Release of paused RNA-Polymerase II at specific introns and chromatin domains favors spontaneous DNA double strand break formation and predicts cancer translocations [RNA-seq]

Transcriptional activation leads to transient accumulation of DNA Double Strand Breaks (DSBs), which might promote formation of chromosomal translocations. We report here the genomic distribution of DSBs in non-transformed mammary cells grown under unperturbed conditions, using genome-wide Breaks Labeling In Situ and Sequencing (BLISS). We found thousands of high-confidence DSBs, which were enriched at the promoters of a subset of moderately- to highly-transcribed genes (fragile promoters) and co-localized with Topoisomerase II beta. Analyses of DSB-predictive features identified gene length and paused RNA Polymerase II, but not transcription, as critical factors (84% prediction accuracy). Analyses of DSB signaling and repair factors showed high levels of XRCC4, but not of gammaH2AX and NBS1, and no RAD51. Finally, the observed DSBs were predictive of a significant fraction of the recurrent translocations found in human breast cancers. These data suggest that, in normal cells, basal transcription from a specific class of promoters entails accumulation of TOP2B, leading to unresolved DSBs and increased probability of chromosomal translocations. Overall design: To test the efficiency of BLISS in capturing genomic DSBs, we first studied induced DSBs, and identified the genomic AsiSI sites digested upon tamoxifen (4-OHT)-induced nuclear translocation of the AsiSI restriction endonuclease fused with the estrogen receptor-binding domain (AsiSIER). At this purpose, we performed ChIPseq experiments using a panel of antibodies against various DSB sensing and repair factors (i.e., NBS1, gammaH2AX, XRCC4 and RAD51), before and after AsiSI induction, and investigated the BLISS-positivity of the AsiSI sites showing increased ChIP signals of DSB sensing and repair factors, following 4-OHT administration. Since many DSB ends sequenced by BLISS were spread all along the genome (in addition to the AsiSI sites), even in the absence of 4-OHT treatment, we identified these endogenous DSB clusters by developing a bioinformatic pipeline that identifies genomic regions showing the concomitant presence of: i) a significant enrichment of BLISS signals over the total and local background at any given genomic site, and ii) =2 fold-enrichment of ChIPseq signals of at least one of three repair factors (NBS1, XRCC4, RAD51) within ±1 kb from the center of the BLISS signal, as compared to the input DNA (Tier1 DSBs). Finally, ChIP assays against Topoisomerase II beta and phosphorylated RNA Polymerase II were used to investigate the mechanisms underlying the generation of endogenous DSBs.