Skip to main content
. 2017 Aug 9;6:e28440. doi: 10.7554/eLife.28440

Figure 4. The relationship between TF binding conservation and motif conservation depends on enhancer context.

Figure 4.

(a) Orthologous CRMs between D. melanogaster (D. mel) and D. virilis (D. vir) were divided into 3 categories based on their occupancy in D. mel; CRMs bound by 1 TF (Singletons), 2 or 3 TFs (Low Bound) or by 4 or 5 TFs (High Bound). CRMs in each category were divided into CRMs with conserved occupancy (circle) and non-conserved occupancy (triangle) in D. vir and the density of transcription factor binding sites (TFBS) for the corresponding factor in D. vir was calculated (Materials and methods). Asterisks depict the differences in TFBS density between CRMs with conserved and non-conserved binding, while horizontal lines indicate average densities across the genome for comparison. The lack of conserved binding for Bap singleton CRMs results in the missing value for conserved Bap TFBS (also in [b]). (b) Spearman correlation between interspecies variation in motif presence (ΔSTAP) and variation in TF occupancy (ΔChIP) at orthologous CRMs for each TF, in the different occupancy classes. A high Spearman correlation indicates inter-species changes in TF binding due to inter-species changes in the presence of the TF’s motif (i.e. binding changes are highly sequence-dependent). (c) Relationship between scaled ΔSTAP and ΔChIP for Singletons (left, 150 peaks) and High Bound (right, 488 peaks) orthologous enhancer pairs bound by Tin in D. melanogaster. Highlighted in red (0 and 28 peaks for Singletons and High Bound, respectively) are enhancer pairs with little changes in Tin occupancy (low ΔChIP) but high changes in Tin motifs (ΔSTAP) and in green (2 and 21 peaks for Singletons and High Bound, respectively) are enhancer pairs with high changes in Tin occupancy (high ΔChIP) with little changes in their Tin motifs (ΔSTAP). The former (red) are consistent with TF collective occupancy, while the latter (green) are consistent with Tin cooperative recruitment by other partner TFs. Dashed lines delineate the thresholds used. Note, many orthologous enhancers fulfill these criteria in the high-bound enhancers while very few or none do in the singletons class (p=0.06 and 0.0005 comparing (High Bound greater than Singletons) green and red populations (one sided Fisher test), respectively).

DOI: http://dx.doi.org/10.7554/eLife.28440.008