Molecular origins of mutational spectra produced by the environmental carcinogen N-nitrosodimethylamine and SN1 chemotherapeutic agents

Abstract

DNA-methylating environmental carcinogens such as N-nitrosodimethylamine (NDMA) and certain alkylators used in chemotherapy form O ⁶-methylguanine (m6G) as a functionally critical intermediate. NDMA is a multi-organ carcinogen found in contaminated water, polluted air, preserved foods, tobacco products, and many pharmaceuticals. Only ten weeks after exposure to NDMA, neonatally-treated mice experienced elevated mutation frequencies in liver, lung and kidney of ∼35-fold, 4-fold and 2-fold, respectively. High-resolution mutational spectra (HRMS) of liver and lung revealed distinctive patterns dominated by GC→AT mutations in 5'-Pu-G-3' contexts, very similar to human COSMIC mutational signature SBS11. Commonly associated with alkylation damage, SBS11 appears in cancers treated with the DNA alkylator temozolomide (TMZ). When cells derived from the mice were treated with TMZ, N-methyl-N-nitrosourea, and streptozotocin (two other therapeutic methylating agents), all displayed NDMA-like HRMS, indicating mechanistically convergent mutational processes. The role of m6G in shaping the mutational spectrum of NDMA was probed by removing MGMT, the main cellular defense against m6G. MGMT-deficient mice displayed a strikingly enhanced mutant frequency, but identical HRMS, indicating that the mutational properties of these alkylators is likely owed to sequence-specific DNA binding. In sum, the HRMS of m6G-forming agents constitute an early-onset biomarker of exposure to DNA methylating carcinogens and drugs.

Graphical Abstract

Distinct DNA methylating agents, including N-nitrosodimethylamine and temozolomide generate nearly identical mutational spectra in repair proficient and deficient mice, and in cell culture, suggesting a convergent chemical mechanism and mutational process.

Figure 1.

Experimental workflow from toxicant exposure to high-resolution mutational spectra (HRMS). (A) Compounds evaluated for mutagenic properties include N-nitrosodimethylamine (NDMA), N-methyl-N-nitrosourea (MNU), streptozotocin (STZ) and temozolomide (TMZ), all of which form a putative methyldiazonium ion (B) prior to reaction with DNA to form adducts such as O⁶-methylguanine (m6G) (C), a mutagenic adduct that mispairs with thymine during DNA synthesis. Portions of MNU and STZ are differentially colored in blue to show shared structural features, and the red functional group on each structure denotes the methyl group transferred to DNA. (D) Neonatal C57BL/6J-gptΔ mice were treated with NDMA on days 8 and 15 post-birth for a total dose of 10.5 mg/kg. Lung, liver and kidney were harvested 10 weeks following the second injection. In some experiments, derivatives of the C57BL/6J-gptΔ mouse were used, where the RaDR locus was incorporated with or without the Mgmt gene, which encodes a protein that repairs m6G. (E) In parallel, mouse embryo fibroblasts (MEFs) generated from C57BL/6J- gptΔ mice were treated with the direct-acting mutagens, MNU, STZ or TMZ, as indicated. DNA isolated from these samples was analyzed: (F) for point mutations in the gpt gene, or (G) subjected to duplex sequencing to generate an HRMS for each agent.

Figure 2.

Organ-specific mutagenicity of NDMA and HRMS of NDMA-treated livers and lungs. (A–C) NDMA induced mutant frequencies of the liver, lung and kidney in gptΔ C57BL/6J wild-type mice. (D) The proportion of single-nucleotide substitution mutations in the liver and lung was measured directly by using duplex sequencing (DS). (E) DS revealed the NDMA-induced high resolution mutational spectrum in liver and lung. The spectrum was dominated by GC→AT mutations with a lower frequency of AT→GC transitions. The sequencing data were plotted as the mutant base (X) accompanied by its 5’ and 3’ neighbors (N); that is, 5’-NXN-3’. There are 16 combinations for each of the six types of base substitution mutations, resulting in a total of 96 possible triplex contexts. The data shown are averaged from the livers of five males and five females and the lungs of two males and two females treated with NDMA. (F) The probability LOGO (pLOGO) was generated from all 15-base pair sequence contexts adjacent to the mutated base, GC→AT. Guanine in gray highlight represents the fixed G position. For panels A-C, statistical comparisons were done with Mann–Whitney U-test, *P = 0.0159, **P = 0.0079. For panel E, bars reflect averages, error bars denote 1 SD. For panel F, the 15-base sequence contexts with G→A mutations fixed at the zero position were extracted and from all datasets. Shown is the compilation of all sequence contexts. Inter-individual replicate sequences were included in the analysis. The red bar (log-odds value of ± 3.05) indicates the P = 0.05 statistically significant threshold following Bonferroni correction. The foreground sequences (fg) represent NDMA-induced mutations (liver n(fg) = 3751; lung n(fg) = 3451). The background sequences (bg) represent the genome sequenced (liver n(bg) = 13 041; lung n(bg) = 10 437).

