Regulator trafficking on bacterial transcription units in vivo

Abstract

The trafficking patterns of the bacterial regulators of transcript elongation sigma(70), rho, NusA, and NusG on genes in vivo and the explanation for promoter-proximal peaks of RNA polymerase (RNAP) are unknown. Genome-wide, E. coli ChIP-chip revealed distinct association patterns of regulators as RNAP transcribes away from promoters (rho first, then NusA, then NusG). However, the interactions of elongating complexes with these regulators did not differ significantly among most transcription units. A modest variation of NusG signal among genes reflected increased NusG interaction as transcription progresses, rather than functional specialization of elongating complexes. Promoter-proximal RNAP peaks were offset from sigma(70) peaks in the direction of transcription and co-occurred with NusA and rho peaks, suggesting that the RNAP peaks reflected elongating, rather than initiating, complexes. However, inhibition of rho did not increase RNAP levels within genes downstream from the RNAP peaks, suggesting the peaks are caused by a mechanism other than rho-dependent attenuation.

Fig. 1. Bacterial regulators of transcript elongation

(A) Regulator trafficking during the transcription cycle. RNAP binds σ⁷⁰ to form holoenzyme which can specifically bind promoter DNA and initiate transcription. Once the nascent RNA has reached a certain length, RNAP releases its strong contacts to σ⁷⁰ and transitions into a more stable elongation complex (EC). ECs can be targeted by NusA, NusG, ρ, and σ⁷⁰ to modulate transcription. (B) ChIP-chip profiles of RNAP and regulators across the E. coli genome. Log₂(IP/input) ratios for σ⁷⁰ (orange) and RNAP (β’; blue) are shown above plot of E. coli genes (rightward and leftward transcription relative to origin separated above and below center). Regions identified as background RNAP interaction are shown as black bars below the RNAP profile (bkgd, see text). Genes encoding rRNA (blue) and tRNA (green) genes are indicated. An expanded region around 0.95 Mb is shown for RNAP, σ⁷⁰, NusA (red), NusG (green), and ρ (violet) with the locations of known (vertical lines with black horizontal arrows) or predicted (vertical lines with gray horizontal arrows) promoters indicated and the baseline set at the Tukey bi-weight mean (Supplemental Experimental Procedures). The middle shaded region shows an example of a region exhibiting low background signals. serS (bold) is one of the 109 high-quality TUs (Fig. 1D). (C) Histogram of RNAP (β’) log₂(IP/input) signals (blue) with overlaid histogram from background regions (black). A blowup of the highest signal region showing the point selected for Occ_app=1 (mean of top ten 3-probe clusters) is shown in an inset. (D) Locations of the 109 high-quality TUs selected for analysis. The numbers correspond to their position on the genetic map; gene names and map positions are listed in Table S1.

Fig. 2. Apparent occupancy profiles of RNAP and regulators on representative TUs

Occ_app for the rrnE TU and seven representative TUs from among the 109 TUs selected for the absence of interferring upstream or downstream signals (Fig. 1D and Table S1). Occ_app was calculated as described in the Supplemental Experimental Procedures using two rounds of sliding-window smoothing (500 bp window for RNAP, NusA, NusG, and ρ; 175 bp window for σ⁷⁰). Genes are depicted as labeled open arrows; promoters, as vertical lines capped with arrows; and known intrinsic terminators, as hairpins. Note that the scales of Occ_app and TU length (in kb, denoted by hatchmarks) differ in each panel. Protein-encoding genes are colored blue, and the rRNA TU is colored yellow. Regulators are colored as in Figure 1. Vertical dotted lines are the center of the σ⁷⁰ peak. For the rrn TU, there are two promoters (and two σ⁷⁰ peaks). (A) serS, a monocistronic TU encoding seryl-tRNA synthetase (B) rrnE, one of seven E. coli rRNA TUs. Due to near-sequence-identity among the rRNA TUs, these signals represent the average of all seven rRNA TUs. (C) atpIBEFHAGDC, the nine-gene TU encoding the F0,F1 ATP synthase. (D) rpsFpriBrpsRrplI, encoding the ribosomal protein S6, DNA replication primosome protein N, ribosomal protein S18, and ribosomal protein L9. (E) gltBDF, encoding glutamate synthase large and small subunits and a periplasmic protein involved in nitrogen metabolism. (F) acnB, a monocistronic TU encoding aconitase B. (G) cyoABCDE, encoding cytochrome bo terminal oxidase and heme O synthase. (H) carAB, encoding carbamoyl phosphate synthetase.

Fig. 3. Mid-TU regulator signals correlated with RNAP signals

(A) Diagram illustrating calculation of mid-TU signals. For each of the 109 high-quality TUs, the log₂(IP/input) signals for all probes within a 200-bp window surrounding the center of the TU were averaged to yield an estimate signal due to elongating RNAP or regulator associated with the elongating RNAP. Regulators are colored according to Figures 1 and 2. (B) Correlation of σ⁷⁰ and RNAP mid-TU signals. Only TUs for which the mid-TU point was more than 500 bp from the σ⁷⁰ peak were included (to avoid influence of signal from the σ⁷⁰ peak; n=80); r=0.68; p<0.001. (C) Correlation of NusA and RNAP mid-TU signals (n=109); r=0.97, p<0.001. (D) Correlation of NusG and RNAP mid-TU signals (n=109); r=0.71, p<0.001 (E) Correlation of ρ and RNAP mid-TU signals (n=109); r=0.78, p=<0.001 (F) The correlation coefficient between the RNAP signal and each of the regulator signals plotted versus mean mid-TU signal for the regulator. Mean signals for NusA, NusG, σ⁷⁰, and ρ are 73%, 95%, 33%, and 51% of mean RNAP signals, respectively.

