Shout-out to @ydjeghmoum.bsky.social who single-handedly led this beautiful work during her PhD! Thanks @erc.europa.eu and the FRM for funding this work.

Finally, transcription specifically suppressed non-allelic recombination, while it authorized allelic repair. Consequently, transcription prevent chromosomal rearrangements between highly transcribed paralogous genes, and compensating the detrimental impact of transcription in generating DNA lesions

D-loop suppression by transcription was independent of, and even more potent than conserved recombination regulators known to promote genome maintenance and suppress tumor formation in humans (such as BLM-TOP3-RMI and FANCM homologs).

Transcription orientation mattered: co-linear orientation suppressed D-loops more efficiently than head-on orientation. This suggests the preferential usage of one DSB end for repair in highly-transcribed genes.

By putting two donors in competition and transcribing only one of them, we could show that transcription suppressed D-loops in cis! In fact, it even redirected D-loops from the transcribed to the non-transcribed donor.

This effect of transcription could be induced and repressed in minutes, suggesting it reflected a direct effect of RNA PolII passage. Indeed, secondary effect of transcription (RNA, RNA:DNA hybrids, nuclear relocalization, endogenous TFs, …) were not involved in D-loop suppression.

We used a highly efficient DNA break induction system in budding yeast, and quantified D-loops (the earliest DNA joint molecule formed upon homology identification) formed at a donor whose transcriptional level could be manipulated. Transcription caused a massive drop of D-loops.