Genome Instability & DNA Repair

1. The Mutator Phenotype — Why Losing a Caretaker Matters

A normal human cell makes roughly 10¹⁷ DNA lesions across a human lifetime: ~70,000 base modifications per cell per day from oxidation, alkylation, deamination, and depurination, plus ~10 double-strand breaks per cell per day from replication-fork collapse and reactive oxygen. The spontaneous per-base, per-replication mutation rate of an unrepaired template would be ~10⁻⁵; with the full repair machinery intact it is ~10⁻⁹ — a four-order-of-magnitude error correction.

Loehr and Loeb (PNAS, 1991, and a quarter-century of follow-up) argued that the normal mutation rate is too low to generate the dozens of driver alterations seen in cancer within a human lifetime. Their resolution: the mutator phenotype — an early event in tumourigenesis that knocks out a caretaker pathway, raising the mutation rate per cell division by 10×–1000× and accelerating acquisition of subsequent drivers.

The lesions that fall on DNA are diverse, and so are the pathways that remove them. A first-pass mapping of insult to repair pathway:

Stratton (Cell 2009, “The cancer genome”) framed the modern view: every cancer genome is a palimpsest of the mutagens that wrote it and the repair pathways that failed to erase them. The mutational signatures of Part II are the forensic readout of this section’s biology.

2. Mismatch Repair (MMR)

Replicative DNA polymerases (Pol δ, Pol ε) are accurate (~10⁻⁷ per base) but not perfect. After proofreading, residual mismatches and small insertion/deletion loops at repetitive tracts (microsatellites) must be excised post-replicatively. That job falls to the mismatch repair (MMR) machinery.

The reaction

• Recognition. MutSα (MSH2–MSH6 heterodimer) recognises base–base mismatches and 1-nt indels; MutSβ (MSH2–MSH3) recognises larger indel loops.
• Recruitment. MutLα (MLH1–PMS2) is recruited; PMS2 carries the latent endonuclease that nicks the daughter (newly synthesised) strand, guided by strand-discrimination signals (in eukaryotes, pre-existing nicks at Okazaki junctions; in bacteria, hemimethylated GATC).
• Excision. EXO1 5′->3′ exonuclease removes a tract of the daughter strand spanning the mismatch.
• Resynthesis. Pol δ fills in, ligase I seals.

Lynch syndrome & MSI-high tumours

Heterozygous germline loss of MLH1, MSH2, MSH6, or PMS2 causes Lynch syndrome (hereditary non-polyposis colorectal cancer, HNPCC): colorectal-cancer risk ~40–80% by age 80, plus endometrial, ovarian, gastric, urothelial, small-bowel, sebaceous-gland (Muir-Torre), and CNS (Turcot) tumours. Bi-allelic germline loss is constitutional MMR deficiency (CMMRD), with paediatric brain tumours, leukaemias and bowel cancers.

MMR-deficient tumours fail to repair replication slippage at microsatellites, producing length variation at simple-repeat tracts — microsatellite instability (MSI-high). The Bethesda 5-marker panel (BAT-25, BAT-26, NR-21, NR-24, MONO-27) or NGS-based MSI inference (e.g. MSIsensor) is now standard. ~15% of CRC, ~30% of endometrial, and a small fraction of many other cancers are MSI-high. Sporadic MSI is most often caused by epigenetic biallelic MLH1 promoter hypermethylation, often alongside BRAF V600E.

Therapeutic consequence (Section 10): MSI-high tumours carry hundreds to thousands of mutations and produce abundant frameshift neo-antigens at mononucleotide tracts in coding regions. They are exquisitely sensitive to anti-PD-1 (Le et al., NEJM 2015), giving rise to the first tissue-agnostic FDA approval (pembrolizumab, 2017).

3. Base Excision Repair (BER)

BER deals with the high-frequency, low-distortion lesions of everyday metabolism: oxidised bases (8-oxo-G), deaminated bases (uracil from cytosine, hypoxanthine from adenine), alkylated bases (3-methyladenine, 7-methylguanine from MMS, temozolomide, endogenous SAM), and abasic sites. The machinery is small, fast, and modular.

