APOBEC Mutagenesis

APOBEC mutagenesis is a pervasive endogenous mutational process in mammalian genomes and human cancers driven by the APOBEC3 (apolipoprotein B mRNA-editing enzyme, catalytic polypeptide-like 3) family of zinc-dependent single-stranded DNA (ssDNA) cytidine deaminases clustered in a tandem array of seven genes (APOBEC3A, 3B, 3C, 3D, 3F, 3G, and 3H) on human chromosome 22q13.1. Evolved in placental mammals as an arm of cell-intrinsic innate immunity to restrict exogenous retroviruses (such as HIV-1), DNA viruses (HPV, HBV, EBV), and endogenous retrotransposons (LINE-1 and Alu elements) via lethal hypermutation of viral cDNA intermediates, APOBEC3A (A3A) and APOBEC3B (A3B) can gain aberrant access to nuclear genomic ssDNA exposed on the lagging-strand template during DNA replication, within transcription-coupled R-loops, or on resected 3' single-stranded overhangs during homologous recombination repair.

Chemically, APOBEC3A and APOBEC3B catalyze the hydrolytic deamination of deoxycytidine (dC) at its C4 exocyclic amine group to produce deoxyuridine (dU) specifically within 5'-TCW-3' trinucleotide motifs (where the mutated cytosine is preceded by a 5' thymine and followed by W = A or T, i.e., 5'-TCA-3' or 5'-TCT-3'). Processing of these genomic uracil lesions by the base-excision repair enzyme uracil-DNA glycosylase (UNG) and the error-prone translesion polymerase REV1 produces two paired mutational footprints across the human cancer genome: COSMIC Single-Base Substitution Signature 2 (SBS2, C→T transitions) and Signature 13 (SBS13, C→G and C→A transversions). Present in >50% of all human tumor types—and accounting for >60%–75% of the total somatic mutation burden in bladder, cervical, head and neck, breast, and lung carcinomas—APOBEC mutagenesis generates both genome-wide dispersed mutations (omikli) and dense, strand-coordinated hypermutation showers (kataegis) at chromosomal rearrangement breakpoints.

Active-Site Zinc Hydrolysis and Loop 7 Dinucleotide Recognition

Members of the AID/APOBEC superfamily share a conserved cytidine deaminase (CDA) core domain folded into a five-stranded mixed β-sheet surrounded by six α-helices, housing the canonical H-X-E-X_23–28-P-C-X_2–4-C catalytic zinc-binding motif. Within the active site of APOBEC3A (a single-domain enzyme, 199 amino acids) and the catalytically active C-terminal domain (CTD) of APOBEC3B (a double-domain enzyme, 382 amino acids):

1. Zinc Coordination: A catalytic Zn²⁺ ion is tetrahedrally coordinated by His70, Cys101, Cys106 (in A3A numbering), and an activated water molecule (H₂O / OH⁻).
2. Proton-Shuttle Hydrolytic Deamination (Glu72): The conserved Glu72 carboxylate side chain acts as a general acid–base proton shuttle. First, Glu72 abstracts a proton from the Zn²⁺-bound water molecule to generate a nucleophilic hydroxide ion (OH⁻) while donating a proton to the N3 ring nitrogen of the target deoxycytidine (dC). The Zn²⁺-stabilized hydroxide then attacks the electrophilic C4 carbon of the pyrimidine ring to form a tetrahedral carbinolamine intermediate. Finally, Glu72 shuttles a proton onto the departing amino group, releasing ammonia (NH₃) and leaving deoxyuridine (dU) in the ssDNA backbone.

Why do APOBEC3A and APOBEC3B strictly require single-stranded DNA (ssDNA) and selectively target 5'-TCW-3' motifs (whereas APOBEC3G targets 5'-CCC-3' and AID targets 5'-WRC-3')? High-resolution X-ray crystal structures of APOBEC3A and APOBEC3B-CTD bound to ssDNA hairpins solved by Reuben Harris, Hideki Aihara, and Shi (2017) demonstrated that double-stranded B-DNA is sterically excluded from the narrow active-site groove. Instead, APOBEC3A/3B force the single-stranded DNA substrate into a sharp U-shaped ("kinked" or hairpin) conformation, flipping the target dC nucleotide (position 0) and its immediate 5' neighbor -1 Thymine (5'-T) deep into adjacent specificity pockets. Loop 7 (residues 130–135, specifically Tyr130, Asp131, and Tyr132 in A3A) forms direct base-specific hydrogen bonds with the Watson–Crick edge (O2, N3-H, and O4) of the -1 Thymine, rejecting purines (A/G) and cytosine at the -1 position. Furthermore, APOBEC3A displays a ~10-fold higher catalytic turnover for 5'-YTCA-3' (Y = pyrimidine C or T) at the stem-loop apex of short 3–4 nt ssDNA hairpins, whereas APOBEC3B preferentially deaminates 5'-RTCA-3' (R = purine A or G) on extended linear ssDNA tracts.

