Mutational spectra of aflatoxin B1 in vivo establish biomarkers of exposure for human hepatocellular carcinoma

Abstract

Aflatoxin B₁ (AFB₁) and/or hepatitis B and C viruses are risk factors for human hepatocellular carcinoma (HCC). Available evidence supports the interpretation that formation of AFB₁-DNA adducts in hepatocytes seeds a population of mutations, mainly G:C→T:A, and viral processes synergize to accelerate tumorigenesis, perhaps via inflammation. Responding to a need for early-onset evidence predicting disease development, highly accurate duplex sequencing was used to monitor acquisition of high-resolution mutational spectra (HRMS) during the process of hepatocarcinogenesis. Four-day-old male mice were treated with AFB₁ using a regimen that induced HCC within 72 wk. For analysis, livers were separated into tumor and adjacent cellular fractions. HRMS of cells surrounding the tumors revealed predominantly G:C→T:A mutations characteristic of AFB₁ exposure. Importantly, 25% of all mutations were G→T in one trinucleotide context (CGC; the underlined G is the position of the mutation), which is also a hotspot mutation in human liver tumors whose incidence correlates with AFB₁ exposure. The technology proved sufficiently sensitive that the same distinctive spectrum was detected as early as 10 wk after dosing, well before evidence of neoplasia. Additionally, analysis of tumor tissue revealed a more complex pattern than observed in surrounding hepatocytes; tumor HRMS were a composite of the 10-wk spectrum and a more heterogeneous set of mutations that emerged during tumor outgrowth. We propose that the 10-wk HRMS reflects a short-term mutational response to AFB₁, and, as such, is an early detection metric for AFB₁-induced liver cancer in this mouse model that will be a useful tool to reconstruct the molecular etiology of human hepatocarcinogenesis.

Keywords: cancer; duplex sequencing; mouse model; mutagenesis; mycotoxins.

Fig. 1.

Role of AFB₁ in development of HCC. AFB₁ is activated by metabolism to form an electrophilic epoxide, which binds to DNA to form mutagenic AFB₁-DNA adducts. The two adducts shown, AFB₁-N⁷-Gua and AFB₁-FAPY, are mutagenic in vivo and cause the type of mutation that is seen most frequently in HCCs (the G:C→T:A transversion). Shortly after dosing, it is proposed that a “founder” or “exposure” mutational spectrum forms. As the tissue ages, subsequent mutational processes continue to mature the founder spectrum into the mutational spectrum seen in end-stage cancer.

Fig. 2.

Experimental work flow. (A) Male gptΔ B6C3F1 mice were treated as neonates with AFB₁ and killed at either 10 wk (A-10) or 72 wk (A-72) of age. D-10 and D-72 are the corresponding DMSO solvent controls. (B) Liver tissues from killed mice at 72 wk of age were subjected to collagenase perfusion, and hepatocytes or tumor cells were isolated (A-72H and A-72T, respectively; D-72 is the corresponding DMSO control). (C) Overview of the DS method used to identify mutational spectra obtained from the mice shown in A and B. Adapted from Schmitt et al. (24). Details are provided in the main text. CAT, chloramphenicol acetyl transferase.

Fig. S1.

The 6-thioguanine–selected mutational spectrum of AFB₁ in the gpt coding sequence of the λ-gptΔ B6C3F1 mouse. The spectrum was generated from the data of Woo et al. (16) plotted in three-base contexts as done in the present paper (e.g., Fig. 3A). Mice were treated with 6 mg/kg AFB₁ at day 4 of life and killed at 10 wk for mutation analysis by a protocol that selected for mutants that confer resistance to 6-thioguanine. A total of 131 mutants are plotted.

Fig. 3.

(A) HRMS of mice 10 and 72 wk after treatment with AFB₁. The MF distributions enumerate base substitutions in each of the 96 possible three-base contexts (the center base in each context is the site of the mutation). Sample designations are presented in Fig. 2. (B) Cosine similarity provides a quantitative metric to express how similar the HRMSs are to one another.

Fig. S2.