The reaction (short-patch BER)

1. Glycosylase recognises and excises the damaged base — OGG1 for 8-oxo-G, UNG for uracil, MPG/AAG for 3-meA, MUTYH for A:8-oxo-G mispairs — leaving an apurinic/apyrimidinic (AP) site.
2. APE1 (AP-endonuclease 1) cleaves the phosphodiester bond 5′ of the AP site, leaving a 3′-OH and 5′-dRP terminus.
3. Pol β removes the 5′-dRP via its lyase activity and adds a single nucleotide.
4. XRCC1 scaffolds Pol β with LIG3, which seals the nick.

Long-patch BER (2–10 nt) handles oxidised dRP termini; here Pol δ/ε and FEN1 take over, with LIG1 ligating.

Single-strand breaks & the role of PARP1

BER intermediates are obligate single-strand breaks (SSBs). Direct SSBs (from ROS, ionising radiation, abortive topoisomerase I) feed into the same end-stage machinery. The chief SSB sensor is PARP1 (poly-ADP-ribose polymerase 1): it binds nicked DNA via its zinc fingers and uses NAD⁺ to synthesise long branched poly-ADP-ribose (PAR) chains on itself and on histones, generating a local signal that recruits XRCC1 and the BER/SSBR machinery within seconds. PARP1 then auto-PARylates and dissociates — a self-limiting recruitment cycle.

\[ \text{PARP1}_{\text{DNA-bound}} + n\,\text{NAD}^{+} \;\xrightarrow{\;\text{PARylation}\;}\; \text{PARP1-(ADP-ribose)}_{n} + n\,\text{nicotinamide} \;\;\xrightarrow{\;\text{PARG}\;}\; \text{PARP1}_{\text{free}} \]

Two consequences are central to oncology. First, PARP1-dependent BER prevents the accumulation of SSBs which, encountered by a replication fork in S-phase, collapse into one-ended double-strand breaks — the substrate of HR. Second, clinical PARP inhibitors (Section 9) not only block enzymatic activity but trap PARP1 on DNA (Murai, Pommier, Cancer Res 2012) — this physical block on the fork, not enzymatic loss alone, explains their potency.

4. Nucleotide Excision Repair (NER)

NER removes bulky, helix-distorting lesions: UV-induced cyclobutane pyrimidine dimers (CPDs) and 6–4 photoproducts; tobacco-PAH–DNA adducts (benzo[a]pyrene-diol-epoxide); cisplatin 1,2-d(GpG) intra-strand crosslinks. Two flavours, sharing a common core:

• Global-genome NER (GG-NER) — surveys the entire genome; lesion recognition by the XPC–RAD23B–CETN2 complex (with UV-DDB / DDB1–DDB2 helping at UV lesions).
• Transcription-coupled NER (TC-NER) — triggered when RNA Pol II stalls at a lesion; CSA, CSB and UVSSA recruit the core machinery to clear the block.

Common core (post-recognition)

1. TFIIH (containing the helicases XPB and XPD) opens a ~30-nt bubble.
2. XPA verifies the lesion; RPA coats the undamaged strand.
3. Two structure-specific endonucleases cut: XPF–ERCC1 5′ of the lesion, XPG 3′.
4. A 24–32-nt oligonucleotide carrying the lesion is released; Pol δ/ε (with PCNA, RFC) fills in; LIG1 (or LIG3–XRCC1) ligates.

Xeroderma pigmentosum, Cockayne syndrome, trichothiodystrophy

Mutations in the seven complementation groups XPA–XPG plus the variant XPV (= POLH, see Section 7) cause Xeroderma Pigmentosum: ~10,000-fold increased risk of UV-induced skin cancers (BCC, SCC, melanoma) starting in early childhood, and ~20-fold risk of internal cancers. CSA/CSB defects yield Cockayne syndrome (developmental, neurodegenerative, no cancer predisposition); certain XPB/XPD/TTDA alleles yield trichothiodystrophy. The XP genome bears the clearest signature SBS7a/b (UV C>T at TpC dipyrimidines) we have.

