Role of Nrf2 and Oxidative stress on Fenofibrate-Induced Hepatocarcinogenesis in Rats

Abstract

Regional specific relationships between oxidative stress and the development of glutathione S-transferase placental form (GST-P)−positive or GST-P−negative lesions in rats, induced by fenofibrate (FF), a peroxisome proliferator, were examined using a two-stage hepatocarcinogenesis model in F344 rats. Animals were initiated with a single ip injection of 200 mg/kg N-diethylnitrosamine (DEN) and from 2 weeks later were fed a diet containing 3000 or ppm FF for 28 weeks. Animals were subjected to a two-third partial hepatectomy at week 3 and sacrificed at week 28. The development of hepatocellular proliferative lesions, which were mainly attributed to GST-P−negative lesions, was significantly increased in the FF-treated groups. Immunohistochemically, GST-P−positive lesions were devoid of intracytoplasmic nuclear factor-erythroid 2−related factor 2 (Nrf2) expression, whereas GST-P−negative lesions expressed higher levels of cytoplasmic Nrf2. On the other hand, nuclear accumulation of Nrf2 was observed in some cells of GST-P−positive lesions that were negative for Nrf2 in the cytoplasm and in GST-P−negative lesions of the DEN-FF group that were positive for Nrf2 in the cytoplasm. The mRNA expression levels of Gpx2 or Gsta2, Nrf2-inducible enzymes, were increased in GST-P−positive tumors or GST-P−positive lesions, respectively. These results suggest that the activation of Nrf2, due to nuclear translocation, occurs in the GST-P−positive lesions. In addition, the development of continuous oxidative stress was identified by mRNA expression analyses as well as by measurements of GST activity and 8-hydroxydeoxyguanosine. These results suggest that the relative inhibition of nuclear translocation of Nrf2 in GST-P−negative lesions aggravated the condition of oxidative stress in the liver of rats given FF, resulting in enhanced tumor promotion in FF-induced hepatocarcinogenesis.

Article

Fenofibrate (FF), a member of the fibrate class of hypolipidemic drugs, has been extensively used in many countries to treat hypertriglyceridemia and mixed hyperlipidemia (Staels et al., 1998). It belongs to the broad class of chemicals known as peroxisome proliferators (PPs), which act through the peroxisome proliferator−activated receptor α (PPARα). Information about FF has been released by the U.S. Food and Drug Administration, which shows FF to be carcinogenic to rodent species when administered at high doses; 200 mg/kg administered to rats for 24 months or to mice for 21 months increased the incidence of hepatocellular carcinomas (HCCs) in both sexes. However, FF showed no mutagenic potential in the following four tests: Ames, mouse lymphoma, chromosomal aberration, and unscheduled DNA synthesis. Therefore, FF is regarded as a nongenotoxic carcinogen, and attention has focused on possible indirect mechanisms to explain PP-induced hepatocarcinogenesis (Rao and Reddy, 1987).

The activation of PPARα induces cell proliferation and suppresses apoptosis (Boitier et al., 2003; Peters et al., 1997). PPARα also mediates the hepatocarcinogenic potential of PPs in rodents since PPARα knockout mice are nonresponsive and do not develop hepatocarcinogenesis after long-term treatment with PPs (Gonzalez, 1997; Peters et al., 1997). In addition, as an indirect mechanism, it is considered that oxidative stress is involved in hepatocarcinogenesis (Klaunig and Kamendulis, 2004). This hypothesis is based on the observation that PPARα agonists markedly induce hydrogen peroxide (H2O2)-generating enzymes, such as acyl-coenzyme A (CoA) oxidase and cytochrome P450 4A, resulting in increased levels of H2O2, leading to lipid peroxidation and oxidative DNA damage (Seo et al., 2004). We have previously identified changes that indicate DNA damage, such as elevations of 8-hydroxydeoxyguanosine (8-OHdG) and expression of DNA repair enzymes, in the liver of rats in the early stage of repeated FF toxicity and also during preneoplastic foci formation, a stage which is linked to oxidative stress (Nishimura et al., 2007, 2008). Glutathione S-transferase placental form (GST-P) is a reliable marker for preneoplastic lesions in rats by the great majority of carcinogens (Sato, 1989), while it is well known that preneoplastic and neoplastic lesions in the liver induced by PPARα agonists are not always stained with GST-P (Rao et al., 1988). The region-specific differences in expression of genes involved in oxidative stress between GST-P−positive and GST-P−negative proliferative lesions remain unclear.