Replication-Over-Uracil (SBS2) vs UNG/REV1 Translesion Synthesis (SBS13)

Once APOBEC3A or APOBEC3B converts cytosine to uracil (dU) on single-stranded genomic DNA (5'-TCW-3' → 5'-TUW-3'), the cell's response splits into two competing pathways that generate the coupled COSMIC SBS2 and SBS13 signatures discovered by Ludmil Alexandrov, Serena Nik-Zainal, and Michael Stratton (2013):

1. Pathway 1 — Direct Replication Over Uracil (SBS2, C→T): If the replication fork arrives before base-excision repair excises the uracil, high-fidelity replicative DNA polymerases (Pol δ or Pol ε) read dU as an instructional thymine analog and incorporate deoxyadenosine (dA) opposite dU. In the subsequent round of DNA replication, the dU:dA pair resolves into a permanent T:A (C→T transition) mutation (SBS2).
2. Pathway 2 — Uracil Excision by UNG and REV1 dCMP Transferase (SBS13, C→G): When nuclear uracil-DNA glycosylase (UNG2) cleaves the N-glycosidic bond of dU, it leaves a non-instructional apurinic/apyrimidinic (abasic, AP) site. If a replication fork stalls at this AP site on single-stranded DNA before APE1 and Pol β can restore the duplex, the cell recruits the translesion synthesis (TLS) deoxycytidyl transferase REV1 (together with Pol ζ). Because REV1 uses an internal arginine template (Arg357) to selectively insert deoxycytidine (dC) opposite abasic sites ("the REV1 C-rule"), replication converts the original C:G pair into a G:C (C→G transversion) (or, when Pol η/Pol ι inserts dT, a C→A transversion), generating SBS13.

Lagging-Strand Asymmetry and Kataegis at Double-Strand Breaks

Unlike ultraviolet (UV) pyrimidine dimers (SBS7) or tobacco benzo[a]pyrene adducts (SBS4), which exhibit transcriptional strand bias due to transcription-coupled nucleotide excision repair (TC-NER), APOBEC mutations (SBS2/SBS13) exhibit strong replicative strand asymmetry. As established by Dmitry Gordenin, Roberts, and Seplyarskiy, discontinuous synthesis of Okazaki fragments on the lagging-strand template transiently exposes ~100–200 nt stretches of unshielded single-stranded DNA between the CMG helicase and Pol α/Pol δ, resulting in a 1.4- to 2.2-fold enrichment of 5'-TCW-3' C→T/G mutations on the lagging-strand template relative to the leading-strand template.

In addition to dispersed lagging-strand mutations, cancer genomes frequently harbor kataegis (from the Greek for "thunderstorm")—localized hypermutation showers comprising 5 to >100 mutations spaced 10 to 1,000 bp apart where every mutated cytosine resides on the exact same DNA strand (processive strand coordination). As proved by Peter Campbell, Serena Nik-Zainal, and Maciejowski, kataegis occurs when chromosomal double-strand breaks (DSBs)—especially during chromothripsis or telomere crisis dicentric bridge breakage—undergo 5'→3' end resection by EXO1 and DNA2 helicase/nuclease, exposing kilobase-length 3' single-stranded DNA overhangs that are processively deaminated by APOBEC3A or APOBEC3B before double-strand break rejoining.