Pharmacological relevance: NER is the principal pathway that resolves cisplatin intra-strand adducts. ERCC1 expression is a long-studied biomarker of platinum resistance in NSCLC and ovarian cancer; ERCC1-low tumours respond better to cisplatin (Olaussen et al., NEJM 2006; superseded by HRD-based biomarkers but historically pivotal).

5. Homologous Recombination (HR)

Double-strand breaks (DSBs) are the most lethal lesion: a single unrepaired DSB kills a cell. Eukaryotes have two complementary pathways. In S/G2, where a sister chromatid is available as an error-free template, homologous recombination (HR)dominates. Choice between HR and NHEJ (Section 6) is determined at the earliest step — resection — by the antagonism of 53BP1/RIF1/Shieldin (anti-resection, pro-NHEJ) and BRCA1/CtIP/MRN (pro-resection, pro-HR), with cell-cycle and chromatin context biasing the decision.

The reaction

1. Sensing. The MRN complex (MRE11–RAD50–NBN) binds the broken end and recruits ATM (Section 8 of Part III).
2. End resection. MRE11/CtIP nick–and–trim creates short 3′ ssDNA; long-range resection by EXO1, BLM–DNA2 produces kilobase-long 3′ ssDNA tails. BRCA1 opposes 53BP1 to license resection.
3. RPA coating. The ssDNA is rapidly coated by RPA, which both protects it and triggers ATR signalling (Section 8).
4. RAD51 loading. BRCA2 (with PALB2 as the BRCA1-BRCA2 bridge) escorts RAD51 onto RPA-coated ssDNA, displacing RPA and forming the RAD51–ssDNA presynaptic filament.
5. Strand invasion. The RAD51 filament searches for, and invades, the homologous duplex, forming a displacement loop (D-loop).
6. Synthesis & resolution. DNA synthesis copies from the intact template; Holliday junctions (or D-loops) are resolved by GEN1, MUS81–EME1, SLX1–SLX4, or dissolved by the BLM–TOP3A–RMI1–RMI2 dissolvasome to give crossover or non-crossover products.

BRCA1 and BRCA2are the central caretakers of HR. BRCA1 (1863 aa, RING + tandem BRCT) is a scaffold that promotes resection and assembles multi-subunit complexes; BRCA2 (3418 aa, eight BRC repeats) is the dedicated RAD51 loader. Heterozygous germline loss causes Hereditary Breast/Ovarian Cancer (HBOC) — lifetime breast-cancer risk ~60–80% (BRCA1 > BRCA2) and ovarian risk ~20–60%, plus elevated prostate, pancreatic, and male breast cancer in BRCA2 carriers. Tumours arise after somatic LOH of the wild-type allele.

At the genome level, HR-deficient tumours carry the SBS3 single-base substitution signature, the ID6 indel signature (long deletions with microhomology — MMEJ scars), and large-scale copy-number aberrations captured by the HRD score = LOH + telomeric AI + large-scale state transitions (Telli et al., Clin Cancer Res 2016).

6. Non-Homologous End Joining (NHEJ)

NHEJ is the dominant DSB-repair pathway in G1 (when no sister chromatid template exists), in post-mitotic neurons, and in lymphoid V(D)J recombination. It is fast (minutes), template-independent, and intrinsically error-prone — small insertions and deletions at the join are typical.

The reaction

1. End binding. The Ku70–Ku80 (XRCC6–XRCC5) heterodimer threads onto each broken DNA end as a ring — a sub-second event, the highest-affinity DNA-end binding known (K_d ~10⁻⁹ M).
2. Kinase recruitment. Ku recruits DNA-PKcs (PRKDC), forming the DNA-PK holoenzyme; the two ends are tethered across the synaptic complex.
3. End processing. Artemis (DCLRE1C) is activated by DNA-PKcs phosphorylation and trims overhangs and opens hairpins (essential for V(D)J coding-joint formation). PNKP, APLF, TDP1/2, and pol μ/λ tidy idiosyncratic end chemistries.
4. Ligation. The XRCC4–LIG4–XLF (NHEJ1) complex, with PAXX, ligates. LIG4 is the dedicated NHEJ ligase; LIG4 hypomorphs cause LIG4 syndrome (microcephaly, immunodeficiency, leukaemia).

