Sex and base excision repair as key susceptibility factors for alkylation-induced mutations, toxicity and hepatic cancer

Abstract

N-Nitrosodimethylamine (NDMA) is present in food, water, and drugs and is considered a probable human carcinogen by the International Agency for Research on Cancer. The mechanism of action of NDMA involves the generation of carcinogenic methyl lesions such as 3-methyladenine (3MeA) on DNA bases. Alkyladenine DNA Glycosylase (AAG; a.k.a. N-methylpurine DNA glycosylase, MPG) removes 3MeA to initiate Base Excision Repair, leaving an abasic site that is resolved by backbone cleavage, nucleotide insertion, and backbone ligation. The intermediate steps following base removal produce potentially toxic and mutagenic abasic sites and single-strand DNA breaks. Here, we explored differences between males and females regarding downstream DNA damage, toxicity, mutations and cancer arising from 3MeA in the livers of WT, Aag^−/−, and Aag-overexpressing (AagTg) mice. We found that males were more susceptible to NDMA-induced mutations (WT and Aag^−/−) and cancer (all genotypes). In contrast, AagTg females were more prone to micronucleus induction. As we showed in our prior analyses where data were pooled for males and females, Aag^−/− mice were significantly more susceptible to NDMA-induced mutations and cancer, and AagTg mice displayed significantly greater toxicity. Building on these findings, our analyses of sex-related differences show that Aag deficiency and maleness are both susceptibility factors for NDMA-induced hepatic cancer, while Aag overexpression drives toxicity, potentially with a greater effect in females. This study reveals a deeper understanding of the underpinnings for a well-known increased risk of hepatic cancer in men versus women by demonstrating a higher susceptibility of male mice to both mutations and cancer.

1. Introduction

N-Nitrosodimethylamine (NDMA) is a potent carcinogen in animal models and is a Group 2 A probable human carcinogen according to the International Agency for Research on Cancer (IARC) [1]. Humans are exposed to NDMA through widespread sources. NDMA can be found in drinking water as a consequence of commonly used water treatment processes [2–6], as well as from ineffectively contained hazardous waste [7–10]. NDMA can also be found in many foods, including processed meats, which are considered to be carcinogenic according to IARC [7–10]. Recently, NDMA has been found as a contaminant in numerous drugs that have since been recalled [11–15]. It is therefore clear that NDMA presents a significant public health concern. Here, we explored differences in susceptibility to NDMA-induced DNA and tissue damage, mutations, and hepatic cancer for males versus females.

The main mode of action for NDMA is through the induction of mutagenic DNA lesions that ultimately drive cancer. Metabolism of NDMA via CYP2E1 generates the highly unstable methyldiazonium ion, which readily adds methyl groups on DNA bases, producing lesions such as 3-methyladenine (3MeA) [7]. In animal models, the risk of cancer for different tissues reflects their CYP2E1 metabolic capacity, and CYP2E1 is expressed at the highest levels in the liver [16]. We recently demonstrated the mutagenic potential of 3MeA in vivo [17]. To defend against 3MeA-driven mutations, 3MeA is removed by the Alkyladenine DNA Glycosylase (AAG; a.k.a., N-methylpurine glycosylase, MPG) [18], which initiates base excision repair (BER). Following excision of 3MeA by AAG, the resulting abasic site is processed by APE1 to create a nick in the DNA backbone 5’ to the lesion. In short-patch BER, DNA polymerase β removes the abasic ribose and inserts a nucleotide. Finally, the nick is sealed by DNA ligase III. The BER process is highly coordinated to minimize the presence of toxic BER intermediates, such as requisite single-strand breaks [19].

Unrepaired 3MeA can block DNA replicative polymerases, leading to replication fork stalling and toxicity [20–22]. However, in vitro studies have demonstrated that some translesion synthesis (TLS) polymerases can bypass 3MeA [22–25], and our prior work showed that 3MeA is a point mutagen in vivo [17]. An alternative pathway for coping with replication-blocking lesions is to invoke homologous recombination (HR). For example, if the leading strand is stalled, DNA replication can continue by using the newly synthesized lagging strand as a template, and vice versa [20]. While HR is mostly error-free, misalignments can lead to insertions, deletions, and translocations. Consistent with the ability for replication blockage by 3MeA to cause point mutations and larger scale sequence rearrangements, AAG-deficient mice unable to repair 3MeA are highly susceptible to NDMA-induced liver cancer [17]. Conversely, mice overexpressing AAG have reduced levels of NDMA-induced cancer compared to wild-type (WT) mice [17]. However, Aag-overexpressing mice also suffer increased toxicity, presumably due to single strand breaks that are formed during BER. Thus, the fate of the cell following alkylation exposure depends on a balance between efficient AAG excision and coupled with the downstream BER machinery’s ability to efficiently resolve DNA strand breaks.

Interestingly, the mutagenic and carcinogenic consequences of alkylating exposures depend not only on the levels of DNA repair enzymes, but also on sex. Generally, the incidence and mortality rate of non-reproductive cancers are elevated in males compared to females [26,27]. Indeed, the occurrence of hepatocellular carcinoma is 2–3 times higher in males versus females [27]. Though lifestyle plays an intricate role in the elevated rates of cancer in men, intrinsic factors such as sex hormones greatly influence carcinogenesis, especially in a sexually dimorphic organ such as the liver. Sex also influences the tumor-genicity of environmental carcinogens [28]. In the context of nitrosamine exposure, several studies have demonstrated that females develop fewer liver tumors than males [29–32].

