Persistent interferon signaling and clonal expansion mark early events in DNA methylation damage-induced liver cancer

Abstract

N-Nitrosodimethylamine (NDMA), a probable human carcinogen, induces toxic and mutagenic O ⁶-methylguanine lesions that are repaired by the O ⁶-methylguanine methyltransferase (MGMT). To elucidate mechanisms of NDMA-induced liver cancer progression, we performed longitudinal analyses of phenomic, transcriptomic, and phosphoproteomic changes in wild-type and MGMT-deficient mice, observing amplified responses in the deficient genotype. Early molecular rewiring indicative of a DNA damage response was detected by phosphoproteomic and transcriptomic profiling within days post-exposure. Transcriptomic analyses identified a persistent and robust interferon response as the dominant activated pathway. This chronic interferon signaling, which remained unresolved, correlated with extensive clonal expansion, an early hallmark of oncogenesis. Spatial transcriptomics further revealed pathway alterations favoring tumorigenesis within clonally expanded cells. These findings delineate the cascade of molecular events triggered by acute early-life NDMA exposure, culminating in cancer development months later. Our study unveils potential predictive biomarkers and strategies for disease mitigation.

Extended Data Fig. 1:. MGMT deficiency exacerbates NDMA-induced liver injury and tumorigenesis.

a, Liver tumor counts at 10 months post-treatment in WT and Mgmt^−/− mice, grouped by sex. n ≥ 6 per group. b, Mouse body weight at 10 months post-treatment (males and females combined). n ≥ 9. Box plots indicate median, upper and lower quartiles, and whiskers showing maximum/minimum values. c, Histopathological scores for liver morphologies, biliary hyperplasia and fibrosis, in WT and Mgmt^−/− mice at 10 weeks post-treatment, assessed by H&E staining (sexes combined). n ≥ 8. d, Histopathological scores of karyomegaly in WT and Mgmt^−/− mice at 10 weeks post-treatment, assessed by H&E staining (sexes combined). n ≥ 8. e, Quantification of mean nuclear size in liver cells by DAPI fluorescence. WT saline- (gray) and NDMA-treated (black). Mgmt^−/− saline- (light green) and NDMA-treated (dark green). See Fig. 2C for images. n ≥ 7. f, Protein levels were measured via western blot to support T-cell activation (CD3ε & CD11C) in Mgmt^−/− and WT livers at 5 days and 10 weeks post-exposure. n = 4 (2 males, 2 females). g, Relative spleen weight (to body weight) and absolute body weight at 10 weeks post-treatment. n ≥ 23. Box plots as described in (B). h, Histopathological scores for biliary hyperplasia and karyomegaly and number of foci hepatocellular alterations in WT and Mgmt^−/− mice at 10 months post-treatment, assessed by H&E staining (sexes combined). n ≥ 16. Data are presented as mean ± s.e.m. Statistical comparisons performed using one-way ANOVA with Šídák’s multiple comparisons test (a,b,e–g) and Kruskal-Wallis and Dunn’s test (c,d,h). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Extended Data Fig. 2:. MGMT suppresses DNA damage retention, toxicity, and compensatory proliferation.

a, Quantification of mean gH2AX fluorescence intensity per nucleus in WT livers, comparing saline-treated (gray) and NDMA-treated (black) groups. n ≥ 7 per group. b, Quantification of mean gH2AX fluorescence intensity per nucleus in Mgmt^−/− livers, comparing saline-treated (light green) and NDMA-treated (dark green) groups. n ≥ 7 per group. c, Western blot analysis of gH2AX in whole-cell liver lysates at 1 day, 5 days, and 10 weeks post-exposure. Band intensities normalized to TPS and saline WT controls. n = 4 per group (2 males, 2 females). d, Quantification of apoptosis via cleaved caspase-3 positive cells as a percentage of total cells in WT livers at indicated timepoints, measured by IHC. Inset shows representative cleaved caspase-3 staining. n ≥ 7. e, Quantification of apoptosis via immunofluorescence of pan-nuclear gH2AX-positive cells as a percentage of total cells in WT livers. Inset provides representative pan-nuclear gH2AX staining. n ≥ 7. Data are presented as mean ± s.e.m. Statistical comparisons performed using one-way ANOVA with Šídák’s multiple comparisons test (a,b,d,e,f) and two-way ANOVA (c). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Extended Data Fig. 3:. Immunoblotting and phosphoproteomics reveal an NDMA-induced DDR.

a, STRING database analysis of top-expressed phosphoproteins at 1 day post-exposure. DNA repair clusters as shown in Fig. 3i from biological processes in gene ontology are shown here.

