Toxicity Profiles in Mice Treated with Hepatotumorigenic and Non-Hepatotumorigenic Triazole Conazole Fungicides: Propiconazole,Triadimefon,and Myclobutanil

Abstract

Conazoles comprise a class of fungicides used in agriculture and as pharmaceutical products. The fungicidal properties of conazoles are due to their inhibition of ergosterol biosynthesis. Certain conazoles are tumorigenic in rodents; both propiconazole and triadimefon are hepatotoxic and hepatotumorigenic in mice, while myclobutanil is not a mouse liver tumorigen. As a component of a large-scale study aimed at determining the mode(s) of action for tumorigenic conazoles, we report the results from comparative evaluations of liver and body weights, liver histopathology, cell proliferation, cytochrome P450 (CYP) activity, and serum cholesterol, high-density lipoprotein and triglyceride levels after exposure to propiconazole, triadimefon, and myclobutanil. Male CD-1 mice were treated in the feed for 4, 30, or 90 days with triadimefon (0, 100, 500, or 1800 ppm), propiconazole (0, 100, 500, or 2500 ppm) or myclobutanil (0, 100, 500, or 2000 ppm). Alkoxyresorufin O-dealkylation (AROD) assays indicated that all 3 chemicals induced similar patterns of dose-related increases in metabolizing enzyme activity. PROD activities exceeded those of MROD, and EROD with propiconazole inducing the highest activities of PROD. Mice had similar patterns of dose-dependent increases in hepatocyte hypertrophy after exposure to the 3 conazoles. High-dose exposures to propiconazole and myclobutanil, but not triadimefon, were associated with early (4 days) increases in cell proliferation. All the chemicals at high doses reduced serum cholesterol and high-density lipoprotein (HDL) levels at 30 days of treatment, while only triadimefon had this effect at 4 days of treatment and only myclobutanil and propiconazole at 90 days of treatment. Overall, the tumorigenic and nontumorigenic conazoles induced similar effects on mouse liver CYP enzyme activities and pathology. There was no specific pattern of tissue responses that could consistently be used to differentiate the tumorigenic conazoles, propiconazole, and triadimefon, from the nontumorigenic myclobutanil. These findings serve to anchor other transcriptional profiling studies aimed at probing differences in key events and modes of action for tumorigenic and nontumorigenic conazoles.

Introduction

Conazoles are azole antifungal agents used in agricultural and pharmaceutical products (Zarn et al., 2003). They are applied as fungicides in fruit, vegetable, and cereal crop protection programs, and also in lawn care and wood preservation. Medically, they are widely used to treat local and systemic fungal and yeast infections (Georgopapadakou and Walsh, 1996); some with anti-estrogenic activity are also being used clinically for treatment of prostate and breast cancer (Ronis et al., 1994; Zarn et al., 2003). The fungicidal properties of conazoles are due to their abilities to inhibit ergosterol biosynthesis (Van den Bossche et al., 1989; Ronis et al., 1994; Debeljak et al., 2003). Since ergosterol is an essential component of fungal membranes, its inhibition leads to cell death. These chemicals are representative of a larger class of ergosterol biosynthesis inhibiting fungicides (EBIFs), which act primarily by inhibiting lanosterol 14α-demethylase (encoded by the CYP51 gene) enzyme activity in sterol biosynthesis, a metabolic pathway in ergosterol production. Conazoles contain either a 1,2,4-triazole ring or imidazole moiety that prevent enzymatic activity through complexing with the heme iron of cytochrome P450 (CYP) enzymes.

Conazole fungicides can also inhibit lanosterol 14α-demethylase in exposed animals. In mammals, this enzyme has a similar function to that in fungi; however, it resides in the analogous cholesterol biosynthesis pathway (Iglesias and Gibbons, 1989; Stromstedt et al., 1996; Gibbons, 2002). Since cholesterol is a substrate for subsequent steps in the production of other sterols, e.g., sex steroid hormones, the disruption of this pathway can lead to endocrine changes and abnormalities in reproduction, development, and fertility (Georgopapadakou and Walsh, 1996; Zarn et al., 2003). Several conazoles have also been reported to cause neurotoxic and developmental effects in experimental animals (Moser et al., 1995, 2001). Some conazoles are hepatotoxic and hepatotumorigenic in mice (EPA 2004), while others induce thyroid follicular cell tumors in rats (Hurley, 1998); a few are active in both species for these effects (EPA, 1996). Conazole (e.g., ketoconazole) hepatotoxicity has also been reported in humans (Lewis, 2000). However, cellular targets in metabolic pathways and functional consequences involved in conazole tumorigenesis are not well understood.

