Enhanced Spontaneous and Benzo(a)pyrene-Induced Mutations in the Lung of Nrf2-Deficient gpt Delta Mice

Introduction

Nrf2 is an essential transcription factor for inducible and constitutive expression of several phase II detoxification enzymes, including glutathione S-transferase-α (GST-α) and GST-π and UDP-glucuronosyl transferase 1A6 ( 1). Nrf2 also regulates the expression of antioxidant enzymes, including NAD(P)H:quinone oxidoreductase-1 and heme oxygenase-1, in response to oxidative stress ( 2, 3). Keap1 acts to harness Nrf2 to the cytoplasm, and Nrf2 in this complex rapidly undergoes ubiquitination and proteasomal degradation via Keap1-Cullin 3 E3 interactions ( 4). However, oxidative or electrophilic modification of Keap1 triggers Nrf2 stabilization ( 5, 6). During oxidative conditions, Nrf2 translocates into the nucleus and activates cytoprotective gene expression by heterodimerizing with small Maf family members and binding to antioxidant-responsive elements (ARE) or electrophile-responsive element in regulatory regions of cytoprotective genes.

Nrf2-mediated induction of cytoprotective enzymes plays an important role in mitigating the adverse effects of mutagens and oxidants. In Nrf2-deficient mice, which have attenuated basal and inducible expression of these enzymes ( 7): (a) DNA adduct formation is accelerated after diesel exhaust exposure ( 8); (b) hepatotoxicity is enhanced after acetaminophen administration ( 9); and (c) benzo(a)pyrene (BaP)–induced DNA adduct and neoplasm formation in forestomach is more prevalent than in wild-type mice ( 10, 11). Taken together, Nrf2 attenuation or malfunction may be an important aspect of diseases caused by environmental mutagens or oxidants, although the mechanism linking Nrf2 deficiency and mutation frequency is not well understood.

Transgenic guanine phosphoribosyltransferase (gpt) delta mice are a model system for detecting in vivo mutations ( 12). In this mouse system, the gpt gene is integrated into the genome as a target gene for detecting mutations, and when the gpt gene is rescued from genomic DNA to Escherichia coli, gpt mutants can be randomly selected as rescued E. coli colonies that form on plates containing 6-thioguanine (6-TG). To assess whether Nrf2 deficiency increases the mutational risk following exposure to BaP, the current study uses nrf2^−/−::gpt mice to analyze mutagenic activity in vivo. Furthermore, alterations in the mutation spectrum between nrf2^+/− and nrf2^−/− mice were assessed after exposure to BaP.

Materials and Methods

Mice. C57BL/6J nrf2 knockout mice ( 7) and gpt delta mice (C57BL/6J background; ref. 12) were as described previously, and gpt mice were obtained from Japan SLC. Nrf2-deficient mice (nrf2^−/−) were crossed with gpt delta transgenic mice (nrf2^+/+::gpt/gpt), and the resultant F1 mice (nrf2^+/−::gpt/0) were crossed again with Nrf2-deficient mice (nrf2^−/−) to produce nrf2 knockout gpt mice that are homozygous (nrf2^−/−) or heterozygous (nrf2^+/−) to the nrf2 knockout allele (nrf2^+/−::gpt and nrf2^−/−::gpt, respectively). Genotyping for nrf2 was accomplished by PCR amplification of genomic DNA isolated from tails. PCR primers were as follows: 5′-TGGACGGGACTATTGAAGGCTG-3′ (sense for both genotype) and 5′-GCCGCCTTTTCAGTAGATGGAGG-3′ (antisense for wild-type mice) and 5′-GCGGATTGACCGTAATGGGATAGG-3′ (antisense for LacZ). The presence of the gpt transgene was confirmed by PCR as previously described ( 12). Nine male Nrf2-deficient gpt delta mice (nrf2^−/−::gpt) and nine male heterozygous nrf2 knockout gpt delta mice (nrf2^+/−::gpt), both 7 to 9 weeks old, were obtained from this breeding scheme. Experiments were done according to protocols approved by the Institutional Animal Care and Use Committee at National Institute for Environmental Studies.

