How Cells Choose Between DNA Repair Pathways—and Why It Matters for Cancer

Every day, each cell in the human body sustains tens of thousands of lesions in its DNA, and among the most dangerous of these are double-strand breaks, the complete severing of both strands of the double helix. A new review published in Nature Reviews Molecular Cell Biology by Michelle Swift, Cody Rogers, Hardeep Kaur, Dipanjan Chowdhury, Patrick Sung and colleagues synthesizes decades of work on one of the central questions in genome maintenance: when a chromosome snaps, how does the cell decide whether to mend it by homologous recombination, a high-fidelity process that uses an intact sister chromatid as a template, or by non-homologous end joining, a faster but error-prone strategy that simply glues the broken ends back together? The answer, the authors argue, hinges on a tightly choreographed molecular contest over the broken DNA ends themselves, and losing that contest can have profound consequences for cancer development and treatment.

The pivotal event in this decision is DNA end resection, the enzymatic trimming of the 5-prime strands at the break to generate single-stranded DNA tails. If resection proceeds, the cell is effectively committed to homologous recombination, because the single-stranded overhangs become coated with the RPA protein and later with RAD51, which searches for homologous sequence on the sister chromatid. If resection is blocked, the ends remain available for the Ku70-Ku80 heterodimer, which recruits the DNA-dependent protein kinase catalytic subunit and the ligase complex that carries out classical non-homologous end joining. Resection is therefore not merely a step in one repair route; it is the molecular switch that determines which route is taken, and its regulation is correspondingly elaborate.

Resection unfolds in two phases. Short-range processing is initiated by the MRE11-RAD50-NBS1 complex, known as MRN, whose nuclease activity is stimulated by the phosphorylated cofactor CtIP. MRN makes an internal incision near the break and removes the blocking Ku protein, a step that single-molecule imaging studies have illuminated in striking detail. Long-range resection then extends the single-stranded tracts over thousands of nucleotides through two principal routes: one driven by the nuclease EXO1, and the other by the helicase-nuclease pair BLM or WRN working with DNA2. These pathways, first defined in budding yeast through the Sgs1-Dna2 and Exo1 systems, are conserved in human cells, where the RecQ helicases cooperate with DNA2 to process breaks and stalled replication forks alike.

Standing athwart this machinery is a famous antagonism between two tumor-suppressive and genome-protective forces: the BRCA1-BARD1 complex and the 53BP1 axis. 53BP1 is recruited to breaks by marks laid down by the RNF8 and RNF168 ubiquitin ligases, and it recognizes dimethylated histone H4 lysine 20 together with ubiquitinated H2A lysine 15 on damaged nucleosomes. Once bound, 53BP1 recruits RIF1 and the shieldin complex, which protect DNA ends and actively counteract resection, in part through CST-Pol-alpha-primase-mediated fill-in synthesis that restores paired ends. Bunting and colleagues showed in 2010 that loss of 53BP1 restores homologous recombination in BRCA1-deficient cells, establishing that these two pathways act in direct opposition at the same breaks.

Recent biochemical work has sharpened the picture of how BRCA1-BARD1 wins this contest in the appropriate cellular context. Two 2024 studies in Nature demonstrated that BRCA1-BARD1 directly stimulates EXO1-dependent and BLM-DNA2 or WRN-DNA2 long-range resection, while also protecting stalled replication forks from excessive nuclease degradation. The BARD1 subunit reads the H2AK15 ubiquitin mark and unmethylated H4K20 characteristic of newly replicated chromatin, which explains why the complex is preferentially recruited to breaks in the S and G2 phases of the cell cycle, when a sister chromatid is available as a template. BRCA1 also promotes the removal of 53BP1 from damage sites and accelerates CtIP-mediated resection, tipping the balance decisively toward homologous recombination precisely when the template-based pathway is feasible and safest.

The review also emphasizes an emerging layer of regulation involving RNA. Transient RNA-DNA hybrids, or R-loops, form at double-strand breaks and are required for efficient repair, yet their persistence is dangerous: accumulated hybrids interfere with resection and homologous recombination and can promote chromosomal translocations. Helicases such as senataxin, DDX5, DHX9 and the DEAD-box protein DDX1, together with RNase H enzymes recruited by BRCA2, resolve these structures at the right time. Remarkably, RNA transcripts can also stimulate repair directly: the discovery of DR-loops, three-stranded intermediates containing both DNA-DNA and RNA-DNA pairing, showed that RNA can help guide the homology search, and recent work suggests RNA transcripts may even serve as templates for repair in human cells, particularly in post-mitotic neurons where sister chromatids are unavailable.

Why does all this matter for medicine? The clearest example is PARP inhibitor therapy, built on the synthetic lethality between BRCA1 or BRCA2 deficiency and inhibition of poly(ADP-ribose) polymerase, a concept demonstrated in landmark 2005 studies and now a mainstay of treatment for BRCA-mutated breast, ovarian, pancreatic and prostate cancers. Tumors frequently escape by restoring the homologous recombination pathway without fixing the original BRCA mutation. Loss of 53BP1, or of its effectors such as RIF1, shieldin components, REV7 or the CST complex, releases the brake on resection and rescues BRCA1-deficient cells from PARP inhibitor sensitivity. Longitudinal profiling of patients has revealed that reversion mutations in BRCA1 and BRCA2 can co-occur with alterations in TP53BP1, RIF1 and PAXIP1, revealing polyclonal and mechanistically diverse routes to drug resistance within the same patient.

The pathway-choice framework also points to new therapeutic vulnerabilities. Homologous-recombination-deficient tumors depend on polymerase theta, an enzyme that mediates an alternative end-joining route, and first-in-class Pol-theta inhibitors have now entered development as potential synthetic lethal partners alongside PARP inhibitors. Conversely, understanding how the CST complex, through its subunits CTC1, STN1 and TEN1, directly blocks both EXO1 and BLM-DNA2 resection suggests that nuclease inhibition is a second, independent mechanism by which cells protect DNA ends, one that operates at telomeres and at replication-coupled breaks as well as at damage-induced double-strand breaks. Each of these nodes is a potential drug target or biomarker in its own right.

What emerges from this synthesis is a picture of repair pathway choice as a dynamic, multi-layered decision governed by cell-cycle position, chromatin state, the availability of a homologous template, the enzymatic machinery assembled at the break, and even the RNA molecules transcribed nearby. The authors caution that despite concerted efforts by many laboratories, the field is only beginning to appreciate the mechanisms that underpin the choice between homologous recombination and end joining. As those mechanisms come into focus, they are increasingly legible in the clinic: in the genetic signatures of PARP inhibitor resistance, in biomarkers such as 53BP1 expression that predict treatment response, and in the design of the next generation of therapies that exploit the repair weaknesses cancer cells cannot hide.

Subject of Research: Mechanisms governing the choice between homologous recombination and non-homologous end joining in DNA double-strand break repair and their relevance to cancer and therapeutic resistance.

Article Title: Mechanisms and disease relevance of DNA break repair pathway choice

Article References: Swift, M. L., Rogers, C. M., Kaur, H., Chowdhury, D., & Sung, P. (2026). Mechanisms and disease relevance of DNA break repair pathway choice. Nature Reviews Molecular Cell Biology. https://doi.org/10.1038/s41580-026-01026-3

Image Credits: AI Generated

DOI: 10.1038/s41580-026-01026-3

Keywords: DNA double-strand breaks, homologous recombination, non-homologous end joining, DNA end resection, BRCA1-BARD1, 53BP1, shieldin, PARP inhibitors, R-loops, genome stability, synthetic lethality, drug resistance