In view of the widespread usage of conazole fungicides, and potential for cumulative human exposures by multiple routes, an improved understanding of the mode(s) of action leading to these tumorigenic effects in rodents is important for assessing potential human cancer risks. Agricultural chemical carcinogens have generally proven to be nongenotoxic in standardized tests (Carmichael et al., 1997); for conazoles, the mostly negative mutagenicity test results are consistent with this (INCHEM, 1987, 1992; EPA, 1996, 2004; Kevekordes et al., 1996). Some conazoles possess properties often associated with tumor promotion in mice; in particular, potentials to alter liver CYP enzyme activities and/or stimulation of cell proliferation (Butterworth et al., 1995; Carmichael et al., 1997; Klaunig and Kamendulis, 2004). Some conazoles are known to induce or inhibit rodent CYP enzymes from the CYP1, CYP2, and CYP3 families (Ronis et al., 1994; Sun et al., 2005, 2006), which function in the metabolism of structurally diverse exogenous chemicals (Honkakoski and Negishi, 2000; Suzuki et al., 2000).

The present studies were aimed at comparing the effects of 2 conazoles known to induce liver tumors in mice with a conazole that has not been shown to induce any treatment-related tumorigenic effects at similar doses. The tumorigenic agents of interest were triadimefon and propiconazole, both of which have been classified by the EPA (1996, 2004) as “possible human carcinogens” (class C-based on limited evidence in animals). Triadimefon increases thyroid follicular cell tumors in treated rats, and liver adenomas in both sexes of 2 strains of treated mice (INCHEM, 1981; EPA, 1996). Propiconazole has been shown in several studies to induce liver lesions, adenomas, and carcinomas in treated mice (INCHEM, 1987; EPA, 2004). Propiconazole has also been shown to be a rat liver tumor promoter (INCHEM, 1987; Hakoi et al., 1992). While myclobutanil has been determined to induce liver pathological effects (e.g., hypertrophy, single-cell necrosis, and vacuolization) in mice, negative results were obtained for liver tumorigenicity (INCHEM, 1992). All 3 of these chemicals have been found to be negative in standardized microbial and mammalian genotoxicity assays (INCHEM, 1987, 1992; EPA, 1996, 2004). Our experimental design in the present studies has included high-dose treatments (MTD) equivalent to those reported in mouse studies to increase the incidences of propiconazole- and triadimefon-induced liver tumors, and to give no increase in tumors by myclobutanil.

We compared the effects of these tumorigenic and non-tumorigenic conazoles on mouse liver CYP enzyme levels, body and liver weights, cell proliferation, levels of serum cholesterol and lipids, and histopathology. This work is part of a larger toxicogenomic study (Ward et al., 2006) aimed at determining the mode(s) of action for these selected hepatotumorigenic conazoles.

Materials and Methods

Chemicals

Chemicals obtained from Sigma Chemical Co. (St. Louis, MO) for these studies included the following: ethoxyresorufin, pentoxyresorufin, methoxyresorufin, resorufin, β-nicotinamide adenine dinucleotide phosphate (NADP, 98%), glucose-6-phosphate (G-6-P, 98%), glucose-6-phosphate de-hydrogenase (G-6-P DH), KCl (99%), MgCl₂ (98%), KHPO (98%), CuSO₄·5H₂O (99%), DL-dithiothreitol (DTT, Cleland’s Reagent, 99%), EDTA (99.8%), glycerol (99%), sucrose (99.5%), protein standard (bovine serum albumin), Folin-Ciocalteau reagent, and K-Na tartrate (99%). Na₂CO₃ and Na₂HPO₄ were purchased from Fisher Scientific Co. (Pittsburgh, PA).

