Double-Strand Break Repair Pathways

A double-strand break occurs when both phosphodiester backbones of the DNA duplex are cleaved at nearby or identical positions, physically separating a chromosome into two fragments. DSBs arise from ionizing radiation, reactive oxygen species, collapsed replication forks encountering a nicked template, and programmed cellular processes such as V(D)J recombination, meiotic crossing-over, and CRISPR-Cas9 cleavage. Unlike a mismatch or a bulky adduct, which can be repaired by reference to the intact opposite strand, a DSB destroys the local information entirely on both strands, so any high-fidelity repair pathway must import genetic information from elsewhere.

Non-homologous end joining (NHEJ) solves this by not requiring a template at all: it recognizes the two broken ends, trims them minimally if needed, and ligates them directly. Because it uses little or no homology, NHEJ can occur throughout the cell cycle, including G1, where no sister chromatid exists. The tradeoff is fidelity — small insertions or deletions (indels) of a few nucleotides are common at the junction, and if the two ends being joined come from different chromosome breaks, NHEJ can produce translocations.

Homology-directed repair (HDR), of which homologous recombination (HR) is the dominant and most accurate form, instead uses an intact homologous DNA molecule — almost always the newly replicated sister chromatid — as a copying template. The broken end is resected into single-stranded DNA, that ssDNA searches for and invades the homologous duplex, and DNA polymerase extends the invading strand using the sister as template, restoring the original sequence exactly. Because a sister chromatid is only available after S phase, HR is restricted to S and G2. A related but distinct pathway, microhomology-mediated end joining (MMEJ), also called alternative end joining, uses short (5–25 bp) microhomologies exposed by limited resection and is inherently mutagenic, deleting the intervening sequence; it serves as a backup when both NHEJ and HR are compromised. The central regulatory event that sorts a broken end into one route or another is DNA end resection — the nucleolytic conversion of blunt or minimally processed ends into long 3′ single-stranded overhangs, which commits the break to HR (or MMEJ) and forecloses classical NHEJ.

Step 1 — Break detection and the choice of first responder. Within seconds of a DSB, the MRN complex (MRE11–RAD50–NBS1 in mammals; MRX in yeast) binds the broken ends and recruits and activates the kinase ATM, which phosphorylates histone H2AX on serine 139 (producing γH2AX) over megabase-scale chromatin domains flanking the break. This creates a visible, quantifiable focus used experimentally as a DSB marker. Independently, the Ku70/Ku80 heterodimer, a ring-shaped protein that threads onto the free DNA end with very high affinity (sub-nanomolar, essentially irreversible without processing), competes for the same ends. Which protein "wins" the end in the first minutes strongly biases pathway choice.

Step 2a — The NHEJ route (if resection is blocked). Ku70/80 rings encircle each broken end and recruit the catalytic subunit DNA-PKcs, forming DNA-PK holoenzyme. Two DNA-PK-bound ends synapse (bridge together), and DNA-PKcs autophosphorylation allows the ends to be processed: nucleases (Artemis, activated by DNA-PKcs) trim overhangs, polymerases (Pol μ, Pol λ) fill in gaps, and the XRCC4–DNA Ligase IV complex, stabilized by XLF (Cernunnos) and PAXX, seals the nick. Because ends are often incompatible (non-complementary overhangs, damaged termini), some nucleotides are typically lost or added, generating small indels — the hallmark scar of NHEJ. The entire process can complete within 30 minutes and requires no sequence homology, so it operates in G0, G1, S, and G2.

Step 2b — The HR route (if resection proceeds). CtIP, activated by cyclin-dependent kinase (CDK) phosphorylation (hence restricted to S/G2), cooperates with MRN's endonuclease activity to make an initial internal nick, from which MRE11's 3′→5′ exonuclease resects a short stretch back toward the break, while long-range resection in the 5′→3′ direction is carried out by EXO1 or by BLM helicase paired with DNA2 nuclease. The result is a long 3′-overhanging single strand, often hundreds to thousands of nucleotides. Replication protein A (RPA) immediately coats this ssDNA, protecting it and removing secondary structure. BRCA2, delivered with partner protein PALB2, displaces RPA and loads RAD51 recombinase onto the ssDNA, forming a right-handed helical nucleoprotein filament — the presynaptic filament — that stretches and searches the genome.

Step 3 — Homology search and strand invasion. The RAD51–ssDNA filament scans the genome (facilitated by chromatin proximity, since sister chromatids are held together by cohesin right next to the break) for a matching duplex sequence. Upon finding homology, RAD51 catalyzes strand invasion: the ssDNA base-pairs with its complement in the donor duplex, displacing the other donor strand as a D-loop (displacement loop).

