DNA repair pathways: base excision, nucleotide excision, mismatch repair, and double-strand break repair

Intuition Beginner

Your DNA is under constant attack. UV light from the sun, reactive chemicals from metabolism, and even the copying process itself all damage DNA thousands of times per day in every cell. Without repair, this damage would accumulate into mutations that could kill the cell or cause cancer.

Cells have specialised repair teams, each handling a different type of damage. Base excision repair fixes small chemical changes to individual bases — like swapping out a single damaged brick in a wall. Nucleotide excision repair cuts out a whole stretch of damaged DNA — like removing a section of warped flooring and replacing it. Mismatch repair catches copying errors made during DNA replication, fixing bases that were paired with the wrong partner.

The most dangerous damage is a double-strand break, where both strands of the DNA backbone snap. Cells have two ways to fix these. One uses an undamaged copy of the same sequence as a template for accurate repair. The other stitches the broken ends back together directly, which is faster but can introduce small errors.

Visual Beginner

Think of DNA as a ladder. Base excision repair replaces a single damaged rung. Nucleotide excision repair cuts out a short section of damaged rungs and replaces them all. Mismatch repair fixes a rung where the two halves do not match. Double-strand break repair fixes a ladder that has been snapped clean across.

Worked example Beginner

Oxidation converts guanine (G) to 8-oxoguanine (8-oxoG), which pairs with adenine (A) instead of cytosine (C). If unrepaired, this causes a G-C to T-A mutation after the next round of replication. Base excision repair catches this:

1. The enzyme OGG1 detects the damaged 8-oxoG base and removes it, leaving an empty spot (an AP site).
2. APE1 cuts the DNA backbone at the empty spot.
3. DNA polymerase beta inserts the correct base (G).
4. DNA ligase seals the backbone.

Four enzymes, four steps, and the original sequence is restored. This single pathway handles the most common oxidative lesion in your genome.

Check your understanding Beginner

Formal definition Intermediate+

DNA repair pathways are enzymatic cascades that detect specific classes of DNA damage, excise the damaged material, and restore the original sequence using the complementary strand or a homologous template.

Base excision repair (BER)

BER repairs small, non-helix-distorting base modifications: deaminated bases (uracil from cytosine, hypoxanthine from adenine), oxidised bases (8-oxoguanine, thymine glycol), and alkylated bases (3-methyladenine). The pathway:

1. DNA glycosylase recognises and removes the damaged base by cleaving the N-glycosidic bond, creating an apurinic/apyrimidinic (AP) site. Different glycosylases target different lesions: UNG (uracil), OGG1 (8-oxoG), SMUG1 (oxidised uracil), MPG (alkylated bases), MBD4 (T from 5-methyl-C deamination).
2. AP endonuclease (APE1) cleaves the phosphodiester backbone 5' to the AP site, generating a 3'-OH and a 5'-deoxyribose phosphate (dRP).
3. DNA polymerase beta removes the dRP via its lyase activity and inserts the correct nucleotide (short-patch BER, replacing one nucleotide). For refractory dRP moieties, long-patch BER uses Pol $\delta /\epsilon$ and FEN1 to replace 2-10 nucleotides.
4. DNA ligase III-XRCC1 (short-patch) or DNA ligase I (long-patch) seals the nick.

Nucleotide excision repair (NER)

NER repairs bulky, helix-distorting lesions: cyclobutane pyrimidine dimers (CPDs), 6-4 photoproducts from UV radiation, and bulky chemical adducts. Two sub-pathways share a common core:

• Global genome NER (GG-NER): Scans the entire genome. XPC-RAD23B detects the thermodynamic destabilisation caused by the lesion (not the lesion itself).
• Transcription-coupled NER (TC-NER): Repairs lesions that block RNA polymerase II on the transcribed strand. CSB/CSA proteins detect the stalled polymerase.

After damage recognition, the shared core mechanism proceeds:

1. TFIIH (containing XPB and XPD helicases) unwinds approximately 30 nucleotides around the lesion.
2. XPA verifies the damage and stabilises the open complex. RPA coats the undamaged strand.
3. XPF-ERCC1 makes the 5' incision; XPG makes the 3' incision, excising a 24-32 nt oligonucleotide.
4. DNA polymerase $\delta /\epsilon$ fills the gap; DNA ligase I seals it.

Mismatch repair (MMR)

MMR corrects base-base mismatches and small insertion-deletion loops that escape polymerase proofreading. The central problem is strand discrimination: identifying which strand is newly synthesised.