Diet Preparation of Conazole-Adulterated Feeds

All treated feeds were prepared and analyzed by Bayer CropScience at Stilwell, KS under GLP conditions during a 21-week period. Every 2 weeks the feeds were prepared and shipped to the U.S. EPA in Research Triangle Park, NC for use in the rodent bioassays. Briefly, triadimefon (Bayleton; 96.1% active ingredient; Bayer CropScience, Stilwell, KS), propiconazole (Orbit; 94.2% active ingredient; Syngenta Crop Protection, Greensboro, NC), and myclobutanil (Eagle; 95.8% active ingredient; Dow AgroSciences LLC, Midland, MI) were dissolved in acetone and added dropwise to Purina Mills Certified Rodent Diet 5002 Meal in a feed blender (Hobart mixer). The mixed feed was stored at room temperature in covered plastic containers until it was shipped to US EPA at which time it was stored at 4°C. Feed samples used for chemical analysis consisted of 100 g (or greater) lots taken at least twice from different locations within the container. Quantitative analyses of the levels of conazoles in the treated feeds were accomplished by LC-MS/MS analyses (using deuterated internal standards) of methanol extracts of the adulterated feed samples. The target concentrations and measured mean concentrations (± S.D.) of each conazole determined over the 21-week period are summarized in Table 1.

Animal Treatments and Weight Measurements

The experimental design was based on treating groups of mice over 4, 30, or 90 days with a powdered feed mixture containing either a tumorigenic (propiconazole, triadimefon) or nontumorigenic (myclobutanil) conazole. All animals were housed in an AAALAC-International accredited USEPA animal facility, and all procedures involving the use of animals were approved by the Institutional Animal Care and Use Committee. Male CD-1 mice, 30 days old, were obtained from Charles River Laboratories (Raleigh, NC) and acclimated for 7 days in the animal facility prior to commencing exposure. Animals were ear-tagged with unique identifying numbers, and randomly assigned and housed 2 per polycarbonate cage with Alpha-Dry bedding in a room under a 12:12-hour light:dark cycle, with controlled temperature (22°C) and humidity (45%).

Water and conazole-feed mix in open-top stainless steel containers were provided ad libitum. Animals were monitored daily and cages and bedding were changed twice/week. Containers of feed were weighed at regular intervals to determine feed and chemical consumption. Mice were individually weighed at the start of the experiment and regularly (typically every 1–2 weeks) thereafter.

Groups of mice were provided conazole adulterated feeds at 1 of 3 different concentrations consistent with those reported in previous cancer bioassay and toxicity studies (INCHEM, 1987, 1992; EPA, 1996, 2004). The 2500 ppm propiconazole feed concentration and the 1800 ppm triadimefon feed concentration have been reported to be tumorigenic while the 2000 ppm myclobutanil feed concentration was a nontumorigenic maximum tolerated dose in a previous chronic bioassay. Negative control groups were fed only the Purina Mills 5002 rodent diet. At the end of each treatment period the animals were euthanized by CO₂ asphyxiation.

Histopathological Examination

For liver histopathology analyses, a section of liver from each mouse was taken for fixation in 10% neutral buffered formalin and processed by routine methods to 5 micron paraffin sections, stained with hematoxylin and eosin, and analyzed with light microscopy. Liver alterations were scored based on the severity of hepatocyte hypertrophy, which was the only alteration present in the majority of samples. A few slides did, however, also have hepatocyte vacuolation in addition to the hypertrophy. The lesion scoring was based on the following criteria: 0—no lesion present, 1—centrilobular hypertrophy, 2—centrilobular and midzonal hypertrophy, 3—panlobular hypertrophy, 4—panlobular hypertrophy with cytoplasmic vacuolation.

Cell Proliferation

Liver cell proliferation was measured in 4-, 30-, and 90-day treated mice. Tissue samples in paraffin blocks were sectioned and stained for proliferating cell nuclear antigen (PCNA) by the Zymed-PCNA Staining Kit (Experimental Pathology Laboratories, Inc., Durham, NC). The % labeling indices, (LI), were determined by counting the number of positive-stained PCNA in 900–1000 hepatocyte nuclei/animal. A Cytology/Histology Recognition Information System (CHRIS, Sverdrup Technology, Inc., Ft. Walton Beach, FL) was used to quantitate the LI from PCNA-stained slides. Each image was examined and edited for accuracy (Medinsky et al., 1999). All liver slides were read without knowledge of treatment dose or duration.