We previously interrogated the biological consequences of NDMA exposure in mice lacking AAG expression (Aag^−/−) or overexpressing AAG (Aag transgenic; AagTg), which are prone to accumulating either 3MeA lesions or BER intermediates, respectively. We analyzed a suite of endpoints spanning the days immediately after exposure up to many months later. By integrating analyses of DNA and tissue damage, animal lethality, mutations, and cancer, we were able to track the progression from initial exposure to frank disease. We found that unrepaired 3MeA (as in Aag^−/− mice) is somewhat toxic but highly mutagenic, whereas excess strand breaks (as in AagTg mice) are poorly mutagenic and highly toxic [17]. In that study, we pooled results from males and females to specifically analyze the biological impacts of AAG expression. Here, we have taken that research a step further by revealing sex-specific differences between males and females. Males were generally more susceptible to NDMA-induced long-term liver damage and cancer in all genotypes, consistent with prior work [29–32]. Concordant with their increased susceptibility to cancer, males were also more vulnerable to NDMA-induced sequence rearrangement mutations. These effects were primarily elevated in Aag^−/− mice and suppressed in AagTg mice compared to WT. Surprisingly, males were not only more susceptible to NDMA-induced DNA sequence rearrangements, but also spontaneous rearrangements in controls for all genotypes. These studies reinforce a model wherein physiological conditions favoring mutations and disfavoring toxicity are highly carcinogenic, while physiological conditions that are less favorable to mutations and more favorable to toxicity are associated with suppression of cancer.

2. Materials and methods

The data presented here aside from RNA sequencing are derived from male and female pooled results presented in our previous publication Kay et al. [17], which contains details regarding materials and methods. Here, we briefly summarize previously published experimental approaches and point the reader to our previous manuscript for details. RNA sequencing is described in full.

2.1. Animals

The RaDR^R/R;gpt^g/g, RaDR^R/R;gpt^g/g;Aag^−/−, and RaDR^R/R;gpt^g/g; AagTg mice utilized in this study were on a C57BL/6 background [17]. Standard chow and drinking water were provided ad libitum in an AAALAC-certified animal care facility. All animal procedures were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and approved by the Massachusetts Institute of Technology Committee on Animal Care.

2.2. NDMA treatment

NDMA used in this study was synthesized in-house and characterized as described by He et al. [33]. It was stored in sealed ampoules under argon at −20°C until use. Mice were administered two intraperitoneal injections of NDMA diluted in saline, 3.5 mg/kg NDMA at postnatal day 8 and 7 mg/kg NDMA at postnatal day 15 for a total dose of 10.5 mg/kg NDMA as previously described [17]. Control mice were treated with equal volumes of saline at the respective timepoints. Depending on the assay, mice were euthanized and necropsied at either 24 h, 48 h, 10 weeks, or 9–11 months post-second injection (average 10.2–10.7 months for saline groups, 10.0–10.3 months for NDMA groups). At the time of necropsy, all visible macroscopic surface lesions on the liver were enumerated. Timepoint details are included in the figure legends.

2.3. RNA sequencing

We performed RNA sequencing on liver samples collected from mice 24 h after the second NDMA (or saline) injection. Total RNA was extracted from ≤ 30 mg mouse liver tissue stored in RNAlater at −20°C using the RNeasy Plus Mini Kit (Qiagen, cat. no. 74134). Tissue was homogenized in Buffer RLT Plus with β-mercaptoethanol (10 μL/ml) using a BeadBug bead-milling homogenizer at 400 rpm for two 30-second intervals. Lysates were centrifuged, and supernatant was processed through gDNA Eliminator and RNeasy spin columns according to the manufacturer’s protocol. RNA was eluted in 30–50 μL RNase-free water and quantified by NanoDrop spectrophotometry (Thermo Scientific).

2.3.1. Library preparation and sequencing

3’ Digital gene expression libraries were created as previously described37,38. RNA samples were cleaned using SPRI beads (1.8x) and quantified and quality assessed using an Advanced Analytical Fragment Analyzer. 10 ng of total RNA was used for library preparation. 3’DGE-custom primers 3V6NEXT-bmc#1–24 were added to a final concentration of 1 μM. (5′-/5Biosg/ACACTCTTTCCCTACACGACGCTCTTCCGATCT[BC6]N10T30VN–3′ where 5Biosg = 5’ biotin, [BC6] = 6 bp barcode specific to each sample/well, N10 = Unique Molecular Identifiers, Integrated DNA technologies), to generate two subpools of 24 samples each. After addition of the oligonucleotides, Maxima H Minus RT was added per manufacturer’s recommendations with Template-Switching oligo 5V6NEXT (10uM, [5V6NEXT: 5’-iCiGiCACACTCTTTCCCTACACGACGCrGrGrG–3’ where iC: iso-dC, iG: iso-dG, rG: RNA G]) followed by incubation at 42°C for 90 min and inactivation at 80°C for 10 min. Following the template switching reaction, cDNA from 24 wells containing unique well identifiers were pooled together and cleaned using RNA Ampure beads at 1.0x. cDNA was eluted with 17 μL of water followed by digestion with Exonuclease I at 37°C for 30 min and inactivated at 8 °C for 20 min.

Second strand synthesis and PCR amplification was done by adding the Advantage 2 Polymerase Mix (Clontech) and the SINGV6 primer (10 pmol, Integrated DNA Technologies 5’-/5Biosg/ACACTCTTTCCCTACACGACGC–3’) directly to the exonuclease reaction. 8 cycles of PCR were performed followed by clean up using regular SPRI beads at 0.6x and eluted with 20 μL of EB. Successful amplification of cDNA was confirmed using the Fragment Analyzer.

Illumina libraries were then produced using standard Nextera tagmentation substituting P5NEXTPT5-bmc primer (25 μM, Integrated DNA Technologies, (5’-AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCG*A*T*C*T*–3’) where * = phosphorothioate bonds) in place of the normal N500 primer. Final libraries were cleaned using SPRI beads at 0.7x and quantified using the Fragment Analyzer and qPCR before being loaded for paired-end sequencing using the Illumina NextSeq500 in paired-end mode (28 F,8I,50 R nt reads).