In human tumors, episodic bursts of APOBEC3A/3B expression act as a double-edged sword for tumor evolution:

• Oncogenic Driver Mutations (PIK3CA E542K / E545K): In cervical (HPV-positive, where viral E6/E7 oncoproteins directly induce APOBEC3A/3B transcription), head and neck (HNSCC), bladder, and estrogen-receptor-positive breast cancers, APOBEC activity generates recurrent hotspot driver mutations residing within canonical 5'-TCW-3' motifs—most notably the helical-domain activating mutations Glu542Lys (c.1624G>A, 5'-TCA-3' on the antisense strand) and Glu545Lys (c.1633G>A, 5'-TCT-3') in the phosphatidylinositol 3-kinase catalytic subunit PIK3CA.
• Subclonal Heterogeneity, Drug Resistance, and Neoantigen Load: Ongoing subclonal APOBEC mutagenesis elevates intratumoral genetic diversity (quantified via subclonal allele frequency Jensen–Shannon divergence, JSD) and fuels acquired resistance to targeted therapies (such as tamoxifen, CDK4/6 inhibitors, and EGFR tyrosine kinase inhibitors like osimertinib). Conversely, high SBS2/SBS13 tumor mutational burden generates abundant MHC Class I neoantigens (hydrophobic amino acid substitutions), sensitizing APOBEC-high bladder, lung, and head-and-neck tumors to PD-1 / PD-L1 immune checkpoint blockade, while making APOBEC-active tumor cells synthetically lethal to ataxia telangiectasia and Rad3-related (ATR) kinase inhibitors that collapse replication forks stalled at UNG-generated abasic sites.

Why does APOBEC mutagenesis always produce both C→T transitions (SBS2) and C→G transversions (SBS13) at the exact same 5'-TCW-3' motifs?

Both SBS2 and SBS13 originate from the exact same primary biochemical lesion: deamination of cytosine to uracil (dU) within a single-stranded 5'-TCW-3' motif (W = A or T) by APOBEC3A or APOBEC3B. If the replication fork copies directly past the unexcised uracil, replicative polymerase incorporates adenine opposite dU, fixing a C→T transition (SBS2). If uracil-DNA glycosylase (UNG) excises the uracil base first, it leaves an abasic (AP) site; during translesion synthesis across that AP site, the deoxycytidyl transferase REV1 inserts a cytosine opposite the abasic lesion, fixing a C→G transversion (SBS13).

Why do individuals homozygous for the germline APOBEC3B deletion paradoxically have higher APOBEC mutation burdens in breast cancer?

A common 29.5-kb germline deletion (ΔAPOBEC3B, allele frequency ~37% in East Asians and ~90% in Oceanic populations) deletes the entire coding sequence of APOBEC3B on Chr22q13.1. However, the deletion breakpoint fuses the protein-coding region of APOBEC3A directly to the stable 3' untranslated region (3'-UTR) of APOBEC3B. Because this chimeric APOBEC3A–3B mRNA escapes microRNA-mediated suppression and APOBEC3A is a ~10- to 50-fold more potent nuclear ssDNA mutator than APOBEC3B, carriers of the deletion experience stronger episodic APOBEC3A hypermutation (5'-YTCA-3' hairpin bias) and a higher incidence of SBS2/SBS13-driven breast cancer.

How is replication strand bias of APOBEC mutagenesis different from transcriptional strand bias of UV or smoking mutations?

Bulky DNA lesions caused by UV light (SBS7) or tobacco polycyclic aromatic hydrocarbons (SBS4) are repaired preferentially on the transcribed template strand by transcription-coupled nucleotide excision repair (TC-NER), creating a transcriptional strand bias. By contrast, APOBEC3A and APOBEC3B cannot deaminate double-stranded DNA; they require transient single-stranded DNA (ssDNA). During DNA replication, discontinuous Okazaki fragment synthesis leaves the lagging-strand template exposed as ssDNA much longer than the leading-strand template, causing APOBEC C→T and C→G mutations to accumulate preferentially on the lagging-strand template.

1. Nik-Zainal, S., Alexandrov, L. B., Wedge, D. C., et al. (2012). "Mutational processes molding the genomes of 21 breast cancers." Cell, 149(5), 979–993. https://doi.org/10.1016/j.cell.2012.04.024
2. Burns, M. B., Lackey, L., Carpenter, M. A., et al. (2013). "APOBEC3B is an enzymatic source of mutation in breast cancer." Nature, 494(7437), 366–370. https://doi.org/10.1038/nature11881
3. Roberts, S. A., Lawrence, M. S., Klimczak, L. J., et al. (2013). "An APOBEC cytidine deaminase mutagenesis pattern is widespread in human cancers." Nature Genetics, 45(9), 970–976. https://doi.org/10.1038/ng.2702
4. Shi, K., Carpenter, M. A., Banerjee, S., et al. (2017). "Structural basis for single-stranded DNA recognition and deamination by APOBEC3A and APOBEC3B." Nature Structural & Molecular Biology, 24(2), 131–139. https://doi.org/10.1038/nsmb.3344