In bacteria (E. coli):

1. MutS binds the mismatch.
2. MutL bridges MutS to MutH.
3. MutH nicks the unmethylated (newly synthesised) strand at the nearest hemimethylated GATC site. Dam methylase has not yet methylated the new strand.
4. Exonuclease degrades the nicked strand past the mismatch. DNA polymerase III resynthesises; DNA ligase seals.

In eukaryotes:

1. MutS-alpha (MSH2-MSH6) recognises base-base mismatches and 1-2 nt loops; MutS-beta (MSH2-MSH3) recognises larger loops.
2. MutL-alpha (MLH1-PMS2) is recruited and activated by PCNA/RFC. Its latent endonuclease introduces nicks in the discontinuous strand near the mismatch.
3. EXO1 degrades the nicked strand past the mismatch.
4. DNA polymerase $\delta$ resynthesises; DNA ligase I seals.

MMR improves replication fidelity approximately 100-fold, from $\sim 10^{-7}$ to $\sim 10^{-9}$ errors per base per replication.

Double-strand break repair (DSBR)

Homologous recombination (HR): Uses the sister chromatid as a template. Restricted to S/G2 phase.

1. The MRN complex (Mre11-Rad50-Nbs1) binds the DSB. BRCA1 promotes 5'-to-3' end resection with CtIP, generating 3' single-stranded DNA overhangs.
2. RPA coats the ssDNA to prevent secondary structure.
3. BRCA2 loads Rad51 onto the ssDNA, displacing RPA and forming a nucleoprotein filament.
4. Rad51 performs strand invasion into the homologous duplex, forming a displacement loop (D-loop). DNA synthesis extends the invading 3' end using the sister chromatid as template.
5. Holliday junctions are resolved by resolvases (GEN1, MUS81-EME1) or dissolved by the BLM-TOP3A-RMI1/2 complex, producing crossover or non-crossover products.

Non-homologous end joining (NHEJ): Directly ligates broken ends. Active throughout the cell cycle.

1. Ku70/Ku80 heterodimer rapidly binds each DSB end.
2. DNA-PKcs is recruited, forming the DNA-PK holoenzyme, which autophosphorylates.
3. Artemis nuclease processes damaged or incompatible ends if necessary.
4. XRCC4-DNA Ligase IV (with XLF/Cernunnos) performs the final ligation.

NHEJ is fast (minutes) but error-prone; small insertions and deletions are common at the junction. When ends from different chromosomes are joined, the result is a chromosomal translocation.

Pathway choice and regulation

The choice between NHEJ and HR is regulated by CDK activity and the competing actions of 53BP1 and BRCA1. In G1, 53BP1 binds DSB ends and shields them from resection, directing repair toward NHEJ. In S/G2, CDK-dependent phosphorylation activates CtIP, and BRCA1 antagonises 53BP1, permitting end resection and committing the break to HR. Loss of 53BP1 in BRCA1-deficient cells partially restores HR, revealing the competitive balance between the two pathways.

Key mechanism Intermediate+

The NER dual-incision mechanism and damage recognition by XPC.

XPC-RAD23B does not directly contact the damaged bases. Instead, it detects the thermodynamic destabilisation of the DNA duplex caused by the lesion. A UV-induced cyclobutane pyrimidine dimer bends the helix by approximately 30 degrees and unwinds it by approximately 9 degrees. XPC senses this disruption by probing the base-pairing stability: it inserts a beta-hairpin loop that flips out bases on the undamaged strand, testing whether the duplex is intact. At a lesion site, the disrupted base pairing allows the hairpin to insert, triggering recruitment of TFIIH.

Once TFIIH is recruited, the XPB and XPD helicases unwind approximately 30 nucleotides around the lesion. XPA arrives and verifies the damage by sensing the distorted structure. RPA coats the undamaged strand. The dual incision follows: XPF-ERCC1 (a structure-specific endonuclease) cuts the damaged strand at the 5' junction of the bubble, and XPG cuts at the 3' junction, releasing a 24-32 nt oligonucleotide containing the lesion. The resulting gap is filled by Pol $\delta /\epsilon$ and sealed by ligase I.

The precision of the dual-incision coordinates is set by the structure-specific recognition: XPF cuts at the single-strand/double-strand junction on the 5' side, and XPG cuts at the corresponding junction on the 3' side. The 24-32 nt excised fragment is remarkably consistent across substrates, reflecting the fixed geometry of the open bubble created by TFIIH.