Mouse treatment. BaP (Wako Pure Chemical) was dissolved in tricaprylin [CH₃(CH₂)₆COOCH₂]₂CHOCO(CH₂)₆CH₃ (Sigma-Aldrich). Five nrf2^+/−::gpt mice and four nrf2^−/−::gpt mice were treated with 1 mg BaP dissolved in 50 μL tricaprylin given in a single intratracheal instillation under anesthesia with halothane for mutation analysis as previously reported ( 13). Vehicle (50 μL tricaprylin) was given to five nrf2^+/−::gpt mice and four nrf2^−/−::gpt mice as BaP-untreated groups. For immunoblot analysis, three nrf2^+/− or nrf2^−/− mice were used for each group. Mice were sacrificed 1 and 14 days after BaP administration under anesthesia with ethyl ether for Western blotting and mutation analysis, respectively. Lungs were removed, quickly frozen in liquid nitrogen, and stored at −80°C until the DNA was isolated.

gpt mutation assay. Genomic DNA was extracted from the lungs using the RecoverEase DNA Isolation kit (Stratagene). Lambda EG10 phages were recovered from the genomic DNA using Transpack Packaging Extract (Stratagene). E. coli (YG6020 expressing Cre recombinase) were infected with the recovered phage harboring the gpt gene and the chloramphenicol (Cm) acethyltransferase (cat) gene (a selection marker), and these genes were rescued as a plasmid ( 14). The gpt mutants can be detected as colonies arising on plates containing Cm and 6-TG. The bacteria were then spread onto M9 salts plates containing Cm and 6-TG, which were incubated for 72 h at 37°C for selection of the colonies harboring a plasmid carrying a mutated gpt gene and cat gene. The 6-TG–resistant colonies were streaked onto selection plates for confirmation of the resistant phenotype. The cells were then cultured in Luria-Bertani broth containing 25 μg/mL of Cm at 37°C and collected by centrifugation. The bacterial pellets were stored at −80°C until DNA sequencing analysis was done. Mutant frequencies for the gpt gene were calculated by dividing the number of colonies growing on (M9 + Cm + 6-TG) agar plates by the number of colonies growing on (M9 + Cm) agar plates, which is the number of colonies harboring the plasmid. To ensure determination of the mutant frequency, mutant colonies were selected from over 300,000 colonies ( 15).

PCR and DNA sequencing analysis of 6-TG–resistant mutants. A 739-bp DNA fragment containing the gpt gene was amplified by PCR using primer 1 and primer 2, as described previously ( 13). The reaction mixture contained 5 pmol of each primer and 200 mmol/L of each deoxynucleotide triphosphate. PCR amplification was carried out using Ex Taq DNA polymerase (Takara Bio) and done with a Model PTC-100 Thermal Cycler (MJ Research). After the PCR products were purified, sequencing reactions were done by using a DYEnamic ET Terminator kit (Amersham Biosciences). The sequencing primers (primer A and primer C) were as described previously ( 13).

Immunoblot analysis of GSTs. Frozen lung was homogenized with 2 mL of 50 mmol/L HEPES buffer (pH 7.5) containing 150 mmol/L NaCl, 1 mmol/L DTT, and 0.2 mmol/L phenylmethylsulfonyl fluoride by glass-Teflon homogenizer chilled with ice. The homogenates were subjected to two steps of centrifugation at 4°C (15,000 × g for 15 min followed by 100,000 × g for 60 min) according to Chanas et al. ( 16). Resulting 100,000 × g supernatants (cytosol fractions) were stored at −80°C until use. After the cytosol fractions mixed with sample buffer containing 1% SDS were heated at 95°C, 9 μg protein (for detecting GST A1/2) or 3 μg protein (for detecting GST A3 and GST P1/2) from each sample was subjected to SDS-PAGE with 15% polyacrylamide gel ( 17). Proteins separated on the gel were transferred to Immobilon-P membrane (Amersham Biosciences). GST A1/2, GST A3 ( 18– 20), and GST P1/2 were immunochemically detected using anti-mouse GST A1/2 and A3 rabbit sera (kindly provided by Dr. J.D. Hayes, University of Dundee, United Kingdom) and GST P1/2 rabbit serum (kindly provided by Dr. I. Hatayama, Aomori Prefecture Institute of Public Health and Environment, Japan), respectively, and goat anti-rabbit IgG antibody labeled with horseradish peroxidase ( 16). ECL-plus and Typhoon 9400 BioImage analyzer (Amersham Biosciences) were used to visualize bands.