Step 4 — DNA synthesis and resolution. A DNA polymerase (Pol δ in mammals) extends the invading 3′ end using the intact sister chromatid as template, copying the sequence that was lost at the break. Two sub-pathways diverge here:

• Double-strand break repair (DSBR) pathway: the second broken end is captured, forming a double Holliday junction, which is resolved by structure-specific nucleases (resolvases like GEN1, or the MUS81–EME1 complex) or dissolved by the BLM–TOPOIIIα–RMI1/2 (BTR) complex, which can produce crossover or non-crossover products.
• Synthesis-dependent strand annealing (SDSA): the invading strand is displaced after limited synthesis and anneals back to the other resected end, yielding exclusively non-crossover products. SDSA dominates in mitotic somatic cells, minimizing loss-of-heterozygosity events; crossover-generating resolution is more prominent in meiosis.

Pathway choice: the resection switch. The decisive fork in the road is whether resection is permitted. Two antagonistic protein modules compete at the break:

• BRCA1–CtIP promotes resection, favoring HR.
• 53BP1, together with RIF1 and the shieldin complex (SHLD1/2/3–REV7), binds the break and physically blocks resection, favoring NHEJ.

CDK activity, high in S/G2 and low in G1, phosphorylates CtIP and other resection factors, so resection — and thus HR — is only licensed after DNA replication, precisely when a sister chromatid template exists. In G1, 53BP1-shieldin dominates unopposed, and Ku/DNA-PK/Ligase IV completes rapid NHEJ. This cell-cycle-gated switch is why HR functions as a high-fidelity, replication-coupled repair system while NHEJ is the constitutive, always-available backup.

Identifying NHEJ: radiosensitive rodent mutants and biochemical reconstitution. In the 1980s, Penny Jeggo and colleagues isolated Chinese hamster ovary (CHO) cell mutants that were hypersensitive to ionizing radiation and specifically defective in rejoining DSBs, screening for cells that failed to survive X-ray doses that wild-type cells tolerated easily. These mutants fell into distinct complementation groups (later named XRCC4, XRCC5=Ku80, XRCC6=Ku70, XRCC7=DNA-PKcs), each defined by a gene whose product, when reintroduced, restored radioresistance and DSB rejoining. Independently, the same genes were found to be essential for V(D)J recombination in developing lymphocytes: mouse scid (severe combined immunodeficiency) mutants, discovered by Melvin and Gayle Bosma in 1983, could not complete V(D)J joining and were later shown to carry a mutation in DNA-PKcs, directly linking the NHEJ machinery to programmed immune-system DSB repair. Biochemical reconstitution experiments in vitro (Lieber, Lindahl, Jackson labs, 1990s–2000s) then demonstrated that purified Ku70/80, DNA-PKcs, and XRCC4-Ligase IV are sufficient to join two DNA ends with compatible or minimally processed overhangs, confirming the pathway's minimal core machinery.

Identifying HR: yeast epistasis groups and physical detection of recombination intermediates. The existence of a homology-based repair pathway was first inferred genetically in Saccharomyces cerevisiae. Radiation-sensitive yeast mutants fell into the RAD52 epistasis group (including RAD51, RAD52, RAD54, RAD55, RAD57), distinct from the RAD3 (nucleotide excision repair) and RAD6 groups, established through classic epistasis analysis: double mutants combining two genes from the same group showed no greater sensitivity than either single mutant, indicating they act in the same pathway, whereas combining genes from different groups produced synergistic (multiplicative) sensitivity, indicating independent pathways. This work, largely from the labs of Robert Mortimer, Lawrence Grossman, and later Rodney Rothstein, mapped out HR as a genetically separable repair system well before its biochemistry was understood. The mechanistic centerpiece — RAD51's biochemical activity — was nailed down when Stephen Kowalczykowski and colleagues (and independently Patrick Sung) purified RAD51 protein in the 1990s and showed in vitro that it forms helical nucleoprotein filaments on single-stranded DNA and catalyzes ATP-dependent strand invasion into homologous duplex DNA, reconstituting D-loop formation with purified components — a direct biochemical demonstration of the search-and-invade mechanism inferred from genetics. Robin Holliday's 1964 model of a crossed-strand recombination intermediate (the Holliday junction) was validated decades later when such four-way junctions were directly visualized by electron microscopy in recombining DNA and their resolution by junction-resolving enzymes (e.g., RuvC in bacteria, GEN1 in humans) was demonstrated biochemically, confirming the physical reality of the intermediate proposed purely from genetic recombination data.

Establishing pathway choice: BRCA1 versus 53BP1. The regulatory antagonism between BRCA1 and 53BP1 was established through a striking genetic epistasis experiment. BRCA1-deficient mouse cells are HR-defective, genomically unstable, and embryonic lethal; several independent groups (Bunting et al., 2010, in Andre Nussenzweig's lab, and follow-up work by the Jackson and Durocher labs) showed that deleting 53BP1 in a BRCA1-null background substantially restores HR proficiency, genomic stability, and viability, even though 53BP1 and BRCA1 have no direct enzymatic relationship. The interpretation: BRCA1's essential job at a DSB is to overcome 53BP1's resection block; remove the blocker, and resection (and thus HR) can proceed even without BRCA1's normal resection-promoting activity. This "epistatic rescue" experiment, combined with later work identifying RIF1 and the shieldin complex (SHLD1/2/3–REV7) as the effectors downstream of 53BP1 that physically occlude resection nucleases, established the resection switch as the mechanistic core of pathway choice — and explained why BRCA1/2-mutant tumor cells are HR-deficient and can be selectively targeted by PARP inhibitors.

