Chromatin signature of embryonic pluripotency is established during genome activation

Abstract

After fertilization the embryonic genome is inactive until transcription is initiated during the maternal-zygotic transition. This transition coincides with the formation of pluripotent cells, which in mammals can be used to generate embryonic stem cells. To study the changes in chromatin structure that accompany pluripotency and genome activation, we mapped the genomic locations of histone H3 molecules bearing lysine trimethylation modifications before and after the maternal-zygotic transition in zebrafish. Histone H3 lysine 27 trimethylation (H3K27me3), which is repressive, and H3K4me3, which is activating, were not detected before the transition. After genome activation, more than 80% of genes were marked by H3K4me3, including many inactive developmental regulatory genes that were also marked by H3K27me3. Sequential chromatin immunoprecipitation demonstrated that the same promoter regions had both trimethylation marks. Such bivalent chromatin domains also exist in embryonic stem cells and are thought to poise genes for activation while keeping them repressed. Furthermore, we found many inactive genes that were uniquely marked by H3K4me3. Despite this activating modification, these monovalent genes were neither expressed nor stably bound by RNA polymerase II. Inspection of published data sets revealed similar monovalent domains in embryonic stem cells. Moreover, H3K4me3 marks could form in the absence of both sequence-specific transcriptional activators and stable association of RNA polymerase II, as indicated by the analysis of an inducible transgene. These results indicate that bivalent and monovalent domains might poise embryonic genes for activation and that the chromatin profile associated with pluripotency is established during the maternal-zygotic transition.

Figure 1. Large-scale changes in chromatin modifications during maternal-zygotic transition

a Western blot analysis of zebrafish embryos, from early blastula stages (8-cell; 1.25 hours post fertilization (hpf)) to the onset of gastrulation (shield; 6 hpf). Note the shift of RNA pol II at oblong stage due to phosphorylation of the C-terminal domain, indicating engagement in transcription. b–e. Average density profiles for RNA polymerase II (b), H3K36me3 (c), H3K4me3 (d) and H3K27me3 (e) prior to (256-cell: black line) and shortly after the maternal-zygotic transition (MZT) (dome/30%: red line). These profiles show the average normalized and smoothed log₂ ChIP enrichment (MA2Cscore) for all 822 analyzed RefSeq genes on the array. Transcription units are shown as metagenes (i.e. relative distance from transcription start site (TSS) to transcription termination site (TTS)), whereas upstream and downstream sequences are shown in absolute distance (bp). f. Example profiles for dact2 before and after the maternal-zygotic transition (MZT). Absence of signal in all four traces indicates absence of probes due to repetitive sequences.

Figure 2. Bivalent chromatin domains in zebrafish embryos

a Pie chart showing all inactive genes (see Supplementary Discussion 1 and Supplementary Figure 6 for criteria) and the relative contributions of bivalent, monovalent and not marked genes. b. Example profiles for cmyb, an inactive gene with a bivalent promoter after the maternal-zygotic transition (MZT) (dome/30%). Absence of signal in all four traces indicates absence of probes due to repetitive sequences. c. Sequential ChIP analysis. Shown is fold enrichment over a negative control. In black, H3K27me3 (1^st ChIP) > H3K4me3 (2^nd ChIP); hoxa13b-ups served as a negative control because this region is occupied by H3K27me3 but not H3K4me3. In gray, H3K4me3 > H3K27me3; β-actin1 served as a negative control because this region is occupied by H3K4me3 but not H3K27me3. Enrichment values are shown as the mean of two or three independent experiments +SEM. See Methods section for details.

Figure 3. Many inactive genes are monovalently marked with H3K4me3

a ChIP-chip profiles for zgc:110784, an inactive gene with a monovalent promoter. b. ChIP-chip profiles for prkcb1, an inactive gene with a monovalent promoter that is associated with RNA pol II around the TSS. c. Plot of normalized ChIP signals for H3K4me3 and RNA pol II reveals genes with undetectable levels of RNA pol II but high levels of H3K4me3.

Figure 4. H3K4me3 occupancy in the absence of a sequence-specific transcriptional activator

a Embryos carrying an integrated transgene (UAS₁₄:EGFP) were either injected with GAL4-VP16 mRNA or left un-injected. Quantitative RT-PCR analysis detected GFP RNA only after the maternal-zygotic transition (MZT), and only in the presence of GAL4-VP16 (data not shown). At dome stage/30%-epiboly, embryos were subjected to ChIP-quantitative PCR (ChIP-qPCR). Arrows indicate qPCR amplicons. b–c. Representative ChIP-qPCR analysis of H3K4me3 (b) and RNA polymerase II (c) in the presence (green line) and absence (black line) of GAL4-VP16. H3K27me3 was not detected at the transcription start site (data not shown).