Otx2 and Oct4 drive early enhancer activation during embryonic stem cell transition from naive pluripotency

Abstract

Embryonic stem cells (ESCs) are unique in that they have the capacity to differentiate into all of the cell types in the body. We know a lot about the complex transcriptional control circuits that maintain the naive pluripotent state under self-renewing conditions but comparatively less about how cells exit from this state in response to differentiation stimuli. Here, we examined the role of Otx2 in this process in mouse ESCs and demonstrate that it plays a leading role in remodeling the gene regulatory networks as cells exit from ground state pluripotency. Otx2 drives enhancer activation through affecting chromatin marks and the activity of associated genes. Mechanistically, Oct4 is required for Otx2 expression, and reciprocally, Otx2 is required for efficient Oct4 recruitment to many enhancer regions. Therefore, the Oct4-Otx2 regulatory axis actively establishes a new regulatory chromatin landscape during the early events that accompany exit from ground state pluripotency.

Graphical abstract

Figure 1

Otx2 and Oct4 Bind to Similar Regulatory Regions (A) Heatmap of tag densities of Otx2 binding over 12 hr following exit from pluripotency (−2i). In each heatmap, the Otx2 tag density at that time point (indicated in red) is plotted for 5 kb on either side of its binding peak summit, and the Otx2 tag densities at other time points are plotted on top of these summit coordinates. The most significant de novo motifs discovered in the Otx2 binding regions at each of the time points are shown above the heatmaps, and the consensus Otx2 binding motif is shown below. (B) De novo motif discovery identifies an enriched motif in the Otx2 binding regions (top) that resembles the Oct4 consensus site (bottom). (C) Heatmap of normalized Otx2 and Oct4 tag densities at the indicated times following exit from pluripotency (−2i), plotted onto 4 kb regions centered on the Otx2 peak summits at 12 hr. Data are partitioned based on whether there is co-occurrence of Otx2 and Oct4 binding (Otx2+/Oct4+) or Otx2 binding alone (Otx2+/Oct4−). (D) Average tag density profiles of Otx2 and Oct4 ChIP-seq analyses mapped onto Oct4 binding region summits. (E) Re-ChIP analysis of Otx2 (first ChIP) and Oct4 (second ChIP) at the indicated loci in the presence of 2i (+) or upon release for 12 hr (−). Data are presented as means ± SEM (n = 2). (F) Gene ontology terms enriched among the genes associated with either the Otx2+/Oct4+ or the Otx2+/Oct4− binding regions. Data are shown graphically according to their p values (x axis) and the associated category (y axis). See also Figure S1.

Figure 2

Interplay between Oct4 and Otx2 Expression (A and C) qRT-PCR analysis of Otx2 and Oct4 (also known as Pou5f1 but referred to here as Oct4 for clarity) mRNA expression in mouse ESCs (Rex1GFPd2) during the indicated time courses following “2i” withdrawal. Data are normalized by the geometric mean of two reference genes (Ppia and Gapdh) and are presented as means ± SEM and are the average of at least three biological replicates (n ≥ 3). (B and D) Immunoblots showing Otx2 and Oct4 levels during the indicated time courses of “2i” withdrawal. Dotted lines in (D, upper) indicate where gels have been cut to remove irrelevant lanes and rejoined. Quantification of the data from (D) is shown below and are normalized for Erk2 levels and presented as means ± SEM (n = 3). (E and F) The expression of Otx2 and Oct4 was analyzed after knockdown of Otx2, Oct4 or in the presence of a nontargeting control siRNA (ctrl) at the indicated times following “2i” withdrawal by qRT-PCR (E) or western blotting (F). (G) Western blot of Otx2 and Oct4 levels in wild-type (Rex1GFPd2) mESCs and a clonally derived cell line containing a doxycyclin-inducible Otx2 transgene (Otx2/dox-2). Cells are released of, from “2i” for 12 hr. Oct4 depletion (oct4i), the presence (+) or absence (−) of doxycycline (dox) and the proteasome inhibitors, lactacystin and MG132, are indicated. Quantification is shown on the right and is normalized for Erk2 levels and presented as means ± SEM (n = 2). (H) Model illustrating the impact of Oct4 on Otx2 expression at the RNA and protein levels. See also Figure S2.

Figure 3

Differential Properties of Otx2+/Oct4− and Otx2+/Oct4+ Binding Regions (A) Distribution of Otx2 ChIP-seq regions relative to the indicated genomic features. Otx2 binding regions are partitioned into Oct4− (left) and Oct4+ (right) subgroups. The promoter-specific association of Otx2+/Oct4− subgroups is statistically significant (p < 0.01). (B and C) The average normalized tag densities from Otx2 ChIP-seq (B) or FAIRE-seq (C) analyses (y axes) associated with the summit of Otx2 binding regions from the Otx2+/Oct4− and Otx2+/Oct4+ subgroups are shown for the indicated times following “2i” withdrawal (x axes). (D and E) Average tag density profiles of H3K27Ac- (D) and H3K4me1-ChIP-seq (E) analyses mapped onto Otx2 binding region summits in the presence of 2i or following 2i withdrawal for 12 hr. The Otx2 peaks are partitioned into Oct4− (left) and Oct4+ (middle) categories. The right images show the average normalized tag densities of K3K27Ac (D) and H3K4me1 (E) ChIP-seq data (y axis) associated with the summit of Otx2 binding regions from the Otx2+/Oct4− and Otx2+/Oct4+ subgroups for the indicated times following “2i” withdrawal (x axis). (F and H) Heatmaps of Otx2 and Oct4 ChIP-seq tag densities in the 4 kb regions surrounding the summit of Otx2 binding regions at the indicated times following exit from pluripotency (−2i). Data are partitioned into regions showing increased (up; F) or decreased (down; H) binding of Oct4 at 16 hr after 2i withdrawal. (G and I) Tag density profiles of H3K27Ac ChIP-seq data mapped onto the summits of Otx2 binding regions showing increased (G) or decreased (I) Oct4 binding (see F and H) in the presence of 2i or following 2i withdrawal for 12 hr. (J) Gene ontology terms enriched among the genes associated with either the Otx2+/Oct4+(up) or the Otx2+/Oct4+(down) binding regions. Data are shown graphically according to their p values (x axis) and the associated functional category (y axis). See also Figure S3.