Nuclear factor-erythroid 2−related factor 2 (Nrf2) is a redox-sensitive transcription factor that plays a pivotal role in the inducible expression of genes encoding detoxifying systems, including phase II drug-metabolizing enzymes (Jaiswal, 2004). These defensive enzymes are coordinately induced through the antioxidant responsive element (ARE) and are tightly regulated by Nrf2 (Nguyen et al., 2003). The roles of Nrf2 in the regulation of expression of many detoxifying and antioxidant enzymes under conditions of oxidative stress have been verified in experiments using Nrf2-deficient mice. The expression of these enzymes is dramatically attenuated, and these mice are much more susceptible to carcinogen-induced toxicity and carcinogenesis (Enomoto et al., 2001; Ramos-Gomez et al., 2001). With regard to the relationship between PPARα and Nrf2, Anderson et al. (2004) confirmed that Wy-14643, a PPARα agonist, alters the expression of stress-inducible genes by an Nrf2-independent mechanism. However, it is not clear whether or not Nrf2 is involved in the development of hepatocellular preneoplastic foci and tumors induced by PPARα agonists.

Thus, in the present study, we focused on hepatocellular proliferative lesions consisting of GST-P−positive and GST-P−negative lesions. We investigated region-specific gene expression analyses using laser microdissection (LMD) and protein localization using immunohistochemistry to reveal the relationship between oxidative stress and the development of hepatocellular proliferative lesions. To rapidly induce large numbers of these lesions, we induced an N-diethylnitrosamine (DEN)−initiated hepatocarcinogenesis model in rats, which had undergone partial hepatectomy (PH). We also performed gene expression analysis and measured total GST activity and 8-OHdG levels using whole liver tissues to estimate cellular redox status.

MATERIALS AND METHODS

Chemicals.

FF (purity, > 99%) and DEN (purity, > 99%) were purchased from Wako Pure Chemical Industries, Ltd (Osaka, Japan) and Nacalai Tesque, Inc. (Kyoto, Japan), respectively. 3,3’-Diaminobenzidine (DAB) was purchased from Wako Pure Chemical Industries, Ltd. VECTASTAIN Elite ABC kit was purchased from Vector Laboratories, Inc. (Burlingame, CA). Anti−GST-P rabbit polyclonal antibody was purchased from Medical & Biological Laboratories (MBL; Aichi, Japan). Anti-Nrf2 rabbit and anti-Gpx2 goat polyclonal antibodies were purchased from Abcam (Cambridge, UK). RNAlater was purchased from Qiagen (Hilden, Germany). Trizol reagent, SuperScript III reverse transcriptase, random primers, and Quant-iT RNA Assay Kit were purchased from Invitrogen Corporation (Carlsbad, CA). SYBR green PCR master mix was purchased from Applied Biosystems (Foster City, CA). LCM Staining Kit, RNAqueous-Micro, and MessageAmp II aRNA Amplification were purchased from Ambion (Austin, TX). BCA Protein Assay Kit was purchased from Pierce Biotechnology (Rockford, IL). All other chemicals were of analytical grade and obtained commercially.

Animals and chemicals.

Male F344/N Slc rats aged 5 weeks were purchased from Japan SLC, Inc. (Shizuoka, Japan). The rats were housed in stainless steel cages with three or four animals per cage and allowed ad libitum access to tap water and a commercial powdered basal diet (MF; Oriental Yeast Industries Co., Ltd, Tokyo, Japan). All the animals were handled under standard conditions (room temperature, 23 ± 3°C; relative humidity, 55 ± 15%; 12-h light/dark cycle). The rats were acclimatized for 1 week before the treatment with DEN (Sigma-Aldrich Chemical Co., St Louis, MO). Animal care and experiments were carried out in accordance with the Guide for Animal Experimentation of the Tokyo University of Agriculture and Technology.

Experimental procedure.