Statistical analysis. All data are expressed as mean ± SD. Statistical significance of mutant frequency was evaluated using the Student's t test. P < 0.05 was considered statistically significant. Statistical comparisons of mutational spectra were done using the Adams-Skopek test ( 21).

Results and Discussion

The frequency of spontaneous mutations in the lung, liver, and testis was compared among gpt delta mice (nrf2^+/+), heterozygous mice (nrf2^+/−), and homozygous mice (nrf2^−/−). In the lung and liver, the mutation frequency was significantly elevated in nrf2^−/− mice, compared with nrf2^+/+ mice ( Fig. 1A ; Supplementary Table S1). The mutant frequency in the lung was approximately three times higher in nrf2^−/− mice (1.40 ± 0.28 × 10⁻⁵) than nrf2^+/− and nrf2^+/+ mice (0.48 ± 0.05 × 10⁻⁵ and 0.50 ± 0.16 × 10⁻⁵, respectively), whereas the mutant frequency was significantly higher in both nrf2^−/− and nrf2^+/− mice (1.24 ± 0.13 × 10⁻⁵ and 1.47 ± 0.15 × 10⁻⁵, respectively) than nrf2^+/+ mice in liver (0.72 ± 0.24 × 10⁻⁵). In contrast, no difference in mutation frequency was observed in testis among the three genotypes ( Fig. 1A). Whereas the DNA repair system is quite efficient in the testis ( 22), metabolically active tissues, such as the liver and lung, seem to be unable to efficiently repair the DNA adducts produced by reactive oxygen species and/or endogenous mutagens without the presence of Nrf2. These results suggest that Nrf2 acts to suppress spontaneous mutagenesis in the lung and liver.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

The mutant frequency of 6-TG selection (A) in the lung, liver, and testis of gpt delta mice (nrf2^+/+, yellow column, n = 3), and nrf2^+/− (light blue column, n = 5) and nrf2^−/− (pink column, n = 4) gpt delta mice and (B) in the lungs of nrf2^+/− (blue column, n = 5) and nrf2^−/− (red column, n = 4) gpt delta mice after BaP treatment. Data of nrf2^+/− and nrf2^−/− lungs in (A) are replicated as BaP-untreated nrf2^+/− and nrf2^−/−, respectively, in (B). Columns, mean; bars, SD. *, P < 0.05; **, P < 0.01, statistical significance among the groups was determined using the Student's t test.

We aimed to quantitatively determine how Nrf2 deficiency affects mutagenicity in vivo in the lung using a single intratracheal instillation of BaP as a model environmental mutagen/carcinogen ( 13). BaP in cigarette smoke or ambient air is readily oxidized to reactive intermediates, such as BaP diol epoxide, by phase I detoxifying enzymes ( 23), and these intermediates are subsequently metabolized to hydrophilic conjugates by phase II detoxifying enzymes that are under Nrf2 regulation. However, unconjugated reactive intermediates, which often form, lead to DNA adduct formation ( 24). DNA adducts cause mispairing of DNA bases and induce gene mutations through the DNA replication process ( 25, 26). This process has been confirmed by in vitro experiment using BaP adduct-containing DNA as a template ( 27, 28). Indeed, a single intratracheal instillation of BaP into gpt delta mice resulted in a statistically significant and dose-dependent increase in the mutant frequency in the lungs of gpt delta mice, and the most frequent mutation induced by BaP was G:C to T:A transversion ( 13), which is characteristic of BaP mutagenesis ( 25, 26).

