Molecular Biology
Homologous Recombination Repair
Error-free double-strand break repair — RAD51, BRCA1/BRCA2, D-loop and Holliday junctions
Homologous recombination repair (HRR) is the error-free pathway that mends DNA double-strand breaks by copying an intact homologous template — almost always the sister chromatid. After a break, the MRN complex resects the ends, RPA coats the exposed 3' single-stranded tails, and BRCA2 loads the RAD51 recombinase into a filament that invades the sister duplex, forming a displacement loop (D-loop) that DNA polymerase extends to restore the lost sequence exactly. Because it needs a sister chromatid, HRR operates only in S and G2 phase, and its accuracy is what distinguishes it from error-prone non-homologous end joining. The mechanistic model was built by Robin Holliday (1964) and refined by the Szostak double-strand-break-repair model (1983); Mary-Claire King mapped BRCA1 to chromosome 17q21 in 1990. When HRR fails — as in BRCA1/BRCA2-mutant cancers — cells become lethally dependent on PARP, the basis for olaparib, the first FDA-approved PARP inhibitor in 2014.
- RepairsDNA double-strand breaks
- AccuracyError-free (vs NHEJ)
- Active phaseS and G2 only
- RecombinaseRAD51 (RecA homolog)
- LoadersBRCA1 · PALB2 · BRCA2
- DrugOlaparib — PARP inhibitor 2014
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
Why homologous recombination repair matters
- It fixes the deadliest lesion in the genome. A single unrepaired DNA double-strand break can kill a cell or, if misjoined, generate a chromosomal translocation that causes cancer. Human cells sustain an estimated 10 to 50 double-strand breaks per cell per day from replication stress, reactive oxygen species, and background radiation. HRR is the only pathway that repairs them with the original sequence intact.
- It is the one guardian of genome stability that is truly error-free. By copying an identical sister chromatid, HRR restores not just the break but the exact base sequence surrounding it — no deletions, no insertions, no scars. This fidelity is why cells route breaks into HRR whenever a sister chromatid is available.
- It underlies inherited cancer risk. Women carrying a pathogenic BRCA1 variant face a roughly 55–72% lifetime risk of breast cancer and 39–44% risk of ovarian cancer; BRCA2 carriers carry a 45–69% breast and 11–17% ovarian risk. Both genes encode HRR proteins — their loss cripples error-free repair and drives genomic instability.
- It created the first genomically targeted cancer drugs. Because HRR-deficient tumor cells cannot compensate when PARP is inhibited, olaparib, niraparib, rucaparib, and talazoparib selectively kill BRCA-mutant cancers while sparing normal, HRR-proficient tissue — the clinical debut of synthetic lethality as a drug strategy.
- It generates genetic diversity in meiosis. The same machinery, redirected by the SPO11 nuclease and DMC1 recombinase to break and invade the homologous chromosome rather than the sister, produces the crossovers that shuffle parental alleles into every egg and sperm. HRR is thus both a repair pathway and the engine of sexual inheritance.
- It restarts stalled and broken replication forks. Beyond frank breaks, HRR proteins protect and restart forks that stall at obstacles. BRCA2 and RAD51 shield nascent DNA from MRE11 degradation at stalled forks — a fork-protection role distinct from break repair, and a second reason their loss is so destabilizing.
- It is a CRISPR editing bottleneck. Precise gene knock-in with CRISPR-Cas9 relies on HRR to copy a supplied donor template into a Cas9-induced break. Because HRR only works in S/G2 and competes with NHEJ, homology-directed knock-in efficiency is often just a few percent — the central practical limit on precise genome editing.
How homologous recombination repair works
HRR proceeds through an ordered series of DNA intermediates, each catalyzed by a dedicated set of proteins. Detection and resection. A double-strand break is first sensed by the MRN complex (MRE11–RAD50–NBS1), which recruits and activates the ATM kinase to phosphorylate histone H2AX into γH2AX, spreading a repair signal over megabases of flanking chromatin. MRE11, an endonuclease, nicks the 5'-terminated strand internally; then MRE11's 3'→5' exonuclease chews back toward the break while long-range resection nucleases — EXO1 and the DNA2 helicase-nuclease acting with BLM — extend the resection outward, generating 3' single-stranded overhangs thousands of nucleotides long. CtIP, in complex with MRN and stimulated by CDK phosphorylation, licenses this resection and is the key commitment step that diverts the break away from NHEJ.
