Fig. 4. Promiscuity and stringency in transcriptional activation depends on recipient and RE GC compositions.

a The fraction of active promoters for different species show recipient promiscuity (P. aeruginosa > E. coli > B. subtilis) in proportion to genomic GC content. Transcription activity is defined in log scale units normalized by the activity of endogenous promoters, such as that a value of 0 indicates the average endogenous activity (Eq. (4)), n = 20,327, 28,608, and 24,473 for B. subtilis, E. coli, and P. aeruginosa, respectively. b The fraction of active promoters across a range of activity thresholds is shown for different promoter GC content. Low GC promoters show more promiscuous activity than high-GC promoters in our regulatory library. c Transcriptional profile as a function of residual activity (i.e., measured minus expected activity based on a regression model with the variables σ70 motif and mRNA 5’end stability) barely shows GC association, in contrast to (b). d Two examples of regulatory sequences showing expected σ70 motifs (spike plot) and experimentally identified TSS locations (red arrows). Low GC content promoters (right) are more likely than higher GC promoters (left) to contain a σ70-like motif. GC-rich recipients (e.g., P. aeruginosa) are more likely to contain σ70-like signals. e Density of number of TSSs per regulatory sequences is higher in GC-rich recipients. Low GC promoters are more likely to contain multiple TSS signals. b, c and low panel in e show results for E. coli. Data for B. subtilis and P. aeruginosa are shown in Supplementary Fig. 8.