Exercises Intermediate+

Translesion synthesis, cancer genomics, and the repair-deficiency landscape Master

Translesion synthesis polymerases

When the replication fork encounters a lesion that has not been repaired, the replicative polymerase stalls — its tight active site cannot accommodate distorted bases. The cell then deploys specialised Y-family translesion synthesis (TLS) polymerases with relaxed active sites that can accommodate damaged bases. This is a calculated trade-off: a mutation is preferable to a stalled fork that could collapse into a double-strand break.

Pol $\eta$ (eta), encoded by POLH, bypasses UV-induced cyclobutane pyrimidine dimers accurately — it inserts two adenines opposite the dimer, preserving the original sequence. Mutations in POLH cause XP-V (xeroderma pigmentosum variant), where NER is functional but UV lesion bypass is inaccurate, elevating mutagenesis.

Pol $\iota$ (iota) has unusual base-pairing preferences and inserts nucleotides opposite certain lesions with reduced fidelity. Pol $\kappa$ (kappa) extends from nucleotides inserted opposite minor-groove adducts. Pol $\zeta$ (Rev3/Rev7, a B-family polymerase) specialises in extending from the mispaired termini generated by Y-family polymerase insertion, completing the bypass. Rev1 acts as a scaffold, recruiting TLS polymerases to the stalled fork via its ubiquitin-binding domain, which recognises monoubiquitinated PCNA.

TLS regulation centres on PCNA ubiquitination. When the fork stalls at a lesion, Rad6-Rad18 (E2-E3 ligase) monoubiquitinates PCNA at K164. This modification recruits TLS polymerases through their ubiquitin-binding motifs. If the lesion is a CPD, Pol $\eta$ performs accurate bypass. If the lesion is not a Pol $\eta$ substrate, Pol $\iota$ or Pol $\kappa$ insert a nucleotide and Pol $\zeta$ extends — this error-prone TLS introduces a mutation but prevents fork collapse. The cell has thus built a hierarchy of bypass accuracy matched to lesion type.

XP genes, HNPCC/Lynch syndrome, and BRCA1/2: the repair-deficiency cancer spectrum

Each DNA repair pathway has a corresponding hereditary cancer syndrome when germline mutations inactivate a key component. These syndromes reveal the in vivo role of each pathway through the specific cancer types and mutational signatures they produce.

Xeroderma pigmentosum (XP): Germline mutations in NER genes (XPA, XPB, XPC, XPD, XPE, XPF, XPG, or POLH). Patients have approximately 2,000-fold increased risk of skin cancer and approximately 10,000-fold increased risk of melanoma before age 20. The cancer spectrum is confined to sun-exposed tissues because the defective pathway (NER) primarily repairs UV-induced damage. Internal cancer rates are near normal.

Lynch syndrome (HNPCC): Germline mutations in MMR genes — most commonly MSH2 or MLH1, less frequently MSH6 or PMS2. MMR loss produces a mutator phenotype: the per-base error rate rises from $\sim 10^{-9}$ to $\sim 10^{-7}$, producing $\sim 640$ new mutations per cell division instead of $\sim 6$. This accelerated mutation rate drives rapid accumulation of driver mutations. Lifetime colorectal cancer risk reaches $\sim 80\%$ for MLH1 carriers ($\sim 4\%$ in the general population). Endometrial cancer risk is $\sim 60\%$ for female carriers. The diagnostic signature is microsatellite instability (MSI) — variable-length short repeats in tumour DNA, caused by unrepaired strand slippage at repetitive sequences. MSI-high tumours have a high neoantigen burden and respond well to immune checkpoint inhibitors (pembrolizumab is approved for any MSI-high solid tumour regardless of tissue origin).

Hereditary breast and ovarian cancer (BRCA1/2): Germline mutations in BRCA1 or BRCA2 impair homologous recombination. BRCA1 promotes 5' end resection at DSBs; BRCA2 loads Rad51 onto ssDNA. Loss of either protein disables accurate HR and forces DSBs through error-prone NHEJ. Lifetime breast cancer risk is $\sim 60$-$85\%$ for BRCA1 carriers and $\sim 45$-$75\%$ for BRCA2 carriers. Ovarian cancer risk is $\sim 15$-$40\%$. The mutational signature of HR deficiency (COSMIC Signature 3) shows large numbers of small indels with microhomology at breakpoints, reflecting NHEJ-mediated repair.