Figure 3.

Macroscopic and histological changes at 10 weeks post NDMA treatment. (A) Neonatal gptΔ C57BL/6J mice were administered a total dose of 10.5 mg/kg NDMA split between day 8 (1/3 dose) and day 15 (2/3 dose). Samples were collected 10 weeks post-NDMA treatment. (B–D) Organ to body weight ratios. (E–J) Representative hematoxylin and eosin (H&E) sections of liver and (K–M) histopathology scores from male and female C57BL/6J gptΔ mice treated with saline (control) or NDMA as described in material and methods. (F) NDMA-treated male mice exhibited areas of periacinar hepatocellular degeneration (swelling) with or without karyomegaly (arrowheads) and (G) bizarre mitotic figures (arrows). (I) Sections of the liver from an NDMA-treated female mice contained multifocal areas of hepatocellular degeneration with or without karyomegaly (arrowheads), bizarre mitoses (J, arrow), and lymphocytic and histiocytic infiltrates (open arrow). Original magnification x400, scale bars = 50 μm. Whitney U-test, *P = 0.0159, **P = 0.0079.

Figure 4.

NDMA-induced mutations and mutational spectra in livers of MGMT-deficient mice. (A) The biochemical mechanism of MGMT, the repair protein for m6G. NDMA generates an electrophile in vivo that methylates DNA to form, among other DNA lesions, m6G. A nucleophilic cysteine thiol residue on MGMT attacks the methyl group of m6G, resulting in a methylated MGMT protein and an undamaged guanine. (B) Mgmt^−/− gptΔ C57BL/6J mice were administered NDMA (10.5 mg/kg total dose) using the regimen of Figure 1D, and liver samples were collected 10 weeks following treatment. NDMA-induced point mutant frequencies in the livers of male and female MGMT-deficient and -proficient mice are shown. (C) The proportion of single-nucleotide mutations in livers of saline control and NDMA-treated Mgmt^−/− mice measured directly by using DS. (D) The pLOGO analysis produced from all 15-base pair sequence contexts adjacent to the mutated base, GC→AT. Guanine in gray highlight represents the fixed G position. (E, F) HRMS from the livers of three male and three female MGMT-deficient mice treated with NDMA. For panel B, statistical comparisons were done with the Mann–Whitney U-test, *P = 0.0159, **P = 0.0079. For panel D, the 15-base sequence contexts with G→A mutations fixed at the zero position were extracted from all datasets. Shown is the compilation of all sequence contexts. Inter-individual replicate sequences were included in the analysis. The red bar (log-odds value of ±3.05) indicates the P = 0.05 statistically significant threshold following Bonferroni correction. The foreground sequences (fg) represent NDMA-induced mutations (liver n(fg) = 52 253). The background sequences (bg) represent the genome sequenced (liver n(bg) = 10 437). For panels E and F, bars reflect averages (n = 3 for each group), error bars denote 1 SD.

Figure 5.

Cosine similarity matrix of the mutational spectra of NDMA treated WT liver (NDMA-WT-liver), WT lung (NDMA-WT-lung) and MGMT-deficient mice (NDMA-Mgmt); MEFs treated with temozolomide (TMZ), N-methyl-N-nitrosourea (MNU), and streptozotocin (STZ); and human mutational signature SBS11. Before performing the cosine similarity comparisons, all mutational spectra were baseline corrected by subtracting the corresponding background (vehicle-treated) spectrum, and then normalized to reflect mutation frequency per trinucleotide.

Figure 6.

HRMS of MEFs treated with alkylating agents MNU, STZ and TMZ. The plots depict the HRMS data and their dose-dependent mutagenicities at the gpt locus (insets) for MNU (A), STZ (B) and TMZ (C). (D) COSMIC mutational signature SBS11 of human cancer.