Cholesterol, Triglycerides, High-Density Lipoproteins

Serum samples from 5 mice per control and conazole treatment group were analyzed for total cholesterol, triglycerides and high-density lipoprotein (HDL). Blood was collected by cardiac puncture at necropsy and placed in serum separation tubes (Becton Dickinson), allowed to clot, centrifuged at 1000 × g, 4°C for 20 minutes and stored in 2 ml microtubes. Serum panel analyses for triglycerides, cholesterol, and high-density lipoproteins were measured by LabCorp, Research Triangle Park, NC, using a Roche Hitachi 717 Chemistry Analyzer.

Liver Microsome Preparation

Liver microsomes were prepared in accordance with published procedures (Matsuura et al., 1991), after some modifications. Briefly, fresh livers were washed with cold 1.15% KCl and 0.25 M sucrose and minced into small pieces. The liver pieces were transferred to a cold Beckman centrifuge tube, to which 15 ml of cold 0.25 M sucrose was added. Liver samples were then homogenized on ice with a tissue disruptor and centrifuged at 9,000 × g for 20 minutes at 4°C in a Beckman L8 70 Ultracentrifuge. The supernatant was transferred to a Beckman centrifuge tube and centrifuged at 105,000 × g for 60 minutes at 4°C. The supernatant was discarded and the pellets were scraped into a homogenizing vessel containing cold storage buffer (pH 7.5, K₂HPO₄: 10 mM, DTT: 0.1 mM, EDTA: 1 mM, glycerol: 20% (v/v)) diluted 1:1 with cold 0.25 M sucrose. Microsomal material was resuspended by hand manipulation and the microsomal suspensions were aliquoted into Nunc vials (Nunc-Nalgene, Rochester, NY). The samples were stored in liquid nitrogen until assays were performed. Microsomal protein levels were determined by the Lowry assay (Lowry et al., 1951) using bovine serum albumin as the protein standard.

AROD Assays

CYP enzyme activities were assessed with alkoxyresorufin O-dealkylation (AROD) assays, as described (Burke et al., 1994), with some modifications. These assays were based on activity measures of ethoxyresorufin O-dealkylation (EROD), pentoxyresorufin O-dealkylation (PROD), and methoxyresorufin O-dealkylation (MROD). Reaction mixtures (3 ml) were prepared in 4-sided, clear methacrylate cuvettes containing phosphate buffer (Na salt, 0.1 M, pH, 7.4), MgCl₂ (3.3 mM), alkoxyresorufin (4.9 μM), NADP (78 mM), G-6-P (198 mM), and G-6-P DH (24 U/ml) and were incubated at 37°C for 2 minutes. Microsomes were added to the mixtures to initiate the reactions. The final concentration of microsomes was 0.1 mg protein/ml. The fluorescence of the mixtures was measured at 37°C on a Perkin-Elmer LS-50 fluorometer with an excitation wavelength of 550 nm and an emission wavelength of 585 nm. Data were collected every 3 seconds for 5 minutes. AROD activities were expressed as rates of resorufin formation, and were calculated based on the fluorescence of a standard curve of resorufin.

Statistical Analyses

Statistical analyses of body weights, liver weights and AROD assay results were performed by the Student’s t-test and/or Dunnett’s multiple comparisons test using SigmaStat (SPSS, Chicago, IL). For statistical analyses of histopathological changes, cell proliferation rates and serum lipid levels, the Student’s t-test, One-Way Analysis of Variance, Dunnett’s multiple comparisons test and/or Tukey-Kramer Honest Significant Differences test were carried out using JMP statistical software (SAS, Cary, NC). The Shapiro–Wilk test for normality was first conducted on all treatment groups for cell proliferation. For each test, differences were considered statistically significant when p < 0.05.

Results

Feed Concentration and Consumed Dose

Each of the conazole-treated feeds were prepared with specific target concentrations: myclobutanil, 100 ppm (low dose), 500 ppm (mid dose), and 2000 ppm (high dose); propiconazole, 100 ppm (low dose), 500 ppm (mid dose), and 2500 ppm (high dose); triadimefon, 100 ppm (low dose), 500 ppm (mid dose), and 1800 ppm (high dose). The actual feed concentrations determined by LC/MS were within 5 % of the targets (Table 1). The consumed mean daily doses (mg conazole consumed/kg/d) were calculated from the feed consumption measurements. The mean daily doses for each conazole were similar at the low and mid doses, ranging from 13.3–15.1 mg/kg/d for the low dose and from 71.7–78.1 mg/kg/d for the mid dose. The mean daily high doses varied with the feed concentrations and were within 5% of the target doses.