Extended Data Fig. 4:. NDMA induces persistent transcriptome dysregulation and an IFN response.

a, Upregulated genes from NDMA-treated WT and Mgmt^−/− liver samples (1 day post-exposure) were analyzed using Enrichr for Gene Ontology Biological Process enrichment. The top 10 enriched pathways are shown, ranked by Fisher exact test p-values. b, Upregulated genes from NDMA-treated WT and Mgmt^−/− liver samples (2 days post-exposure) were analyzed using Enrichr for MSigDB Hallmark gene set enrichment. The top 10 pathways are displayed, ranked by Fisher exact test p-values. c, Volcano plots present differentially expressed genes in WT liver (NDMA vs saline) across selected timepoints (1 day, 5 days, and 10 weeks), with −log10 p-value plotted against log₂ fold change. Genes circled in red belong to the GSEA Hallmark p53 pathway data set; Cyp2e1 is circled in blue. Differential expression was determined using an adjusted p-value cutoff of < 0.05 and absolute log₂ fold change cutoff of > 0.58. d, CYP2E1 protein abundance was assessed by western blot in liver samples at 5 days and 10 weeks post-exposure. Band intensities normalized to TPS and saline WT controls. n = 4 per group (2 males, 2 females). Statistical comparisons (two-way ANOVA with Šídák’s multiple comparisons. Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant. e, Heatmaps display expression of a curated interferon response gene list in WT and Mgmt^−/− livers at 1 day, 10 weeks, and 10 months post-exposure, demonstrating persistence of IFN signaling in Mgmt^−/− samples. f, Gene Set Enrichment Analysis (GSEA) on Mgmt^−/− liver at 1 day and 10 weeks post-NDMA exposure highlights enrichment of the Hallmark p53 pathway compared to saline controls.

Extended Data Fig. 5:. Upstream regulators of the IFN response are persistently activated in Mgmt^−/− mice.

a, Ingenuity Pathway Analysis (IPA) of RNA-sequencing data from Mgmt^−/− mice identified a few activated (> 2.0 z-score) and inhibited (< 2.0 z-score) upstream regulators throughout their lifetime with pertaining to DNA repair genes (labeled in each plot). Analysis included transcripts with p-adjusted values < 0.05 and log₂ fold change > 0.58. b, Phosphorylated IRF7 (pIRF7) protein levels were measured via western blot at 10 weeks post-exposure. Band intensities normalized to TPS and saline WT controls. n = 4 per group (2 males, 2 females). Statistical comparisons performed by two-way ANOVA with Šídák’s multiple comparisons test. Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Extended Data Fig. 6:. NDMA induces recombination events indicative of persistent genomic instability.

a, Quantification of sequence rearrangement mutations in liver samples of WT (left, gray/black bars) and Mgmt^−/− (right, green bars) mice following NDMA or saline exposure at multiple timepoints, shows a peak at 10 days. RaDR foci (eGFP expression) per cm² quantified using machine learning image analysis. n ≥ 10 per group. b, Measurements of total left lateral liver lobe size (cm²) in WT (left) and Mgmt^−/− (right) mice across NDMA and saline groups. Significant reductions are observed in NDMA-treated WT mice at 5 and 10 days, and in Mgmt^−/− mice from 10 to 70 days post-exposure. n ≥ 10 per group. c, Quantitation of sequence rearrangement mutations in the pancreas, shown for WT (left) and Mgmt^−/− (right) mice. RaDR foci per cm² quantified by machine learning. Significant increases are detected in Mgmt^−/− mice after NDMA exposure at 50- and 70-days post-dose, but not in WT controls. n ≥ 10 per group. d, Direct comparison of pancreatic sequence rearrangement mutations post-NDMA exposure reveals significantly increased RaDR foci per cm² in Mgmt^−/− (green bars) compared to WT (gray/black bars) at 50 and 70 days. n ≥ 10 per group. e, Representative whole-mount fluorescence images of pancreatic tissue from WT and Mgmt^−/− RaDR-GFP mice 70 days after NDMA or saline treatment. Insets show higher magnification (scale = 1 mm) of boxed regions (scale = 5 mm at 2x) where areas are with GFP-positive foci indicating mutations. Statistical comparisons performed using one-way ANOVA with Šídák’s multiple comparisons test (a–d). Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Extended Data Fig. 7:. NDMA induces clonal expansion in Mgmt^−/− livers.