2.3.2. RNA data processing and analysis

Sequencing data were aligned to the mouse genome reference mm10 (GRCm38) using Spliced Transcripts Alignment to a Reference (STAR) aligner (RRID:SCR_004463), and gene expression was quantified using the Expression analysis of the digital expression (ESAT) pipeline. Differential expression was performed using DESeq2(median-of-ratios normalization)39. Log₂ fold changes were shrunk using lfcShrink to shrink dispersion40. Genes were filtered and adjusted with p-value < 0.05 and log2 fold change > 0.58. Mean counts of Aag, Ape1, Polβ, and Ligase III are shown in in Supplemental Table 1, and comparisons of Aag expression among each group are listed in Supplemental Table 2.

Expression analysis of the digital expression libraries targeting transcript ends was performed using the ESAT. ESAT takes a set of alignment files (SAM or BAM) with genome alignment coordinates, a file containing transcript coordinates (BED or text file) and outputs read counts for each transcript provided. ESAT can be found here on GitHub: (src/java/umms/esatJar/esat.latest.jar).

The package DESeq2 was used for differential expression analysis and can be found here: (http://www.bioconductor.org/help/work-flows/rnaseqGene/). DESeq2 uses negative binomial generalized linear models which uses estimates of dispersion and logarithmic fold changes to incorporate data-driven prior distributions. Lastly, normalization was performed. RNA-seq data have been deposited in the NCBI data repository Gene Expression Omnibus (GEO) under accession number GSE332646.

2.4. Sequence rearrangement mutation analysis (RaDR)

Sequence rearrangement mutations 10 weeks post-second injection were analyzed using the Rosa26 Direct Repeat (RaDR) transgenic mouse model [17]. Briefly, the left liver lobe was freshly excised from mice and put into ice-cold PBS with 0.01% trypsin inhibitor. The lobe was then placed on a slide and the dorsal surface was imaged with a 2x objective in the FITC channel. A machine learning algorithm was utilized to quantify the number of fluorescent foci within the lobe, and foci counts were normalized to the area of the lobe [17].

2.5. Point mutation analysis (GptΔ assay)

Liver samples from mice 10 weeks post-dosing were flash frozen in liquid nitrogen and stored at −80°C. λ-EG10 phage containing genomic DNA was extracted from the liver samples, then packaged and infected into Escherichia coli YG6020 [34]. These bacteria were cultured on chloramphenicol and 6-thioguanine selective media plates. Mutations in the gptΔ gene were measured by the number of 6-thioguanine resistant colonies.

2.6. γH2AX staining and analysis

Formalin-fixed paraffin-embedded (FFPE) livers from mice 24 h post-dosing were sectioned at 4-μm thickness prior to staining. Liver sections were deparaffinized, rehydrated, and subjected to heat-induced epitope retrieval (HIER) in citrate buffer (pH 6.0). After blocking, the sections were incubated with γH2AX antibody (1:200; Cell Signaling Technologies) overnight at 4°C. Washed sections were incubated with AlexaFluor 488 secondary antibody (1:400; Invitrogen) then mounted with ProLong Gold AntiFade Mountant with DAPI (ThermoFisher). Stained sections were imaged with a 60x objective under DAPI and FITC filters. Percentages of nuclei with 5 or more γH2AX foci, pan-nuclear γH2AX staining, and super-bright pan-nuclear staining were analyzed blinded with ImageJ [17].

2.7. Micronucleus (MN) assay

Livers collected from mice 48 h post-dosing were excised and immediately placed in ice-cold Liver Preservation Buffer, then shipped to Litron Laboratories for analysis with the Prototype In Vivo MicroFlow^® BASIC ML Kits (Litron Laboratories). Livers were processed for flow cytometry as described in Avlasevich et al., 2018 [35], and analyzed for SYTOX nuclear stain (488 nM excitation) and anti-Ki67-eFluo^R 660 (633 nM excitation) on a FACSCanto^™ II flow cytometer (BD Biosciences, San Jose, CA) [17]. Percent of micronucleated and proliferating hepatocytes were quantified as number of MN / number of nuclei × 100 and number of Ki67 + nuclei / number of nuclei × 100, respectively.

2.8. Cleaved caspase-3 staining and analysis

FFPE liver sections collected from 24 h post-dosing were prepared as in γH2AX Staining. Sections were deparaffinized, rehydrated, and subjected to HIER in Buffer H, pH 8.8 (Thermo Scientific). Slides were blocked with Peroxidazed 1 (Biocare Medical) then Background Sniper (Biocare Medical) prior to incubation with cleaved caspase-3 antibody (Cell Signaling Technology). Washed sections were incubated with Rabbit-on-Rodent HRP Polymer (Biocare Medical), then washed and incubated with DAB (Betazoid DAB Chromagen kit, Biocare Medical), then washed and counterstained with hematoxylin. Stained slides were scanned with a Leica Aperio AT2 slide scanning microscope at 40x and analyzed for percentage of apoptotic events using QuPath software version 0.2.1 [36].