Filament assembly. The exposed single-stranded DNA is instantly bound by RPA (replication protein A), which removes secondary structure but must be displaced for repair to proceed. BRCA1, positioned at the break with PALB2, recruits BRCA2, the dedicated RAD51 mediator. BRCA2's eight BRC repeats and its DNA-binding domain hand RAD51 monomers onto the ssDNA, evicting RPA and nucleating a right-handed helical presynaptic filament. RAD51 paralog complexes (BCDX2: RAD51B–RAD51C–RAD51D–XRCC2, and CX3: RAD51C–XRCC3) stabilize the growing filament, which stretches the DNA about 1.5-fold and rotates the bases outward to present them for homology reading.
Homology search and strand invasion. The presynaptic filament samples duplex DNA throughout the nucleus, testing for complementarity. On finding the identical sister sequence, RAD51 catalyzes strand invasion: the 3' single-stranded tail base-pairs with its complement in the donor duplex and displaces the other donor strand as a displacement loop (D-loop). DNA polymerase δ, loaded via PCNA, extends the invading 3' end using the sister as template, resynthesizing the sequence lost at the break.
Resolution. Three outcomes are possible. In synthesis-dependent strand annealing (SDSA) — the dominant somatic route — the extended invading strand is unwound from the D-loop and anneals back to the resected second end of the original break; gap-filling and ligation finish the repair with no crossover. Alternatively, the second end is captured to form a double Holliday junction, a pair of four-way DNA crossovers. This can be dissolved by the BLM helicase with topoisomerase IIIα and RMI1/RMI2 to yield strictly non-crossover products, or resolved by structure-specific nucleases (GEN1, or SLX1–SLX4–MUS81–EME1) that cut the junctions to give crossover or non-crossover products. Ligation by DNA ligase I seals the final nicks, and the break is restored to its exact original sequence.
Common misconceptions
- HRR uses the homologous chromosome. The name misleads. In somatic cells HRR overwhelmingly copies the sister chromatid, an identical replicated copy held nearby by cohesin — not the homologous chromosome, which carries different alleles. Using the homolog would risk loss of heterozygosity. Interhomolog recombination is deliberately reserved for meiosis.
- HRR works throughout the cell cycle. It cannot. HRR requires a sister chromatid, which only exists after DNA replication, so it is confined to S and G2. In G1 and in non-dividing cells, double-strand breaks are handled almost entirely by non-homologous end joining. CDK activity gates the pathway by phosphorylating CtIP and other resection factors.
- NHEJ is simply the "bad" pathway and HRR the "good" one. NHEJ is error-prone but fast, template-independent, and essential — it is the only option in G1 and it assembles antibody genes during V(D)J recombination. The two pathways compete, and the balance (set by 53BP1/RIF1 versus BRCA1) is finely tuned to the cell-cycle context, not a quality ranking.
- RAD51 is the same as bacterial RecA. RAD51 is the eukaryotic homolog of RecA and performs the analogous strand-exchange chemistry, but its regulation is far more elaborate: it depends on BRCA2 for loading, on RAD51 paralogs for filament stability, and on anti-recombinases (RECQ5, FBH1, PARI) for controlled disassembly. RecA in E. coli is loaded by RecBCD/RecFOR and needs no BRCA2.
- Every HRR event makes a Holliday junction and a crossover. Most somatic repair proceeds through SDSA, which never forms a stable double Holliday junction and is inherently crossover-free. Even when a double Holliday junction does form, dissolution by BLM yields non-crossover products. Crossovers are actually the minority outcome of somatic HRR and are actively suppressed to avoid chromosome rearrangements.
- PARP inhibitors work by blocking HRR. They do the opposite — they block single-strand-break repair via PARP. The lethality arises only because the cancer already lacks HRR (through BRCA mutation), so the double-strand breaks that accumulate when PARP is trapped cannot be fixed. In an HRR-proficient cell, PARP inhibition is well tolerated. That is the definition of synthetic lethality.