a, At 10 weeks post-NDMA exposure large clonal expansion events are significantly greater in male Mgmt^−/− livers compared to females. A single large clonal expansion event was defined as RaDR-GFP foci exceeding 2000 pixels in area. Representative whole-mount eGFP images highlight the significant difference between female and male Mgmt^−/− livers. n ≥ 14 per group. Statistical comparisons performed using Mann-Whitney test. Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant. b, 2-photon microscopy of a liver section (100 μm thick) from a ~13-month post-NDMA Mgmt^−/− female mouse visualizes both eGFP-positive clonal expansion events (green, lower boxes) and tumor areas detected by tissue autofluorescence (yellow, upper box). Insets display high-magnification images of the indicated regions. c, 3-dimensional reconstruction of sequential 2-photon images from (b) visualizes the spatial distribution of eGFP clonal expansion events (green) and tumors (yellow) within a Mgmt^−/− liver. The vascular network is displayed in blue. See Supplementary Video 2.

Extended Data Fig. 8:. RaDR-GFP imaging.

a, Schematic illustration of the RaDR-GFP mouse experiment. Freshly collected liver tissue is collected and mounted on a glass slide for GFP imaging at 2x magnification, with excitation at 480 nm. This protocol enables direct visualization of recombination reporter signals in liver tissue. b, Representative GFP fluorescence images of whole livers from WT and Mgmt^−/− RaDR mice at 1 day, 10 days, and 10 weeks post-saline treatment, imaged at consistent exposure and emission settings. In rare cases (~2%), spontaneous recombination of the direct repeat substrate occurs early in development which can lead to EGFP expression in many daughter cells. Early recombination events can be distinguished from the majority by the presence of a jagged edge with a decrease in brightness and contrast. Out of 327 mice, this was observed in 7 samples, and these were eliminated from analysis. Insets highlight regions of interest.

Extended Data Fig. 9:

Western Blot Images.

Extended Data Fig. 10:

Western Blot Images.

Fig. 1:. MGMT deficiency exacerbates NDMA-induced liver injury and tumorigenesis.

a, Experimental timeline displaying two intraperitoneal injections of saline or NDMA at P8 (3.5 mg/kg) and P15 (7 mg/kg) in WT and Mgmt^−/− mice, with indicated collection timepoints. b, Representative gross images of livers from WT and Mgmt^−/− male mice 10 months post-treatment. Black arrows denote visible tumors. c, Representative H&E-stained liver sections at 10 months post-exposure (ii) show trabecular subtype hepatocellular carcinoma in NDMA-treated WT mice, and (iv) exhibit expansile neoplasms consistent with hepatocellular carcinoma (solid subtype) in Mgmt^−/− NDMA-treated mice (magnification 40x). Representative sections are depicted in photomicrographs in (g). d, Quantification of liver tumors at 10 months. n ≥ 15. e, Representative H&E sections (magnification 200x, scale bars = 100 μm) and histopathology scores for hepatocellular necrosis and inflammation at 1 day post-treatment. (ii) NDMA-treated WT mice showed periacinar hepatocellular degeneration (swelling; black dash line) with few macrophages and mild portal inflammation (arrows). (iv) Mgmt^−/− NDMA-treated mice exhibited focal/multifocal periacinar degeneration with or without single-cell necrosis (arrowheads) and mild portal/periportal inflammation (arrows). n ≥ 7. f, At 10 weeks post-NDMA treatment (ii) WT livers exhibited hepatocellular degeneration with increased cytoplasmic alterations, eosinophilia hypertrophy, and/or kariocytomegaly (black arrow). (iv) Mgmt^−/− mice exhibited multifocal hepatocellular degeneration (black arrow), with or without single-cell necrosis (white arrow), and mild to moderate lymphocytic and histiocytic inflammation in portal tracts (stealth black arrowhead). Biliary hyperplasia (black arrowheads) was common in portal tracts. Magnification 200x, scale bars = 100 μm. n ≥ 8. g, (ii) High-magnification H&E sections (400x) at 10 months (ii) show a hepatocellular carcinoma – trabecular subtype (arrows) in an NDMA-treated WT male mouse. (iv) An NDMA-treated Mgmt^−/− mouse shows a hepatocellular carcinoma (solid subtype). Hepatocellular necrosis and inflammation scores show significant increases in Mgmt^−/− NDMA treated mice compared to controls. n ≥ 16. Histopathology scoring was based on H&E stains across indicated panels (e–f). Data are presented as mean ± s.e.m. Statistical comparisons by Kruskal-Wallis and Dunn’s test (d–f). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Fig. 2:. MGMT suppresses DNA damage retention, toxicity, and compensatory proliferation.

