Transient pairing of homologous Oct4 alleles accompanies the onset of embryonic stem cell differentiation

Abstract

The relationship between chromatin organization and transcriptional regulation is an area of intense investigation. We characterized the spatial relationships between alleles of the Oct4, Sox2, and Nanog genes in single cells during the earliest stages of mouse embryonic stem cell (ESC) differentiation and during embryonic development. We describe homologous pairing of the Oct4 alleles during ESC differentiation and embryogenesis, and we present evidence that pairing is correlated with the kinetics of ESC differentiation. Importantly, we identify critical DNA elements within the Oct4 promoter/enhancer region that mediate pairing of Oct4 alleles. Finally, we show that mutation of OCT4/SOX2 binding sites within this region abolishes inter-chromosomal interactions and affects accumulation of the repressive H3K9me2 modification at the Oct4 enhancer. Our findings demonstrate that chromatin organization and transcriptional programs are intimately connected in ESCs and that the dynamic positioning of the Oct4 alleles is associated with the transition from pluripotency to lineage specification.

Figure 1. Assessing potential homologous and heterologous interactions between Oct4, Sox2, and Nanog gene loci during ESC differentiation

(A–E) Representative images of 3-D multi-color DNA FISH analysis of gene position in ESCs, at 0, 1, 2, 3, and 4 days after the withdrawal of LIF. Individual ESC nuclei outlined with dashed line. Red arrows in (D) denote paired Oct4 alleles. Scale bar = 5μm. (F) Quantification of allelic interactions between Oct4, Sox2, and Nanog alleles. n=100 per locus pair at each time point. p-values for the 2-sample z-test statistic are reported. **= p<0.01. See also Figures S1 and S2.

Figure 2. The Timing of Oct4 Allelic Pairing Reflects the Kinetics of ESC Differentiation

(A) Quantification of homologous pairing at representative genomic loci in ESCs before (Day 0) and after 3 days of differentiation (−)LIF. n=100 per locus pair at each time point, error bars = +/− SEM of 3 biological replicates. **= p<0.01. (B) Assessment of the transcriptional status of paired Oct4 alleles, by RNA and DNA FISH. The percentage of paired Oct4 alleles with two active alleles (top panel, red arrows), one active allele (middle panel, red arrow), or zero active alleles (bottom panel) is indicated. Individual ESC nuclei are outlined with dashed line. n=93 paired loci. Scale bar = 5μm. (C) Schematic diagram of differentiation scheme with Oct4.468/R26-rtTA ESCs. (D) Assessment of Oct4 gene expression by quantitative RT-PCR over 3 days of differentiation, using (−)LIF, RA, or shRNA differentiation paradigms. Gene expression is normalized to β-actin expression for each sample. (E) Quantification of Oct4 allelic interactions over 3 days of differentiation, using (−)LIF, RA, or shRNA differentiation paradigms. n=100 for each time point. Error bars = +/− SEM of 3 biological replicates. p-values for the 2-sample z-test statistic are reported. **= p<0.01. (F) Representative images of paired Oct4 alleles in Oct4.468/R26-rtTA ESCs differentiated in (−)LIF, RA, or shRNA conditions. See also Table S1 and Table S2.

Figure 3. Oct4 allelic associations occur during post-implantation development in vivo

(AC) Representative images of a 10μm sagittal cryosection of an E7.75 mouse embryo, showing A) OCT4 immunofluoresence (IF), B) DAPI stained nuclei, and C) OCT4 IF overlayed with DAPI. Scale bar = 100μm. (D–F) Enlargement of corresponding boxed areas in panel (B), showing Oct4 DNA FISH in the D) anterior, E) middle, and F) posterior regions of the ectoderm/neuroectoderm. Nuclei with paired Oct4 alleles are outlined with dashed lines. Scale bar = 5μm. (G) Quantification of allelic associations in E7.75 embryos. Error bars = +/− SEM of 3 biological replicates. p-values for the 2-sample z-test statistic are reported. **= p<0.01. See also Figure S3.

Figure 4. A DNA fragment encompassing the 5′ regulatory region of the Oct4 gene is sufficient to induce pairing with endogenous Oct4 loci

(A) Schematic representation of a 10kb genomic region at the Oct4 locus on mouse chromosome 17, which was divided into 4 separate genomic DNA fragments: OCF2, OCF3, OCF4, and OCF5. (B) Oct4 genomic fragments were inserted at the Col1A1 locus in KH2 ESCs. (C) Targeted insertion of Oct4 fragments at the Col1A1 locus was verified by DNA FISH. Representative images of each cell line are shown, with the transgenic (Tg) and WT Col1A1 loci labeled. Inset images show an enlarged region of the Tg-Col1A1 locus. Scale bar = 5μm. (D) Schematic representation of distance measurements made between the endogenous Oct4 alleles (Oct4(En)), the WT-Col1A1 allele, and the transgenic Col1A1 allele (Tg-Col1A1). E) Quantification of allelic interactions between endogenous Oct4 loci (Oct4(En)), WT-Col1A1, and Tg-Col1A1 in each cell line. n=100 each time point, error bars = +/− SEM of 3 biological replicates. p-values for the 2-sample z-test statistic are reported. **= p<0.01. (F–G) Representative images of paired loci in OCF2 ESCs. Pairing of Tg-Col1A with (F) one endogenous Oct4 locus was observed more frequently than simultaneous pairing of Tg-Col1A with both endogenous Oct4 loci (G). Scale bar = 5μm. See also Figure S4.