2.9. Histological analysis

FFPE livers from mice necropsied 24 h and ~10 months post-dosing were sectioned at 4-μm thickness and stained with hematoxylin and eosin. Liver sections from male and female mice of each group were scored by a board-certified veterinary pathologist (SEC) blinded to the sample identity. To assess the progression of hepatic lesions in this model, a previously developed histological grading system was applied [17], which considered either the extent of lesions or the number of foci present across all examined hepatic lobes. Lesions were graded on a numerical scale from 0 to 4, where 0 = normal, 1 = minimal (< 5% / 1–5 foci), 2 = mild (5–20% / 6–12 foci), 3 = moderate (20–65% / 12–18 foci), and 4 = severe (> 65% / > 18 foci). Hepatocellular degeneration, stellate cell hyperplasia, and biliary hyperplasia were scored based on the extent of the lesion within hepatic lobes, whereas hepatocellular necrosis and inflammation were scored according to the number of foci observed. Foci of altered hepatocytes were classified according to the morphologic criteria described by Thoolen et al. [37] and enumerated. Hepatic tumors were classified based on established diagnostic histomorphological features [37]. Histological sections were examined using an Olympus BX41 microscope equipped with an iKona digital camera.

2.10. Quantification and statistical analyses

Statistical analysis was performed using GraphPad Prism software [17]. Tumor multiplicity, sequence rearrangement mutation frequencies, gpt mutant fractions, cleaved caspase–3 staining, and histopathology scores were compared by Mann-Whitney U-test. γH2AX staining, MN frequency, and Ki67 were compared with unpaired Student’s t-test. A P value was considered significant if less than 0.05.

3. Results

We previously studied the consequences of NDMA exposure on DNA damage, mutations, cell death, histopathological changes, and cancer in livers of WT, Aag-deficient (Aag^−/−), and Aag-overexpressing (AagTg) mice [17]. Mice were exposed twice to NDMA, once on postnatal day 8 (3.5 mg/kg) and again on day 15 (7 mg/kg), and we monitored carcinogenic, molecular, cellular, and physiological changes as they unfolded over time [17,38]. Here, we have expanded the results published in Kay et al. [17] by reporting on the impact of sex on NDMA-induced biological responses.

3.1. BER enzyme expression levels are similar among sexes

To distinguish the influences of sex and AAG on NDMA-induced outcomes, we performed RNA sequencing for Aag and other BER enzyme expression levels on livers collected from mice 24 h post-dosing. These analyses showed no significant differences in Aag expression between males and females for any genotype/exposure group (Supplemental Tables 1 and 2). Levels of APE1, Polβ, and Lig3 in males and females were within 45% of each other for each genotype/exposure group. We also confirmed increased Aag expression in AagTg mice and negligible expression in Aag knockouts (within range of assay noise). Although differences in mRNA expression do not necessarily translate to equivalent differences in protein levels, RNA sequencing provides at least some evidence that sex-specific differences in BER enzyme expression were not the primary driver of sex-related differences in NDMA-induced outcomes. Interestingly, we also found that Aag expression was reduced after NDMA exposure among AagTg mice only, with both males and females showing approximately 2-fold lower expression compared to saline-treated AagTg animals of the same sex.

3.2. Maleness and Aag deficiency increase risk of NDMA-induced hepatic tumors

Previously, we showed that Aag^−/− mice were significantly more prone to NDMA-induced liver tumors ~10 months post-exposure compared to WT, and that the overexpression of Aag suppressed hepatic tumorigenesis [17]. Upon classifying these data by sex, we consistently found that males were more prone to hepatic cancer than females. For WT mice exposed to NDMA, 80% of males necropsied developed macroscopically evident liver tumors, compared to just 46% of females (Table 1). Males also showed an increase in the number of NDMA-induced tumors per liver compared to females (p = 0.05; Table 1 and Fig. 1), suggesting that tumorigenesis may be accelerated in male versus female mice.

Table 1.

Hepatic tumors per mouse and cancer incidence in NDMA-treated mice 9–11 months post-injection (n = 11–15 per group).

Genotype	Sex	Median # of tumors/mouse	Incidence (% mice with tumors)
WT	M	1	12/15 (80%)
	F	0	6/13 (46%)
Aag−/−	M	8	13/13 (100%)
	F	1	9/11 (81%)
AagTg	M	1	8/12 (67%)
	F	0	2/12 (17%)

Fig. 1.

Liver tumors following NDMA exposure at 9–11 months post-injection. A) Data points represent the number of liver tumors per mouse. All saline-NDMA comparisons in mice of the same sex and genotype were statistically significant (p < 0.05), except for AagTg females; comparisons not shown for ease of reading. Lines show the median; each point represents one mouse (n = 11–15 per group). Mann-Whitney U-test: *p < 0.05. B) Representative low- and high-magnification (upper and lower panels, respectively) H&E-stained sections of hepatocellular lesions. In WT mice, the hepatic parenchyma is expanded by a hepatocellular carcinoma (trabecular subtype) in males and a hepatocellular adenoma (solid subtype) in females. Representative liver sections from Aag^−/− mice show a hepatocellular carcinoma (trabecular subtype) in males and a hepatocellular adenoma (solid subtype) in females. Livers from AagTg mice exhibit hepatocellular carcinoma (trabecular subtype) in males and foci of hepatocellular alteration admixed with oval cell hyperplasia in females. Low-magnification scale bars = 500 μm, high-magnification scale bars = 50 μm.

The sex-related difference in cancer susceptibility was most apparent in Aag^−/− mice, particularly when considering the number of tumors per liver. Males harbored significantly more macroscopic NDMA-induced hepatic tumors than females, with a median of 8 tumors per male mouse compared to just 1 tumor per female (Table 1 and Fig. 1).

Having previously shown in mixed-sex mice that AAG overexpression suppressed NDMA-induced hepatic cancer compared to WT [17], we went on to separate these data by sex. We found suppression of NDMA-induced tumors by AAG overexpression was far more evident in female versus male mice. Only 17% of NDMA-exposed AagTg females developed macroscopic tumors, compared to 67% of AagTg males (Table 1). Tumor counts were also significantly lower in AagTg females (median = 0) versus males (median = 1; Table 1 and Fig. 1). In fact, AagTg females were so resistant to liver cancer that the difference in tumor multiplicity between saline- and NDMA-treated AagTg females was not statistically significant, unlike all other saline-NDMA comparisons.

