A Genome-Wide Analysis of Poised Promoters in Bacteria

Abstract

As the first and usually rate-limiting step of transcription initiation, bacterial RNA polymerase binds to double stranded DNA (the closed complex formation) and subsequently opens the two strands of DNA (the open complex formation). Poised promoters in bacteria are sequences where RNAP binds with high binding affinity, but which do not have detectable levels of transcription initiation due to too slow transition from closed to open complex. Existence of a considerable number of poised promoters in genome has been often hypothesized, but poised promoters have not been systematically studied, since a large scale analysis of promoter kinetics is not experimentally feasible. To computationally address promoter poising on a genome-wide scale we use a recently developed biophysical model of transcription initiation [1]. We show that promoter poising is significantly reduced by i) Existence of −35 box interactions ii) Binding specificities of (physically independent) RNAP domains that interact with −10 box single-stranded and double-stranded DNA. We show that the later (dominant) effect is not due to generic properties of protein-DNA interactions, and argue that RNAP is designed to reduce promoter poising in genome. However, despite this reduction, we obtain that the number of poised promoters is still significant, and corresponds to ∼30% of strongly bound sequences in bacteria [2]. This number roughly matches with lower bound of reported false positives in RNAP ChIP-chip experiments, which suggests that poised promoters are a major contributor to false positives in searches of bacterial promoters. [1] M Djordjevic and R Bundschuh, Biophysical Journal 94: 4233 (2008).[2] M Djordjevic, submitted (2009).