 protein by immunostaining using anti-FLAG antibody (left); and localization of Venus fluorescence (middle) and DNA (right). (F) Karyotypes of ES[MC1R(20):tetNr5a2(7)] showing 88% euploidy. (G) A representative picture of the metaphase spread. See Supplementary Experimental Procedures for information on other clones.
Figure 2
Figure 2. Global expression profiles of TF-inducible ES cell lines
(A) Principal component analysis (PCA) of gene expression profiles of 152 different cell types: 54 TF-inducible ES lines with induced over-expression of various TFs (48 hr in Dox−, marked red), the same 54 TF-inducible ES lines (48 hr in Dox+, marked red), and 44 different cell types, which represent ES cells differentiating into three cell lineages (trophoblast, primitive endoderm (PE), and neural, marked blue, green, and yellow, respectively). All cell lines with induced TFs show gene expression profiles (encircled) very similar to that of undifferentiated ES cells, although a few TFs caused some changes towards differentiation (shown by arrows). The explanation of PCA and the details of these 44 cell types are given in the previous publication (Aiba et al., 2009). Only representative cell types are labeled: Klf4 (Dox−), Sox9 (Dox−), Tcf3 (Dox−), Cdx2 (Dox−), and Eomes (Dox−). The trophoblast lineage is represented by Z0 – Z5 (ES cells differentiating into trophoblast cells from day 0 to day 5 after repressing Pou5f1 expression), TS (trophoblast stem cells), and PL (E13.5 placenta). The PE lineage is represented by F0 – F5 (Embryonal carcinoma cells differentiating into primitive endoderm from day 0 to day 5). The neural lineage is represented by N2 – N6 (ES cell differentiating into neural lineage from day 2 to day 6), P0 – P4 (Embryonal carcinoma cells differentiating into neural cells), NS (neural stem/progenitor cells), and DC (differentiated neuron and glia). 3T (NIH3T3 fibroblast cells). MB (Mouse embryo fibroblast cells). (B) A heatmap showing the results of hierarchical clustering analysis of all the microarray data (54 ES cell lines). Only the top 3000 genes whose expression are most significantly altered are used for the analysis. Both genes and ES cell lines are clustered according to their similarity of global gene expression patterns measured by Pearson correlation of log-transformed expression values. The list of genes and their expression change for this heatmap is given in Table S2. (C) (D) Significance of correlations between gene expression response to the induction of TFs in TF-inducible ES cell lines (data from this paper) and gene expression in published data sets (Aiba et al., 2009; Su et al., 2002). Gene expression in published data sets was log-transformed and centered: the mean value was subtracted for each gene. (C) Cell types in the data set for trajectories of early lineage differentiation (Aiba et al., 2009): Extraembryonic (TS, PL); Trophoblast (Z2 – Z5); Fibroblasts (3T, MB, MD, ST); Primitive endoderm (F2 – F5, G1 – G5); Neural/Primitive ectoderm (N2 – N6, NS, DC); and Other (E1 – E5, EG, F0, F1, G0, IF, IN, N0, N1, P0, P4, TG, Z0, Z1). (D) Tissues in the GNF database (Su et al., 2002): Placenta; Heart and muscles (skeletal); Lymph node, thymus, immune (B220+ B-cells, CD4+T-cells, CD8+T-cells); Umbilical cord, uterus; Blastocyst; and Other (adipose tissue, adrenal gland, amygdala, bladder, bone, bone marrow, brown fat, cerebellum, cerebral cortex, digits, dorsal root ganglia, dorsal striatum, embryo day 10.5, embryo day 6.5, embryo day 7.5, embryo day 8.5, embryo day 9.5, epidermis, eye, fertilized egg, frontal cortex, hippocampus, hypothalamus, kidney, large intestine, liver, lung, mammary gland (lact), medial olfactory epithelium, olfactory bulb, oocyte, ovary, pancreas, pituitary, preoptic, prostate, salivary gland, small intestine, snout epidermis, spinal cord lower, spinal cord upper, spleen, stomach, substantia nigra, testis, thyroid, tongue, trachea, trigeminal, vomeralnasal organ).
Figure 3
Figure 3. Extent of transcriptome perturbation by TFs and pair-wise comparison of expression changes
