Two independent transcription initiation codes overlap on vertebrate core promoters

Abstract

A core promoter is a stretch of DNA surrounding the transcription start site (TSS) that integrates regulatory inputs and recruits general transcription factors to initiate transcription. The nature and causative relationship of the DNA sequence and chromatin signals that govern the selection of most TSSs by RNA polymerase II remain unresolved. Maternal to zygotic transition represents the most marked change of the transcriptome repertoire in the vertebrate life cycle. Early embryonic development in zebrafish is characterized by a series of transcriptionally silent cell cycles regulated by inherited maternal gene products: zygotic genome activation commences at the tenth cell cycle, marking the mid-blastula transition. This transition provides a unique opportunity to study the rules of TSS selection and the hierarchy of events linking transcription initiation with key chromatin modifications. We analysed TSS usage during zebrafish early embryonic development at high resolution using cap analysis of gene expression, and determined the positions of H3K4me3-marked promoter-associated nucleosomes. Here we show that the transition from the maternal to zygotic transcriptome is characterized by a switch between two fundamentally different modes of defining transcription initiation, which drive the dynamic change of TSS usage and promoter shape. A maternal-specific TSS selection, which requires an A/T-rich (W-box) motif, is replaced with a zygotic TSS selection grammar characterized by broader patterns of dinucleotide enrichments, precisely aligned with the first downstream (+1) nucleosome. The developmental dynamics of the H3K4me3-marked nucleosomes reveal their DNA-sequence-associated positioning at promoters before zygotic transcription and subsequent transcription-independent adjustment to the final position downstream of the zygotic TSS. The two TSS-defining grammars coexist, often physically overlapping, in core promoters of constitutively expressed genes to enable their expression in the two regulatory environments. The dissection of overlapping core promoter determinants represents a framework for future studies of promoter structure and function across different regulatory contexts.

Figure 1. Dynamics of transcription initiation at 1bp resolution throughout zebrafish early embryonic development

a, CAGE signal at “shifting” promoter of cyclin 1 (ccni) gene. Colouring from blue to red reflects maternal to zygotic transition. Corresponding zebrafish developmental stages are depicted on the left, with timescale denoting hours past fertilization (hpf). b, Expression profiles obtained by self-organizing map (SOM) clustering of individual CAGE transcription start sites (CTSS). Each box represents one cluster, with beanplots showing distribution of relative expression at different time points for all CTSSs belonging to that cluster (number above the box). The developmental stages at x-axis in all boxes are shown at the bottom.

Figure 2. Sequence signature of a large set of “shifting” promoters changes dramatically during maternal to zygotic transition

a, Dinucleotide density (see Extended Data Fig. 10) at 911 “shifting” promoters sorted and aligned according to the distance and orientation of the TSS shift (schematics on the top; P1, P2, P3 – individual promoters; M – maternal stage; Z – zygotic stage). Promoters were centred at either maternal (left) or zygotic (right) dominant TSS. Blue arrowhead: TA enrichment at the expected position of the TATA-box; red arrowhead: boundary between GC and TA enrichment ~50bp downstream of zygotic TSS. b, Distribution of match (%) to TATA-box in the region −35 to −22 bp upstream of maternal (blue) and zygotic (red) dominant TSS (P-value - two-tailed Wilcoxon rank-sum test). c, Motif obtained by motif discovery upstream of maternal dominant TSS.

Figure 3. Transition from maternal W-box motif-dependent, to zygotic nucleosome positioning signal-related transcription initiation is pervasive

a, Dinucleotide density at 8369 constitutively expressed promoters sorted by the distance between maternal and zygotic dominant TSS. Promoters were centred at either maternal (left) or zygotic (right) dominant TSS. Blue arrowhead: position of maternal TSS-associated W-box; red arrowhead: SS∣WW boundary ~50bp downstream of zygotic TSS; asterisks: GC enrichment in the internucleosomal region. b, Predicted maternal and zygotic codes in sf3a2 promoter. Dinucleotide density and sequence of the wild-type (wt) and mutated (mut) sf3a2 promoter is shown on top. TSSs detected by CAGE in wild type zebrafish in maternal and zygotic stage are shown in blue and red, respectively. The W-boxes associated with maternal TSSs are marked in orange, and the introduced point mutations disrupting them in red. Single locus CAGE TSSs in stable transgenic lines for endogenous sf3a2, wild type sf3a2 transgene and mutant sf3a2 transgene are shown in different shades of purple (** P < 0.01, *** P < 0.001, one-tailed Welch’s two sample t-test, n_mut = 4, n_wt = 3). c, Subtracted H3K4me3 coverage (Δ) of reads mapping to (+) and (−) strand (schematic on top) in three developmental stages at the same set of promoters from panel a. d, Density of AA dinucleotide in +/− 100 bp region for promoters from panel a.

Figure 4. H3K4me3-marked nucleosome positioning reveals dynamic changes in underlying sequence signature and relation to TSS during MZT

a, Frequency of dinucleotides centred on +1 nucleosome of constitutively active promoters in maternal (512 cells) and zygotic (prim 6) stage. Centres of nucleosomes were estimated from subtracted H3K4me3 coverage (gray). Density of maternal and zygotic transcription start sites is shown in light blue and light red, respectively. b, H3K4me3 signal at promoters of constitutively present transcripts sorted by the time of activation of their zygotic component. Horizontal lines separate groups of promoters that activate zygotic component at a denoted developmental stage. c, H3K4me3 signal at TBP-dependent promoters in non-injected embryos (top), embryos injected with mismatch morpholino (middle) or TBP-targeting morpholino (bottom), sorted by TBP expression fold-change between knockdown and wild type embryos. d, Summary of transcription initiation, TSS configuration and nucleosome positioning dynamics throughout MZT.