The uniformity we found in the combination of sex and genetics upon the biological responses to NDMA exposure was striking when we compared Aag^−/− and AagTg cohorts to WT mice. Consistent with the increased susceptibility to NDMA-induced liver tumors among males and among Aag^−/− mice, Aag^−/− males developed significantly more tumors compared to WT males, whereas the difference between tumor multiplicity in WT and Aag^−/− females did not reach statistical significance (p = 0.06) (Fig. 1). Similarly, tumor multiplicity was significantly reduced in NDMA-treated AagTg females compared to WT females, while this difference was not statistically significant for AagTg males versus WT males.

Microscopic analyses also demonstrated increased susceptibility to hepatic neoplasms for Aag^−/− mice and males, and comparative resistance among AagTg mice and females. It should be noted that not all livers were available for histopathologic examination, as portions of livers were allocated for other assays, so overall tumor burden and distribution of hepatocellular neoplasms across lobes could not be fully assessed by histopathology. Nevertheless, among NDMA-exposed mice with at least one liver lobe evaluated histopathologically, hepatocellular neoplasms were observed in 5 of 12 WT males and 2 of 12 WT females; 8 of 11 Aag^−/− males and 1 of 11 Aag^−/− females; and 1 of 12 AagTg males and none of the 13 AagTg females. The most common neoplasms identified in NDMA-exposed mice were hepatocellular adenomas and hepatocellular carcinomas. Representative images of neoplastic or pre-neoplastic lesions for each sex and genotype of NDMA-exposed mice are shown in Fig. 1B. None of the saline-treated control mice developed hepatocellular neoplasia at 10 months post-exposure.

Altogether, these data demonstrated sex bias in NDMA-induced tumors, with male mice displaying increased vulnerability irrespective of genotype.

3.3. Males are more sensitive to long-term NDMA-induced hepatocellular degeneration and hyperplasia

In addition to neoplasms, we evaluated other histopathological indicators of liver damage at ~10 months post-exposure. Hepatocellular degeneration was frequently observed (Fig. 2A), characterized by hypertrophic or swollen hepatocytes with cytoplasmic eosinophilia with or without karyomegaly. These changes are not specific to NDMA and can be caused by a variety of toxicants including other alkylating agents [39, 40]. NDMA significantly promoted degeneration in WT and AagTg mice compared to saline controls (Fig. 2A). Remarkably, male mice appeared to be more susceptible to this type of liver injury for all genotypes, though this comparison did not reach statistical significance for Aag^−/− (p = 0.09). These findings are consistent with prior literature pointing to livers of male mice being more susceptible to hepatic injury [27].

Fig. 2.

Histopathological scoring of hepatocellular changes at 9–11 months post-second NDMA exposure. A) Hepatocellular degeneration. B) Ito cell hyperplasia. C) Foci of hepatocellular alteration. Bars represent median scores from individual mice (n = 8–16 per group), calculated from the average of four independent regions on one slide. Mann-Whitney U-test: *p < 0.05.

Histopathological analyses of livers from mice ~10 months post-NDMA exposure also showed that males were significantly more susceptible than females to Ito cell hyperplasia across all genotypes (Fig. 2B). Interestingly, induction of Ito cell hyperplasia by NDMA was only statistically significant for WT males and AagTg males relative to saline controls. Ito hyperplasia refers to the proliferation of Ito cells (a.k. a., stellate cells), which can lead to increased fibrosis and ultimately contribute to severe liver disease [41,42]. Observations of hepatocellular degeneration and Ito cell hyperplasia clearly showed that males were more susceptible to liver damage than females across all three genotypes.

Finally, we analyzed foci of hepatocellular alterations, which are aggregates of hepatocytes with changed morphology. There is a positive correlation between hepatocellular alterations and cancer, and it is thought that such hepatocellular alterations precede tumor formation [43]. Aag^−/− male mice displayed the greatest increase in NDMA-induced hepatocellular alterations, which was significantly elevated in comparison to WT males (Fig. 2C). Consistent with the increase in susceptibility to liver tumor formation (Table 1 & Fig. 1), males were generally more susceptible to NDMA-induced hepatocellular alterations compared to females, though this was only statistically significant in Aag^−/− mice (Fig. 2C).

Altogether, tissue pathology at ~10 months post-NDMA exposure revealed enhanced liver damage in male mice compared to females, suggesting that hormonal differences may be an important mechanistic component underlying the observed sex bias in NDMA-induced liver cancer. Given that mutagenesis was strongly predictive of cancer induction [17], we evaluated male-female differences in mutation susceptibility.

3.4. Spontaneous and NDMA-induced mutations are elevated among males

Our experimental mice were homozygous for two transgenic reporter genes, RaDR and gptΔ, that allow for measurement of DNA sequence rearrangement and point mutations in mouse livers, respectively. Briefly, RaDR mice contain two expression cassettes for enhanced green fluorescent protein (EGFP) at the ubiquitously expressed Rosa26 locus, each lacking critical sequence information either on the 5’ end or the 3’ end of the cDNA such that the protein products do not fluoresce [44]. HR with misalignment between the repeats can restore full length sequence, leading to the expression of fluorescent EGFP and enabling quantification of de novo recombination events in situ with fluorescence microscopy. The gptΔ transgene contains λ phage that when mutated, packaged, and infected into E. coli confers resistance to 6-thioguanine [45,46]. The frequency of 6-thioguanine resistant-colonies reflects the number of mutant transgenes present in the tissue. Sequence rearrangement and point mutations are important drivers of cancer, and the RaDR^R/R;gpt^g/g transgenic mouse model uniquely enables measurement of both mutation types in the same tissue from a single animal.