Relative frequency of each trinucleotide sequence context in the 6.4-kb sequencing target. The occurrence of each trinucleotide context was tallied in the λ-EG10 6.4-kb sequencing target. Of the 4³ = 64 possible trinucleotide contexts, only 32 are shown; each sequence count also includes the occurrences of its reverse complementary sequence (e.g., ACA denotes the counts for both ACA and its complementary TGT trinucleotide sequences).

Fig. S3.

Mutational patterns derived from the livers of AFB₁-treated mice 10 wk after carcinogen administration. Spectra from four biological replicates are shown to demonstrate the reproducibility of the method. The yellow stripe highlights the G:C→T:A hotspot in the 5′-CGC-3′ context. The underlined G is the position of the mutation. Averaged data are shown in the bottom spectrum, along with error bars denoting SD. Avg., average.

Fig. S4.

Distribution of CGC→CTC mutations in the 6.4-kb target sequence observed at 10 wk for each of the individual mice treated with AFB₁. The numbers 1642, 1643, 1644, and 8114 denote individual mice. The areas marked as gpt, cat, and ColE1 represent features of the EG10 fragment and are highlighted only for orientation purposes. The red bars indicate the position where a CGC→CTC mutation was observed. The areas marked as gpt, cat, and ColE1 represent the locations of the corresponding genes within the 6.4-kb EG10 fragment for orientation purposes.

Fig. S5.

Comparison of A-72T with linear combinations of A-10 and D-10 spectra by cosine similarity. Linear combinations of A-10 and D-10 spectra were compared with the tumor spectrum A-72T using cosine similarity. The extreme values reflect the cosine similarity between A-72T and D-10 (0.76) when the A-10 contribution is 0% and between A-72T and A-10 (0.66) when the A-10 contribution is 100%. The linear combination composed of 29% A-10 and 71% D-10 (denoted with the dotted line) gave the maximum cosine similarity with A-72T, which was 0.85.

Fig. S6.

Distribution of total number of mutations, and the number of unique mutations, in each of the tumor samples from the AFB₁-treated mice at 72 wk. The numbers 6210, 6211, 6212, and 6213 denote individual tumor-bearing mice. Total base substitution mutations were plotted by their three-base contexts (A) as well as by the respective positions and intensities of those mutations within the 6.4-kb transgene analyzed in the λ-gptΔ B6C3F1 mouse (B). The areas in B marked as gpt, cat, and ColE1 represent features of the EG10 fragment and are highlighted only for orientation purposes. The vertical bars in B indicate the position where a mutation was observed; the multiplicity of each mutation is represented by the height of each bar (relative to the scale shown on the y axis). In each case, unique mutations in three-base contexts (C) and across the 6.4-kb transgene (D) are plotted. In three of the four mice (6210, 6211, and 6213), there are mutations with a high clonality, because they occur hundreds of times; mouse tumor 6212 showed less evidence of clonality. All sequenced mutations are included in A and B. As shown in Table 1, up to 95% of the mutations were clonal in origin in the tumor samples; the mutational spectrum in which clonal mutations (mutations that occurred repeatedly at the same nucleotide position in the 6.4-kb target) were counted only once is shown in C and D.

Fig. S7.

Distribution of the total number of mutations and unique mutations in the four AFB₁-induced tumors. This composite figure was compiled from data on individual tumors presented in Fig. S6.

Fig. 4.

Comparison of the mutational patterns of human liver cancer with the murine AFB₁ exposure spectrum. (A) Dendrogram showing the results of unsupervised clustering of 314 human HCCs and murine spectrum A-10. The red cluster indicates the 13 human HCC samples with closest cosine similarity to A-10 (the blue vertical stripe). (B) HRMS of A-10 and the mutational spectra of the five human HCCs most similar to A-10 identified in A. The yellow stripe highlights the G:C→T:A hotspot in the 5′-CGC-3′ context. All five humans harbored TP53 mutations, and four specifically carried the TP53.R249S mutation. (C) Cosine similarity matrix of the red cluster from A with the murine spectrum A-10. The numbers in the matrix (also the darkness of the shade of blue) indicate the cosine similarity between compared samples. Asterisks on the bottom of the matrix indicate TP53 status (*known mutation in TP53, but not at position 249; **TP53.R249S mutation; no asterisk indicates TP53 wild type or status unknown).