a, Genomic DNA from liver tissue was analyzed for O⁶-methylguanine DNA adducts by triple quadrupole mass spectrometry. Adduct levels (fmol) were normalized to DNA (μg). Data points represent 1 male and 1 female sample combined. Saline controls received a single dose at P15. b, Quantification of the mean gH2AX fluorescence intensity per nucleus in NDMA-exposed WT (gray/black bars) to Mgmt^−/− (green bars) livers. n ≥ 7. c, Representative immunofluorescent images of gH2AX (green) and DAPI (blue) stained liver sections across indicated post-treatment time points. Persistent DNA damage (gH2AX foci) and cell enlargement are observed in Mgmt^−/− NDMA-treated livers compared to WT. Related DAPI quantification provided in Fig. S1E. Scale bar = 20 μm. d, Quantification of hepatocyte micronuclei formation by flow cytometry at 2 days post-NDMA exposure. n ≥ 10 per group. e, Apoptosis quantified by IHC for cleaved caspase-3 (CC3) in Mgmt^−/− mouse livers. Representative CC3 staining shown in inset. n ≥ 2. f, Quantification of apoptosis was measured by immunofluorescence of pan-nuclear gH2AX-positive cells for Mgmt^−/− mice. Inset shows representative pan-nuclear gH2AX staining. n ≥ 7. g, Proliferation measured via flow cytometry for % Ki67-positive cells at 2 days post-exposure. n ≥ 10. h, PCNA protein quantification by western blot in liver whole-cell lysates at 5 days and 10 weeks post-exposure. Band intensities normalized to TPS and saline WT controls. n = 4 per group (2 males and 2 females). Data are presented as mean ± s.e.m. Statistical comparisons performed using one-way ANOVA with Šídák’s multiple comparisons test (b,d–h). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant. DAPI, 4′,6-diamidino-2-phenylindole.

Fig. 3:. Immunoblotting and phosphoproteomics reveal an NDMA-induced DDR.

a, Whole-cell liver lysates collected at 1 day, 5 days, and 10 weeks post-exposure were analyzed by western blot for phosho-p53. Band intensities normalized to TPS and saline WT controls. n = 4 per group (2 males, 2 females). b, Liver lysates from identical timepoints were immunoblotted for p21, normalized as above. n = 4 per group (2 males, 2 females). c, Liver lysates from identical timepoints were immunoblotted for RAD51, normalized as above. n = 4 per group (2 males, 2 females). d,e, Principal Component Analysis (PCA) of phosphoproteomics at 1 day (d) and 10 weeks (e) post-exposure displayed distinct sample clustering by treatment group and genotype. n = 4 (2 males and 2 females). f,g, Heatmaps show the most differentially expressed phosphorylated proteins at 1 day (f) and 10 weeks (g) post-exposure, identified by phosphoproteomics. DNA repair proteins are highlighted in bold (f). Protein selection based on multi-group ANOVA by treatment (p < 0.05), with standard deviation < 0.05; 354 total proteins (f), 214 total proteins (g). h, Phosphoproteomic quantification of gH2AX (combined S140 and S137 sites) at 1 day post-exposure. n ≥ 2 per group. i, STRING database analysis of top-expressed phosphoproteins at 1 day and 10 weeks post-exposure. Functional enrichment at 1 day highlights DNA repair clusters; 10 weeks displays enrichment for non-DNA repair biological processes from gene ontology. Statistical comparisons performed using one-way ANOVA with Šídák’s multiple comparisons test (a–c,h). Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Fig. 4:. NDMA induces persistent transcriptome dysregulation and an IFN response.

