Comprehensive transcription terminator atlas for Bacillus subtilis

Abstract

The transcriptome-wide contributions of Rho-dependent and intrinsic (Rho-independent) transcription termination mechanisms in bacteria are unclear. By sequencing released transcripts in a wild-type strain and strains containing deficiencies in NusA, NusG and/or Rho (10 strains), we produced an atlas of terminators for the model Gram-positive bacterium Bacillus subtilis. We found that NusA and NusG stimulate 77% and 19% of all intrinsic terminators, respectively, and that both proteins participate in Rho-dependent termination. We also show that Rho stimulates termination at 10% of the intrinsic terminators in vivo. We recapitulated Rho-stimulated intrinsic termination at 5 terminators in vitro and found that Rho requires the KOW domain of NusG to stimulate this process at one of these terminators. Computational analyses of our atlas using RNAstructure, MEME suite and DiffLogo, combined with in vitro transcription experiments, revealed that Rho stimulates intrinsic terminators with weak hairpins and/or U-rich tracts by remodelling the RNA upstream of the intrinsic terminator to prevent the formation of RNA structures that could otherwise compete with the terminator hairpin. We also identified 56 putative examples of 'hybrid Rho-dependent termination', wherein classical Rho-dependent termination occurs after readthrough of a Rho-stimulated intrinsic terminator.

Extended Data Figure 1.. NusA depletion and benchmarking terminators.

a. Western blot for all samples used for Term-seq was performed once. Top panel, image after probing for NusA. Bottom panel, image after probing for σ^A as a loading control. Lanes: 1, purified NusA_6His (top Panel) or σ^A (bottom panel); 2, PLBS730 +IPTG (WT); 3, PLBS730 −IPTG (nusA_dep); 4, PLBS731 + IPTG (ΔnusG); 5, PLBS731 −IPTG (nusA_dep ΔnusG); 6, PLBS890 +IPTG (Δrho); 7, PLBS890 −IPTG (nusA_dep Δrho); 8, PLBS891 +IPTG (ΔnusG Δrho); 9, PLBS891 −IPTG (nusA_dep ΔnusG Δrho). b. Venn diagram showing the number and overlap of intrinsic terminators identified in PLBS727 (WT in this study) and those identified previously in PLBS730 +IPTG (WT in previous study). c-f. RNA-seq coverage data from WT and Δrho strains across the previously identified classical Rho-dependent terminators for ylaL (c), spoVB (d), rapD (e) and slrA (f). Arrows at the bottom indicate the direction of transcription. Top tracks are the 3’ ends identified by Term-seq across each region.

Extended Data Figure 2.. A-rich tracts and Rho-stimulated intrinsic termination in vivo.

a. A-rich tract motifs generated for the FI, SA, SG, and SR intrinsic terminator subpopulations. A hierarchical clustering analysis can be found at the top of the motifs. b. IGV screenshot of the region upstream of the tbcS terminator. Top track is the 3’ end identified by Term-seq. Bottom tracks are the RNA-seq coverage data for the WT and Δrho strains. Arrow at the bottom indicates the direction of transcription. c. IGV screenshot of RNA-seq and ribosome profiling data across the transcript containing two sfp pseudogenes. Arrows at the bottom indicate the direction of transcription.

Extended Data Figure 3.. Verification of intrinsic terminator function in vitro.

a. Single-round in vitro termination assay with the tbcS intrinsic terminator. Experiments were performed with templates containing the WT terminator or those in which the DNA corresponding to the U-rich tract (ΔU) or the U-rich tract and the terminator hairpin (ΔU + HP) was deleted. Experiments were performed in the absence (–) or presence of Rho (R). b. Single-round in vitro termination assay with the yybG intrinsic terminator. Experiments were performed with WT or ΔU + HP templates in the absence (–) or presence of Rho (R), NusA (A) and/or NusG (G). c. Single-round in vitro termination assay with the bstB intrinsic terminator. Experiments were performed with WT, ΔHP or ΔU + HP templates in the absence (–) or presence of Rho (R). These qualitative intrinsic terminator validation experiments were performed once.

Extended Data Figure 4.. Rho-stimulated intrinsic terminators with antiterminators.

a. IGV screenshot of a genomic window centered around the yybG intrinsic terminator. Top track is the 3’ end identified by Term-seq. Bottom tracks are the RNA-seq coverage data for the WT and Δrho strains. %T in each strain is shown on the right of each track. Arrow at the bottom depicts the direction of transcription. b. Model of the yybG intrinsic terminator. c. Single-round in vitro termination assay with the yybG intrinsic terminator. Experiments were performed in the absence (–) or presence of Rho (R). Positions of terminated (Term) and full-length (FL) transcripts are marked. %T ± standard deviation is shown below each lane. Loss of termination in vitro when the terminator was deleted established that this is an authentic intrinsic terminator (Extended Data Fig. 3b). d-f. same as for panels a-c, except that it is the bstB (yuaE) intrinsic terminator. Loss of termination in vitro when the terminator was deleted established that this is an authentic intrinsic terminator (Extended Data Fig. 3c). g. RNA sequence of the ribD, sfp, tbcS, yybG and bstB leader terminators (inverted arrows). Red sequences can participate in the formation of alternative antiterminator (AT) structures. In vitro transcription experiments were performed 3 times. Values are averages ± standard deviation.

