(A) Overlap between RLFS motif harboring intragenic regions and detected Pol2 stalling sites in B-lineage and ES cells. The high overlap of RLFS-positive regions is statistically significant compared to random regions (empirical P is indicated for 30% and 28% overlaps, respectively). (B) Overlap of detected Pol2 stalling sites also increases based on the strength of antisense signal level for B-lineage and ES cell convT regions divided into quartiles. (C) The influence of RLFS at TSS on ES cell DRIP-seq signal level is shown (Wilcoxon rank sum test P is indicated). Input signal levels are shown as control. (D) ES cell DRIP-seq signal is plotted similarly as in C, from convT-positive and -negative TSS regions. The DRIP-signal is higher in convT-positive TSS (Wilcoxon rank sum test P is indicated, TSS with convT N = 11774, TSS without convT N = 12092, refer to Figure 3—source data 2 for statistical analysis based on separate DRIP-seq replicates). (E) The percentages of breakpoint regions with no RSS motifs overlapping intragenic Pol2 stalling sites found in B-lineage cells are shown as barplots. The mean overlap observed in random sampling is indicated in grey bars (further statistical analysis is presented in Supplementary file 3). Categories with increasing cut-off for recurrence (1: non-recurrent in dim color, >1 and above: recurrent in darker color) were tested. (F) Overlap with RLFS, convT and annotated TSS is shown, as in E, for ETV6-RUNX1 NR-breakp (see also Supplementary file 3). (G) A schematic model illustrating how transcription from both strands (convT) or RLFS can locally arrest the Pol2 complex leading to recruitment of DNA damage-sensing complexes to R-loops, such as AID or BRCA (Alt et al., 2013, Hatchi et al., 2015), in an RSS-independent manner. (H) NR-breakp hotspot with the highest recurrence (TPI1 locus) is shown. DRIP-seq signal (shown in tones of red overlaid with input control signal in blue), and RLFS motifs indicated as a magenta bar track represent two levels of independent data that were integrated with GRO-seq data (signal from REH and ES cells is shown) to characterize properties of convT and Pol2 stalling regions. The breakpoint data (NR-breakp in brown) and detected convT (in purple) and Pol2 stalling in B-lineage cells (in blue) are shown. At the the recurrent breakpoint sites antisense transcription of neighboring gene (SPSB2 primary transcript) leads to a broad convT region, as indicated in the figure. Elevated DRIP-signal indicates formation of DNA-RNA hybrids (see also Figure 3—figure supplement 3).