(A) Scatter-plots comparing expression profiles of representative ES cell lines between Dox+ and Dox− conditions. Red spots represent genes that show higher than 2-fold expression (up-regulated) and green spots represent genes that show lower than 2-fold expression (down-regulated) in Dox− condition than in Dox+ condition with statistical significance of FDR<0.05. The number of up- and down-regulated genes are also shown. (B) The number of genes whose expression was affected significantly (FDR<0.05 and expression changes >2-fold) by induction of various TFs in ES cells. ES cell lines are ordered according to the expression levels of endogenous TF gene in undifferentiated ES cells based on published RNA-Seq data (Table S11) (Cloonan et al., 2008).
Figure 4
Figure 4. Analysis of the CDX2-inducible ES cell line
(A) Time-course analysis of CDX2 protein expression by Western blotting (day 0, 0.5, 1, 2, 3, 5, and 7 after removal of Dox). Antibody against CDX2 protein recognizes both endogenous and exogenous (i.e., transgene-derived) CDX2 protein. Antibody against FLAG recognizes only exogenous CDX2 protein. Antibody against ACTB is used for the loading control. (B) Alkaline phosphatase activity was tested in Cdx2 over-expressing (lower row) and control cells (upper row) with or without Dox for 6 days in culture (Over view; left panel, Magnified; right panel). (C) Cdx2 over-expressing cells induce trophectoderm markers CDC42 (upper) and integrin alpha 7 (ITGA7; lower) by 6 days.
Figure 5
Figure 5. ChIP-Seq analysis of Cdx2-inducible ES cell line
(A) Chromatins were prepared from Cdx2-inducible ES cells cultured for 48 – 60 hours in the Dox+ and Dox− conditions. Chromatin immunoprecipitation (ChIP) was carried out by using anti-FLAG M2 affinity gel. ChIP product was tested by Western blotting using anti-FLAG antibody. Nuclear extract from ES cells cultured for 48 – 60 hours in Dox+ and Dox− condition was used for the Western blot. (B) CDX2 ChIP-Seq peaks in the Hoxa7 gene region. UCSC Mouse Mm9 browser view of Hoxa7 gene locus after mapping CDX2 ChIP-Seq tags locations in the wiggle format. CDX2 ChIP-Seq peaks are shown in red. (C) CDX2-binding motifs identified with CisFinder using 200 bp sequences centered at ChIP sites. (D) Over-representation of CDX2 binding motifs in ChIP sites. Genomic sequences (2000 bp) centered at CDX2 binding sites found by ChIP-seq were extracted from the genome and searched for the occurrence of CDX2 motifs. Binding motif was characterized by the position-frequency matrix (PFM) generated using CisFinder software (see the text). Motif fit was evaluated by log-likelihood method assuming false positive rate of 1 per 10Kb of a random sequence. (E) Functional CDX2-target genes. (F) Cdx2 ChIP-Seq result was verified by qPCR. Primers flanking a promoter region of Hbb-b1 and Pou5f1 as well as a gene desert region in chromosome 3 were used as negative controls. Primers flanking of Actb gene promoter were used for normalization. The relative enrichment of CDX2 binding was indicated as fold change. (G) Blue line: Correlation between gene expression changes caused by Cdx2-induction and the proportion of genes with CDX2 binding sites identified using a sliding window of 500 genes. Red line: Correlation between gene expression changes caused by Cdx2-induction and the proportion of genes with pTF binding sites identified using a 500-bp sliding window (more than 2 TFs out of POU5F1, SOX2, NANOG, STAT3, and SMAD1; data from Figure 5H). (H) Proportion of genes with binding sites for each of 14 TFs among genes whose expression was up-regulated or down-regulated by the induction of Cdx2. (I) Time course analysis of endogenous POU5F1, SOX2, NANOG, and ACTB protein expression by Western blotting using antibodies against each protein. Cdx2-inducible ES cell line was cultured for 0, 0.5, 1, 2, 3, 5, and 7 days in the Dox− condition. (J) ChIP-qPCR analysis for POU5F1 binding on its target genes in Cdx2-inducible ES cells. Primers flanking a gene desert region in chromosome 3 were used as a negative control. The relative enrichment of POU5F1 binding was represented as a fold change.