The experimental design is shown in Figure 1. We used a two-stage liver carcinogenesis model. After acclimatization, 40 animals were divided into four groups, consisting of 12 (group 1: DEN-alone group), 12 (group 2: DEN-FF group), 8 (group 3: nontreated group [NT group]), and 8 (group 4: FF-alone group) animals. Animals of groups 1 and 2 underwent ip injection of DEN (200 mg/kg) dissolved in saline to initiate hepatocarcinogenesis (2 weeks). After 2 weeks, animals of groups 2 and 4 and of groups 1 and 3 began a diet containing 3000 or ppm FF for 28 weeks, respectively. The dosage in our study was selected based on the results of our previous study (Nishimura et al., 2007). To enhance hepatocellular proliferation, animals of groups 1 and 2 were subjected to two-third PH 1 week after the start of FF treatment. Soon after PH, two rats from group 1 and one rat from group 2 died due to bleeding from the liver. Body weight and food consumption were measured once a week. Necropsy was performed under anesthesia with ether at the end of week 28 after starvation for 16 h. All remaining lobes of the liver were removed and weighed.

The liver samples were sectioned, and one section was fixed in formalin solution for histopathological examination and the other was fixed in methacarn solution, described below, for region-specific expression analysis of mRNAs using LMD. The remaining portions of normal-appearing liver tissues were frozen in liquid nitrogen for the measurement of total GST activity and levels of 8-OHdG in DNA. A portion of sample was stored at 80°C in RNAlater (Qiagen) for mRNA expression analysis.

For light microscopy, formalin-fixed liver tissues were embedded in paraffin and then sectioned. Hematoxylin and eosin (H&E) staining was conducted according to routine histopathological methods. Various immunohistochemical investigations and H&E staining to count preneoplastic foci and to assess the incidence, multiplicity, and histopathological classification of liver tumors were carried out. In addition, the Schmorl reaction was conducted to detect lipofuscin deposition.

Immunohistochemistry for GST-P, Nrf2, and Gpx2.

After deparaffinization (and target retrieval by autoclaving in citrate buffer in the case of Nrf2 and by a hot water bath [95°C] in the case of Gpx2) and incubation in 0.3% H2O2 and normal serum, sections were subjected to immunohistochemistry with a rabbit polyclonal anti-rat GST-P antibody (MBL; 1:1000, 4°C, overnight) and a goat polyclonal anti-Gpx2 antibody (Abcam; 1:100, 4°C, overnight) using a VECTASTAIN Elite ABC kit and DAB. All specimens were lightly counterstained with hematoxylin. The numbers and areas of GST-P−positive foci and the number of Nrf2-positive foci > 0.2 mm in diameter, and total areas of liver sections, were measured with the use of WinRoof software (Mitani Corp., Fukui, Japan).

Determination of enzyme activities of GST.

The cytosolic fraction was prepared from liver tissue by homogenization in ice-cold 1.15% KCl buffer (pH 7.4) containing 0.2mM ethylenediaminetetraacetic acid, 0.1mM dithiothreitol, 0.1mM phenylmethylsulfonyl fluoride, and 20% glycerin and centrifugation at 700 × g, 4°C, for 10 min. Following centrifugation of the supernatants at 10,000 × g, 4°C, for 20 min, the supernatants were additionally centrifuged at 105,000 × g, 4°C, for 90 min. Total glutathione transferase enzyme activity was determined using 1-chloro-2,4-nitrobenzene (CDNB) according to the procedure of Habig et al. (1974). Cytosol (25 μl) was added to 200 μl 0.1M phosphate buffer, pH 6.5, followed by 225 μl distilled water with 1mM glutathione, 1mM CDNB, mixed, and optical absorbance was read at 340 nm at 10-s intervals over 5 min. Activity was defined as nanomoles CDNB conjugate formed/min/mg protein.

Measurement of 8-OHdG levels in liver DNA.

The measurement of 8-OHdG levels in liver DNA was performed as described previously (Nishimura et al., 2007, 2008). Nuclear DNA was extracted using a DNA Extractor WB Kit (Wako Pure Chemical Industries, Ltd). During the extraction, an iron chelator was used to prevent DNA oxidation. The DNA was digested to deoxynucleotides using nuclease P1 and alkaline phosphatase, and the level of 8-OHdG (8-OHdG/105 deoxyguanosine) was assessed by high-performance liquid chromatography using an electrochemical detection system (Coulochem II; ESA Biosciences, Inc., MA).

RNA isolation and gene expression analyses using whole liver tissues.