Animal Body and Liver Weights

The effects of conazole treatments on animal body weights over the course of 90 days are indicated in Figure 1. Myclobutanil-treated mice did not reveal any differences in body weight gain as compared with the untreated control group. Mice treated with propiconazole did not have significant differences in weight gain over the first 7 weeks; however, at 9 and 12 weeks, the high-dose group was significantly lower in weight (p < 0.05) as compared with controls. For triadimefon-treated mice, the mid-dose group had a significant reduction in weight compared to controls (p < 0.05) at 5 weeks (only), and the high-dose group had significant reductions in weight compared to controls (p < 0.05) after 9, 12, and 13 weeks.

For the subgroups of mice analyzed for liver weights after 4, 30, and 90 days of treatment, Table 1 presents liver weights and % body weights. All 3 conazoles at high doses induced significant increases in liver weights and liver % body weights at all 3 exposure periods. After 90 days, mice treated with high doses of myclobutanil, propiconazole and triadimefon had liver weights averaging 148%, 195%, and 152% of control values, respectively. Myclobutanil also caused an increased liver % body weight (p < 0.05) after the mid-dose treatment for 30 days.

Histopathology

All 3 conazoles induced dose-dependent increases in liver cell hypertrophy (Table 2). Following the 4-day treatment period, centrilobular hypertrophy, and in some instances midzonal hypertrophy, were also observed in most exposed animals. Following the 30- and 90-day treatments, cell hypertrophy was more pronounced and also included panlobular and cytoplasmic vacuolation for all the conazoles. In 90-day treatment animals, the 3 conazoles were similar in their induction levels of these pathological changes.

Cell Proliferation

The hepatic cell proliferation data was statistically analyzed using a pooled estimate of error variance due to the unequal intersample variances. Propiconazole and myclobutanil were positive (p < 0.05) for this endpoint only after 4 days of treatment at the high dose (Table 3), while triadimefon was positive (p < 0.05) at the mid and high doses only after 30 days. After 90 days of exposure, there was no increased cell proliferation for any of the chemicals. As the control values for cell proliferation were not significantly different among the 4-, 30-, and 90-day treatment groups, we also compared across treatment periods. Cell proliferation within a group did not differ as a function of treatment duration. However, cell proliferation observed after treatment with high-dose myclobutanil for 4 days was significantly greater than after 30 and 90 days. Cell proliferation after exposure to the middle dose of myclobutanil for 30 days was significantly greater than after 4 and 90 days.

Cholesterol, Triglycerides, and High-Density Lipoproteins

The 3 conazoles exerted various effects on the serum cholesterol, triglyceride, and HDL levels after the high-dose treatment depending on duration (Table 4). The control values for each of these lipids did not significantly differ among the 4-, 30-, and 90-day time points. After 4 days of treatment, only triadimefon caused a significant reduction (0.69–0.71 that of control) in cholesterol and HDL levels. There were no effects from any of the chemicals on triglyceride levels at this time. Following 30 days of treatment, all of the conazoles caused significantly reduced levels of cholesterol and HDL (0.54–0.76 that of control), with concomitant increased levels of triglycerides (1.49–1.82 that of control). In 90-day treatment animals, those receiving triadimefon had control levels of cholesterol, HDL and triglycerides; whereas, those receiving myclobutanil or propiconazole again had a significant reduction in cholesterol and HDL (0.52–0.63 that of control), and no differences in the triglyceride levels.

Liver Microsomal AROD Activities

The effects of myclobutanil, propiconazole, and triadimefon on mouse liver microsomal AROD activities are shown in Table 5. After 4 days of treatment, all 3 conazoles at mid and high doses induced significant (p < 0.05), dose-dependent increases in PROD. At the high doses, 12- to 15-fold increases over control were measured. High doses of the 3 conazoles also induced significant increases (p < 0.05) in EROD of 2.1- to 2.7-fold over control. High doses of propiconazole and triadimefon also induced significant increases (p < 0.05) in MROD of 1.9- and 2.2-fold increases over control, respectively. Myclobutanil did not significantly increase activity levels for MROD.