Therefore, nrf2^−/−::gpt and nrf2^+/−::gpt mice were given a carcinogenic dose (1 mg; ref. 29) of BaP through trachea, which resulted in a significant increase in the mutation frequency in lungs of both nrf2^+/− and nrf2^−/− mice. Importantly, BaP-treated nrf2^−/− mice had a 2-fold higher mutant frequency (2.93 ± 0.56 × 10⁻⁵) than BaP-treated nrf2^+/− mice (1.47 ± 0.31 × 10⁻⁵; Fig. 1B; Supplementary Table S2). The increment of mutant frequency by BaP treatment was higher in nrf2^−/− mice than in nrf2^+/− mice.

We thought that the expression level of Nrf2-regulated cytoprotective enzymes may explain both the higher basal mutant frequency in the Nrf2-deficient mouse and that following treatment of Nrf2-deficient mouse with BaP. Because Chanas et al. have shown that the class π GST isozymes are expressed at substantially lower levels in the livers of Nrf2-deficient mice than in wild-type mice ( 16). Because a thorough study of pulmonary GSTs in Nrf2- deficient mice has not been described in the literature, we decided to examine whether expression of GSTs was actually suppressed in the lungs of BaP-treated and BaP-untreated nrf2^−/− mice by immunoblotting. In this study, we have examined expression of GST A1/2, GST A3, and GST P1/2, as these GSTs are known to be under the regulation of the Nrf2-ARE system ( 7) and are essential for the detoxification of BaP ( 30). Showing very good agreement with the report by Chanas et al. ( 16), which analyzed the expression of these enzymes in the mouse livers, the expression level of GST A1/2 was suppressed in the lungs of nrf2^−/− mice compared with that in the nrf2^+/− mice ( Fig. 2 ), and the level of GST A3 was also low in Nrf2-deficient mice (data not shown). Under the experimental condition, GST A1/2 level was not elevated substantially by the BaP treatment in the lungs of nrf2^+/− mice. Similarly, the expression level of GST P1/2 was also suppressed in the lungs of nrf2^−/− mice compared with that in the nrf2^+/− mice. GST P1/2 level was elevated by the BaP treatment in the lung of nrf2^+/− mice, but there was no such difference in nrf2^−/− mice. As the change in this immunoblotting experiment was relatively small, we repeated this experiment and found that the result was reproducible (data not shown). These results thus suggest that Nrf2 keeps the mutation frequency at low level in the lungs of mice by directing the expression the GSTs. As the changes in this GST immunoblotting experiments was relatively small, we speculate that lack of the other Nrf2 target genes may also contribute to the mutant frequency in Nrf2-deficient mice.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Immunodetection of GSTs. Cytosol fractions were extracted from the lungs of nrf2^+/− (+/−) and nrf2^−/− (−/−) mice, separated on SDS/PAGE, and electrophoretically blotted to Immobilon-P membrane. GST A1/2 and GST P1/2 were detected immunochemically using specific antibodies and ECL-plus system. BaP, cytosol fractions extracted from BaP-treated mouse lungs; Cont, cytosol fractions extracted from BaP-untreated mouse lungs.