Extended Data Figure 5.. Sequencing data from each strain and Rho ATPase assay.

a. Principal component analysis (PCA) plot of transcriptomics data collected from each Term-seq replicate. b. Bar graph showing the nmol of P_i that was released by Rho in the presence of a polyC transcript and in the absence (–) or presence of BCM. ATPase assays were performed 3 times. Values are averages ± standard deviation.

Figure 1.. In silico profile of classical Rho-dependent terminators.

a. Flow chart of the Term-seq method. POT, point of termination. b. Violin and box plots showing the %T distribution of all classical Rho-dependent terminators in the wild-type (AGR), nusA_dep (GR), ΔnusG (AR), Δrho (AG), ΔKOW (ANGNR), ΔKOW Δrho (ANGN), nusA_dep Δrho (G), nusA_dep ΔnusG (R), ΔnusG Δrho (A), and nusA_dep ΔnusG Δrho (–) strains. Designations in the figure correspond to factors present in the strain. Box, first to last quartiles; whiskers, 1.5× interquartile range; center line, median; points, outliers; violin, density distribution. A two-tailed Wilcoxon rank-sum test was conducted to compare each of these datasets in a pair-wise fashion. These p-values are in Supplementary Table 3. Source data is in Supplementary Table 1. c. IGV screenshot downstream of the yrbD classical Rho-dependent terminator. Top track, 3’ ends identified by Term-seq. Bottom tracks, RNA-seq coverage data for the WT (AGR), nusA_dep (GR), Δrho (AG), and nusA_dep Δrho (G) strains. Similar to b, designations correspond to factors present in the strain. Arrow indicates the direction of transcription. d. Violin and box plots illustrating the distribution of C > G bubble density in the 250 nt upstream (up) and downstream (down) of 3’ ends attributed to classical Rho-dependent terminators, and a control set of 250 nt sequences obtained at random from the genome (con). Box, first to last quartiles; whiskers, 1.5× interquartile range; center line, median; points, outliers; violin, density distribution. A two-tailed Wilcoxon rank-sum test was conducted to compare each of these datasets in a pair-wise fashion. These p-values are in Supplementary Table 3. Source data is in Supplementary Table 1. e. Same as for panel d, except focusing on the number of YC dimers in each of these regions. All Term-seq experiments were performed in triplicate.

Figure 2.. In silico profile of intrinsic terminators.

a. Violin and box plots showing the %T distribution of all intrinsic terminators in the wild-type (AGR), nusA_dep (GR), ΔnusG (AR), Δrho (AG), ΔKOW (ANGNR), ΔKOW Δrho (ANGN), nusA_dep Δrho (G), nusA_dep ΔnusG (R), ΔnusG Δrho (A), and nusA_dep ΔnusG Δrho (–) strains. The designations correspond to the factors present in the strain. Box, first to last quartiles; whiskers, 1.5× interquartile range; center line, median; points, outliers; violin, density distribution. A two-tailed Wilcoxon rank-sum test was conducted to compare each of these datasets in a pair-wise fashion. These p-values are in Supplementary Table 3. Source data is in Supplementary Table 2. B. Same as for a, except focusing on the putative Rho-stimulated intrinsic terminators. c. Violin and box plots showing the distribution of predicted hairpin strength as reported in ΔG (kcal/mol) for the factor-independent (FI), NusA-stimulated (SA), NusG-stimulated (SG), and Rho-stimulated (SR) intrinsic terminators. Box, first to last quartiles; whiskers, 1.5× interquartile range; center line, median; points, outliers; violin, distribution of density. A two-tailed Wilcoxon rank-sum test was conducted to compare each of these datasets in a pair-wise fashion. These p-values are in Supplementary Table 3. Source data is in Supplementary Table 2. d. U-rich tract motifs generated for intrinsic terminator subpopulations specified in panel c. A hierarchical clustering analysis can be found to the right of the motifs. Source data is in Supplementary Table 2. All Term-seq experiments were performed in triplicate.

Figure 3.. rut sites are encoded downstream of Rho-stimulated intrinsic terminators.

a. Violin and box plots illustrating the distribution of C > G bubble density in the 250 nt upstream (up) and downstream (down) of all intrinsic terminator 3’ ends, upstream (up) and downstream (down) of Rho-stimulated intrinsic terminator 3’ ends, or at random from the B. subtilis genome (con). Data pertaining to all intrinsic terminators (all IT) or those pertaining to only Rho-stimulated intrinsic terminators (Rho-stimulated IT) are marked, where IT is an abbreviation for intrinsic terminators. Box, first to last quartiles; whiskers, 1.5× interquartile range; center line, median; points, outliers; violin, density distribution. A two-tailed Wilcoxon rank-sum test was conducted to compare each of these datasets in a pair-wise fashion. These p-values are in Supplementary Table 3. Source data is in Supplementary Table 4. b. Same as for panel a, except focusing on the number of YC dimers in each of these regions. All Term-seq experiments were performed in triplicate.