Total RNA from normal-appearing liver tissue was isolated from six animals of each group using Trizol reagent (Invitrogen Corporation) according to the manufacturer's protocol. Quantitative real-time reverse transcriptase (RT)-PCR analyses using the SuperScript III First-Strand Synthesis System (Invitrogen Corporation), SYBR green PCR master mix (Applied Biosystems), and an ABI Prism 7000 Sequence Detection System (Applied Biosystems) were performed. Taking into account our previous reports (Nishimura et al., 2007, 2008), we assessed gene expression of the following 14 genes: acyl-CoA oxidase 1 (Aco), cytochrome P450, 4A1 (Cyp4a1), apurinic/apyrimidinic endonuclease 1 (Apex1), X-ray repair complementing defective repair in Chinese hamster cells 5 (Xrcc5), glutathione peroxidase 2 (Gpx2), DNA damage−inducible 45 alpha (Gadd45a), MutL homolog 1 (Mlh1), nibrin (Nbn), 8-oxoguanine DNA glycosylase (Ogg1), UDP glycosyltransferase 1 family, polypeptide A6 (Ugt1a6), glutathione S-transferase Yc2 subunit (Yc2), glutathione S-transferase, alpha type2 (Gsta2), glutathione S-transferase, mu type 2 (Gstm2), and glutathione S-transferase, mu type 3 (Gstm3). The PCR primers were designed using the Primer Express software (Applied Biosystems). The primer sequences are listed in Table 1. To obtain the relative quantitative values for gene expression, β-actin (Actb) was used as an endogenous control, and its expression levels were calculated according to the 2−ΔΔCt method (Livak and Schmittgen, 2001).

Tissue fixation for LMD.

Methacarn solution, consisting of 60% (vol/vol) absolute methanol, 30% chloroform, and 10% glacial acetic acid was used as fixative for LMD, based on the report of Shibutani et al. (2000). Methacarn solution was freshly prepared before fixation and stored at 4°C until use. At necropsy, liver slices were fixed in methacarn solution for 4−5 h at 4°C with gentle agitation. For embedding, liver slices were dehydrated three times for 1 h in fresh 99.5% ethanol at 4°C, immersed in xylene for 1 h and then three times for 30 min at room temperature, and then immersed in hot paraffin (60°C) four times for 1 h, for a total of 4 h. Paraffin-embedded tissues (PETs) were stored at 4°C until use.

Preparation of tissue specimens and microdissection.

Three serial 20-μm thick sections for evaluation of region-specific gene expression analyses using LMD and two serial 2-μm thick sections for identification of GST-P−positive and GST-P−negative lesions for microdissection were prepared from methacarn-fixed rat liver PETs, respectively. Four sets of these sections were prepared. The 20-μm thick sections were mounted on PEN-foil film (Leica Microsystems Japan) and then dried in an incubator overnight at 37°C. The 2-μm thick sections were subjected to routine H&E staining and immunohistochemical staining for rat GST-P to identify proliferative lesions for the microdissection. The sections for LMD were deparaffinized and stained with cresyl violet using LCM Staining Kit (Ambion) according to the manufacturer's protocol and air-dried. GST-P−positive and/or GST-P−negative foci/tumor and corresponding surrounding tissues were microdissected with the use of an LMD system (Leica Microsystems Japan) from five animals of each of the DEN-alone and DEN-FF groups. The presence of GST-P−positive and/or GST-P−negative lesions in the sections for LMD was confirmed by setting the microscopic figures in the sections for LMD against those in the serial sections of H&E staining and GST-P immunohistochemical staining.

Total RNA isolation and region-specific quantitative real-time RT-PCR assays.

Total RNA was isolated using RNAqueous-Micro (Ambion) (and if RNA amplification was required, MessageAmp II aRNA Amplification [Ambion] was applied), according to the manufacturer's instructions. cDNAs were synthesized from 100 ng (if RNA amplification was not performed) or 1 μg (if RNA amplification was performed) of total RNA as the template. To assess the mRNA levels of genes involved in oxidative and metabolic stress in the preneoplastic or neoplastic lesions, we assessed the mRNA levels of the following six genes with the use of real-time RT-PCR: Aco, Cyp4a1, Apex1, Xrcc5, Gpx2, and Gsta2. To obtain the relative quantitative values for gene expression, β-actin (Actb) was used as an endogenous control, and its expression levels were calculated according to the 2−ΔΔCt method (Livak and Schmittgen, 2001).

Statistical evaluation.