Following 30 days of treatment (Table 5), the 3 conazoles were similar in inducing significant (p < 0.05), dose-dependent increases in PROD, with high-dose increases over control levels reaching 17-fold for myclobutanil, 14.9-fold for triadimefon, and 37.5-fold for propiconazole. At the high dose, the 3 conazoles induced EROD levels ranging from 3.1- to 4.6-fold. Only the high dose of triadimefon induced a significant increase (2.5-fold) in MROD.

Following 90 days of treatment (Table 5), all 3 conazoles had significant (p < 0.05) dose-dependent increases in PROD; levels reached 14.6-fold for myclobutanil, 10.5-fold for triadimefon, and 35.9-fold for propiconazole. At high doses, these chemicals induced from 2.9- to 4.0-fold increases in EROD. High doses of both propiconazole and triadimefon induced 2.0-fold increases over control in MROD. The high dose of myclobutanil induced a 1.5-fold increase in MROD. Overall, while constitutive activity levels of PROD, EROD, and MROD remained fairly constant over the different treatment periods, conazole high-dose induction of these activities appeared to be higher at 30/90 days than at 4 days. The most striking example of time-dependent increase in activity pertained to that for propiconazole-induced PROD, the levels of which rose dramatically from 4 to 30 days, after which they remained high over 90 days (Figure 2).

Discussion

A number of conazole fungicides are known to be hepatotumorigenic in mice, yet little is known about the underlying mechanisms leading to these effects. Most agrochemical hepatocarcinogens in mice have proven to be nongenotoxic, and are believed to act through pathways that trigger cell proliferation and/or liver enzyme induction (Carmichael et al., 1997). While it cannot be ruled out that hepatotumorigenic conazoles may be directly/indirectly DNA-reactive in the liver, there are few exceptions (Vijayaraghavan and Nagarajan, 1994) to the negative results reported for these chemicals in conventional assays for chromosome aberrations, gene mutations, and other genotoxicity endpoints. Therefore, the data suggest that conazoles induce mouse liver tumors through modes of action that are associated with induction of liver enzyme activity, perturbed regulation of cell growth and death, and/or as yet to be defined mechanisms.

The present studies in mice were aimed at comparing myclobutanil, propiconazole, and triadimefon for their relative activities to induce histopathological, cell proliferative, xenobiotic CYP enzymic, and other biochemical changes. Our results confirm earlier reports that all of these conazoles induce histological changes in the liver (INCHEM, 1987, 1992; EPA, 1996, 2004; Sun et al., 2005), and they describe patterns of histopathological effects which are similar in many respects.

The hepatic responses of increased organ weight and histological effects were qualitatively and quantitatively similar for all 3 conazoles. There was no difference between the 3 conazoles in severity score for hepatic histopathology after 90 days. In an analysis of 138 carcinogenicity studies conducted in various mouse strains by the agrochemical industry over the period 1983–1993, a clear relationship between hepatomegaly at 1 year after exposure and a positive tumorigenic outcome at 18 months or 2 years after exposure was demonstrated (Carmichael et al., 1997). Our 90-day conazole treatment results are not consistent with this point because the degree of hypertrophy associated with nontumorigenic (myclobutanil) and tumorigenic (propiconazole, triadimefon) agents was not different.

Mitogenic hepatotoxicants such as phenobarbital, a nongenotoxic rodent liver carcinogen, are characterized by transient increases in cell proliferation (Butterworth et al., 1995; Yoshikawa, 1996). All 3 conazoles evaluated in the present studies appeared to increase cell proliferation—this effect occurring at either 4 or 30 days. The magnitude of the cell proliferation response for these conazoles is considerably smaller than the response reported for female CD1 mice treated with fenbuconazole or phenobarbital after 1 week of treatment, where responses of 852- and 1170-fold increases are observed (using an alternative 5-bromo-′-deoxyuridine labeling method), respectively (Juberg et al., 2006). In our experiments, the observation of propiconazole-induced cell proliferation only after 4 days of treatment suggests that this chemical might be acting as a weak mitogen rather than a cytotoxicant which characteristically generates a sustained compensatory increased labeling index as part of the regenerative process (Butterworth et al., 1995). The cell proliferative response following triadimefon treatment at mid and high doses was increased after 30 days. That triadimefon increases cell proliferation only after a longer duration of treatment may suggest a difference in the toxicity or metabolic pathway than propiconazole. The nontumorigen, myclobutanil, which gave positive 4-day results similar to that observed for propiconazole, suggests that an early weak mitogenic burst alone is not sufficient to result in a tumorigenic response.

