Mechanism of recurrent DNA break clusters (RDCs) generation in mouse and human cells

Abstract

Our lab utilized the high-throughput, genome-wide, translocation sequencing (HTGTS) to identify recurrent DNA double-stranded breaks (DSBs) cluster (RDC) containing genes (RDC genes) in the genome of mouse neural stem and progenitor cells (NSPCs) and, in recent studies of mouse neural progenitor cells (NPCs) derived from differentiation of mouse embryonic stem cells (ES cells) into NPCs in culture (ESC-NPCs). Many but not all RDC genes were detected upon mild, aphidicolin (APH)-induced ("ectopic") replication stress of NSPCs or ESC-NPCs and mapped to very long, late replicating, transcribed neural genes that were mostly associated with specific roles in synapse function and/or neural cell adhesion. Most RDC genes have been associated in mice, humans, or both with neuropsychiatric disorders and cancer. Mechanisms of RDC formation had remained unknown, but we and others hypothesized that frequent DSBs in RDC genes are generated by head-on collisions between transcription and, under replication stress, extending DNA replication forks. This general mechanism had been proposed as an underlying mechanism for the generation of genomic common fragile sites (CFSs), that have been implicated in genomic instability associated with cancer. In this context, a subset of the robust primary NSPC and ESC-NPCs RDC genes overlapped with certain genes that have copy number variations (CNVs) in mouse ES cells and with a substantial number of genes in human fibroblasts that are the location of CFSs. In addition, 70% of mouse RDCs genes are orthologs of RDC genes found in human NPCs derived from human-induced pluripotent stem cells. Recently, we have extended RDC studies to a human osteosarcoma cell line (U2OS) that harbors robust RDCs exclusively upon being subjected to ectopic replication stress. We find that a significant fraction of U2OS RDC genes overlap with mouse RDC genes and genes harboring CFSs in human fibroblasts and various cancer cell lines. These studies support our hypothesis that many RDC genes become detectable only in the presence of ectopic replication stress, and that the subset of RDCs that appear in mouse primary NSPCs in the absence of ectopic replication stress arise from unknown stressors potentially associated with the ex vivo culture or in vivo development. We have performed new experiments to elucidate the relationship between transcription, replication, and replication stress in RDC gene generation. In both mouse NPCs and human U2OS cells, we found that all RDCs genes were transcribed, but there was no direct relationship between RDC formation and RDC gene transcription levels. To confirm a role for transcription in RDC formation, in ES cells, we bi-allelically deleted transcriptional promoters of ectopic replication stress-dependent NPC RDC genes with widely varying transcription levels and then differentiated these promoter-deleted ES cells into NPCs. In all three RDC genes tested, abrogation of transcription via promoter deletion obviated RDC occurrence in APH-treated NPCs. This result unequivocally demonstrated that transcription or transcription-related processes are directly involved in RDC formation. We utilized Okazaki-Seq (OK-Seq) to measure the DNA replication profile of DNA sequences genome-wide, at high resolution, in U2OS cells, either in the absence or presence of APH ectopically induced replication stress. We identified a genomic region within RDC genes, which normally harbors the DNA replication fork termination zone (TZ), where converging replicating forks are terminated. This region is shifted or otherwise aberrant in RDC genes of U2OS cells exposed to ectopic replication stress. The aberrant TZ region of RDC genes correlated both with frequency of RDC breaks and also with a shift of their DSB junction pattern, visualized by ends captured by HTGTS shifting from one direction to the other. We made similar OK-Seq findings in studies of mouse NPCs subjected to ectopic replication stress. We hypothesize that during impaired replication fork termination, and potential de novo initiation in one direction or the other in the vicinity of the normal TZ region, that, transcription encounters de novo-initiated forks in head-on direction regardless of its orientation relative to normal replication fork direction within the gene. This finding suggests a new model to support the transcription-replication fork collision model for CFS and RDC gene formation in which transcription would encounter aberrant replication forks moving in both directions in the region of the termination zone. More broadly, our findings indicate that this general mechanism promotes generation of DNA broken ends within CFSs and RDC genes that can serve as substrates for genomic instability deletions, and translocations.