Enhancers regulate progression of development in mammalian cells

Abstract

During development and differentiation of an organism, accurate gene regulation is central for cells to maintain and balance their differentiation processes. Transcriptional interactions between cis-acting DNA elements such as promoters and enhancers are the basis for precise and balanced transcriptional regulation. We identified modules of combinations of binding sites in proximal and distal regulatory regions upstream of all transcription start sites (TSSs) in silico and applied these modules to gene expression time-series of mouse embryonic development and differentiation of human stem cells. In addition to tissue-specific regulation controlled by combinations of transcription factors (TFs) binding at promoters, we observed that in particular the combination of TFs binding at promoters together with TFs binding at the respective enhancers regulate highly specifically temporal progression during development: whereas 40% of TFs were specific for time intervals, 79% of TF pairs and even 97% of promoter-enhancer modules showed specificity for single time intervals of the human stem cells. Predominantly SP1 and E2F contributed to temporal specificity at promoters and the forkhead (FOX) family of TFs at enhancer regions. Altogether, we characterized three classes of TFs: with binding sites being enriched at the TSS (like SP1), depleted at the TSS (like FOX), and rather uniformly distributed.

Figure 1.

The Workflow. (A) Motifs (PWMs) for TF binding sites were collected from a database. (B) Upstream sequences were collected for each transcript. Promoters (± 100 bp of the TSS) and enhancers (defined by accumulation of binding motifs and phylogenetic conservation, 2000–10 000 bp upstream of the TSS) were selected. (C) Statistical and combinatorial analysis of TF binding sites of promoters and enhancers. (D) Characterization of single motifs with respect to their distributions in the observed sequences (0–10 000 bases upstream of the TSS) and network analysis. (E) Assembly of promoter–enhancer modules. A promoter–enhancer module consisted of a pair of TFs binding at the promoter and a pair of TFs binding at the enhancer. (F) Gene expression data was taken from microarray studies of the development of several mouse tissues and of the differentiation of human stem cells. A time series analysis was performed to identify genes being differentially expressed at distinct (developmental) time intervals and tissues/cell types. (G) Differentially expressed genes of each time interval were tested to be enriched of genes with predicted binding sites of single TFs, TF pairs and promoter–enhancer modules. Promoter–enhancer modules were used to predict differential expression of developmental time intervals.

Figure 2.

Tissue and temporal specificity for each regulatory element during differentiation of human stem cells. The number of tissues versus the number of time intervals is plotted for (A) TFs, (B) pairs of co-occurring TFs in promoter regions, (C) pairs of co-occurring TFs in enhancer regions and (D) promoter–enhancer modules. The percentage of the different regulatory elements is indicated at each entry in the grid (i.e. 9% of TFs are specific for a single tissue and a single time interval). The total number of regulatory elements is given in brackets.

Figure 3.

Frequency distribution of the number of time intervals for human stem cell differentiation. The histograms show the frequency distribution of the number of time intervals for (A) TFs, (B) pairs of co-occurring TFs in promoter regions, (C) pairs of co-occurring TFs in enhancer regions, and (D) promoter–enhancer modules.

Figure 4.

Distribution of binding sites for different groups of TFs. For different groups of TFs the distribution of binding sites with respect to the TSS is shown exemplarily for the TFs SP1, FOXA1, and TP53: (A) The distribution for SP1 which represents the distribution of binding sites for TFs preferentially binding at the TSS (TSS-enriched-BS), (B) the distribution of FOXA1 which represents the distribution of binding sites for TFs with a depletion of binding sites at the TSS (TSS-depleted-BS), and (C) the distribution for TP53 representing uniformly distributed binding sites (uniformly distributed-BS).

Figure 5.

Networks of TF pairs (human). The network of pairs of co-occurring TFs are shown for (A) the promoter regions and (B) the enhancer regions. (C) A network of TFs mapped onto a PPI network (42). TFs showing preferential binding around the TSS (TSS-enriched-BS) are marked in red, TFs with a depletion of binding sites around the TSS (TSS-depleted-BS) in blue (dark) and TFs showing no preferential binding (uniformly distributed-BS) in green (light).