All results were presented as mean ± SD. Since the experimental design was 2 × 2 (as indicated in Fig. 1), the data shown in Tables 2 and 3 were analyzed by the two-way ANOVA. After performing the ANOVA, Tukey-type multiple comparison test was adopted to make a comparison between the groups. The incidences and multiplicities of neoplastic lesions observed in the DEN-alone and DEN-FF group were analyzed by the Fisher's exact test and F-test, respectively. The differences of region-specific mRNA expression in multigroups between the surrounding tissue of DEN-alone group and other regions, between the GST-P−positive foci of DEN-alone group and GST-P−positive or −negative foci of DEN-FF group, and between the surrounding tissue of DEN-FF group and other regions of DEN-FF group were analyzed by Dunnett's multiple comparison test, following the test of the homogeneity of variance between the groups by using Bartlett's test. When the data were homogenous, Dunnett's test was used, and when heterogeneous, Dunnett's rank sum test was used. A p value of less than 0.05 was considered statistically significant.

RESULTS

Body Weights, Food Consumption, and Liver Weights

The results are shown in Table 2. All rats survived until their scheduled sacrifices except for three animals that did not survive PH. Two-way ANOVA showed that final body weights, food intake, and relative liver weights were attributed to the FF treatments. With respect to the comparison between the DEN-alone group and DEN-FF group, final body weights significantly decreased and relative liver weights significantly increased in the DEN-FF group as compared with the DEN-alone group.

Histopathological Examinations, Schmorl Reaction, and Immunohistochemistry for GST-P

The criteria used for histopathological evaluation of the hepatocellular proliferative lesions were those stated in Guides for Toxicologic Pathology, published by the Society of Toxicologic Pathology (Goodman et al., 1994), and the results of histopathological examinations and quantitative analyses of hepatocellular altered foci (HAF) and GST-P−positive foci are shown in Table 2. By histopathological examination, HAF, consisting of eosinophilic or basophilic hepatocytes, hepatocellular adenomas (HCAs), and HCCs were found in the livers of DEN-alone and DEN-FF groups. The results of the two-way ANOVA showed that the number of HAF and the multiplicity of HCAs were affected by the factors such as FF treatments and both treatments with DEN initiation and PH, and interactions of these factors were observed in the case of multiplicity of HCAs. The multiplicity of carcinoma was also affected by both treatments with DEN initiation and PH. In addition, Tukey-type multiple comparison test showed that the number of HAF and multiplicity of HCAs significantly increased in the DEN-FF group as compared with the DEN-alone group. Fisher's exact test showed that the incidence of HCAs significantly increased in the DEN-FF group as compared with the DEN-alone group. On the other hand, although the multiplicity of HCCs was affected by both treatments with DEN initiation and PH, no significant differences were observed between the DEN-alone and DEN-FF groups. With respect to the nontreated control and FF groups, a few number of HAF was observed in the FF group. However, the number of HAF in the FF group was very scarce, and these values were about 1/30 of those in the DEN-FF group.

In addition, histopathological examinations and immunohistochemical examinations for GST-P demonstrated an obvious increase in GST-P−negative tumors in the DEN-FF group but not in the GST-P−positive lesions in the DEN-FF group compared with the DEN-alone group. The two-way ANOVA showed that the number of GST-P−negative tumors was affected by the factors such as FF treatments, both treatments with DEN initiation and PH, and their interactions. Although the number and area of GST-P−positive foci and the number of GST-P−positive tumors were affected by the treatments with DEN initiation and PH, no significant differences were observed between the DEN-alone and DEN-FF groups. The number and area of GST-P−positive foci in the FF group were very scarce, and these values were about 1/10 of those in the DEN-FF group.

The Schmorl reaction revealed that most hepatic altered foci and tumors in the DEN-FF group had small amounts of lipofuscin, while high levels of depositions of this pigment were observed in the surrounding normal hepatocytes (Figure 3C).

Measurement of GST Activity in Liver Tissues and 8-OHdG Levels in Liver DNA

The results are shown in Table 2. The results of two-way ANOVA showed that both results of total GST activity and 8-OHdG levels in liver DNA were affected by the FF treatments. In addition, total GST activity was significantly decreased in the DEN-FF group compared with that in the DEN-alone group. The 8-OHdG levels in liver DNA of the DEN-FF group were significantly increased compared to the corresponding values in the DEN-alone group. Since there were no changes in the 8-OHdG values between the DEN-alone and the NT group, it was recognized that treatment with DEN had no effect on the formation of 8-OHdG, indicating that the production of 8-OHdG was induced by the continuous administration of FF.