PARP inhibitors and synthetic lethality

The clinical exploitation of HR deficiency is the paradigmatic example of synthetic lethality in oncology. PARP (poly-ADP-ribose polymerase) repairs single-strand breaks through the BER pathway. When PARP is pharmacologically inhibited (olaparib, rucaparib, niraparib, talazoparib), unrepaired SSBs accumulate. During replication, the fork encounters an SSB and converts it into a one-ended DSB. In HR-proficient cells, Rad51-mediated strand invasion repairs the DSB accurately. In HR-deficient cells (BRCA1/2 loss of heterozygosity), the DSB is shunted to NHEJ, producing genomic catastrophe and cell death.

The selectivity is elegant: normal cells (heterozygous for BRCA mutation, HR-proficient) survive PARP inhibition; tumour cells (homozygous BRCA loss, HR-deficient) die. PARP inhibitors have become standard-of-care for BRCA-mutant ovarian and breast cancer and are being tested in other HR-deficient tumours (pancreatic, prostate).

Resistance to PARP inhibitors emerges through several mechanisms: (a) secondary mutations in BRCA1/2 that restore the reading frame and reactivate HR; (b) loss of 53BP1, which restores end resection and partially reactivates HR despite BRCA1 loss; (c) upregulation of drug efflux pumps; (d) loss of PARP1 expression. The restoration-of-reading-frame mechanism is particularly striking: the selective pressure of PARP inhibitor treatment favours revertant mutations that restore BRCA function, demonstrating that the tumour evolves under therapy exactly as predicted by the synthetic lethality model.

Chromothripsis and catastrophic genome rearrangement

Chromothripsis is a recently discovered phenomenon in which a single chromosome (or chromosome arm) undergoes tens to hundreds of DSBs in a single catastrophic event, followed by error-prone reassembly by NHEJ. The result is a chromosome with dozens of rearrangements — deletions, inversions, duplications, and translocations — all occurring in one pulse rather than accumulating gradually.

The evidence for single-event reconstruction comes from the oscillating copy-number pattern: segments of the affected chromosome alternate between one copy (deleted segments) and two copies (retained segments), with the retained segments randomly arranged (not in their original order). This pattern is inconsistent with gradual accumulation of independent events and instead indicates that all breaks occurred simultaneously and were reassembled in a random order by NHEJ.

Chromothripsis is observed in $\sim 2$-$3\%$ of all cancers and $\sim 25\%$ of osteosarcomas. The trigger is thought to be chromosome missegregation during mitosis, producing micronuclei — small nuclear envelopes containing one or a few chromosomes. The micronuclear envelope is fragile and prone to rupture, exposing the enclosed chromosome to cytoplasmic nucleases and replication stress that generates the catastrophic DSB burst. The damaged chromosome is then reincorporated into the main nucleus during the next mitosis, and NHEJ reassembles the fragments.

COSMIC mutational signatures

The catalogue of mutational signatures maintained by COSMIC (Catalogue Of Somatic Mutations In Cancer) provides a systematic framework for identifying which DNA repair deficiencies and mutagenic exposures drove each tumour's evolution. Each signature is a probability distribution over 96 trinucleotide-context mutation classes (6 substitution types $\times$ 4 flanking 5' bases $\times$ 4 flanking 3' bases).

1. Signature 1 (age): C-to-T transitions at CpG sites, driven by spontaneous deamination of 5-methylcytosine. Present in virtually all cancers, reflecting the baseline endogenous damage rate.

2. Signature 3 (HR deficiency): Large numbers of small (1-20 bp) deletions with microhomology at breakpoints. Marks BRCA1/2-mutant tumours.

3. Signature 4 (tobacco): C-to-A transversions. Found in lung, head and neck, and bladder cancers of smokers.

4. Signature 7 (UV): C-to-T transitions at dipyrimidine sites, with a CC>TT tandem mutation component. Diagnostic of UV-induced skin cancers.

5. Signature 6 / 15 / 20 / 26 (MMR deficiency): Elevated C-to-T transitions and T-to-G transversions with associated indels at microsatellite repeats. These signatures identify MSI-high tumours.

6. Signature 8 / 18 (BER deficiency): Associated with oxidative damage. Signature 18 specifically correlates with 8-oxoG accumulation.

The clinical application is direct: sequencing a tumour genome and extracting its mutational signatures identifies the underlying repair deficiency, guiding therapy selection (PARP inhibitors for Signature 3, immunotherapy for MSI-high signatures, UV avoidance for Signature 7).

