Investigating the role of RNA structures in transcriptional pausing using in vitro assays and in silico analyses

Abstract

Transcriptional pausing occurs across the bacterial genome but the importance of this mechanism is still poorly understood. Only few pauses were observed during the previous decades, leaving an important gap in understanding transcription mechanisms. Using the well-known Escherichia coli hisL and trpL pause sites as models, we describe here the relation of pause sites with upstream RNA structures suspected to stabilize pausing. We find that the transcription factor NusA influences the pause half-life at leuL, pheL and thrL pause sites. Using a mutagenesis approach, we observe that transcriptional pausing is affected in all tested pause sites, suggesting that the upstream RNA sequence is important for transcriptional pausing. Compensatory mutations assessing the presence of RNA hairpins did not yield clear conclusions, indicating that complex RNA structures or transcriptional features may be playing a role in pausing. Moreover, using a bioinformatic approach, we explored the relation between a DNA consensus sequence important for pausing and putative hairpins among thousands of pause sites in E. coli. We identified 2125 sites presenting hairpin-dependent transcriptional pausing without consensus sequence, suggesting that this mechanism is widespread across E. coli. This study paves the way to understand the role of RNA structures in transcriptional pausing.

Keywords: Escherichia coli; RNA polymerase; Transcription; hisL; pause.

Figure 1.

Transcriptional pausing at the hisL and trpL pause sites. (A) The hisL pause site is located in the hisL leader peptide of the hisLGDCBHAFI operon. The red rectangle represents the region of the sequence in which is found the pause site within the operon. The stop codon of the leader peptide is shown in red. The −1 position refers to the pause site. The dashed line represents the genomic segment used for in vitro transcription assays in this study. (B) Schematic of a cotranscriptionally folded RNA hairpin located upstream of the DNA-RNA hybrid region. The hairpin is shown interacting closely with exit channel of the RNA polymerase (RNAP, pale red). NusA (teal) can establish a bridge between the hairpin loop and the RNAP, resulting in an extended pause half-life. The consensus sequence G₋₁₀Y₋₁G₊₁ is presented in hollowed black residues within the hybrid region. The different mutant sets used in the study are shown and are colour coded. Double mutants (hisLS3, hisLS6 and hisLS8) are shown with colours corresponding to single mutants. (C) Representative transcription kinetics of the hisL pause. The full-length (FL, 77 nt) and the paused transcripts (PT, 54 nt) are shown. Only the paused transcripts are shown when using NusA and for the tested mutants. (D) Relative half-lives of hisL pause and the different mutants used. The mutants are depicted by their respective numbers (e.g. 1 for hisLS1). The colours are related to the panel B and hatched bars represent combinations of the corresponding mutation sets. (E) Relative half-lives of trpL pause and the different mutants used. The nomenclature used is the same as the one described in panel D. The predicted structure of the hairpin and the different mutants used are shown on the right. The compensatory mutant trpLS3 is shown with colours corresponding to single mutants trpLS1 and trpLS2.

Figure 2.

Study of pause sites leuL, pheL, thrL1, thrL2, pheM and pyrL. (A) Predicted structures of hairpins formed for the different pause regions. The mutants used in transcription assays are shown in each case. Additional mutations were performed in the DNA consensus sequence for the leuL pause. (B) Relative half-lives for the different pauses. The mutants are colour coded using the same nomenclature as shown in panel A. Hatched bars represent the combinations of the corresponding mutants. No combination of mutants was performed for thrL1 and thrL2 since the destabilization of the 5’ side of the hairpin (thrL1S1 and thrL2S1 mutants) did not significantly decrease the half-life of their respective pause.

Figure 3.

Study of the rfaQ and rnpB pause sites. (A and B) Relative pause half-lives of rfaQ (A) and rfaQ (B) pauses and the different mutants tested. The predicted structures of the respective hairpins are shown on the right. The different mutations used are shown on the structure. The mutants are depicted by their respective numbers (e.g. 1 for rfaQS1 or rnpBS1).

Figure 4.

The three NET-Seq datasets are enriched for different characteristics. (A) Pauses overlap between datasets. The size of the circles is proportional to the number of identified pauses. The number of pauses identified in more than one study is indicated in the shared regions. The legend of the Control group is shown only for the colour scheme used in the other panels. (B) Frequency of all consensus variants among the three datasets. Numbers over the bars represent the fold change relative to the Control group. The residue Y in the consensus sequence denotes a pyrimidine. The datasets are identified using the same colour scheme as in panel A. (C) Distribution of predicted structures energy in all three sets, sorted in four energy bins (Unstructured: 0 to −1 kcal/mol, Weak: −1 to −3 kcal/mol, Stable: −3 to −6 kcal/mol, Hyperstable: −6 kcal/mol to lower energies). Numbers over the bars represent the fold change relative to the Control group. The datasets are identified using the same colour scheme as in panel A. Red and green pauses represent the presence or the absence of a NusA effect, respectively. (D) Distribution of consensus sequence for pausing among energy bins in the datasets. The datasets are identified using the same colour scheme as in panel A.