Fig. 4. Aggregate apparent occupancy for highly expressed TUs

(A) Aggregate normalized Occ_app for 42 highly transcribed TUs (curve A; Table S1) was calculated by averaging Occ_app from the TUs (2× rolling-averaged; 300-bp window). Occ_app for each TU was normalized as a fraction of the highest Occ_app in each TU prior to averaging. TUs were aligned to the peak of σ⁷⁰ Occ_app. σ⁷⁰ (orange), blue (RNAP). Curve A is the aggregate for all 42 TUs; lines B and C represent subsets of this aggregate: B, TUs lacking a promoter-proximal RNAP peak; C, 29 TUs exhibiting a promoter-proximal RNAP peak. Lines B and C are shown in panels B and C with the aggregate signals for the other regulators. (B) Aggregate normalized Occ_app for RNAP and regulators on the 13 TUs that lacked an obvious promoter-proximal RNAP peak (also shown as B for RNAP in panel A). Numbers in the figure correspond to the distance of the peak to the σ⁷⁰ peak. Colors as in Figure 1. (C) Aggregate normalized Occ_app for RNAP and regulators on the 29 TUs that exhibited an obvious promoter-proximal RNAP peak (also shown as C for RNAP in panel A). Numbers in the figure correspond to the distance of the peak to the σ⁷⁰ peak.

Fig. 5. Gene-averaged regulator/RNAP ratios

(A) Gene-averaged NusA/RNAP ratios computed using average NusA IP/input values for each gene divided by the average RNAP IP/input values and plotted as a function of RNAP log₂(IP/input) values (Table S3). Each gene is represented by one data point. (B) Gene-averaged ρ/RNAP ratios. (C) Gene-averaged NusG/RNAP ratios. Zoomed in region shows identity of genes with unusually high NusG/RNAP ratios. (D) Gene-averaged NusG/RNAP ratios plotted as a function of distance of gene from TSS using genes for which this distance could be assigned and for which the average RNAP log₂(IP/input) signal was greater than 0.1 (830 genes; Table S3). Zoomed in region shows identity of genes with unusually low NusG/RNAP ratios. (E) Gene-averaged NusG/RNAP ratios computed for the 30 functional classes of genes shown in the panel (Table S5) and plotted as a function of the average distance to the TSS for genes in each functional class (Supplemental Experimental Procedures). Red dot represents the genome average.

Fig. 6. Frequency of ρ and NusA co-occurrence for RNAP peaks associated with genes

(A) Diagram illustrating the identification of RNAP peaks associated with genes and the calculation of traveling ratio (TR; Reppas et al., 2006; Supplemental Experimental Procedures). Vertical dotted lines indicate σ⁷⁰ peak (orange) and RNAP peak (blue). (B) Fraction of promoter-proximal RNAP peaks for which NusA or ρ peaks exist within 300 bp of the RNAP peak, binned by the TR of the gene associated with the RNAP peak. Only RNAP peaks that could be associated with specific genes and only genes longer than 1 kb were included in this analysis. NusA peaks, red columns; ρ peaks, purple columns. (C) Fraction of promoter-proximal RNAP peaks for which NusA or ρ peaks exist within 300 bp, binned by expression level (Allen et al., 2003) of the gene associated with the RNAP peak. Only RNAP peaks that could be associated with specific genes and only genes longer than 500 bp were included in this analysis. (D) Fraction of promoter-proximal RNAP peaks for which NusA or ρ peaks exist within 300 bp, binned by height of RNAP peak. The same RNAP peaks as shown in Fig. 6C were included in this analysis. (E) Effect of ρ inhibition on the aggregate RNAP Occ_app profile for the 29 TUs with promoter-proximal RNAP peaks (Table S1; Fig. 4C). RNAP signal is for cells grown with (lavender) or without (blue) 20 μg bicyclomycin/ml to cause ρ inhibition. The σ⁷⁰ aggregate profile for cells grown without bicyclomycin treatment is shown for reference (light orange). (F) Histogram of TR calculated for the 293 genes shown in Fig. 6B with or without ρ inhibition. RNAP signal with no inhibition, blue; RNAP signal with ρ inhibition, lavender. (G) Occ_app on rho gene with (lavender) or without (blue) ρ inhibition. The effect of bicyclomycin is readily apparent in the readthough of the rho attenuator.

Fig. 7. Model of transcription regulator trafficking during initiation to elongation transition

As RNAP moves away from a promoter, contacts to upstream DNA are presumably lost upon the transition from abortive to productive synthesis (Revyakin et al., 2006). At least some of σ⁷⁰ contacts to RNAP must release during this transition. Release of upstream DNA contacts would free the α CTD and flap tip for interaction with NusA, and thus explain the early association of NusA. ρ appears to target RNA as it emerges from the RNAP exit channel and to bind without terminating transcription. As elongation progresses, NusG may slowly displace σ⁷⁰ from interaction with the clamp helices, and ribosome binding could occlude ρ interaction with RNAP.