V(D)J recombination uses NHEJ machinery

The adaptive immune system generates antibody and T-cell receptor diversity through V(D)J recombination, a programmed DNA rearrangement that deliberately creates and repairs DSBs using the NHEJ pathway. The RAG1/RAG2 recombinase introduces DSBs at recombination signal sequences flanking V, D, and J gene segments. The resulting coding-end hairpins are opened by Artemis (the same nuclease used in general NHEJ), and the broken ends are joined by Ku70/80, DNA-PKcs, and XRCC4-Ligase IV.

The NHEJ machinery's inherent imprecision is functionally important: nucleotides are lost from the coding ends (trimming) and new nucleotides are added by terminal deoxynucleotidyl transferase (TdT) before ligation. This junctional diversity at the V-D and D-J boundaries is the primary source of antibody complementarity-determining region (CDR3) diversity, which determines antigen-binding specificity.

Patients with mutations in NHEJ components (Artemis, DNA-PKcs, XRCC4, Ligase IV) have severe combined immunodeficiency (SCID) because V(D)J recombination cannot be completed — lymphocyte development is arrested at the progenitor stage. This clinical observation directly demonstrates that NHEJ is the dedicated DSB repair pathway for immune receptor gene assembly.

Synthesis. The four repair pathways — BER, NER, MMR, and DSBR — form a hierarchical defence against the full spectrum of DNA damage. BER handles the highest volume (small base modifications from endogenous chemistry); NER handles the bulky adducts; MMR proofreads replication; DSBR handles the most dangerous lesions. The clinical landscape maps directly onto this hierarchy: each pathway has a corresponding hereditary cancer syndrome, a characteristic mutational signature, and increasingly a pathway-specific therapy. The central insight is that repair deficiency does not merely increase mutation rate — it produces a qualitatively distinct mutation spectrum that identifies the broken pathway from the tumour genome alone. The bridge to 17.06.03 pending runs through the transposable-element machinery: transposases create DSBs that are repaired by NHEJ and HR, and the cell deploys DNA methylation and heterochromatin formation (processes intertwined with repair signalling) to suppress transposition.

Connections Master

• Mutation and repair overview 17.06.01. This unit deepens the four repair pathways introduced in 17.06.01 with full mechanistic detail. The layered fidelity cascade (polymerase selectivity, proofreading, MMR) proved in 17.06.01 is here extended to the damage-specific pathways (BER, NER) and the DSB repair choice (HR vs NHEJ).

• Cellular respiration and ROS production 17.04.01. The electron transport chain generates reactive oxygen species that produce the oxidative lesions (8-oxoG, thymine glycol) repaired by BER. The BER pathway is therefore coupled to metabolic rate: tissues with high oxidative phosphorylation activity have the highest BER substrate load.

• DNA replication 17.05.01. MMR acts on mismatches in newly synthesised DNA, using the strand asymmetry of the replication fork for discrimination. HR uses the sister chromatid generated during S phase as its repair template. The replication fork itself is a source of DSBs when it encounters unrepaired lesions or collapses at stalled sites.

• Cell cycle and checkpoints 17.08.01. The DNA damage response (ATM/ATR kinases, p53) coordinates repair with cell-cycle arrest. ATM is activated by DSBs via the MRN complex; ATR is activated by RPA-coated ssDNA at resected breaks or stalled forks. These kinases halt the cell cycle at G1/S and G2/M checkpoints to allow repair before replication or mitosis.

• Transposable elements 17.06.03 pending. DNA transposons create DSBs as part of their mobilisation mechanism, which are repaired by NHEJ (producing the target-site duplications flanking new insertions) or by HR. The host cell deploys the DNA methylation machinery — which interacts with MMR and BER pathways — to silence transposons in germline and somatic cells.

• Immunology 17.10.01. V(D)J recombination in developing lymphocytes uses the NHEJ machinery (Ku70/80, DNA-PKcs, Artemis, XRCC4-Ligase IV) to join RAG-generated DSBs. NHEJ component mutations cause SCID. Class-switch recombination in activated B cells also uses NHEJ after AID-mediated deamination, connecting BER (which processes the deaminated bases) to the DSB repair outcome.

• Population genetics 19.02.05. The per-base mutation rate ($\sim 10^{-9}$ with functional repair) is the parameter that feeds into the Wright-Fisher model as the supply of new alleles. Repair deficiency increases this rate by 100-fold (MMR loss) to 1000-fold (polymerase proofreading loss), fundamentally altering the population-genetic dynamics of mutation-selection balance.

