CpGs outside of TF motifs contribute to transcriptional activity of CGIs. (A) Mutation strategy. Cs in CpGs were mutated to As within 10-bp windows that showed small or no effect on activity when mutated. CpGs were mutated in random combinations within mutant promoter constructs. (B) CpGs outside of regions with a strong effect on transcriptional activity contribute to transcriptional activity. Scatter plot of normalized CpG density versus transcriptional activity relative to WT Pwp2 for promoter mutants. Normalized CpG density correlates significantly with transcriptional activity. The average Spearman's correlation coefficient for all three replicates ±1 SD) and its significance is indicated in the upper left part of the scatter plot. P-values were determined based on an approximate permutation test (see Methods). (C) Mutation strategy to generate an artificial sequence context and strategy for adding back CpGs. We first determined all 10-bp sequence blocks of the Pwp2 promoter with no effect or a weak effect on their activity when mutated and replaced them with random CpG-free sequences to retain correct spacing. Subsequently, different numbers of CpGs were re-introduced into the random CpG-free sequences at the same spatial locations as in the WT Pwp2 promoter. (D) Normalized CpG density itself contributes to CGI activity. Scatter plot of normalized CpG density versus transcriptional activity relative to the activity of WT Pwp2 for constructs in C. Normalized CpG density positively correlates with transcriptional activity. The average Spearman's correlation coefficient of all three replicates (±1 SD) and its significance is indicated in the upper left part of the scatter plot. P-values were calculated as in B.