a, PCA of Mgmt^−/− liver samples collected at days 1, 2, and 5 post-saline or NDMA exposure shows clear clustering separation by treatment group. RNA was extracted from liver tissue and analyzed by RNA-seq; n ≥ 7 per group (males and females combined). b, Bar graphs depict the number of differentially expressed RNA transcripts (up- and downregulated) in WT and Mgmt^−/− liver following NDMA exposure across indicated timepoints. c, Venn diagrams compare overlap of significantly up- and downregulated genes across genotypes and timepoints. d, Pathway enrichment analysis of persistently up- and downregulated genes (intersected across all timepoints from panel c, red circles) in Mgmt^−/− samples using Enrichr for MSigDB Hallmark gene sets. The top 8 pathways are displayed, ranked by Fisher exact test p-value. Pathways involved in IFN response are marked with an asterisk (*). e, Volcano plots display the most differentially expressed genes in Mgmt^−/− livers comparing NDMA vs saline treatment at 1 day, 5 days, and 10 weeks. Plots show –log₁₀ p-value versus log₂ fold change. Genes from the GSEA Hallmark p53 pathway are circled in red or black; Cyp2e1 downregulation is circled in blue. f,i, Bar graphs quantify the number of up- and downregulated genes within the pre-curated IFN (f) and DNA repair (i) gene lists, comparing NDMA effects between WT and Mgmt^−/− across the time course. g,h, Heatmaps show expression profiles for curated IFN response genes (g) and DNA repair genes (h) in response to NDMA treatment in WT and Mgmt^−/− livers at days 1, 2, and 5 post-exposure. Color scale reflects relative gene expression intensity. Statistical analysis: All comparisons use thresholds of adjusted p-value < 0.05 and absolute log₂ fold change > 0.58 (b–i).

Fig. 5:. Upstream regulators of the IFN response are persistently activated in Mgmt^−/− mice.

a, Upregulated genes from WT and Mgmt^−/− livers (5 days post-NDMA treatment) were analyzed for transcriptional regulatory network enrichment using the TRRUST (Transcriptional Regulatory Relationships Unravelled by Sentence-based Text-mining) database via Enrichr. Bar graphs display the top 10 enriched transcription factors as ranked by p-value. b, Ingenuity Pathway Analysis (IPA) of RNA-sequencing data from Mgmt^−/− mice identified activated (> 2.0 z-score) and inhibited (< 2.0 z-score) upstream regulators throughout their lifetime with IFN genes labeled in each plot. c, IPA of RNA-sequencing data for WT mice identified activated and inhibited upstream regulators only at 1 day and 5 days post-exposure. No genes met cutoff thresholds (same as (B) for 10-week or 10-month data. Genes involved in IFN response are labeled. d, IFN pathway protein levels (cGAS, STING, STAT1, pSTAT1) were quantified by western blot from whole-cell liver tissue at 1 day, 5 days, and 10 weeks post-NDMA exposure. Band intensities normalized to TPS and saline WT controls. n = 4 per group (2 males, 2 females). Statistical comparisons performed by two-way ANOVA with Šídák’s multiple comparisons test. Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant. All transcriptomic analyses utilized filtering thresholds of p-adjusted value < 0.05 and absolute log₂ fold change > 0.58.