Historical notes Master

Tomas Lindahl's 1993 Nature paper, "Instability and decay of the primary structure of DNA," systematically catalogued spontaneous DNA damage rates and established that DNA is chemically unstable — decaying at a rate incompatible with life without constant repair. This quantitative framework motivated Lindahl's subsequent discovery and characterisation of base excision repair. He demonstrated that DNA glycosylases recognise and remove damaged bases through a base-flipping mechanism, and he purified the first glycosylase (UNG, uracil-DNA glycosylase) and the AP endonuclease that processes the resulting abasic sites.

Aziz Sancar characterised the nucleotide excision repair pathway through a series of biochemical reconstitution experiments in the 1980s-1990s. Working initially on photolyase (the light-dependent enzyme that directly reverses UV dimers, absent in placental mammals) and then on the UvrABC excinuclease system in E. coli, Sancar defined the dual-incision mechanism that excises a 12-13 nt oligonucleotide containing the lesion. His later work on the eukaryotic NER machinery identified the roles of TFIIH, XPA, XPC, XPF, XPG, and the other XP gene products. Sancar's 2015 Nobel Lecture in Chemistry provides the definitive account.

Paul Modrich defined the mismatch repair pathway through elegant biochemical genetics. His reconstitution of the E. coli MutHLS system in vitro demonstrated that MutS recognises the mismatch, MutL couples recognition to strand discrimination, and MutH nicks the unmethylated strand at hemimethylated GATC sites. His subsequent work on eukaryotic MMR identified the MSH and MLH heterodimers and the PCNA-dependent strand discrimination mechanism. Modrich's quantitative measurements of MMR's contribution to replication fidelity ($\sim 100$-fold improvement) established the multiplicative fidelity cascade.

Lindahl, Sancar, and Modrich shared the 2015 Nobel Prize in Chemistry "for mechanistic studies of DNA repair."

The discovery of the BRCA1 (1994) and BRCA2 (1995) genes by positional cloning linked homologous recombination deficiency to hereditary breast and ovarian cancer. The subsequent demonstration by Bryant and Farmer in 2005 that PARP inhibition is synthetically lethal in BRCA-deficient cells translated the basic biochemistry of DSBR into the first pathway-targeted cancer therapy. Olaparib received FDA approval in 2014 for BRCA-mutant ovarian cancer.

Chromothripsis was discovered in 2011 by Campbell, Stratton, and colleagues through whole-genome sequencing of a chronic myeloid leukaemia patient, revealing a chromosome that had shattered and been reassembled in a single catastrophic event. The finding challenged the prevailing model of cancer evolution as a gradual, stepwise process and introduced the concept of punctuated genomic catastrophe.

The COSMIC mutational signature catalogue was developed by Alexandrov, Stratton, and colleagues starting in 2013. By applying non-negative matrix factorization to thousands of cancer genomes, they extracted approximately 30 distinct signatures, each corresponding to a specific mutagenic process or repair deficiency. This work transformed cancer genomics from descriptive cataloguing into a diagnostic tool for identifying the underlying cause of each tumour's mutational burden.

Bibliography Master

1. Lindahl, T. — Instability and decay of the primary structure of DNA. Nature 362 (1993) 709-715.

2. Sancar, A. — Mechanisms of DNA repair by photolyase and excision nucleases. Angew. Chem. Int. Ed. 55 (2016) 8623-8641.

3. Modrich, P. — Mechanisms in eukaryotic DNA mismatch repair. J. Biol. Chem. 281 (2006) 30305-30309.

4. Friedberg, E. C., Walker, G. C., Siede, W. & Wood, R. D. — DNA Repair and Mutagenesis, 2nd ed. (ASM Press, 2006).

5. Alberts, B. et al. — Molecular Biology of the Cell, 7th ed. (Garland Science, 2022).

6. Lodish, H. et al. — Molecular Cell Biology, 9th ed. (W. H. Freeman, 2021).

7. Bryant, H. E. et al. — Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature 434 (2005) 917-921.

8. Farmer, H. et al. — Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434 (2005) 917-921.

9. Alexandrov, L. B. et al. — Signatures of mutational processes in human cancer. Nature 500 (2013) 415-421.

10. Stephens, P. J. et al. — Massive genomic rearrangement acquired in a single catastrophic event during cancer development. Cell 144 (2011) 27-40.