The conazoles present a complex pattern of time-dependent responses for serum cholesterol, HDL, and triglyceride levels. Only triadimefon showed reductions in cholesterol and HDL effects as early as 4 days with no significant changes in triglyceride levels. After 30 days all 3 lowered cholesterol and HDL and increased triglycerides, while at 90 days both propiconazole and myclobutanil (but not triadimefon) reduced cholesterol and HDL with no effects on triglycerides. The effects on serum cholesterol levels are likely to be related to the hallmark property of these conazoles to inhibit sterol 14α-demethylase, a key enzyme functioning in mammalian cholesterol biosynthesis (Zarn et al., 2003). Other inhibitors of lanosterol 14α-demethylase have been shown to disrupt mitotic cell cycling and lead to polyploidy (Martinez-Botas et al., 1999; Fernandez et al., 2004). Thus, conazole deregulation of cholesterol biosynthesis may be implicated in tumorigenesis through effects on cell cycling and mitotic progression to result in genomic instabilities. However, alternative pathways may also be involved as there are reports of other mouse liver tumorigens not known to inhibit sterol 14α-demethylase that clearly reduce plasma cholesterol.

The insecticide, thiamethoxam, has been found to cause a reduction in serum cholesterol that precedes liver histopathological changes similar to those observed in the present study (Green et al., 2005). These authors interpreted the low cholesterol as an early sign of liver dysfunction that may be causally associated with the later pathological changes, and ultimately cell death, increased cell proliferation and tumor formation. In our studies, triadimefon effects to lower cholesterol and induce histopathological changes occurred co-incidentally at 4 days. Further, propiconazole and myclobutanil-induced histopathological changes were detected earlier (at 4 days) than lowered cholesterol effects (at 30 days). Despite these observed differences between thiamethoxam and conazole chemical effects, together these studies suggest that altered cholesterol levels might play a role in tumorigenesis.

Exposures to propiconazole resulted in the most striking temporal pattern of PROD induction; the nearly 38-fold increases over control evident after 30 and 90 days were 2.5-fold greater than the level observed at 4 days, and represented more than twice the levels of increase induced by myclobutanil and triadimefon for any of the treatment periods. These exceptional results for propiconazole are consistent with previous studies in mice and rats with propiconazole and fluconazole (Leslie et al., 1988; Ronis et al., 1994; Sun et al., 2005, 2006) showing greater induction effects (typically, several-fold higher) on liver PROD, relative to EROD and MROD measures of enzyme activities. Overall, the different conazoles induced similar liver CYP enzyme induction patterns that increased with dose and duration of treatment. All of these agents proved to be most efficient for inducing PROD which, at the high dose, rose to more than twice the activity levels observed for EROD and MROD. After 90 days of treatment PROD activities reached levels 10.5- to 36-fold higher than control for the conazoles. While the tumorigenic potentials of the 3 conazoles suggested that these enzymatic activities bore no direct relationship with the tumorigenic process in propiconazole or triadimefon, the increases in CYP enzymatic activities could nevertheless play an indirect role in that process.

In the mouse Cyp1a1 and Cyp1a2 enzymes are generally associated with EROD and MROD activities while Cyp2b10(20) enzymes are generally associated with PROD activities (Weaver et al., 1994), and all are known to play important roles in xenobiotic metabolism (Guengerich, 2001). The highly conserved Cyp1a1 and Cyp1a2 genes are regulated by the aryl hydrocarbon (Ah) receptor (Chaloupka et al., 1994). The Cyp2b10(20) and Cyp3a11 genes are regulated by the constitutive androstane receptor (CAR) and pregnane X receptors (PXR) (Maglich et al., 2002) and have been widely studied for their high-level inducibility by phenobarbital (Pan et al., 2000). Propiconazole, as well as clotrimazole (Ronis et al., 1994), miconazole (Lake et al., 1998) and N-benzylimidazole (Newman et al., 2005) have been reported earlier to induce some or all of these enzymes.

