High incidence of epithelial cancers in mice deficient for DNA polymerase delta proofreading

Abstract

Mutations are a hallmark of cancer. Normal cells minimize spontaneous mutations through the combined actions of polymerase base selectivity, 3' --> 5' exonucleolytic proofreading, mismatch correction, and DNA damage repair. To determine the consequences of defective proofreading in mammals, we created mice with a point mutation (D400A) in the proofreading domain of DNA polymerase delta (poldelta, encoded by the Pold1 gene). We show that this mutation inactivates the 3' --> 5' exonuclease of poldelta and causes a mutator and cancer phenotype in a recessive manner. By 18 months of age, 94% of homozygous Pold1(D400A/D400A) mice developed cancer and died (median survival = 10 months). In contrast, only 3-4% of Pold1(+/D400A) and Pold1(+/+) mice developed cancer in this time frame. Of the 66 tumors arising in 49 Pold1(D400A/D400A) mice, 40 were epithelial in origin (carcinomas), 24 were mesenchymal (lymphomas and sarcomas), and two were composite (teratomas); one-third of these animals developed tumors in more than one tissue. Skin squamous cell carcinoma was the most common tumor type, occurring in 60% of all Pold1(D400A/D400A) mice and in 90% of those surviving beyond 8 months of age. These data show that poldelta proofreading suppresses spontaneous tumor development and strongly suggest that unrepaired DNA polymerase errors contribute to carcinogenesis. Mice deficient in poldelta proofreading provide a tractable model to study mechanisms of epithelial tumorigenesis initiated by a mutator phenotype.

Fig 1.

Generation of Pold1^D400A mice. (A Left) Structural conservation of the Exo I, II, and III motifs from Klenow (gray), T7 (blue), T4 (purple), RB69 (red), Tgo (green), and D. Tok (yellow) DNA polymerases (Protein Data Bank ID codes , , , , , and , respectively). Cα atoms were superimposed by using SWISS-PDBVIEWER 3.5 (www.expasy.ch/spdbv/mainpage.htm). Catalytic carboxylate side chains and metal ions (yellow dots) are shown. (Right) Alignment of Exo II amino acid sequences. Arrows, conserved aspartic acid residue mutated in this report. (B) Mouse Pold1 gene (Upper) and the vector used for targeted homologous recombination (Lower). (Scale bar = 1 kb.) Boxes, exons; lines, introns; white X, D400A point mutation; NEO, neomycin-resistance gene; TK1 and TK2, thymidine kinase genes; L1 and L2, PCR primers; Probe, Southern blot probe. (C) Confirmation of gene targeting by Southern blot (Upper) and PCR (Lower) analyses. Lanes 1–3, targeted ES cell clones; wt, WT ES clone; −, no DNA control; +, WT Pold1 plasmid control; m, λ HindIII marker; mut, Pold1^D400A mutant allele. (D) Sequencing of RT-PCR products from heterozygous Pold1^+/D400A ES cells. (Middle) Direct sequencing of unfractionated RT-PCR products. (Top and Bottom) Sequencing of WT and Pold1^D400A alleles after subcloning.

Fig 2.

Survival of Pold1^D400A mice. (A) Kaplan–Meier survival estimates for F₂ generation mice (53 Pold1^+/+, 97 Pold1^+/D400A, and 49 Pold1^D400A/D400A). Pold1^D400A/D400A mouse survival was significantly reduced (P < 0.0001 by log-rank and Wilcoxon tests) compared with Pold1^+/+ and Pold1^+/D400A mice, which were statistically indistinguishable from each other (P > 0.2). The major types of tumors observed in Pold1^D400A/D400A mice and their time of appearance are indicated at the bottom. Gray triangles mark the median age of death caused by each tumor type (except skin tumors that were not lethal). Lung CA, lung adenocarcinomas; others include diffuse abdominal lymphomas, sarcomas, and teratomas (see Table 1). (B) Kaplan–Meier survival estimates for F₁ generation mice (100 Pold1^+/+ and 80 Pold1^+/D400A). The two curves are statistically equivalent (P > 0.2). Statistical analyses were performed by using JMP IN 3.2.1 software (SAS Institute, Cary, NC).

Fig 3.

Tumors in Pold1^D400A/D400A mice. (A) Mediastinal thymic lymphoma (TL). H, heart; L, liver. Lungs are collapsed and not visible. (Right) Hematoxylin and eosin (H&E)-stained section of thymic lymphoma (original magnification, ×400). White arrows, nucleoli; black arrows, mitotic figures. (B) Tail with marked irregularities throughout its length and a focal skin tumor (arrow). (Right) Section of skin tumor stained with H&E (original magnification, ×200). KP, keratin pearl; arrow, mitotic figure. (C) Lung adenocarcinoma (arrow). (Right) H&E-stained section showing the tumor (arrowheads) compressing the normal lung parenchyma (original magnification, ×100).

Fig 4.

3′ → 5′ exonuclease activities of polδ p125 proteins in vitro. (A) Activities on single-stranded DNA. Recombinant WT (•) or D400A mutant (▪) polδ p125 proteins were incubated with single-stranded (dT)₃₃[³H]dT_1–3 substrate, and the release of [³H]dTMP from 3′ termini was measured and plotted as a function of polymerase activity. Polymerase units were determined on poly(dA-dT) as described in Materials and Methods. (B) Activities on double-stranded DNA. Recombinant WT (lanes 2 and 4–7) or D400A mutant (lanes 3 and 8–11) polδ p125 proteins were incubated for 5 min (lanes 1–4, 6, 8, and 10) or 10 min (lanes 5, 7, 9, and 11) with double-stranded [5′-³²P]20-mer⋅40-mer substrate, and products were analyzed by urea-PAGE and PhosphorImaging. Incubations contained comparable amounts of WT and mutant enzymes as indicated by the similar yields of +1 product after 5-min incubations in the presence of 500 μM dCTP (lanes 2 and 3). Either Mg²⁺ (lanes 1–5, 8, and 9) or Mn²⁺ (lanes 6, 7, 10, and 11) was included as divalent metal activator. A schematic of the [5′-³²P]20-mer⋅40-mer substrate is shown (Left). 3′ → 5′ exonucleolytic cleavage products are indicated as −1, −2, −3, and −4. Products down to −6 were also observed in the incubations with WT enzyme and Mn²⁺ (data not shown). Lane 1, control incubation (5 min) lacking polδ p125 protein.