Dynamic regulation of the transcription initiation landscape at single nucleotide resolution during vertebrate embryogenesis

Abstract

Spatiotemporal control of gene expression is central to animal development. Core promoters represent a previously unanticipated regulatory level by interacting with cis-regulatory elements and transcription initiation in different physiological and developmental contexts. Here, we provide a first and comprehensive description of the core promoter repertoire and its dynamic use during the development of a vertebrate embryo. By using cap analysis of gene expression (CAGE), we mapped transcription initiation events at single nucleotide resolution across 12 stages of zebrafish development. These CAGE-based transcriptome maps reveal genome-wide rules of core promoter usage, structure, and dynamics, key to understanding the control of gene regulation during vertebrate ontogeny. They revealed the existence of multiple classes of pervasive intra- and intergenic post-transcriptionally processed RNA products and their developmental dynamics. Among these RNAs, we report splice donor site-associated intronic RNA (sRNA) to be specific to genes of the splicing machinery. For the identification of conserved features, we compared the zebrafish data sets to the first CAGE promoter map of Tetraodon and the existing human CAGE data. We show that a number of features, such as promoter type, newly discovered promoter properties such as a specialized purine-rich initiator motif, as well as sRNAs and the genes in which they are detected, are conserved in mammalian and Tetraodon CAGE-defined promoter maps. The zebrafish developmental promoterome represents a powerful resource for studying developmental gene regulation and revealing promoter features shared across vertebrates.

Figure 1.

Mapping of transcription initiation in zebrafish embryo development. (A–C) Genome browser view of ncalda gene with CAGE-seq, ChIP-seq, and RNA-seq tracks from selected developmental stages. Schematic representation of developmental stages is on the left. Vertical bar with blue (maternal) and red (zygotic) bars indicates transcriptional activity of the genome. White arrowhead indicates the onset of zygotic transcription at the mid-blastula transition (MBT). Vertical scales on the left of tracks are tpm values and fixed within experiments. Height of the CTSS bars is proportional to the number of CAGE tags aligned to that position. Transcript clusters (TC) of varying width are labeled with brackets. (A) Full-length transcripts of ncalda indicating two promoter regions (arrow and arrowhead) were detected by CAGE and verified by H3K4me3 peaks and RNA-seq data. (Fert) Fertilized. (B) High-resolution mapping of zygotically active novel alternative TSS (arrow in A) of ncalda gene. (C) High-resolution mapping of continuously active Ensembl annotated TSS (arrowhead in A) of ncalda gene. (D, top) Schematic of gene structures for analysis of distribution of TCs. (Bottom) Number of TCs overlapping with the annotated segments of the genome is shown at the developmental stages indicated by schematics. Colors from blue to red indicate transition from maternal to zygotic transcriptomes. (E) Distribution of sharp and broad TCs at selected developmental stages. Shades of color indicate gene segment, blue to red transition indicates maternal to zygotic transition of transcriptome. P-values of one-tailed Fisher's exact test for selected comparisons are denoted above the bars. (F) Intersection of Ensembl gene 5′-ends detected by CAGE (>1 tpm, shown in red), RNA-seq (>1 rpkm, shown in blue), and H3K4me3 peaks (in green) at prim-6 stage.

Figure 2.

Identification and developmental dynamics of alternative initiation sites. (A) Frequency and developmental dynamics of alternative promoters. Colors reflect maternal to zygotic transition as in Figure 1. Genes with up to three alternative promoters are plotted (see Supplemental Table 7 for details). Shades indicate alternative promoter numbers, color transition indicates maternal to zygotic transition of transcriptome. (B) Clustering of three sets of genes based on their reference promoter activity. Annotated reference promoters (as assigned by Ensembl 71) are on the left and alternative promoters on the right. Genes are clustered in three groups according to the reference promoter being active during maternal (M), zygotic (Z), or maternal and zygotic (M-Z) stages. Total number of genes in each group is indicated in the left. Black rectangles indicate genes where the previously unannotated alternative promoter's activity is preferential over that of the annotated reference promoter. (C) Fluorescent Venus reporter activity driven by alternative (ALT1) and novel (NVL1) core promoters attached to a neural specific enhancer (E) in transgenic embryos. Control (CON) indicates a random DNA fragment replacing a promoter. Maximum projections of embryos overlaid from a single injection experiment are shown (see details in Supplemental Table 8). Bright-field (BF) image of a single zebrafish embryo is shown for reference. (Arrowhead) Cerebellum; (arrow) spinal cord activity.

Figure 3.

Sequence characteristics of developmentally regulated core promoters are evolutionarily conserved. (A) A genome browser view with annotated Tetraodon genes (top of the panel) along with CAGE-seq tracks from two developmental (maternal and zygotic) stages. In yellow boxes, core promoter regions of annotated genes expressed specifically at maternal (M), zygotic (Z), or maternal and zygotic stages (MZ). CTSSs in red and blue indicate sense and antisense direction, respectively. (B) Correlation of dinucleotides of CTSSs (−1,+1) between zebrafish and Tetraodon orthologs represented as fold enrichments vs. expected by chance. Asterisks denote significant correlations (P ≤ 0.05). Only dinucleotides, which occur at TSSs, are shown. (C,D) Sequence logos and their information content of initiator motifs for selected dinucleotides: (C) CC/TC and (D) AA dinucleotides. Human genes with “AA” initiation motifs were plotted from ENCODE cell lines (see Methods). (E) Enriched GO terms of genes with AA initiator. Identical terms are highlighted in gray. Heat map represents the –log (P-values) of enriched GO terms.

Figure 4.

Exonic CAGE tags during development. (A) A genome browser view of the intragenic region of the tp53 gene. Arrow with “ia” indicates the intronic alternative promoter of the delta117p53 variant (Chen and Peng 2009; Chen et al. 2009). Arrows labeled “e” and “u” indicate RNA start sites of exonic and 3′-UTR regions, respectively. (B) Lack of fluorescent Venus reporter activity in maximum projection overlays of 36 hpf embryos injected with an exonic CAGE marked candidate promoter region (EXN1) as compared with an active core promoter (ALT4) and a negative control (CON1) linked to a neural specific enhancer (E) (details in Supplemental Table 8). Insert of a bright-field image of an embryo is shown as reference for view of fluorescence image.

Figure 5.

Distribution and developmental regulation of intronic CAGE tags. (A) Distribution of all intronic TCs aggregated and aligned in windows of 1% length of a normalized intron. TCs in specific stages are shown in colors as indicated. Insets show aggregates of CAGE tags aligned at single bases, up to 10 bases from either side of intron ends. (B) A genome browser view of splice donor site of the acin1b gene and associated intronic 5′-end CAGE tags. (C) Venn diagram of intersection and number of genes with various types of intronic TCs. (D) Enriched GO terms of genes with intron 5′-end CTSSs in zebrafish and human. The heat map represents the –log (P-values) of significantly enriched GO terms. (MBT) Mid-blastula transition.

Figure 6.

Intragenic CAGE tags do not carry core promoter features. (A) Dinucleotide frequency analysis of dominant CTSSs (−1,+1 bp) of gene 5′-end promoter and of intragenic RNA products. Relative abundance of dinucleotides is shown in bubbles of varying size. Number of TCs analyzed from 12 stages are indicated in brackets (repeat incidences in multiple stages are not included). (B) Sequence logos and their information content of dominant CTSSs (−1,+1 bp) at gene 5′-end and intragenic sites in zebrafish and human.