Cancer relevance is two-fold. (i) NHEJ generates the chromosomal translocations that drive lymphomas and sarcomas (BCL2-IgH, MYC-IgH, EWS-FLI1) when two independent DSBs are mis-joined. (ii) Loss of NHEJ components causes severe combined immunodeficiency (RAG-, Artemis-, LIG4-, NHEJ1-SCID); some patients develop lymphomas because mis-rejoined V(D)J intermediates persist.

When canonical NHEJ fails, an error-prone backup — microhomology-mediated end joining (MMEJ / alt-EJ / TMEJ) — takes over. Pol θ (POLQ) anneals 2–20-bp microhomologies after limited resection, gap-fills, and hands off to LIG3. MMEJ is the source of the characteristic mid-sized deletions with flanking microhomology in BRCA-deficient tumours, and POLQ inhibitors (novobiocin analogues, Ideaya/Artios programmes) are now in trials as a synthetic-lethal strategy in HR-deficient cancers (Ceccaldi et al., Nature 2015).

7. Translesion Synthesis (TLS)

When a replicative polymerase (Pol δ or ε) encounters a lesion it cannot accommodate, the fork stalls. Rather than collapse, the cell can swap in a Y-family translesion polymerasethat has a wider active site, low fidelity, no proofreading, and is dedicated to copying past damage. TLS is mutagenic but life-saving: it preserves fork progression at the cost of a few base substitutions.

Mechanism

At a stalled fork, RAD18–RAD6 mono-ubiquitylates PCNA at K164. Mono-ubiquitin-PCNA recruits Y-family polymerases via their UBM/UBZ domains; one inserts opposite the lesion, REV1 acts as a scaffold, and Pol ζ extends past the mismatch. PCNA is then de-ubiquitylated and the replicative polymerase resumes. An alternative branch is poly-ubiquitylated PCNA (K63-linked, by UBC13–MMS2–HLTF/SHPRH) that drives error-free template switching.

Therapeutic angle: TLS polymerases are mutators that also drive resistance to platinum and alkylators. REV1 inhibitors and inhibitors of the REV1–Pol ζ interaction (e.g. JH-RE-06, Sail-Boat compounds) are in development to potentiate cisplatin efficacy and suppress the mutational evolution that produces resistant clones (Wojtaszek et al., Cell 2019).

8. The Replication-Stress Response

Replication stress — defined as slowing or stalling of replication-fork progression — is the single most common form of endogenous genome insult in proliferating cells, and it is dramatically elevated in cancer cells driven by oncogenes that force unscheduled S-phase entry (MYC, cyclin E, RAS). Halazonetis (Nature 2008) and others argued that oncogene-induced replication stress is the proximate cause of genomic instability in early tumourigenesis.

The ATR–CHK1 axis

1. Sensing. When the replicative helicase (CMG) and polymerase uncouple, long stretches of RPA-coated ssDNA accumulate at the fork.
2. ATR activation. RPA-ssDNA recruits ATRIP–ATR; the 9-1-1 (RAD9–HUS1–RAD1) clamp is loaded by RAD17, and TOPBP1 (or ETAA1) activates ATR kinase activity.
3. Effector signalling. ATR phosphorylates CHK1 (S317, S345); CHK1 phosphorylates CDC25A (degradation) and WEE1 (stabilisation), holding CDK2 inactive and enforcing intra-S/G2 arrest.
4. Fork protection. ATR phosphorylates dozens of substrates (~700 SQ/TQ sites in proteomic surveys) that stabilise the fork: SMARCAL1/ZRANB3/HLTF (fork reversal), BRCA2/RAD51 (protection of reversed-fork nascent strands from MRE11-mediated degradation), FANCD2.
5. Origin firing. ATR–CHK1 globally suppresses late-origin firing, preserving the dNTP pool and limiting fork density.

\[ \text{RPA-ssDNA} \xrightarrow{\;\text{ATRIP/9-1-1/TOPBP1}\;} \text{ATR}^{*} \xrightarrow{\;\;} \text{CHK1}^{*} \xrightarrow{\;\;} \begin{cases} \text{CDC25A degradation} \to \text{CDK2 OFF} \\ \text{WEE1 stabilisation} \to \text{CDK1 OFF} \\ \text{origin firing suppressed} \\ \text{fork stabilised} \end{cases} \]