Gene Expression Analyses Using Whole Liver Tissues

The results of real-time RT-PCR are shown in Table 3. The results of two-way ANOVA showed that mRNA expressions of genes except for Mlh1 and Yc2 were affected by the FF treatments. The effect of interactions in addition to FF treatment was observed in the case of Mlh1. In the case of Yc2, the effect of both treatments with DEN initiation and PH in addition to FF treatments was also observed. Comparison between DEN-alone and DEN-FF groups revealed significantly increased expressions of Aco, Cyp4a1, Apex1, Xrcc5, Mlh1, Nbn, Gadd45a, Gpx2, Ugt1a6, and Yc2 and decreased Gsta2, Gstm2, and Gstm3 in the DEN-FF group. On the other hand, the changes observed in the FF-alone group were similar to those of DEN-FF group. In Ogg1 mRNA, no significant differences were observed between the DEN-alone group and DEN-FF groups.

Gene Expression Analyses in Microdissected Regions of Rat Livers

Among the above 14 genes, the expression levels of six genes, Aco, Cyp4a1, Apex1, Xrcc5, Gpx2, and Gsta2, were examined to reveal differences of region-specific expression of genes involved in oxidative stress, DNA repair, and phase II drug-metabolizing enzymes. The mRNA levels of Aco and Cyp4a1, which are involved in the production of H2O2, were significantly increased in GST-P−positive and GST-P−negative foci of the DEN-FF group compared with the GST-P−positive foci of the DEN-alone group, while these expression levels were significantly decreased in GST-P−positive foci of the DEN-FF group compared with the surrounding tissue of the DEN-FF group (Figure 2). The mRNA levels of Apex1 and Xrcc5, which are involved in DNA repair, were significantly increased in GST-P−positive and GST-P−negative foci of the DEN-FF group compared with GST-P−positive foci of the DEN−alone group, while obvious changes were not observed in GST-P−positive and GST-P−negative foci of the DEN-FF group compared with the surrounding tissue of DEN-FF group (Figure 2). The mRNA levels of Gpx2 and Gsta2 were significantly increased in the GST-P−positive foci and/or tumors of the DEN-FF group compared with those of surrounding tissue of the DEN-FF group, while these levels in GST-P−negative foci and tumors of the DEN-FF group were not significantly increased in the surrounding tissues of the DEN-FF group (Figure 3A).

Protein Levels of Transcription Factor Nrf2 and of Nrf2-Inducible Phase II Drug-Metabolizing Enzyme, Gpx2, in the Liver of FF-Treated Rats

Nrf2 has a close connection to oxidative stress and is a central transcription factor involved in the transcriptional activation of many genes encoding phase II drug-metabolizing enzymes via the ARE (Hayes and McMahon, 2001). Therefore, we conducted immunohistochemistry to determine the localization of transcription factor Nrf2 in GST-P−positive or GST-P−negative foci or tumors. The cytoplasm of most cells forming GST-P−positive foci and/or tumors in the DEN-alone and DEN-FF groups was negative for Nrf2, but the accumulation of Nrf2 was observed in the nuclei of some of these cells (Figure 4C). In contrast, the cytoplasm of most cells forming GST-P−negative foci and/or tumors was positive for Nrf2, although some of the nuclei of these cells were also positive for Nrf2 (Figure 4A and B). In the quantitative evaluation of Nrf2-positive HAF (> 0.2 mm), the number and area of Nrf2-positive foci in the DEN-FF group was significantly increased compared to that of the DEN-alone group and was increased twofold in number and eightfold in area compared to the GST-P−positive foci in the DEN-FF group (Figure 4E). Because there may be differences in Nrf2 function between GST-P−positive or GST-P−negative foci, we performed immunohistochemistry for Gpx2, which is transcriptionally regulated by activated Nrf2 (Dewa et al., 2008). Contrary to the fluctuation of Nrf2, the expression of Gpx2 protein was observed in some lesions that expressed GST-P protein in the DEN-alone and DEN-FF groups, while those of GST-P−negative foci and/or tumors were not observed (Figure 4D).

