Pause sequences facilitate entry into long-lived paused states by reducing RNA polymerase transcription rates

Abstract

Transcription by RNA polymerase (RNAP) is interspersed with sequence-dependent pausing. The processes through which paused states are accessed and stabilized occur at spatiotemporal scales beyond the resolution of previous methods, and are poorly understood. Here, we combine high-resolution optical trapping with improved data analysis methods to investigate the formation of paused states at enhanced temporal resolution. We find that pause sites reduce the forward transcription rate of nearly all RNAP molecules, rather than just affecting the subset of molecules that enter long-lived pauses. We propose that the reduced rates at pause sites allow time for the elongation complex to undergo conformational changes required to enter long-lived pauses. We also find that backtracking occurs stepwise, with states backtracked by at most one base pair forming quickly, and further backtracking occurring slowly. Finally, we find that nascent RNA structures act as modulators that either enhance or attenuate pausing, depending on the sequence context.

Fig. 1

Single molecule transcription assay and data analysis. a Experimental geometry. Biotinylated E. coli RNAP halted on the template DNA was tethered through a neutravidin bridge to a biotinylated 1.5 kb DNA ligated to the oligo-coated bead. The selection of which end of the template DNA to ligate to the other beads enables selection between assisting force (illustrated in panel a) and opposing force geometry. The major pause sites (‘his’, ‘a’, ‘b’, ‘c’, ‘d’) are indicated in the sequence of the repeat. b (left) representative traces obtained under assisting forces. Dashed magenta lines were added to highlight the locations of the ‘his’ pause site in the template. The mean residence time histogram shown (right) was calculated by averaging the time spent at each position in the repeat across all traces at all conditions, except RNase data, which was aligned separately (203 traces). Pause sites are marked, and pause-free regions are shaded. For clarity, the mean of measurements up to the 95% percentile is shown to remove to effect of rare pauses occurring outside the pause sites. However, data analysis was performed on the full data set. c Total variation denoising and computation of pause site crossing times. The total variation denoising (red) of the raw data (blue) consists of flat segments separated by discrete jumps. One of these segments occurs in the vicinity of an expected pause site. 1 bp windows are drawn in the ±3 bp range surrounding the expected pause site; the window that took the longest to cross (solid black) is used to define the pause site crossing time for the crossing of this pause site. d Crossing time distributions for different sites measured at 25 pN assisting force. The complementary cumulative distribution function (fraction of events longer than a given crossing time, CCDF) is plotted. The gray shaded area marks the timescales accessible to previous experiments

Fig. 2

Pausing efficiencies are high and force-independent. a Description of the method used to calculate pausing efficiencies, illustrated for the ‘his’ pause at 10 pN assisting force. The reference distribution is rescaled to indicate the overlap between the two distributions below the cutoff time. For the distribution at the ‘his’ site, all events shorter than the cutoff are classified as non-paused (light green), as well as events longer than the cutoff in the same proportion as in the reference distribution (dark green). The remaining events (red) are classified as paused. b Pausing efficiencies at the major pause sites at different forces, calculated using extrapolation of the crossing time distributions above 1 s towards faster times (red), and by our nonparametric method (blue). Error bars indicate 25–75 percentiles for for 100 bootstrapped sets. c Residence time histogram ratios. The ratios of the residence times at three forces to the residence times at 25 pN assisting force are plotted. The ratio for 25 pN assisting force, which equals 1 by definition, is plotted as a reference

Fig. 3

Backtracking dynamics. a Analysis of backtracking events. The trace shown contains a backtracking event occurring at site ‘b’. The backtrack depth, pre-backtrack time and backtrack duration are shown. b Histogram of backtracking events by position (blue). Gray zones in the figure indicate the ± 3 bp region surrounding each major pause site. c Backtracking depths and times measured at opposing forces for site ‘b’. d Histogram of pre-backtrack times measured at site ‘b’ at opposing forces

Fig. 4

Effect on 0.87 µM GreB on transcription dynamics measured at 10 pN opposing force. Top: transcription data in the presence and absence of GreB with backtracking events at site ‘b’. Bottom: effect of GreB on the crossing time distributions at the major pause sites. At short time scales, comprising 80–90% of the measured events, GreB slightly increases crossing times. Therefore mean residence times are longer in the presence of GreB at pause sites (Supplementary Fig. 4b). GreB reduces the crossing times at sites ‘a’ and ‘b’ for the longest events, indicating that these are backtracked events

Fig. 5

Effect of RNase on pausing dynamics. a Residence time histograms collected at 7 pN opposing force with and without RNase. b Effect of RNase on pausing efficiencies. Error bars indicate 25–75 percentiles for 100 bootstrapped sets. c Effect of RNase on residence time distributions at sites ‘P2’, ‘d’, and ‘his’

Fig. 6

Proposed model for transcriptional pausing by E. coli RNAP. TEC: Transcription Elongation Complex; the indices n−1, n, and n + 1 indicate the length of the RNA product. At pause sites, the paused state is rendered kinetically accessible to the polymerase through a slowing down of the on-pathway forward translocation rate. Depending on the sequence context, this paused state can transition slowly to a ≥ 2 bp backtracked state (in sites ‘a’ and ‘b’), be stabilized or destabilized by the nascent RNA (‘his’, ‘d’, and ‘P2’) or be stabilized by other mechanisms (such as in site ‘c’, which exhibited neither backtracking nor RNase sensitivity)