Figure 1.

Context-specific Twist binding sites display differential motif content. (A) Twist binding sites identified by ChIP-chip in D. melanogaster embryos (Zinzen et al. 2009) differ between developmental time points (2–4, 4–6, and 6–8 hpf). The ‘early,’ ‘late,’ and ‘continuous’ classes of binding sites are indicated and comprise sites to which Twist binds exclusively at 2–4 (early), 6–8 hpf (late), or at all three time points (cont.). (B) The genes close to the Twist binding sites also differ between time points. (C) Genes close to the early and late binding sites are enriched in different Gene Ontology (GO) categories, as expected for respective developmental stages. Shown are hypergeometric P-values (see color legend) for selected categories that are significant for at least one time point (see Supplemental Table S2 for all categories and the GO identifiers). (D) Distribution of the Twist (TWI), Tramtrack (TTK), and ME119 motifs according to their distances to the summits of all Twi ChIP-chip peaks of the ‘early’ (purple) and ‘late’ (green) time points (shown are motif counts for each 200-bp bin averaged across all ChIP-chip peaks for each time point). While the distribution of TWI motifs does not differ between early and late binding sites, the TTK motifs are enriched and the ME119 motifs are depleted in early sites. (E) Motifs are differentially enriched between the early and late binding sites, even when corrected for different overall sequence composition using shuffled controls motifs. (Left column) Shown are fold-enrichment values for the motifs that are differentially distributed between the early and late sites (P-value ≤ 0.05; motifs from Stark et al. 2007). Interestingly, the differential motif distribution between the early and late sites stems from consistent motif enrichment and depletion with respect to the average intergenic genome (right columns), as expected for motifs that help define TF binding sites within a large genome.