To further characterize the mutational profile in the lungs of nrf2^−/−::gpt and nrf2^+/−::gpt mice after BaP exposure, we did DNA sequence analysis of 178 gpt mutant lung samples ( Fig. 3 ; Supplementary Table S3). In nrf2^−/− mice, the predominant spontaneous mutations were G:C to T:A transversion (26%, 15 of 58), G:C to A:T transition (24%, 14 of 58), and base deletions (19%, 11 of 58; Supplementary Table S3). A previous report of gpt delta mice (mmh/ogg1::gpt) suggested that accumulation of 8-hydroxyguanine in cells was the primary cause of increase in G:C to T:A transversion ( 31). 8-Hydroxyguanine may also play a role in the induction of G:C to T:A transversion in the lungs of nrf2^−/− mice because the level of antioxidant enzymes were suppressed in nrf2^−/− mice, and subsequently, generation of reactive oxygen spices was probably accelerated.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Comparison of mutant frequencies among the types of mutations in BaP-treated and BaP-untreated nrf2^+/− and nrf2^−/− mice. Light blue column, BaP-untreated nrf2^+/− mice; pink column, BaP-untreated nrf2^−/− mice; blue column, BaP-treated nrf2^+/− mice; red column, BaP-treated nrf2^−/− mice.

The BaP treatment increased base substitutions at G:C pairs and one base deletion both in nrf2^+/− and nrf2^−/− mice ( Fig. 3). Among the G:C substitutions, G:C to T:A and G:C to C:G transversions were markedly elevated in nrf2^−/− mice after BaP treatment. Consistent with our previous studies with gpt delta mice ( 13), the predominant mutation provoked by BaP treatment was a G:C to T:A transversion (a major base substitute induced by the BaP-DNA adduct formation) in both nrf2^+/− (32%, 14 of 44) and nrf2^−/− (34%, 21 of 62) mice (Supplementary Table S2), and the mutant frequency of this transversion was higher in BaP-treated nrf2^−/− mice than BaP-treated nrf2^+/− mice ( Fig. 3). In the lungs of nrf2^−/− mice, DNA adducts are probably accumulated in the higher level than those in nrf2^+/− mice because the expression levels of phase II enzymes that detoxify BaP by forming conjugates ( 32) and antioxidant enzymes are low in nrf2^−/− mice comparing to nrf2^+/− mice. We surmise that this increase of DNA adduct formation might elevate the mutant frequency of G:C to T:A transversion in the Nrf2-deficient condition. Additionally, generation of oxidative DNA adduct due to BaP-derived quinines ( 33) may be accelerated in nrf2^−/− mice and play a role, albeit partly, in elevating the mutant frequency in nrf2^−/− mice. Indeed, BaP adduct formation was accelerated ∼2-fold in Nrf2-deficient mouse forestomach compared with wild-type mice ( 11), supporting our contention that the increase in the amount of DNA adduct enhanced the frequency of these transversions at G:C pairs.

To delineate the mode of mutation in Nrf2-deficient mice, the mutation positions in the gpt gene of BaP-treated and BaP-untreated mice were determined ( Fig. 4A ; Supplementary Table S4). Of the mutations found in BaP-treated mice (shown in lower side of the gpt sequence; Fig. 4A), G:C to T:A transversions at nucleotides 140, 143, and 189 were observed in three or more mice, including both nrf2^+/− and nrf2^−/− mice. Thus, these nucleotides are the hotspots of BaP-induced mutation. These nucleotides coincide with those previously reported (i.e., nucleotides 115, 140, 143, 189, and 413; ref. 13), which are shown with green arrows in Fig. 4A. The frequency of mutation at these hotspots was rather low in BaP-untreated mice (shown in upper side of the gpt sequence; Fig. 4A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Overall distribution of the mutation detected on the gpt gene in the lungs of BaP-treated nrf2^+/− and nrf2^−/− mice and BaP-untreated nrf2^+/− and nrf2^−/− mice. The mutations are summarized in Supplementary Table S4. A, Mutations detected in nrf2^+/− (blue) and nrf2^−/− (pink) mice. The mutations detected in BaP-treated (below gpt sequence) and BaP-untreated mice (above gpt sequence). The number of characters in parenthesis is the number of mutations in one mouse. Δ, one base deletion; half-boxes, deleted nucleotides; V, a position of insertion. Green arrows, guanine nucleotides of BaP-induced mutation hotspots reported previously ( 13); orange characters, mutation hotspots found in this study. B, close-up of hotspots of mutations.