We previously demonstrated that the RaDR substrate detects spontaneous recombination events in mouse liver [44], which we found to be elevated in Aag^−/− mice compared to WT at 10 weeks post-dosing [17]. Surprisingly, here we found that even in the absence of NDMA, rearrangement mutations in the saline-treated controls were significantly elevated in males compared to females. This was true across all three genotypes (Fig. 3A; significance markers not shown in figure for ease of reading).

Fig. 3.

Spontaneous and NDMA-induced mutations in livers at 10 weeks post-second injection. A) Spontaneous and NDMA-induced sequence rearrangement mutations measured as RaDR foci density. All male-female comparisons in saline-treated mice of the same genotype were statistically significant (p < 0.05); comparisons not shown for ease of reading. Bars show the median; each point represents one mouse (n = 12–15 per group). B) NDMA-induced point mutations measured in the GptΔ assay. Bars show the median; each point represents one mouse (n = 12–15 per group). Mann-Whitney U-test: *p < 0.05.

We also previously found that NDMA exposure significantly elevated the levels of mutagenic recombination events in the liver 10 weeks after exposure and that AAG activity was protective against these mutations [17]. Separating the results by sex revealed a strong male bias, with significantly higher frequencies of large-scale mutations in males versus females among NDMA-treated WT and Aag^−/− mice, but interestingly not for AagTg mice (Fig. 3A). We also found that the loss of AAG was more consequential for NDMA-induced HR-driven mutagenesis in females (Aag^−/− females versus WT females p < 0.0001) than in males (Aag^−/− males versus WT males p = 0.06; Fig. 3A). Consistent with our previous findings that Aag overexpression suppressed NDMA-induced sequence rearrangements, both AagTg males and females were significantly protected from NDMA-induced sequence rearrangements compared to WT (Fig. 3A).

In addition to large-scale rearrangements, NDMA exposure generates point mutations [17,34]. We previously found that NDMA induced significantly more point mutations in Aag^−/− mice and significantly fewer point mutations in AagTg mice when compared to WT 10 weeks after exposure [17]. Upon separating these data by sex, AagTg mice still showed fewer point mutations than WT for both males and females, though the differences between Aag^−/− and WT were not statistically significant for either sex (Fig. 3B). It is important to note that, unlike the results for DNA sequence rearrangements, there were no significant sex-related differences in NDMA-induced point mutations within any genotype, although there was a trend toward reduced susceptibility to point mutations in the female Aag^−/− mice compared to males (Fig. 3B). A limitation to the point mutation analysis is the high mouse-to-mouse variation and smaller sample size, which likely contributed to the lack of statistically significant differences.

3.5. Aag overexpression exacerbates NDMA-induced DNA damage

In our prior study, we analyzed the genotoxicity of NDMA by γH2AX staining and micronucleus (MN) quantification at 24 h and 48 h post-NDMA exposure, respectively [17]. Depending on the morphology of γH2AX staining, different biological consequences can be inferred. Specifically, punctate subnuclear γH2AX foci are indicative of DNA strand breaks, pan-nuclear γH2AX staining occurs in response to overwhelming DNA damage, and super-bright pan-nuclear γH2AX staining is thought to reflect toxic replication stress during S-phase [17]. MN can form during mitosis as a result of fragmented chromosomes.

Our earlier findings showed that NDMA increased the frequency of cells with ≥ 5 γH2AX subnuclear foci as well as pan-nuclear γH2AX staining 24 h post-exposure [17]. The NDMA-induced increase in γH2AX foci was largely preserved upon separating the data by sex (Fig. 4 A). In addition, Aag^−/− mice tended to have lower induction of γH2AX foci and pan-nuclear staining relative to WT and AagTg [17]. This was only statistically significant for γH2AX foci in Aag^−/− females versus WT females when the data were stratified by sex (Fig. 4A-B). There were no significant differences between males and females for either γH2AX foci or pan-nuclear staining within any genotype (Fig. 4A-B).

Fig. 4.

γH2AX foci and micronucleus formation in livers following NDMA exposure. A) Percentage of cells with punctate γH2AX foci 24 h post-second NDMA injection. B) Percentage of cells with pan-nuclear γH2AX signals 24 h post-second NDMA injection. C) Percentage of super bright pan-nuclear γH2AX-stained cells 24 h post-NDMA exposure. D) Percentage of micronucleated hepatocytes 48 h post-second NDMA injection. Box plots show the median and interquartile range, with whiskers indicating the range. Micronuclei data are presented as mean ± SD. Each data point represents one mouse (n = 4), and γH2AX values are averaged from two regions per slide. Unpaired Student’s t-test: *p < 0.05.

With regard to NDMA-induced super-bright pan-nuclear γH2AX staining, we previously observed a substantial increase specifically for AagTg mice relative to both WT and Aag^−/− mice 24 h post-exposure [17]. After splitting the data by sex, there was still an apparent increase in super-bright cells relative to control, though not statistically significant for AagTg females (Fig. 4 C). Compared to WT, AagTg males had more super-bright cells, and though the same trend was apparent in females, it did not reach statistical significance (p = 0.08), presumably due to high interindividual variation. While the Aag^−/− males showed a statistically significant increase in NDMA-induced super-bright pan-nuclear stained cells, the magnitude of the effect was small (Fig. 4 C).

Consistent with sex-pooled data [17], we observed significant induction of MN by NDMA 48 h post-exposure for all three genotypes, with AagTg being the most sensitive. Upon separating the data by sex, the AagTg mice still displayed the highest MN levels (Fig. 4D). Sex-separated data revealed an interesting trend of increased susceptibility to NDMA-induced MN in females compared to males, with a statistically significant difference in the AagTg mice (Fig. 4D). There were no sex-related differences in hepatocyte proliferation at this timepoint (Supplemental Figure 1), indicating elevated MN among females was not an artifact of increased cell division. Overall, the only significant difference observed when comparing the levels of DNA damage (either γH2AX or MN) between males and females was the increased susceptibility of AagTg females to NDMA-induced MN.