Figure 4

Otx2 Is Required for Enhanced H3K27Ac and Oct4 Cobinding at a Subset of Its Binding Regions (A–C) ChIP-qPCR analysis of H3K27Ac levels (A), p300 binding (B), or Otx2 and Oct4 binding (C) to genomic regions associated with the indicated genes in mouse ESCs. Cells were maintained in 2i or cultured in the absence of 2i for 12 hr in the presence of siRNAs against Otx2 or control duplexes. Data in (A) are presented as a heatmap (left) or bar chart (right) and depict the fold change in H3K27Ac levels following release from culture in 2i in the presence or absence of siRNAs against Otx2. Data are presented as means ± SEM (n = 3 or n = 2 [indicated by ^∗] in A and C and n = 2 in B). See also Figure S4.

Figure 5

Otx2 and Oct4 Coregulate Target Gene Expression (A) Volcano plot of microarray data showing gene expression changes in mESCs (Rex1GFPd2) cells 12 hr after “2i” withdrawal following Otx2 depletion or treatment with control siRNAs (top). The table depicts the intersection between expression microarray and Otx2 ChIP-seq data sets (bottom). (B) Pie charts showing the proportion of genes up or downregulated upon withdrawal of 2i from ESCs for 12 hr, for subcategories of genes that are either activated (top) or repressed (bottom) by Otx2. (C) Summary of gene expression changes affected by Otx2 and Oct4 depletion observed 12 hr after 2i release. p Values are calculated from a hypergeometric distribution. (D) UCSC genome browser view of Otx2 binding profiles and RNA-seq tags associated with the Slc16a3 locus. Data are shown for the indicated times following release from 2i and for the RNA-seq data, treatment with control (ctrl) or the indicated RNAi (i) duplexes. RPKM values for each time point are shown on the right. (E and F) Heatmap (E) and selected examples (F) of qRT-PCR analysis with the Fluidigm system of mRNA expression of the indicated genes over 24 hr following 2i withdrawal. Cells were treated with control, Otx2, or Oct4 siRNAs where indicated. Data in (E) are row Z score normalized and are presented in (F) as means ± SD (n ≥ 3). Genes associated with Oct4 binding regions are indicated by purple dots. (G) Expression analysis of the genes in the “primary neural tube formation” gene ontology category that are direct targets of Otx2 and Oct4, either in 2i conditions or upon withdrawal from 2i for 12 hr in the presence of control (ctrl) or Otx2 siRNAs. Data are from RNA-seq analysis and are shown as a heatmap (row Z-normalized; left) and the individual profiles of the top six genes in the heatmap are shown graphically (middle). A UCSC genome browser view of the ChIP profiles across the Enah locus is shown on the right. See also Figure S5.

Figure 6

Enhancer Activation Correlates with Otx2 Target Gene Activation (A and B) Average tag density profiles showing indicated ChIP-seq data mapped onto the 2 or 4 kb regions surrounding the summits of Otx2 binding regions associated with Otx2 activated genes and from the Otx2+/Oct4− and Otx2+/Oct4+ subgroups (A) or associated with Otx2 binding regions showing increased (up) or decreased/no change (down) in Oct4 binding in the presence of 2i or for the indicated times following 2i withdrawal (B). Numbers in brackets represent the number of binding regions in each subgroup. (C–E) UCSC genome browser views of Otx2, Oct4, and H3K27Ac binding profiles and RNA-seq signals associated with the Hells (C), Zfp42 (D), and Dnmt3a (E) loci. Otx2 binding regions are boxed. RPKM values for the RNA-seq data are shown adjacent to the relevant lanes. See also Figure S6.

Figure 7

Otx2 Drives Enhancer Activation, Oct4 Recruitment, and Target Gene Expression (A–G) Heatmap (A) and selected examples (B, D, and F) of qRT-PCR analysis with the Fluidigm system of mRNA expression of the indicated genes in wild-type (Rex1GFPd2), mESCs and two clonally derived cell lines, Otx2/dox-1 and Otx2/dox-2, containing a doxycyclin-inducible Otx2 transgene. Cells were maintained in 2i media and either left untreated or treated with increasing concentrations (10 and 100 ng/ml) of doxycycline (dox) for 24 hr. Data in (A) are row Z score normalized and are presented in (B, D, and F) as means ± SEM (n = 3). The vertical color bars represent the following categories of genes based on siRNA depletion experiments; Otx2 upregulated (red), Otx2 downregulated (blue), control nontargets (green), and not regulated by Otx2 (gray). Gene expression following release from 2i for 12 hr is also shown in (B, D, and F). ChIP-qPCR analysis of H3K27Ac levels or Otx2 and Oct4 binding to genomic regions associated with the indicated genes in Otx2/dox-3 mESCs (C, E, and G). Cells were maintained in 2i either in the presence or absence of doxycycline (dox) treatment for 24 hr. Data are presented as means ± SEM (n = 3). (H) Model illustrating the impact of Otx2 on enhancer activation during early stem cell commitment. See also Figure S7.