Figure 5. Mutation of putative OCT4/SOX2 binding sites within the OCF2 DNA fragment abolishes pairing with endogenous Oct4 loci

(A) Schematic representation of the 3.6 kb OCF2 fragment, with putative binding sites for CTCF, E2A, OCT4/SOX2, and YY1. (B) Targeted insertion of mutated OCF2 fragments at the Col1A1 locus was verified by DNA FISH. Representative images of each cell line, with the transgenic (Tg) and WT Col1A1 loci labeled. Inset images show enlarged region of the Tg-Col1A1 locus. Scale bar = 5μm. (C) Quantification of allelic interactions between endogenous Oct4 loci (Oct4(En)), WT-Col1A1, and Tg-Col1A1 in each cell line. n=100 for each time point, error bars = +/− SEM of 3 biological replicates. p-values for the 2-sample z-test statistic are reported. **= p<0.01, *=p<0.05. (D–I) Representative images of paired loci in OCF2ΔCTCF, OCF2ΔYY1, and OCF2ΔE2A ESCs. Percentage of pairing events in each cell line involving two or three loci are indicated. Scale bar = 5μm. See also Figure S5 and Table S3.

Figure 6. OCT4 protein binding at four sites within the OCF2 DNA fragment is disrupted by site-directed mutagenesis

A) Schematic representation of the 3.6 kb OCF2 fragment, with putative binding sites for OCT4/SOX2 proteins indicated. B) Binding of recombinant OCT4 protein to WT or mutated oligos was assessed. Upward shift in the gel band (*→) indicates OCT4 binding. Unbound oligonucleotides are seen as a lower band (→). C) Oct4 ChIP-QPCR was performed in OCF2ΔOCT4/SOX2 ESCs differentiated for 0,3, or 6 days (−)LIF. Allele-specific Taqman probes were used to distinguish the endogenous Oct4 distal enhancer (DE) region (Chr. 17, blue underline “A”) from the mutated transgenic DE region (Chr. 11, red underline “B”). OCT4 enrichment at each site is reported relative to IgG. Error bars = +/− SEM of 3 biological replicates, and p-values for Student’s t-Test statistic are reported. *=p<0.05. D) Oct4 ChIP-QPCR was performed in OCF2ΔYY1 ESCs, using allele-specific Taqman probes to distinguish endogenous Oct4 distal enhancer (DE) region (Chr. 17, blue underline “A”) from mutated transgenic DE region (Chr. 11, green underline “C”). OCT4 enrichment expressed relative to IgG. Error bars = +/− SEM of 3 biological replicates. See also Tables S4.

Figure 7. Intact OCT4/SOX2 protein binding sites are necessary for inter-chromosomal interactions in trans

A–C) Circular chromosome conformation capture followed by paired-end DNA sequencing (4C-seq) utilized to evaluate inter-chromosomal interactions between the WT endogenous Oct4 locus (Chr. 17) and a mutated OCF2 transgenic locus (Chr. 11) in A) “pairing deficient” OCF2ΔOCT4/SOX2 ESCs and B) “pairing competent” OCF2ΔYY1 ESCs. C) The frequency of inter-chromosomal contacts between the WT Oct4 locus (Chr. 17) and a mutated OCF2 transgenic locus (Chr. 11) was determined and is reported as the fold enrichment relative to the background level of inter-chromosomal interactions at a non-interacting genomic region. The change in trans-interactions was compared between Day 0 and Day 3 in each cell line. Error bars = +/− SEM of 2 biological replicates, and p-values for Student’s t-Test statistic are reported. *=p<0.05. D) H3K9me2 ChIP-QPCR was performed in OCF2ΔOCT4/SOX2 ESCs differentiated for 0, 3, or 6 days (−) LIF. Allele-specific Taqman probes were used to distinguish the endogenous Oct4 distal enhancer (DE) region (Chr. 17, blue underline “A”) from the mutated transgenic DE region (Chr. 11, red underline “B”). H3K9me2 enrichment at each site is reported as fold enrichment relative to Day 0 values. Error bars = +/− SEM of at least 2 biological replicates, and p-values for Student’s t-Test statistic are reported. *=p<0.05. E) H3K9me2 ChIP-QPCR was performed in OCF2ΔYY1 ESCs differentiated for 0, 3, or 6 days (−) LIF. Allele-specific Taqman probes were used to distinguish the WT Oct4 distal enhancer (DE) region (Chr. 17, blue underline “A”) from the mutated transgenic DE region (Chr. 11, green underline “B”). H3K9me2 enrichment at each site is reported as fold enrichment relative to Day 0 values. Error bars = +/− SEM of at least 2 biological replicates.