HRR vs non-homologous end joining
| Property | Homologous recombination (HRR) | Non-homologous end joining (NHEJ) |
|---|---|---|
| Template required | Yes — sister chromatid | None — ends ligated directly |
| Accuracy | Error-free (restores exact sequence) | Error-prone (small indels at junction) |
| Cell-cycle window | S and G2 only | All phases; dominant in G1 |
| Key end-processing | Extensive 5'→3' resection (MRN/CtIP, EXO1, DNA2-BLM) | Minimal processing (Artemis trimming) |
| Core proteins | MRN, RPA, BRCA1, PALB2, BRCA2, RAD51 | Ku70/Ku80, DNA-PKcs, XRCC4, LIG4 |
| Commitment switch | BRCA1 removes 53BP1 to permit resection | 53BP1–RIF1–shieldin block resection |
| Speed | Slow (hours) | Fast (minutes) |
| Physiologic role | Genome maintenance, fork restart, meiotic crossovers | General break repair, V(D)J recombination |
The molecular intermediates, in order
| Stage | DNA intermediate | Principal proteins |
|---|---|---|
| Break detection | Blunt/complex double-strand break, γH2AX flanking | MRN, ATM, MDC1 |
| Short-range resection | Nicked 5' strand, initial 3' overhang | MRE11, CtIP (CDK-activated) |
| Long-range resection | Long 3' single-stranded tails, RPA-coated | EXO1, DNA2–BLM, RPA |
| Filament assembly | RAD51 presynaptic nucleoprotein filament | BRCA1, PALB2, BRCA2, RAD51 paralogs |
| Strand invasion | Displacement loop (D-loop) | RAD51, RAD54 |
| DNA synthesis | Extended D-loop | DNA polymerase δ, PCNA |
| Resolution (SDSA) | Annealed product, no junction | Helicases (RTEL1), LIG1 |
| Resolution (dHJ) | Double Holliday junction | BLM-TOPOIIIα-RMI (dissolution) or GEN1 / SLX-MUS81 (resolution) |
Famous experiments and history
- Holliday's junction (1964). Robin Holliday, working on fungal gene conversion in Ustilago, proposed a molecular model in which two DNA duplexes exchange single strands to form a four-way crossover intermediate — now universally called the Holliday junction. The structure was later visualized directly by electron microscopy and X-ray crystallography, confirming a prediction made from genetics alone.
- The Szostak double-strand-break-repair model (1983). Jack Szostak, Terry Orr-Weaver, Rodney Rothstein, and Frank Stahl proposed in Cell that recombination initiates from a double-strand break, is processed by resection, and proceeds through a double Holliday junction — replacing earlier single-strand-break models. This framework, refined into SDSA for somatic repair, remains the textbook model. Szostak shared the 2009 Nobel Prize (for telomeres and telomerase).
- Mary-Claire King maps BRCA1 (1990). After 17 years of skepticism, King's group used linkage analysis in 23 extended families to place a dominant breast-and-ovarian-cancer susceptibility gene on chromosome 17q21, published in Science. The gene, BRCA1, was cloned by Skolnick's Myriad team in 1994; BRCA2 was mapped and cloned by Stratton's group in 1995. Only afterward was their function in HRR established.
- BRCA2 loads RAD51 (2002–2010). Ashok Venkitaraman, Stephen West, and later Ryan Jensen and Patrick Sung's crystallographic and biochemical work showed that BRCA2's BRC repeats bind RAD51 and that BRCA2 physically delivers RAD51 onto RPA-coated ssDNA, directly explaining why BRCA2 loss abolishes HRR. This connected an inherited cancer gene to a precise biochemical step.
- Synthetic lethality with PARP (2005). Two back-to-back Nature papers — from Alan Ashworth's group (Farmer et al.) and Thomas Helleday's group (Bryant et al.) — showed that BRCA1/2-deficient cells are up to a thousand-fold more sensitive to PARP inhibition than wild-type cells. This proof of synthetic lethality led directly to olaparib, which the FDA approved in December 2014 for BRCA-mutant ovarian cancer — the first drug of its class.
Frequently asked questions
How is homologous recombination different from non-homologous end joining?
Both repair DNA double-strand breaks, but they differ in accuracy and timing. Non-homologous end joining (NHEJ) simply ligates the two broken ends back together using Ku70/Ku80, DNA-PKcs, and LIG4/XRCC4. It needs no template, works in any phase of the cell cycle, and is fast — but it frequently loses or adds a few nucleotides at the junction, so it is error-prone. Homologous recombination repair (HRR) instead copies an intact homologous sequence, almost always the sister chromatid generated during replication, so it restores the original sequence exactly and is essentially error-free. Because a sister chromatid only exists after S phase, HRR is confined to S and G2, whereas NHEJ dominates in G1. The choice between them is governed by end resection: 53BP1 and RIF1 block resection to favor NHEJ, while BRCA1 antagonizes 53BP1 to license resection and commit the break to HRR.
What does RAD51 do in homologous recombination?