3.6. Acute toxicity of NDMA is similar between sexes and elevated among Aag-overexpressors

We previously reported that Aag^−/− livers were more susceptible to apoptosis, necrosis, and inflammatory cell infiltration compared to WT, and that AagTg livers experienced even more severe cytotoxicity and inflammation [17]. We also observed whole animal lethality in NDMA-treated AagTg mice, wherein over 12% of AagTg mice died within two weeks of exposure [17], although the cause of death was not confirmed. Given that others have shown that males and females differ in susceptibility to inflammation, including in the liver [27,47,48], we anticipated that males would show increased toxicity phenotypes over females. Instead, however, the levels of apoptotic cells and histopathological scores did not differ between males and females 24 h post-NDMA exposure (Supplemental Figure 2), and the same proportion of AagTg males and females suffered acute lethality within two weeks of injection (Supplemental Table 3).

3.7. Integrated analysis of in vivo physiological responses to NDMA exposure modulated by sex

We integrated our phenotypic observations over time using radar plots (Fig. 5). For each endpoint, results are normalized to the highest score among the groups. For example, NDMA-treated Aag^−/− males had the highest median number of tumors, so all tumor burden medians were normalized to NDMA Aag^−/− males; similarly, NDMA AagTg females had the highest mean induction of MN, so all MN means were normalized to NDMA AagTg females. Although differences in some endpoints were not statistically significant, the radar plots provide a holistic view of overall trends. As we found with sex-pooled data [17], Aag^−/− mice were more susceptible to long-term adverse effects (mutations and hepatic cancer, on the left) whereas AagTg mice experienced higher initial toxicity (DNA damage and lethality, on the right) but fewer persistent effects. Importantly, males were more vulnerable to persistent mutations and eventual hepatic cancer, while females were somewhat more sensitive to early toxicity, although the increased sensitivity for females was only statistically significant for MN induction.

Fig. 5.

Radar plots of key study endpoints in chronological order. For each endpoint, medians or means from each group are normalized to the highest-scoring group (e.g., NDMA-treated Aag^−/− male mice had the highest tumor burden, so all group medians were normalized to the Aag^−/− male median. Similarly, NDMA-treated AagTg female mice showed the highest micronucleus induction, so MN results were normalized to the AagTg female mean).

4. Discussion

It is well established that males are more susceptible to liver cancer than females, having been observed in both humans and rodents [26–32, 49–51]. Furthermore, prior research has shown that males are more vulnerable than females to environmental triggers of liver cancer [28], including nitrosamines [29–32]. Here, we have expanded upon that prior work to delve into the sex-specific molecular, cellular, genetic and physiological changes that determine the consequences of NDMA exposure, providing insights into the mechanisms underlying sex bias in NDMA-induced cancer. First, we robustly confirmed that males are more prone than females to NDMA-induced hepatic cancer (Table 1 & Fig. 1). We also showed that male mice were much more prone to DNA sequence rearrangements 10 weeks post-exposure (Fig. 3A) and they displayed severely increased liver pathology ~10 months post-exposure (Fig. 2). In contrast, signs of early liver damage were slightly higher in the females (Supplemental Figure 2). We anticipate that the sex-related differences we observed in this study were not due to differences in BER enzyme expression because Aag, APE1, Polβ, and Lig3 mRNA levels were similar among males and females for all genotype and treatment groups (Supplemental Tables 1 and 2).

Across all genotypes, males were more susceptible to NDMA-induced liver cancer than females (Table 1 and Fig. 1). The most striking sex-related enhancement in tumor incidence was in AagTg mice, while Aag^−/− mice showed the greatest sex-related enhancement in tumor multiplicity. By combining the variables of sex and genotype, it becomes clear that maleness combined with lower Aag expression creates conducive conditions for tumor development, whereas femaleness combined with increased Aag expression is most protective against cancer.

We saw a strong association between the relative susceptibility to sequence rearrangements (Fig. 3A) and to cancer (Fig. 1 and Table 1). This observation was consistent with the pooled data from males and females wherein sequence rearrangements, which occur relatively soon after exposure, are highly correlated with downstream cancer susceptibility [17]. Approximately 50% of the genome is composed of repetitive sequences [52,53], so a propensity for recombination events at the HR substrate in the RaDR locus likely reflects a biologically relevant mechanism for carcinogenic large-scale rearrangement mutations.

An important contribution of this study is the identification of male sequence rearrangement mutation bias in somatic tissue. Until now, male mutation bias has primarily been shown in germline [54–56] or cancer contexts [49–51,57]. However, these earlier studies often did not specify the type or the mechanism of mutation (e.g., point mutations, frameshifts, sequence rearrangements), and they had not analyzed physiologically normal somatic tissues. A remarkable finding in comparing sex differences in mutagenesis in the present work was that, for all genotypes, females accumulated significantly fewer spontaneous sequence rearrangement mutations (i.e. in the absence of NDMA exposure) compared to their male counterparts (Fig. 3A). Additionally, among NDMA-treated mice, WT and Aag^−/− females developed fewer sequence rearrangements than males. To our knowledge, this is the first demonstration that males are more prone to HR-driven sequence rearrangement mutations in the liver both spontaneously and following exposure to a carcinogenic alkylating agent. We infer from these results that a hitherto unknown mechanism involving HR and BER might underlie the well-known increased rates of liver cancer in males.