Figure 4.. Rho cooperates with NusA and NusG to stimulate intrinsic termination.

a. IGV screenshot of a genomic window centered around the tbcS intrinsic terminator. Top track is the 3’ end identified by Term-seq. Bottom tracks are the RNA-seq coverage data for the nusA_dep ΔnusG Δrho (–), nusA_dep ΔnusG (R), ΔnusG Δrho (A), ΔnusG (AR), nusA_dep Δrho (G), nusA_dep (GR), ΔKOW Δrho (ANGN), ΔKOW (ANGNR), Δrho (AG), and WT (AGR) strains. The in vivo data is organized so that it reflects the elongation factors present in the corresponding in vitro transcription reactions in c. %T in each strain is shown on the right of each track. Arrow at the bottom indicates the direction of transcription. b. Single-round in vitro termination assay with the tbcS intrinsic terminator. Experiments were performed in the absence (–) or presence of NusA (A), NusG (G), NGN only NusG (NGN), and/or Rho (R) as indicated. Positions of terminated (Term) and full-length (FL) transcripts are marked. %T ± standard deviation is shown beside each lane. The direction of transcription is specified by an arrow at the bottom of the gel. c. Model of the tbcS intrinsic terminator. d. Same as panel c, except the experiments were performed in the absence (–) or presence of bicyclomycin (BCM), NusG (A), NusG (G), and/or Rho (R) as indicated. In vitro transcription experiments were performed 3 times. Values are averages ± standard deviation.

Figure 5.. Rho stimulates termination in a wide range of transcription contexts.

a. IGV screenshot of a genomic window centered around the intrinsic terminator in the ribD riboswitch. Top track is the 3’ end identified by Term-seq. Bottom tracks are the RNA-seq coverage for WT and Δrho strains. %T is shown on the right. Arrow shows the direction of transcription. b. Model of the ribD leader terminator. c. Single-round in vitro termination assay with the ribD leader terminator. Experiments were performed in the absence (–) or presence of Rho (R) and/or FMN. Positions of terminated (Term) and full-length (FL) transcripts are marked. %T is shown below each lane. d-e. Identical to panels a-b except it is the sfp intrinsic terminator. f. Single-round in vitro termination assay with the sfp intrinsic terminator. Experiments were performed in the absence (–) or presence of Rho (R) and/or BCM. ΔAT, antiterminator deletion; ΔT, terminator deletion. Positions of terminated (Term) and full-length (FL) transcripts are marked. Positions of classical Rho-dependent termination is shown. %T for the intrinsic terminator is shown below each lane. g. Single-round in vitro termination assay with the sfp intrinsic terminator. Experiments were performed in the absence (–) or presence of Rho (R) and/or NusA and NusG (AG). Positions of terminated (Term) and full-length (FL) transcripts are marked. h. Signal intensity was plotted across each lane in g, Left, intensity across lane – (red) compared to intensity across lane R (blue). Right, intensity across lane AG (red) compared to intensity across lane AGR (blue). Locations of classical Rho-dependent termination are specified by vertical black lines. i. Models of overlapping antiterminator-like hairpin and terminator hairpins for sfp. Overlapping regions are in red. In vitro transcription experiments were performed 3 times. Values are averages ± standard deviation.

Figure 6.. Models of Rho-stimulated intrinsic and hybrid Rho-dependent termination.

a. Left, model illustrates transcription at the intrinsic terminator in the absence of Rho. The non-template DNA (ntDNA) and template DNA strands are red and blue, respectively. NusA (grey) is shown with its NTD bound adjacent to RNA exit channel of RNAP (blue rectangle). NusG (green) is shown with its NGN domain contacting the ntDNA strand in the transcription bubble. The intrinsic terminator (black) and competing antiterminator structure (orange), which forms in the absence of Rho, are shown. Right, model illustrates transcription at the intrinsic terminator in the presence of Rho (purple). Rho has contacted RNA corresponding to the antiterminator, allowing the intrinsic terminator hairpin to form in the RNA exit channel. Rho is shown hydrolyzing ATP and the KOW domain is contacting Rho. b. Model of hybrid Rho-dependent termination. Rho binds to the rut site in nascent RNA (orange). The other RNA is black. Left to right. Transcription elongation complex at an intrinsic terminator. Rho utilizes the KOW domain and ATP hydrolysis to stimulate intrinsic termination, otherwise readthrough occurs. A rut site is encoded downstream of the intrinsic terminator. Once transcribed, Rho contacts the rut site and stimulates transcript release in an ATP hydrolysis-dependent fashion. Rho induces transcript release at several locations downstream of the intrinsic terminator. c. IGV screenshot of the sfp hybrid Rho-dependent terminator. Top, 3’ ends identified by Term-seq. Bottom, RNA-seq coverage data for WT and Δrho strains. Arrows indicate direction of transcription. The hybrid Rho-dependent terminator includes both Rho-stimulated intrinsic and classical Rho-dependent terminators (boxed).