Translocating RNA polymerase generates R-loops at DNA double-strand breaks without any additional factors

Abstract The R-loops forming around DNA double-strand breaks (DSBs) within actively transcribed genes play a critical role in the DSB repair process. However, the mechanisms underlying R-loop formation at DSBs remain poorly understood, with diverse proposed models involving protein factors associated with RNA polymerase (RNAP) loading, pausing/backtracking or preexisting transcript RNA invasion. In this single-molecule study using Escherichia coli RNAP, we discovered that transcribing RNAP alone acts as a highly effective DSB sensor, responsible for generation of R-loops upon encountering downstream DSBs, without requiring any additional factors. The R-loop formation efficiency is greatly influenced by DNA end structures, ranging here from 2.8% to 73%, and notably higher on sticky ends with 3′ or 5′ single-stranded overhangs compared to blunt ends without any overhangs. The R-loops extend unidirectionally upstream from the DSB sites and can reach the transcription start site, interfering with ongoing-round transcription. Furthermore, the extended R-loops can persist and maintain their structures, effectively preventing the efficient initiation of subsequent transcription rounds. Our results are consistent with the bubble extension model rather than the 5′-end invasion model or the middle insertion model. These discoveries provide valuable insights into the initiation of DSB repair on transcription templates across bacteria, archaea and eukaryotes.


Contents
: DNA oligomers for construction of transcription templates Table S2: Single-molecule assay data in Figure 2A Table S3: Single-molecule assay data in Figure 2B Table S4: Information on the number of molecules and replicated experiments analyzed

Figure S12. Extension of RNA-DNA hybrids.
Using the template DT7 with a 5-nt 3'-overhang end, the FRET changes were timed relative to the antibody detection of R-loops in the three different dye-labeling schemes: scheme I (top), scheme II (middle) and scheme III (bottom), as described Fig. 4A.The template and nontemplate strands were each constructed by ligation of two or three pertinent oligonucleotides using a splint DNA.The T7A1 promoter sequence is underlined in the nontemplate strands.
The red-colored T denotes Cy5-labeled thymine.
The green-colored T in the nontemplate strand oligomer for DT3 in scheme II denotes Cy3labeled thymine.

Figure S1 :
Figure S1: FRET changes after transcription resumption Figure S2: Antibody binding to R-loops Figure S3: Additional representative time traces for Figure 1 Figure S4: Gradual increase of high-FRET fraction after antibody binding Figure S5: R-loop degradation by RNase H Figure S6: Effect of antibody on R-loop efficiency Figure S7: Cy3-RNA runoff time in the absence of R-loop formation Figure S8: FRET histograms at the time point of antibody binding to R-loops Figure S9: Additional representative time traces for Figure 3 Figure S10: FRET changes in the template DT3's scheme II Figure S11: Histograms of time difference between antibody binding and FRET ascending Figure S12: Extension of RNA-DNA hybrids Figure S13: Additional representative time traces for Figure 4 Figure S14: Additional representative time traces for Figure 5 TableS1: DNA oligomers for construction of transcription templates TableS2: Single-molecule assay data in Figure2ATableS3: Single-molecule assay data in Figure2BTableS4: Information on the number of molecules and replicated experiments analyzed

Figure S1 .
Figure S1.FRET changes after transcription resumption.The elongation complexes of DT1 were examined before NTP injection (gray bars) and 12 s after NTP injection (red bars).A Gaussian function fitted to the gray bars (black line) peaks at EFRET = 0.69.This value represents the most frequently observed FRET efficiency before NTP injection.

