HOT regions function as patterned developmental enhancers and have a distinct cis-regulatory signature

Abstract

HOT (highly occupied target) regions bound by many transcription factors are considered to be one of the most intriguing findings of the recent modENCODE reports, yet their functions have remained unclear. We tested 108 Drosophila melanogaster HOT regions in transgenic embryos with site-specifically integrated transcriptional reporters. In contrast to prior expectations, we found 102 (94%) to be active enhancers during embryogenesis and to display diverse spatial and temporal patterns, reminiscent of expression patterns for important developmental genes. Remarkably, HOT regions strongly activate nearby genes and are required for endogenous gene expression, as we show using bacterial artificial chromosome (BAC) transgenesis. HOT enhancers have a distinct cis-regulatory signature with enriched sequence motifs for the global activators Vielfaltig, also known as Zelda, and Trithorax-like, also known as GAGA. This signature allows the prediction of HOT versus control regions from the DNA sequence alone.

Figure 1.

HOT regions act as enhancers with diverse activity patterns. (A) Transcriptional reporter to test enhancer activity of candidate regions (Pfeiffer et al. 2008). (B) A large majority of the 108 HOT regions (red) function as active transcriptional enhancers during Bownes stages 3–15 (1–13 h after fertilization), while much fewer of the 41 control regions are active (black). Shown are the positive rates (bar heights), the number of positive regions (numbers above bars), and the hypergeometric P-values. (C) HOT enhancers display diverse spatial patterns in the blastoderm embryo. Shown are seven representatives of manually grouped patterns that reoccurred at least three times and “other” patterns, the number of embryos in each group (each bottom right corner), and the corresponding percentage (see Supplemental Fig. S1 for all patterns).

Figure 2.

HOT enhancers regulate nearby genes. (A) Genes next to HOT regions are up-regulated during early Drosophila development. Shown are median RNA expression levels (reads per kilobase per million reads [RPKM]) (Graveley et al. 2011) for all genes (black line) and for genes assigned to regions bound by one to three factors (COLD; blue), four to 10 factors (WARM; orange), and >10 factors (HOT; red). (B) The HOT enhancer ∼10 kb upstream of Blimp-1 recapitulates three of the four stripes of the Blimp-1 expression pattern (University of California at Santa Cruz Genome Browser screenshot shows chr3L: 5,600,000–5,652,000, including published ChIP-on-chip and ChIP-seq [ChIP coupled with deep sequencing] profiles) (MacArthur et al. 2009; The modENCODE Consortium 2010). (C) The top right embryo shows the in situ hybridization against the Blimp-1 transcript from BDGP (Tomancak et al. 2002), and the bottom right embryo highlights the HOT enhancers' activity by in situ hybridization to the GAL4-reporter. (D) HOT enhancer is required for correct Blimp-1 expression. A BAC construct with an ∼45-kb region surrounding the endogenous Blimp-1 locus in which the coding sequence was replaced by GAL4 and integrated in the fly genome. The top embryo shows in situ hybridization to the wild-type GAL4-reporter, which fully reproduces the endogenous Blimp-1 expression pattern. The bottom embryo shows the same BAC after we deleted the HOT region, leading to a lack of the first, second, and third Blimp-1 stripes.

Figure 3.

Many transcriptional activators appear neutral with respect to HOT enhancer activity. Shown are the expression patterns of the transcriptional activator Twist (A) and the repressor Kruppel (B), surrounded by blastoderm stage embryos for representative HOT enhancers bound by Twist (A, middle) or Kruppel (B, middle) (see the text for details; see Supplemental Fig. S2 for other factors). (*) For repressors such as Kr, “Match” means a fully complementary pattern and “Overlap” means a partially complementary pattern.

Figure 4.

HOT regions are characterized by a unique cis-regulatory signature that is predictive. (A) Distribution of ZLD, GAGA, and ME133 motifs (motif count per 200-nucleotide bin) around the HOT and COLD regions aligned by their center. Lines represent means, and error bars represent standard deviations from three nonoverlapping subsets of the data (30% each). Motifs matching to ZLD and GAGA are strongly enriched in HOT compared with COLD regions, while the ME133 motif is depleted from HOT regions. (B) Heat map showing the most differentially distributed motifs (multiple testing-corrected P-value < 0.01) between HOT and COLD regions (first column), HOT regions and EEEs (second column), and their enrichments in HOT and COLD regions and EEEs compared with the genome average values (third through fifth columns). (C) HOT regions are dependent on ZLD and GAGA motifs. Shown are well-predicted HOT regions (score ≥75) that drop substantially (≥20) after in silico motif mutations or are robust (violet).