DISCUSSION

The present analyses using whole liver revealed that FF continuously exerted increased fatty acid oxidation and DNA-damaging effects, such as elevation of 8-OHdG in liver DNA, and increased mRNA levels of DNA repair enzymes during the stage of tumor formation. In addition, decreased GST activity was observed with concomitant decrease in the expression of some GST subunits in the FF-treated groups. GST substrate, CDNB, has the broadest range of isozyme detectability (e.g., alpha-, mu-, pi-, and other GST isoforms). Therefore, it was considered that the GST activity in the present study was the value containing the activity of the isozyme assayed for by either of the genes assayed by mRNA levels. These findings suggest the possibility that the oxidative stress was induced continuously and decreased activities of its eliminating enzymes further perturb redox balances. Furthermore, under such circumstances, a significant increase in the number of HAF as well as the incidence and multiplicity of HCAs observed in the DEN-FF group compared with the DEN-alone group strongly suggest that FF has a liver tumor-promoting effect through the production of reactive oxygen species. On the other hand, a limited accumulation of lipofuscin, which is known as an indirect indicator of oxidative stress (Reddy et al., 1982), was observed in hepatocellular proliferative lesions induced by FF, in contrast to excess lipofuscin accumulation in surrounding normal liver tissues. This finding suggests that there are different regulations in the elimination of oxidative stress between the proliferative lesions and surrounding tissues. Reddy et al. (1982) noted that the relative amount of oxidative stress decreased with increased cell proliferation and mitosis in neoplastic cells of F344 rats treated with the PP, methyl clofenapate. In contrast, various analyses using whole liver tissues in the present study did not indicate any differences between the FF-alone and DEN-FF groups except for histopathological findings. These findings indicate that treatment with DEN does not make any qualitative differences in the liver of rats given FF. Thus, in the present study, we focused on hepatocellular proliferative lesions consisting of GST-P−positive and GST-P−negative lesions and performed region-specific analyses using LMD and gene expression analyses and immunohistochemistry.

First, we investigated differences in the region-specific mRNA levels of H2O2-generating enzymes, such as Aco and Cyp4a1, and DNA repair enzymes, such as Apex1 and Xrcc5. These genes were shown to be upregulated in our previous studies (Nishimura et al., 2007, 2008) as well as by the analyses of the present study using the whole liver. Our data in the region-specific analyses demonstrated decreased levels of Aco and Cyp4a1 mRNA, which are regulated by PPARα, in GST-P−positive foci compared with the surrounding tissue of the DEN-FF group, although these levels significantly increased in the GST-P−positive and GST-P−negative foci of the DEN-FF group compared with the GST-P−positive foci of the DEN-alone group. These data suggest the possible involvement of differences in PPARα activation between GST-P−positive and GST-P−negative foci in the liver of rats given FF. Indeed, there have been several reports demonstrating repression of peroxisomal bifunctional enzyme (enoyl-CoA hydratase/l-3-hydroxyacyl-CoA dehydrogenase [BE]), expression in hepatic lesions of rats induced by PPs (Yokoyama et al., 1993), and repressed PPARα expression in BE-negative foci (Kudo et al., 2006). With respect to the DNA repair enzymes, mRNA analyses using LMD in liver tumors of mice induced by dicyclanil has demonstrated no remarkable upregulation of an oxidative DNA damage repair gene, such as Ogg1, combined with the negative regulation of apoptosis and the induction of oxidative stress (Moto et al., 2006). This means that there is a high possibility of reduced DNA repair ability in these hepatocellular proliferative lesions compared with the normal surrounding areas. In contrast, in our study, the expressions of Apex1 and Xrcc5 in the hepatocellular proliferative lesions of the DEN-FF group were higher than those of GST-P−positive foci of the DEN-alone group. However, there were no marked differences in the expressions of Apex1 and Xrcc5 between the hepatocellular proliferative lesions and surrounding tissue of the DEN-FF group. These results may suggest the possibility that FF-induced hepatocellular proliferative lesions, including GST-P−positive and GST-P−negative lesions, still preserve DNA repairing ability during the stage of tumor formation. In other words, our data suggest that continuous oxidative stress was induced in all areas of the liver and resulted in increased 8-OHdG levels in the liver DNA of rats in the DEN-FF group. On the other hand, these results may lead the supposition that these genes have a slight effect on the growth of hepatocellular proliferative lesions of the DEN-FF group. However, the details were unknown, but the fact that PPARα knockout mice do not induce any increase in the expression of DNA repair genes (Rusyn et al., 2004) and do not develop hepatocarcinogenesis (Gonzalez, 1997; Peters et al., 1997) will suggest that there are some relationship between the fluctuated DNA repair genes and tumor development of FF.