Fig. 6:. NDMA induces recombination events indicative of persistent genomic instability.

a, Representative whole-mount fluorescence images of left lateral liver lobes from WT RaDR-GFP mice at multiple timepoints post-NDMA or saline exposure. eGFP-positive foci indicate large-scale sequence rearrangements/recombination events. Insets show higher magnification (scale = 1 mm) of boxed regions (scale = 5 mm at 2x). Images are representative of male mice. b, Representative whole-mount fluorescence images of left lateral liver lobes from Mgmt^−/− RaDR-GFP mice, showing extensive mutation burden and prominent clonal expansion of GFP-positive foci after NDMA treatment. Images are representative of male mice. c, Quantitation of sequence rearrangement mutations (eGFP-positive RaDR foci per cm²) in WT (gray/black bars) and Mgmt^−/− (green bars) livers post-NDMA exposure over time. Foci quantified via machine learning-assisted image analysis. n ≥ 10 per group. d, Measurements of total left lateral liver lobe area (cm²) in WT (gray/black bars) and Mgmt^−/− (green bars) mice at various timepoints following NDMA exposure. n ≥ 10 per group. Statistical comparisons performed using one-way ANOVA with Šídák’s multiple comparisons test (c,d). Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Fig. 7:. NDMA induces clonal expansion in Mgmt^−/− livers.

a, Quantitation of sequence rearrangement mutations that formed clonal expansion colonies in Mgmt^−/− (green dots) livers starting at 30 days post-NDMA exposure compared to WT (gray/black bars). A single clonal expansion event was defined as RaDR-GFP foci exceeding 500 pixels in area. n ≥ 6 per group. b, At 10 weeks (70 days) post-NDMA exposure, Mgmt^−/− livers exhibit significantly larger clonal expansion colonies (RaDR-GFP foci >2000 pixels) compared to WT. A representative eGFP image of a Mgmt^−/− male mouse liver at 10 weeks post-NDMA exposure highlights clonal expansion colonies. c, 2-photon imaging of a ~13-month post NDMA treated Mgmt^−/− female liver shows both eGFP clonal expansion and tumor expression from autofluorescence (yellow), with corresponding insets displaying high-resolution views. Representative gross liver images highlight visible tumors (red arrows). d, 3D reconstruction of sequential 2-photon images from (c) visualizes the spatial distribution of eGFP clonal expansion events (green) and tumors (yellow) within a Mgmt^−/− liver. The vascular network is displayed in blue. See Supplementary Video 1. Statistical comparisons performed using Kruskal-Wallis and Dunn’s test (a,b). Data are presented as mean ± s.e.m. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. NS, not significant.

Figure 8.. Spatial transcriptomics of clonal outgrowths and tumors.

a, Spatial transcriptomic profiling was performed on male Mgmt^−/− liver samples 10 weeks post-treatment using 10x Genomics Visium technology. Representative male Mgmt^−/− liver samples 10 weeks post-treatment were analyzed. For each sample, the same liver section was used for H&E staining (row 1) and spatial transcriptomics mapping (row 2), while an adjacent serial section was imaged for eGFP fluorescence (row 3). In saline-treated livers, transcriptomic clusters were uniform, but NDMA treatment revealed distinct clustering, accompanied by histological changes including portal hepatitis and biliary hyperplasia, which overlapped with an intensely GFP-positive colony. One featured clonal expansion (~0.5 mm²) was analyzed for differentially expressed RNA transcripts; upregulated genes from this region were enriched for pathways in the MSigDB Hallmark collection, with the top 9 pathways shown. b, Spatial transcriptomic profiling of male Mgmt^−/− livers 10 months post-treatment. Analysis used the same tissue section for H&E staining (row 1) and spatial transcriptomics (row 2); a serial section was imaged for eGFP (row 3). Saline controls continued to show uniform transcriptomic clustering, while NDMA treatment resulted in a hepatocellular adenoma with a few dilated bile ducts. Differentially expressed genes from the tumor region were subjected to MSigDB Hallmark pathway enrichment, with the top 9 pathways shown. c, Radar plots summarize key endpoints from this study at 1 day and 10 weeks post-exposure. Medians or means from each data set were normalized to the highest scoring group in that analysis method (e.g., NDMA-treated Mgmt^−/− mice had the highest tumor burden, so medians of all groups were normalized to the median Mgmt^−/−).