Figure S2 .
Figure S2.Antibody binding to R-loops.(A) Antibody-R-loop association time.R-loops were generated through 30-min multiple-round transcription of the immobilized stalled elongation complexes, and the time of Alexa-488 signal appearance since the antibody injection was measured.A single-exponential decay curve (red line) revealed a decay time constant of 6.6 s. (B) Comparison of Alexa-488 images before (left) and after (right) the antibody binding to Rloops.The antibody was injected to the immobilized stalled elongation complexes of DT3 (left) or injected to the products after 30-min multiple-round transcription (right).The DNA templates are visualized using the Cy5 signal (open circles), and some of them exhibit binding to the antibody (black dots).(C) The average time delays from NTP injection to the first antibody binding on DT1, DT2 and DT3.(D) Average time traces of Alexa-488 fluorescence of the high-FRET complexes.For the display, the first stable antibody binding was post-synchronized at time zero with fluorescence signals normalized to one.As expected from the sizes of S9.6 antibody and fully-extended R-loop, two or three copies of the antibody complex could bind Rloops at maximum.

Figure S3 .
Figure S3.Additional representative time traces for Figure 1.Additional time traces for Rloop, runoff and retention events are presented.Each time trace displays Cy3 and Cy5 fluorescence at Cy3 excitation (top), Cy3-Cy5 FRET (middle), and Alexa-488 fluorescence at its excitation (bottom).

Figure S4 .
Figure S4.Gradual increase of the high-FRET fraction after antibody binding.(A) The Cy3-Cy5 FRET efficiencies were measured at the time point of the first antibody binding to DT1. (B) A high-FRET fraction increases over time since the antibody binding at 0 s.The fraction with EFRET > 0.7 on the y-axis is plotted against the antibody binding time on the x-axis.The error bars represent the standard deviations of three independent experiments.

Figure
Figure S5.R-loop degradation by RNase H. (A) Shown are the representative time traces obtained from the experiments in which RNase H was added some time after NTP injection (top), from the control experiments where RNase H was not added (middle)or from the experiment of RNase H injection on stalled elongation complex (bottom).In the traces, Cy3 and Cy5 fluorescence at Cy3 excitation and the corresponding FRET are colored in green, red and black, respectively.(B) The change of a high FRET population over time.The fraction with EFRET > 0.7 on the y-axis is plotted against the time since the RNase H injection on the x-axis.The error bars represent the standard deviations of three independent experiments.

Figure S6 .
Figure S6.Effect of antibody on R-loop efficiency.The high FRET population with EFRET > 0.7 was negligibly affected by the presence of antibody.

Figure S7 .
Figure S7.Cy3-RNA runoff time in the absence of R-loop formation.The Cy3-RNA retention duration was measured in the complexes that did not bind the R-loop-specific antibody during a 10-min period.Some complexes retained Cy3-RNA beyond 575 s, and they are represented by a single blue bar at the 575-s timepoint.The remaining complexes, which showed RNA runoff from templates DT1, DT2, DT3 and DT4, were analyzed.The data points are fitted to a gamma distribution (red line) using the equation y(x) =  1   ()  −1  − .The runoff times (n/r) were then calculated to be 15 s, 17 s, 15 s and 16 s for templates DT1, DT2, DT3 and DT4, respectively.

Figure S8 .
Figure S8.FRET histograms at the time point of the first antibody binding to R-loops.The Cy3-Cy5 FRET efficiencies were measured at the time point of the initial antibody binding to Rloops formed on template DT3 with a blunt end (left), template DT7 with a 5-nt 3'-overhang (center) and the template with a 5-nt 5'-overhang end (right).

Figure S9 .
Figure S9.Additional representative time traces for Figure 3.Each time trace displays Cy3 and Cy5 fluorescence at Cy3 excitation (top), Cy3-Cy5 FRET (middle), and Alexa-488 fluorescence at its excitation (bottom).In scheme II, a transient PIFE occurs due to the brief encounter of transcribing RNAP with fluorophores.

Figure S10 .
Figure S10.FRET changes in the template DT3's scheme II.The Cy3-Cy5 FRET efficiencies were measured before the transcription resumption by NTP injection (gray bars) and after the R-loop formation (red bars) in the template DT3's dye-labeling scheme II.The Gaussian functions (red lines) peak at EFRET = 0.21 and 0.64, respectively.

Figure S11 .
Figure S11.Histograms of time difference between antibody binding and FRET ascending.The average times when EFRET rises above 0.7 after antibody binding are 105 s and 31 s for DT1 and DT2, respectively.