Example 1 — Radiosensitivity complementation cross in CHO cells. Two independently isolated CHO mutant lines, XR-1 and irs2, are both hypersensitive to X-rays. You fuse the two cell lines to generate hybrid cells (somatic cell fusion) and score X-ray survival.

• If the hybrid is radioresistant (rescued), the two mutations complement each other, implying they lie in different genes (different complementation groups).
• If the hybrid remains radiosensitive, the mutations fail to complement, implying they are allelic — mutations in the same gene.

Historically, XR-1 defined the XRCC4 complementation group and irs2 mapped to a separate group; fusing them yielded radioresistant hybrids, confirming XRCC4 and the irs2 gene product act in distinct steps of the same overall NHEJ pathway. This logic — complementation testing by cell fusion, analogous to complementation testing by genetic cross in Drosophila or yeast — is exactly how each new NHEJ component (Ku70, Ku80, DNA-PKcs, XRCC4, Ligase IV, XLF) was assigned to a distinct gene before any of the proteins had been biochemically characterized.

Example 2 — Reasoning a CRISPR editing outcome from cell-cycle phase. A lab delivers Cas9 and a single-guide RNA targeting a specific exon into a population of asynchronously dividing human cells, alongside a single-stranded DNA donor template carrying a precise point mutation flanked by 40-bp homology arms, intended to introduce that mutation via HDR. Reasoning through the expected outcome:

• At any instant, roughly 45–50% of an asynchronous cycling population sits in G1, where 53BP1-shieldin dominates and resection is blocked; Cas9 cuts here will be repaired almost exclusively by NHEJ, producing indels (frameshift knockouts) rather than the intended edit.
• Only cells in S/G2 (roughly 35–45% of the population, cell-type dependent) have active CtIP/resection machinery and a nearby sister chromatid, licensing HDR.
• Even within S/G2 cells, HDR using an exogenous ssDNA donor must compete kinetically with the much faster NHEJ machinery (Ku70/80 binds ends with sub-nanomolar affinity within seconds), and NHEJ often "wins" the race even when HDR is biochemically permitted.

Consequently, unmodified experiments typically yield low precise-editing efficiencies (often single-digit to ~20%) alongside a majority of indel alleles. This is why real editing protocols synchronize cells in S/G2 (e.g., nocodazole or double-thymidine block), transiently inhibit DNA Ligase IV or Ku70/80 (e.g., using small-molecule NHEJ inhibitors such as SCR7, or a "CtIP fusion to Cas9" strategy), or use donor designs and cell-cycle-restricted Cas9 fusions to bias the resection-switch decision toward HR — a direct practical application of the resection-based pathway choice mechanism described above.

Stephen C. Kowalczykowski's biochemical reconstitution of RAD51-mediated strand invasion, alongside the earlier genetic mapping of the RAD52 epistasis group, exemplifies how DSB repair research moved from pure genetics to mechanistic biochemistry. Equally central is the discovery lineage of NHEJ, anchored in Penny Jeggo's radiosensitive CHO mutant screens and the identification of the scid mouse mutation in DNA-PKcs, which unexpectedly linked DSB repair machinery to the adaptive immune system's antibody diversification process.

Cancer therapy and synthetic lethality. Tumors carrying BRCA1 or BRCA2 mutations are HR-deficient and depend heavily on base excision repair and PARP1 to manage single-strand breaks; PARP inhibitors (olaparib, niraparib, rucaparib) trap PARP1 on damaged DNA, converting single-strand breaks into DSBs during replication that these tumor cells cannot repair by HR, producing selective tumor-cell killing (synthetic lethality) while sparing HR-proficient normal tissue.

Radiotherapy and radiosensitization. Because DNA-PKcs and ATM drive DSB repair after ionizing radiation, small-molecule inhibitors of these kinases are in clinical trials as radiosensitizers, aiming to block tumor cells from repairing radiation-induced DSBs.

Genome engineering. CRISPR-Cas9 and related nucleases generate site-specific DSBs; whether the outcome is a gene knockout (via NHEJ-generated indels) or a precise sequence correction (via HDR with a donor template) depends directly on which repair pathway resolves the cut, motivating strategies (cell-cycle synchronization, NHEJ inhibition, fusion of resection-promoting domains to Cas9) engineered around the biology described above.

Immunology. NHEJ components are indispensable for V(D)J recombination, the process that generates the enormous diversity of antibody and T-cell receptor genes; inherited NHEJ defects cause severe combined immunodeficiency (SCID) in humans, directly paralleling the mouse scid phenotype.

Agricultural and conservation biotechnology. Precision gene editing in crops and endangered species relies on the same HR/NHEJ balance; engineers bias editing outcomes toward HDR to introduce specific beneficial alleles (e.g., disease resistance) rather than simple gene disruption.

Ask the tutor: How exactly does the RAD51 filament find its one matching sequence among the entire genome during homology search?