Cancer cells that have lost p53 (almost all of them in some way) rely disproportionately on the ATR–CHK1–WEE1 axis for survival under their chronically elevated replication stress. This sets up the targeted-therapy opportunity: ATR inhibitors (ceralasertib, elimusertib), CHK1 inhibitors (prexasertib), and WEE1 inhibitors (adavosertib/AZD1775) are all clinical candidates, alone and in combination with platinum, gemcitabine, PARP inhibitors, and radiotherapy. Synthetic-lethal partners include ATM-deficient (mantle-cell lymphoma), ARID1A-mutant, SETD2-mutant and CCNE1-amplified tumours (Reaper, Lukas, Forment).

9. Synthetic Lethality — PARP Inhibitors in BRCA-Deficient Tumours

Synthetic lethality is a genetic relationship between two genes, A and B, in which loss of either alone is viable but loss of both is lethal. Coined by Calvin Bridges (1922) in Drosophilaand formalised by Dobzhansky (1946) in D. pseudoobscura, it was repurposed as a cancer-therapy strategy by Hartwell, Friend and colleagues (Science 1997): if a tumour has lost gene A, then a drug inhibiting the synthetic-lethal partner B will kill tumour cells specifically while sparing normal A⁺ tissue. The therapeutic window is the genotype difference between tumour and host.

The flagship clinical realisation is the BRCA1/2–PARP1 pair. Two papers published back-to-back in Nature in 2005 made the prediction: Bryant et al. (Helleday lab) and Farmer et al. (Ashworth/Lord lab) independently showed that BRCA1- or BRCA2-deficient cells were ~1000-fold more sensitive to small-molecule PARP inhibitors than wild-type cells. The clinical proof followed in Fong et al., NEJM 2009 (olaparib phase I, dramatic responses concentrated in BRCA-mutant carriers); olaparib was approved by the EMA and FDA in 2014 for germline-BRCA-mutant ovarian cancer, the first synthetic-lethal drug in oncology.

The mechanism, in three steps

1. PARP1 inhibition prevents BER/SSBR.Without PARP1 activity at SSBs, single-strand breaks accumulate. When a replication fork encounters an unrepaired SSB, the fork collapses into a one-ended double-strand break — the obligate substrate of HR.
2. Trapped PARP1 is the cytotoxic species.Clinical PARP inhibitors don’t simply remove PARP enzymatic activity; they catalyse retention of PARP1 on DNA (allosteric stabilisation of the DNA-bound state). A PARP1–DNA roadblock encountered by a fork is far more toxic than enzymatic loss alone, which is why trapping potency — not enzymatic IC₅₀ — predicts cytotoxicity. Talazoparib has the highest trapping activity and the lowest clinical dose; veliparib has weak trapping and is essentially a catalytic inhibitor only.
3. BRCA-deficient cells cannot repair the resulting DSBs.In wild-type cells, fork-collapse DSBs are repaired by HR. In BRCA1- or BRCA2-deficient tumour cells, RAD51 cannot be loaded onto the resected end; NHEJ acts in S/G2 inappropriately, producing chromosome fusions and aneuploidy; MMEJ via Pol θ produces large microhomology-flanked deletions; and cells eventually die through mitotic catastrophe. Normal heterozygous BRCA1^+/− tissue retains one wild-type allele and is spared.

\[ \underbrace{\text{PARPi}}_{\text{trap PARP1}} \;\Rightarrow\; \text{unrepaired SSBs} \;\xrightarrow{\;\text{S phase}\;}\; \underbrace{\text{1-ended DSBs}}_{\text{HR substrate}} \;\xrightarrow[\;\text{BRCA1/2 loss}\;]{\;\text{HR fails}\;}\; \text{death} \]

BRCAness & the HRD score

BRCAness (Turner, Tutt & Ashworth, Nat Rev Cancer 2004) describes tumours that phenocopy BRCA1/2 loss without BRCA1/2 mutation: defects in PALB2, RAD51C/D, BARD1, BRIP1, ATM, ATR, FANCA-N, CHEK2, CDK12; promoter methylation of BRCA1 or RAD51C; or composite measures of genomic scarring. These tumours are also PARPi-sensitive, and the question ofhow to identify them clinically has driven the development of the HRD score.