The fact that HR-driven sequence rearrangements were lower in females than males may be related to the fact that estrogen receptor α (ERα) facilitates the two major pathways that repair NDMA-induced DNA damage: BER initiated by AAG (which removes the replication-blocking 3MeA lesion) and direct reversal by O⁶-methylguanine-DNA-methyltransferase (MGMT, which removes the aberrant methyl group from O⁶MeG). ERα interacts directly with AAG to enhance excision activity [58], thus reducing the potential for 3MeA to block DNA replication, a situation known to promote mutagenic recombination events [19]. Furthermore, the Mgmt promoter contains ERα-responsive elements, resulting in enhanced MGMT expression following estrogen exposure [59]. In addition, spent MGMT negatively regulates ERα-mediated cell proliferation, preventing cell division before repair of O⁶MeG is complete [60]. While the O⁶MeG lesion does not block replication, it readily pairs with thymine instead of guanine. Such mismatched DNA bases stimulate mismatch repair futile cycling, leading to persistent single strand breaks, which can be recombinogenic [61,62]. Thus, even in Aag^−/− mice, which do not have estrogen-enhanced excision of 3MeA, estrogen is likely protective against mutagenic recombination through enhanced MGMT expression (reducing futile cycling) and negative regulation of cell division (by spent MGMT).

Interestingly, previous studies in mammary cell lines have shown that estrogen delays γH2AX resolution, reduces HR, and induces chromosomal instability following irradiation [63], which is consistent with our observations of increased MN (Fig. 4D) and fewer sequence rearrangements (Fig. 3A) in females. The greater induction of MN in WT and AagTg females may be associated with HR inhibition by estrogen, preventing resolution of strand breaks and leading to chromosomal fragments during mitosis. Another possible explanation for increased MN induction in WT and AagTg females is that the interactions between ERα and AAG may enhance glycosylase excision [58], thus leading to increased BER-intermediate strand breaks that result in double-strand breaks during replication.

We hypothesized in our previous publication that the excess BER-intermediate strand breaks in NDMA-exposed AagTg mice reduced mutations by elevating cytotoxicity, whereas unrepaired 3MeA in Aag^−/− mice promoted sequence rearrangement mutations through template switching [17]. We hypothesize the toxicity of BER-intermediate strand breaks likely caused the observed downregulation of Aag mRNA in NDMA-treated AagTg males and females 24 h after exposure. However, while NDMA-treated females displayed higher induction of MN (Fig. 3D), they did not show overt differences in hepatocellular or whole-animal toxicity shortly after exposure (Supplemental Figure 2 & Supplemental Table 3), suggesting that sex differences in clearance of damaged cells does not explain the reduced recombination in females. Interestingly, results from the Samson laboratory previously showed that in the kidney, retina, and cerebellum, cytotoxicity of alkylating exposures is reduced in females, which suggests tissue-specific differences in the impact of Aag on cytotoxicity [64–66].

Long-term morphological changes and liver damage were more apparent in NDMA-exposed males versus females (Fig. 2). In nearly all cases, the levels of NDMA-induced hepatocellular degeneration, Ito hyperplasia, and foci of hepatocellular alteration were elevated in males. The liver is a sexually dimorphic organ, and hormones play an integral role in modulating responses to environmental insult. For example, one study showed that exposure to diethylnitrosamine in mice significantly increased serum interleukin-6 (IL-6) levels in males compared to females, which was suppressed by treatment with an ERα agonist [30]. IL-6 is an inflammatory cytokine primarily produced by Kupffer cells in the liver, and it can promote survival, proliferation, invasion, angiogenesis, and metastasis through the IL-6/STAT3 axis [27]. Interestingly, knockout of IL-6 abolished the sex-related difference in diethylnitrosamine-induced liver tumor development [30]. The suppression of IL-6 in females is one of many estrogen-mediated effects that has been shown to reduce the occurrence and aggression of hepatocellular carcinoma. In contrast, androgens in males can promote proliferation and survival, leading to the retention of damaged cells and potentially elevating carcinogenesis [27].

Our observations of differences in WT, Aag^−/−, and AagTg males and females in response to NDMA reveal surprising patterns that warrant further investigation. For example, we found remarkable protection against mutations and cancer in females, but no sex-related difference in γH2AX and an increase in NDMA-induced MN among AagTg females compared to males. We also showed here and previously [17] that γH2AX and MN shortly after exposure were highest in AagTg and, for γH2AX foci, lowest in Aag^−/−, following an inverse correlation with mutations and cancer risk. Levels of apoptosis and necrosis 24 h after exposure were lowest in WT, highest in AagTg [17], and similar among males and females for all genotypes (Supplemental Figure 2 A and B), suggesting differences in mutations and cancer were not due to differences in clearance of damaged cells. These patterns indicate that γH2AX and MN can point to a genotoxic mode of action and likely carcinogenicity of test compounds, but they are not effective for stratifying cancer risk. It would be interesting to analyze a larger number of mice for γH2AX, MN, cytotoxicity, and point mutations to determine if slight sex-related differences exist that were not apparent here. It is also important to note that, while estrogen is mostly produced at similar levels in neonates of both sexes, females experience a brief spike in estrogen at two weeks of age [67], around the second NDMA injection. It would be informative to analyze DNA and tissue damage after injection at postnatal day 8 when sex hormone levels are similar and after puberty when sex hormone production is significantly different. This could help to elucidate possible windows of susceptibility to NDMA exposure for both sexes, a topic of great importance for toxicological risk analysis.

Taken together, we demonstrate here the importance of both sex and AAG-mediated DNA repair in response to NDMA exposure. Key findings from this study reveal that male mice are more susceptible to NDMA-induced large-scale mutations and long-term liver damage than females, helping to explain the increased susceptibility to liver cancer observed among men.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.dnarep.2026.103955.

Data availability

Links to our FAIR data repositories are provided.