Pluripotency transcription factor Oct4 mediates stepwise nucleosome demethylation and depletion

Abstract

The mechanisms whereby the crucial pluripotency transcription factor Oct4 regulates target gene expression are incompletely understood. Using an assay system based on partially differentiated embryonic stem cells, we show that Oct4 opposes the accumulation of local H3K9me2 and subsequent Dnmt3a-mediated DNA methylation. Upon binding DNA, Oct4 recruits the histone lysine demethylase Jmjd1c. Chromatin immunoprecipitation (ChIP) time course experiments identify a stepwise Oct4 mechanism involving Jmjd1c recruitment and H3K9me2 demethylation, transient FACT (facilitates chromatin transactions) complex recruitment, and nucleosome depletion. Genome-wide and targeted ChIP confirms binding of newly synthesized Oct4, together with Jmjd1c and FACT, to the Pou5f1 enhancer and a small number of other Oct4 targets, including the Nanog promoter. Histone demethylation is required for both FACT recruitment and H3 depletion. Jmjd1c is required to induce endogenous Oct4 expression and fully reprogram fibroblasts to pluripotency, indicating that the assay system identifies functional Oct4 cofactors. These findings indicate that Oct4 sequentially recruits activities that catalyze histone demethylation and depletion.

FIG 1

ESCs can be differentiated normally with low-level Oct4. (A) Sequence of the murine Pou5f1 enhancer region. CpG positions are boxed. Known transcription factor binding sites are underlined. (B) Oct4 Western blots of lysates from tetON-Oct4 ESCs (21) and differentiated ESCs cultured for 16 days in medium lacking LIF and containing RA, with or without doxycycline. Experiments were performed in the absence and presence of doxycycline (Dox). Oct4, Nanog, and REST protein expression levels were evaluated. β-Actin was used as a loading control. (C) Oct1 and Oct4 ChIP enrichment at the Pou5f1 distal enhancer. Differentiation was for 12 days. Values represent averages of triplicate experiments. Error bars depict the standard deviations of the mean. (D) Gata4 mRNA expression was determined by qRT-PCR under similar conditions to (C). mRNA levels were normalized to β-actin and changes in expression are represented as the fold change (n = 3). Error bars depict the standard deviations. (E) Phase microscopy images of cultured tetON-Oct4 ESCs. The upper left panel shows cells propagated in ESC culture medium. The upper right panel shows cells differentiated for 16 days in medium lacking LIF and containing RA. The bottom panel shows cells also cultured in doxycycline. Images were taken using an Olympus IX51 inverted microscope at ×40 magnification with a Lumenera Infinity 2 camera. (F) Pou5f1 mRNA expression levels as determined by qRT-PCR using primers specific for the endogenous allele (n = 3). Error bars depict the standard deviations. (G) Bisulfite sequencing analysis of the murine Pou5f1 enhancer and promoter regions. A schematic is shown at the top. The Oct4 binding site is boxed. Filled circles indicate DNA methylation, and open circles indicate unmethylated DNA for a particular sequenced clone. Sequence reads on either side of dashed line were separately amplified, miniprepped, and sequenced.

FIG 2

Oct4 prevents DNA methyl mark deposition at Pou5f1. (A) Dnmt3a reactivity with the Pou5f1 distal enhancer region was assessed by Aza-IP using differentiating ESCs. Enrichment in normal ESCs (day 0) lacking a 5-AzadC pulse is arbitrarily set to 1.0, and the relative enrichment for 1 μM RA-treated cells with or without a 5-AzadC pulse is shown. (B) tetON-Oct4 ESCs differentiated for 12 days with RA with or without the addition of doxycycline were used in Aza-IP assays. Enrichment in normal ESCs (day 0) lacking a 5-AzadC pulse is arbitrarily set to 1.0, and the relative enrichment for RA-treated cells with or without a 5-AzadC pulse is shown. (C) tetON-Oct4 ESCs were differentiated by removing LIF and adding RA. A 12-day time course is shown. Oct4 expression was probed by Western blotting. β-Actin was used as a loading control. No doxycycline was present.