Because mutations were accumulated at relatively high level in the gpt gene of nrf2^−/− mice even without BaP treatment, we assumed that we could assess hotspots of spontaneous mutation in these mice. Indeed, G:C to A:T transition at nucleotides 64 was observed in three BaP-untreated mice and one BaP-treated mouse. This mutation is exclusive in nrf2^−/− mice. Thus, nucleotide 64 is the spontaneous mutation hotspot in Nrf2-deficient condition. In contrast, nucleotides 110 and 115 are common hotspots in BaP-treated and BaP-untreated mice; G:C to A:T transition at position 110 was observed in three BaP-untreated and three BaP-treated mice, and G:C to T:A transversion was also induced in three BaP-untreated mice.

Figure 4B shows mutation hotspots in the gpt gene. We found that one of the major trinucleotide sequences with BaP-induced guanine nucleotide mutation in this gene was CGT (nucleotides 110, 143, and 189). This is in good agreement with the previous observation that the instillation of BaP into the lung of gpt delta mice induced mutations frequently in CGT trinucleotide of the gene ( 13). CGG (nucleotides 64, 115, and 413) was another frequently found trinucleotide with guanine nucleotide mutations, but no link was found between this mutation and BaP treatment. It should be noted that guanine centered in CGC at position 140 was a frequent target of BaP-induced mutation in Nrf2-deficient condition, whereas previous experiments showed mutations were little in CGC of wild-type mice ( 13).

Whereas there was no significant difference in the mutation frequency, the position of the mutation was significantly different between BaP-untreated nrf2^−/− mice and BaP-treated nrf2^+/− mice ( Fig. 4A; P < 0.05, Adams-Skopek test). This result suggests that chemical mutagenesis and spontaneous mutation in the nrf2^−/− mice arise through different mechanisms. Thus, the Nrf2 deficiency had a marked affect on the mutational profile that arose either spontaneously or by BaP induction. However, further studies are required to clarify how Nrf2 deficiency alters the mutation profile, and whether nucleotides surrounding the guanine nucleotide are important for the mutation frequency in BaP-treated mice and in BaP-untreated nrf2^−/− mice.

Several lines of recent evidence have pointed towards a role for Nrf2 in prevention of carcinogenesis. One of the salient examples is that Nrf2 could prevent the formation of DNA adduct and gastric tumors from occurring after BaP administration ( 10, 11). Furthermore, Nrf2-deficient mice are sensitive to the alkylating agent [N-nitrosobutyl(4-hydrixybutyl)amine] and rapidly form bladder tumors after administration ( 34). This study shows that Nrf2 can prevent increase in the number of spontaneous and inducible mutations that occur in the gpt gene in mouse lung and liver and can prevent the induction of mutations at the hotspots, such as nucleotides 64 and 140 in the lung. We surmise that through induction of phase II and antioxidant enzyme activities as well as cross-talk with phase I detoxifying system ( 35), Nrf2 can mitigate the effects of mutagens, such as BaP, on adduct formation, leading to protection from neoplasm and tumor formation and ultimately aiding in prevention of pulmonary diseases that arise, such as lung cancer from tobacco smoke ( 36), or from hyperoxic injury ( 37).

The results presented in this study suggest that Nrf2 deficiency is a possible risk factor for development of lung cancer or other lung diseases caused by mutagens or oxidants in ambient air. Whereas molecular mechanisms by which Nrf2 deficiency changes the mutation profile still require clarification, one plausible explanation is that Nrf2 deficiency may allow accumulation of specific reactive oxygen intermediates or electrophiles. We are now examining how exaggerated mutagenesis in the Nrf2-deficient condition quantitatively contributes to the enhanced carcinogenicity. We believe that the Nrf2-deficient gpt delta mice will provide useful information for revealing the relationship between in vivo mutagenesis and carcinogenicity.