RAD51 is the central recombinase, the eukaryotic homolog of bacterial RecA. After resection exposes 3' single-stranded DNA coated by RPA, BRCA2 delivers and loads RAD51 monomers onto that DNA, displacing RPA to build a right-handed helical nucleoprotein filament — the presynaptic filament — that extends the DNA by roughly 1.5-fold and underwinds it. This filament performs the homology search, scanning the genome for a matching duplex, then catalyzes strand invasion: the 3' end base-pairs with the complementary strand of the donor, displacing the other strand as a D-loop. RAD51 paralogs (RAD51B, RAD51C, RAD51D, XRCC2, XRCC3) and mediators (BRCA2, the RAD51 paralog complexes) stabilize the filament, while anti-recombinases like RECQ5 and FBH1 dismantle it to prevent inappropriate recombination. Without functional RAD51 loading, cells cannot complete error-free repair and shunt breaks into mutagenic pathways.
Why is a sister chromatid the preferred template?
The sister chromatid is an exact copy of the broken DNA molecule, produced by replication and held nearby by cohesin, so copying from it restores the original sequence with no loss of heterozygosity. Using the homologous chromosome instead — which carries different alleles — could copy the wrong version of a gene and, if a crossover follows, cause loss of heterozygosity that can unmask a recessive tumor-suppressor mutation. Cells therefore strongly bias template choice toward the sister: cohesin rings encircle sister pairs after S phase, and this proximity plus regulatory suppression of interhomolog exchange keeps somatic HRR nearly always intrasister. Meiosis is the deliberate exception — there, programmed SPO11 breaks are directed to the homolog to generate crossovers and shuffle alleles between generations.
What is a Holliday junction and how is it resolved?
A Holliday junction is a four-stranded, X-shaped DNA structure where two duplexes are physically joined by a crossover of strands. Named after Robin Holliday, who proposed it in 1964, it forms when the invading 3' end is extended by DNA polymerase and the second broken end is captured, ligating both duplexes into a double Holliday junction. There are two fates. Dissolution: the BLM helicase together with topoisomerase III-alpha and RMI1/RMI2 migrates the two junctions toward each other and decatenates them, always producing non-crossover products — the safe, dominant somatic route. Resolution: structure-specific nucleases (GEN1, or the SLX1-SLX4-MUS81-EME1 complex) cut the junction; depending on which strands are cleaved, this yields either crossover or non-crossover products. Most somatic breaks avoid Holliday junctions entirely by using synthesis-dependent strand annealing, which is inherently crossover-free.
How do PARP inhibitors kill BRCA-mutant cancers?
PARP inhibitors exploit synthetic lethality: two individually survivable defects become lethal in combination. PARP1 detects single-strand breaks and initiates base-excision repair. Inhibitors such as olaparib block PARP catalytic activity and, crucially, trap PARP1 on DNA, converting unrepaired single-strand breaks into replication-associated double-strand breaks when the fork collides with them. A normal cell repairs those breaks by error-free homologous recombination and survives. A BRCA1- or BRCA2-mutant cancer cell, already unable to perform HRR, cannot fix them and dies, while surrounding normal tissue (heterozygous, HRR-proficient) is spared. Olaparib became the first FDA-approved PARP inhibitor in December 2014 for BRCA-mutant ovarian cancer; niraparib, rucaparib, and talazoparib followed. The concept was demonstrated in two landmark 2005 Nature papers from the Ashworth and Helleday groups, and the general idea of exploiting HRR deficiency now extends to 'BRCAness' tumors beyond germline BRCA mutations.
Who discovered BRCA1 and BRCA2?
Mary-Claire King's laboratory used linkage analysis across families with early-onset breast and ovarian cancer to localize the first breast cancer susceptibility gene, BRCA1, to chromosome 17q21 in a 1990 Science paper — overturning the prevailing view that breast cancer was not heritably driven by a single locus. Mark Skolnick's team at Myriad Genetics cloned the gene in 1994. BRCA2, on chromosome 13q, was mapped by Michael Stratton's group and cloned in 1995. Only later did the field establish that both proteins act in homologous recombination: BRCA2 directly loads RAD51 onto single-stranded DNA through its eight BRC repeats and DNA-binding domain, while BRCA1, working with PALB2 and the RAD51 paralogs, promotes end resection and RAD51 loading and antagonizes 53BP1. Their loss cripples error-free double-strand break repair, driving genomic instability and cancer, and creating the vulnerability that PARP inhibitors target.