Establishing the mode of action (MOA) for chemicals which induce liver tumors in rodents is fundamental to determining their relevance for human cancer risk (Cohen et al., 2004; Holsapple et al., 2006). Overall, it would appear that the tumorigenic conazoles may act by a combination of different pathways ultimately leading to induction of liver tumors in mice.

Propiconazole was shown to induce hepatomegaly, increased cell proliferation, increased PROD, MROD, and EROD activities, and decreased serum cholesterol and HDL. Several of these and other effects are similar to those occurring after treatment with the mouse hepatocarcinogen phenobarbital. Both propiconazole and phenobarbital cause reductions in body weight, increases in liver weight, and increases in cell hypertrophy and cell proliferation (to different extents) in addition to high levels of PROD activity. Thus, propiconazole’s tumorigenic MOA in mouse liver may, at least in part, be similar to that of phenobarbital, particularly with regard to the induction of CYP2B liver enzymes and potential increase in oxidative stress and enhanced mitogenesis. Based on studies in rodents, there is increasing evidence that the induction of CYP enzymes can lead to uncoupling of the catalytic cycle to result in release of high levels of reactive oxygen species (ROS) (Parke, 1994; Klaunig and Kamendulis, 2004). Overproduction of ROS through “futile cycling” of CYPs, or through other endogenous and exogenous oxidative processes associated with conazole exposure may lead to indirect DNA modifications and altered expression of key regulatory genes (e.g., cell-cycling, cell-to-cell communication, or apoptosis) important in carcinogenesis (Elrick et al., 2005). These events appear to be relevant for phenobarbital where rats treated with phenobarbital produced increased levels of CYP2B1 and CYP3A2 and increased levels of 8-oxo-deoxyguanosine.

Microsomes from these livers also produced high levels of hydroxyl radicals (Imaoka et al., 2004). Propiconazole may also produce other phenobarbital-like effects, such as increased expression of the proto-oncogene c-myc as well as other key response genes associated with increased reactive oxygen species (Imaoka et al., 2004; Elrick et al., 2005) and other epigenetic effects (Watson and Goodman, 2002).

Triadimefon caused reductions in body weight, increased hepatomegaly, increased PROD, MROD, and EROD activities, and decreased serum cholesterol and HDL. Triadimefon was unique in its early response of reduced cholesterol levels (evident at 4 days), which may be associated with its hepatotumorigenesis through effects that lead to genomic instability. As an essential component of cell membranes, cholesterol maintains their integrity, the disruption of which has been implicated in the generation of polyploidy (Fernandez et al., 2004).

Our data do not support the idea that CYP induction and enhanced mitogenesis alone explains the hepatotumorigenic activity of conazoles. Indeed, the nontumorigenic myclobutanil also induced hepatomegaly, increased cell proliferation, increased PROD, MROD, and EROD activities, and decreased serum cholesterol/HDL. We infer from these findings that it is the comparative level and temporal pattern of these effects, in combination with other unknown factors, which result in their modes of action.

In our studies of mice treated with tumorigenic and non-tumorigenic conazoles, we have observed a qualitatively similar response profile for all 3 compounds. All of these chemicals induced liver toxicity, histopathological changes and specific liver CYP enzymes that increased as a function of dose and duration of treatment. However, there were some responses by the tumorigenic chemicals not seen with myclobutanil. Propiconazole and triadimefon reduced body weights, propiconazole had the greatest induction and time-dependent increases in PROD, and triadimefon significantly decreased cholesterol/HDL levels after 4 days of exposure. These data do not reveal a common mode of action for the tumorigenic conazoles that sets them apart from the nontumorigenic conazole. However, some of the conazole-specific differences in cell proliferation, cholesterol and HDL, CYP enzyme activity, and temporal dose-response patterns provide a basis for further study of potential key events distinguishing propiconazole and triadimefon modes of action leading to hepatotumorigenesis. As described in a companion paper (Ward et al., this issue), we have extended the present comparative studies of mice treated with different conazoles to include transcriptional profiling of differentially expressing genes and molecular pathways in order to further assess key events in potential modes of action associated with conazole-induced liver tumorigenesis.