Nuclear levels of Nrf2 increase when peritoneal macrophages are treated with oxidative stressors, such as diethyl maleate and paraquat (Ishii et al., 2000). These changes were also found in the liver when D3T and β-naphthoflavone were administered to rats (Kwak et al., 2001a, 2001b). Our present study revealed different levels of expression of Nrf2 in GST-P−positive and GST-P−negative foci induced by FF. This is the first report to identify the different distribution of Nrf2 in GST-P−positive and GST-P−negative lesions induced by PPs. Immunohistochemistry for Nrf2 in the GST-P−positive lesions showed decreased Nrf2 expression in the cytoplasm, but the nuclei of some cells forming these altered foci and tumors of the DEN-FF group were immunoreactive for Nrf2. In contrast, GST-P−negative foci and tumors showed enhanced Nrf2 expression in the cytoplasm, while there were some cells showing nuclear accumulation of Nrf2 in these proliferative lesions. In addition, our region-specific analyses revealed overexpression of Gpx2 and/or Gsta2, which are known to be regulated by Nrf2 (Dewa et al., 2008; Rushmore et al., 1990), and enhanced Gpx2 protein levels in some GST-P−positive lesions of the DEN-alone and/or DEN-FF groups. Gpx2 mRNA has been reported to be upregulated in the liver GST-P−positive foci by other researchers (Suzuki et al., 2004). These results suggest that the activation of Nrf2, due to nuclear translocation of Nrf2, was induced in GST-P−positive proliferative lesions. On the other hand, in the present study, the cytoplasm of most cells forming GST-P−negative foci and/or tumors was positive for Nrf2, although some of the nuclei of these cells were also positive for Nrf2. Such a relative increase in intracytoplasmic accumulation of Nrf2 in GST-P−negative lesions may result from inactive Nrf2 accumulation in the cytoplasm. In addition, there have been some reports that Nrf2 acts as a positive transcription factor for GST-P expression (Ikeda et al., 2004; Ohta et al., 2007). Fan et al. (2008) previously reported that GST-P−positive foci were devoid of Nrf2 expression and GST-P−negative foci expressed higher levels of Nrf2 in HAF induced by the Solt-Farber protocol. Their interpretation, with respect to Nrf2 expression, is different from our interpretation as our data strongly support the view of Ikeda et al. and of Ohta et al. GST-P also modulates the activities of p38 mitogen-activated protein kinase (Yin et al., 2000). This factor was activated by various stress-associated stimuli, cytokines, and growth factors and control many aspects of mammalian cellular physiology, including cell growth, differentiation, and cell death (Ip and Davis, 1998; Kyriakis and Avruch, 1996). Taking into account these findings and the results of our study, it can be considered that inhibition of transcriptional factors, via decreased GST-P expression, and of nuclear translocation of Nrf2 occurs in GST-P−negative proliferative lesions induced by FF and may enhance the tumor promotion in these proliferative lesions.

In conclusion, our data demonstrated the different localization of Nrf2 protein and of Nrf2-regulated enzymes between GST-P−positive and GST-P−negative lesions in rats initiated with DEN and promoted by FF and suggest that the activation of Nrf2 by nuclear translocation occurs in the GST-P−positive lesions. In contrast, although no obvious regional differences in mRNA levels of DNA repair enzymes were observed in the liver of rats given FF, the regional differences of the expression of eliminating enzymes of oxidative stress such as Gsta2 and Gpx2 were observed in the liver of rats given FF. The development of continuous oxidative stress was identified in whole liver by mRNA expression analyses, as well as by measurements of GST activity and 8-OHdG levels. In addition, the cytoplasm of most cells forming GST-P−negative foci and/or tumors was positive for Nrf2, although some of the nuclei of these cells were also positive for Nrf2. These results suggest that the relative inhibition of nuclear translocation of Nrf2 in GST-P−negative lesions aggravates the condition of the oxidative stress in the liver of rats given FF, resulting in enhanced tumor promotion in FF-induced hepatocarcinogenesis. Additional investigations are now in progress to clarify further the mechanism of tumor promotion in FF-induced hepatocarcinogenesis.

References