Figure 6
Figure 6. Analysis of the CDX2 protein complex pulled down by FLAG-immunoprecipitation
(A) Confirmation of immunoprecipitation using anti-FLAG antibody. Immunoprecipitates and nuclear extract were used for western blotting using anti-FLAG antibody. (B) A silver-stained SDS gel showing the presence of CDX2 major band and other distinct bands. Anti-FLAG M2 affinity gel was used for IP of CDX2 protein complex from Cdx2-inducible ES cells. Nuclear extracts were prepared from Cdx2-inducible ES cells cultured for 48 – 60 hours in the Dox+ and Dox− condition. Some bands are marked with protein names identified by the mass spectrometry of the IP products. M, markers. (C) Mass-spectrometry result was verified by IP-Western assay. Anti-FLAG M2 affinity gel was used to immunoprecipitate (IP) CDX2 protein complex from the nuclear extracts from Cdx2-inducible ES cells cultured for 48 –60 hours in the Dox+ and Dox− conditions. IP products were tested by Western blotting using antibodies against FLAG, KPNB1, HDAC1, and SALL4. Antibody against UBF was used as a control. (D) Reverse IP carried out by using antibodies against either HDAC1 or SALL4. IP products were tested by Western blotting using antibodies against FLAG, HDAC1, SALL4, and UBF. Nuclear extracts were also used as controls. Control UBF was detected in HDAC-IP sample as reported previously, but not detected in the SALL4-IP sample.
Figure 7
Figure 7. Compendium analysis of TF-binding loci and expression profiles after TF-induction
(A) Abundance of TFBS (transcription factor binding sites) in distal (0.3 – 15 Kb upstream and downstream from the TSS) and proximal (<300 bp upstream and downstream from the TSS) portions of the promoter in genes up-regulated or down-regulated (>2-fold changes of gene expression, but at least 200 genes in each group) by the induction of TFs (shown in the first column). CDX2, POU5F1, SOX2, NANOG, STAT3, and SMAD1 bind mostly to distal sites, and the number of binding sites in proximal promoters was not sufficient, and thus was not included for analysis. TFs are ordered according to the expression level of endogenous genes in ES cells from abundant to those that are not expressed in ES cells based on published RNA-Seq data (Table S11) (Cloonan et al., 2008). Cells are color-coded based on the over-representation or under-representation of genes with TFBS compared to the control group of genes that did not respond to the induction of TF (<1.25 fold change). Cells are not color-coded if the difference in the proportion of genes with TFBS is not statistically significant. Data on TFBS and chromatin modifications (Chr) in promoters of genes were compiled from our ChIP-Seq experiment with CDX2 (Figure 5), and published data (H3K4me3 and H3K27me3 data from (Mikkelsen et al., 2007); the rest of the data from (Chen et al., 2008)). K4me3: genes marked with H3K4me3. K4K27me3: genes marked with both H3K4me3 and H3K27me3 (“bivalent domains”). (B) Potential applications of TF-inducible ES Cell Bank, for which a proof-of-principle has been shown in this paper. (C) A model for ES cells in undifferentiated state. Cdx2-target genes (red boxes, e.g., Hoxa7; Figure 5F) are not actively transcribed. Pluripotency Associated Transcription Factors (pTFs, e.g., POU5F1, SOX2, and NANOG) are present and bind to the regulatory regions of pTF-target genes (blue boxes), resulting in the active transcription of these genes. pTF-target genes include genes encoding pTFs, which thus form positive feedback loops and maintain the levels of pTFs. (D) A model for ES cells in the early phase after the forced induction of Cdx2. CDX2 protein binds directly to the regulatory region of Cdx2-target genes, which begin to be actively transcribed. CDX2 suppresses the binding of pTFs to the regulatory regions of pTF-target genes and shut downs the transcription of pTF-target genes.

Similar articles

See all similar articles

Cited by 102 articles

See all "Cited by" articles

MeSH terms

Feedback