The Myriad myChoice HRD score sums three genome-wide DNA-segmentation measures derived from SNP arrays or NGS:

\[ \text{HRD score} \;=\; \text{LOH} \;+\; \text{TAI} \;+\; \text{LST} \]

• LOH: number of loss-of-heterozygosity segments > 15 Mb but not whole-chromosome.
• TAI: number of telomeric allelic-imbalance regions extending to the sub-telomere.
• LST: number of large-scale state transitions (>10 Mb) between segments.

An HRD score ≥ 42 (or ≥ 33 in some companion-diagnostic settings) defines HRD-positive ovarian cancers, the population in which olaparib + bevacizumab maintenance gave a hazard ratio for progression of ~0.33 (PAOLA-1, Ray-Coquard et al., NEJM 2019). Tumour mutational signatures (SBS3, ID6) and the more recent HRDetect classifier (Davies et al., Nat Med 2017) are alternative WGS-based readouts.

Approved PARP inhibitors and their indications

Resistance

No targeted therapy escapes resistance, and PARPi resistance is now well characterised:

• Reversion mutations — secondary BRCA1/2 mutations restoring the reading frame and HR competence (Edwards & Ashworth, Nature 2008).
• Loss of 53BP1/Shieldin — restores resection and HR even in BRCA1-null cells (Bunting et al., Cell 2010).
• Loss of PARG — rescues residual PARylation signalling.
• Replication-fork stabilisation — loss of PTIP, MLL3/4, EZH2 protects nascent strands at reversed forks from MRE11.
• Drug efflux — ABCB1 (MDR1) upregulation.

POLQ inhibitors (targeting Pol θ-mediated MMEJ on which HR-deficient cells depend) and WRN inhibitors (synthetic-lethal in MSI-high cancers; Chan, Behan, Garnett, Nature 2019) are the next generation of synthetic-lethal targets, expanding the framework beyond PARP.

10. Therapeutic Implications

The repair pathways of Sections 2–8 are not merely tumour-suppressor biology — they are also the targets, and the determinants of response, of the great cytotoxic chemotherapies of the 20th century and the targeted DDR inhibitors of the 21st.

Lord & Ashworth (Nature 2012, Science 2017) articulated the modern view: the DNA-damage response is a network of partially redundant pathways whose deficiencies create predictable, druggable vulnerabilities. Stratton (Cell 2009) connected this network to mutational signatures, allowing the readout of repair-pathway status from any tumour’s genome. The clinical practice of oncology has now caught up: ovarian cancer is treated by HRD status, colorectal cancer by MMR status, GBM by MGMT methylation, and an expanding repertoire of synthetic-lethal pairs awaits clinical validation.

Key references. Stratton MR, Campbell PJ, Futreal PA. The cancer genome. Nature 458, 719 (2009). Lord CJ, Ashworth A. The DNA damage response and cancer therapy. Nature 481, 287 (2012). Lord CJ, Ashworth A. PARP inhibitors: synthetic lethality in the clinic. Science 355, 1152 (2017). Bryant HE et al. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature 434, 913 (2005). Farmer H et al. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434, 917 (2005). Fong PC et al.Inhibition of poly(ADP-ribose) polymerase in tumours from BRCA mutation carriers. NEJM 361, 123 (2009). Le DT et al. PD-1 blockade in tumours with mismatch-repair deficiency. NEJM 372, 2509 (2015). Helleday T. The underlying mechanism for the PARP and BRCA synthetic lethality. Mol Oncol 5, 387 (2011). Ceccaldi R et al. Homologous-recombination-deficient tumours are dependent on Polθ-mediated repair. Nature 518, 258 (2015). Murai J, Pommier Y. Trapping of PARP1 and PARP2 by clinical PARP inhibitors. Cancer Res 72, 5588 (2012). Davies H et al. HRDetect is a predictor of BRCA1 and BRCA2 deficiency. Nat Med 23, 517 (2017).