FIG 3

Oct4 associates with Jmjd1c and FACT at Pou5f1. (A) Jmjd1a and Jmjd1c ChIP enrichment was measured at the Pou5f1 enhancer in tetON-Oct4 ESCs, ESCs undergoing RA-mediated differentiation for 11 days, and similarly differentiated ESCs treated with doxycycline for 12 h to induce Oct4. (B) Coimmunoprecipitation assay using undifferentiated tetON-Oct4 ESCs (without doxycycline in the medium). Lysates were immunoprecipitated with anti-Jmjd1c antibodies or with an isotype control, followed by Western blotting with Oct4 antibodies. (C) Differentiation time course of Oct1, Jmjd1c, Dmnt3a, and H3K9me2 protein expression (without doxycycline). Western blot data are shown. Histone H3 and β-actin were used as loading controls. (D) Histone H3 and H3K9me2 ChIP enrichment in 11-day-differentiated ESCs (with or without 12 h of doxycycline treatment to induce ectopic Oct4 expression). (E) A ChIP time course was performed using ESCs differentiated for 11 days with RA and subsequently induced using doxycycline for the indicated times. ChIP was performed at the Pou5f1 enhancer with the indicated antibodies. For each antibody, the ChIP signal is shown as a percentage of the maximum (100%). Each experiment was performed in biological triplicates. Error bars depict the standard deviations. In cases where the same time point showed maximum signal for all biological and technical replicates (Oct4, histone H3, and SSRP1), the signal was 100%, and the standard deviation for that time point was 0. In other cases (Jmjd1c and H3K9me2) the maximum signal was at different times for different replicates, and an error was derived at all time points. (F) A similar ChIP experiment was performed at the 0- and 10-h time points with antibodies against the FACT subunit Spt16. (G) A Western blot was performed with antibodies against Oct4 to monitor production of protein over the same time course. ESCs and endogenous Oct4 are shown as a positive control. β-Actin was used as a loading control. (H) Endogenous Pou5f1 expression was monitored in partially differentiated ESCs and the same cells treated with doxycycline for 12 h using qRT-PCR. Undifferentiated ESCs were used as a control. (I) ChIP was performed at the Pou5f1 enhancer in 11-day RA-differentiated ESCs and after 10 h of induction with doxycycline as in panel D; however, DMOG was also added after 4 h of doxycycline treatment. Cells were therefore exposed to doxycycline for 10 h and to DMOG for 6 h. Jmjd1c, H3K9me2, Spt16, and H3 antibodies were used.

FIG 4

Oct4, Jmjd1c, and FACT coassociate with multiple genomic targets. (A) Relative alignment depth coverage tracks of ChIP-Seq data corresponding to the murine Pou5f1 locus and surrounding region. Oct4 enrichment is shown in blue in 11-day RA-differentiated ESCs in the absence of doxycycline (0 h Oct4) or in red after a 10-h doxycycline treatment (10 h Oct4). Jmjd1c and Spt16 enrichment is shown below in red. Input controls for each time point are shown below in black. Relative depth tracks are scaled the same and represented as per-base FPKMs (fragments per kilobase of exon per million fragments). (B to E) Similar data for Vamp1 (B), Slain2 (C), Nanog (D), and Sox2 (E). A region of upstream noncoding RNA expression (Sox2ot) is also shown. (F) Venn diagram depicting target identification for Oct4, Spt16, and Jmjd1c ChIPseq, as well as overlap. The peak calling was based on a 2-fold cutoff with a Q-value FDR of 0.05. (G) ChIP enrichment for Oct4, Jmjd1c, Spt16, and histone H3 was determined by qPCR similarly to Fig. 3; however, a ChIP primer pair spanning the Oct4 site in the murine Nanog, Vamp1, and Slain2 promoters was used in place of Pou5f1.

FIG 5

Jmjd1c is required to completely reprogram MEFs to pluripotency. (A) Uninfected primary Oct4-GFP MEFs (passage 5) or MEFs cotransduced with empty vector or Jmjd1c shRNA-encoding lentiviruses were infected with hSTEMCCA encoding human Oct4, Sox2, Klf4, and c-Myc. Cells were selected with 4 μM puromycin in ESC medium and plated on irradiated feeder fibroblasts. Images were obtained at 14 days posttransduction. Scale bars, 100 μm. (B) Quantification of total colonies and GFP⁺ colonies observed in a 10-cm dish. Averages were taken from four biological replicates. Error bars indicate ± the standard deviations. (C) Oct4-GFP MEFs or mouse ESCs were probed for Jmjd1c by Western blotting (lanes 8 and 9). In lanes 1 to 7, MEFs infected with hSTEMCCA and reprogrammed for 14 days and then collected and probed for Jmjd1c. Lanes 1 to 6 were additionally infected with Jmjd1c knockdown lentiviruses or scrambled controls. β-Actin was used as a loading control.

FIG 6

(Step 1) Model mechanism for sequential chromatin changes mediated by Jmjd1c and FACT recruited to Oct4 at pluripotency targets. (Step 2) Prior to Oct4 induction, target chromatin is characterized by H3K9me2 and the ability of Dnmt3a to methylate CpG DNA). (Step 3) After Oct4 expression, Oct4 and Jmjd1c associate and mediate H3K9 demethylation. (Step 4) FACT is subsequently recruited and mediates nucleosome depletion. FACT dissociates, and Oct4-occupied DNA refractory to methylation by Dnmt3a remains.