Gene Expression Dose-Response of Liver with a Genotoxic and Nongenotoxic Carcinogen

Abstract

Tumorigenic mechanisms due to chemical exposure are broadly classified as either genotoxic or nongenotoxic. Genotoxic mechanisms are generally well defined; however nongenotoxic modes of tumorgenesis are less straightforward. This study was undertaken to help elucidate dose-response changes in gene expression (transcriptome) in the liver of rats in response to administration of known genotoxic or nongenotoxic liver carcinogens. Male Big Blue Fischer 344 rats were treated for 28-days with 0, 0.1, 0.3, 1.0, or 3.0 mg/kg/day of the genotoxin 2-acetylaminofluorene (AAF) or 0, 10, 30, 60, or 100 mg/kg/day of the nongenotoxin phenobarbital (PB). Transcriptome analysis was performed using the relatively focused Clontech Rat Toxicology II microarray (465 genes) and hybridized with ³²P-labeled cDNA target. The analysis indicated that after 28 days of treatment, AAF altered the expression of 14 genes (9 up-and 5 down-regulated) and PB altered the expression of 18 genes (10 up- and 8 down-regulated). Of the limited genes whose expression was altered by AAF and PB, four were altered in common, two up-regulated, and two down-regulated. Several of the genes that show modulation of transcriptional activity following AAF and PB treatment display an atypical dose-response relationship such that the expression at the higher doses tended to be similar to that of control. This high-dose effect could potentially be caused by adaptation, toxicity, or tissue remodeling. These results suggest that the transcriptional response of the cells to higher doses of a toxic agent is likely to be different from that of a low-dose exposure.

Chemicals causing tumors in animals are generally classified into two broad categories, genotoxic and nongenotoxic, based upon their primary mechanisms of tumorigenicity. The mechanism by which genotoxic chemicals generate tumors has been relatively well studied compared to that of nongenotoxic chemicals. Genotoxic chemicals can cause DNA damage in a variety of ways (strand break(s), alkylation, adduct formation, base elimination, etc.). Subsequently, some damage escapes repair or is repaired incorrectly, leading to mutations that can accumulate with time, resulting in anaplastic transformation of cells and ultimately formation of tumors. In contrast, nongenotoxic chemicals may cause tumor formation via a number of relatively diverse modes of action. These may include chronic cytotoxicity, immunosuppression, decreased tumor suppressor gene function, decreased cell-cell communication, activation of a receptor, decreased apoptosis, and increased secretion of trophic hormones or various combinations of these factors (Silva Lima and Van der Laan 2000). Significantly, many of the precursor events of tumorigenesis for genotoxic and nongenotoxic carcinogens will result in dose-related changes both in specific and generalized patterns of gene expression identifiable by qualitative and quantitative changes in the transcriptome.

This study examined the alteration in the transcriptome of rat livers over a range of nontumorigenic and tumorigenic dosages after 28 days of treatment with the genotoxin 2-acetylaminofluorene (AAF) or the nongenotoxin phenobarbital (PB). AAF, a complete carcinogen, is an aromatic amine that undergoes metabolic activation to a genotoxin that forms bulky DNA adducts mainly on the C8 position of guanine (Heflich and Neft 1994; Lou et al. 2000) and also has the ability to cause promotion (Neumann et al. 1990). AAF reportedly causes tumors in the livers of rats at dose levels of 3.3 mg/kg/day or greater (Williams et al. 1998). In contrast, PB is a well-known potent inducer of hepatic xenobiotic metabolizing oxidative enzymes, which results in a pronounced hepatocellular hypertrophy and ultimately in liver tumor formation in rodents (Carthew, Edwards, and Nolan 1998). The strong induction of enzymes is believed to play a part in the mechanism of rodent liver tumorigenesis by PB and similar enzyme-inducing agents (Carmichael et al. 1997).

In this study, we examined the dose-responsive changes in gene expression from rat livers exposed to genotoxic and nongenotoxic carcinogens, AAF and PB, respectively. The goal of this work was to compare the alterations in gene expression across a range of lower doses between these two types of carcinogens.

MATERIALS AND METHODS

Animals and Treatment

Male Big Blue (Fischer 344–derived) rats were purchased from Stratagene (La Jolla, CA). In accordance with the U.S. Department of Agriculture’s rules on animal welfare, 9 CFR Parts 1 to 4, the animal care and use activities required for the conduct of this research were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of The Dow Chemical Company. Upon arrival to the facility (accredited by the Association for Assessment and Accreditation of Laboratory Animal Care [AAALAC] International), the laboratory veterinarian evaluated the animals. The animals were then housed in stainless steel cages in rooms designed to maintain adequate environmental conditions. Animals were provided free access to Purina Certified Rodent Lab Diet no. 5002 (Purina Mills, St., Louis, MO) in pelleted form and municipal water. Groups of animals (7/dose group) were randomly assigned to treatment groups and were administered 0 (control, corn oil), 10, 30, 60, or 100 mg/kg/day PB or 0.1, 0.3, 1.0, or 3.0 mg/kg/day AAF via oral gavage. After 28 days of treatment, animals were anesthetized with methoxyflurane, sacrificed, and briefly exsanguinated by decapitation. Livers were rapidly excised, weighed, and a portion of the liver was immersed in RNALater (Ambion, Austin, TX) and the remainder was flash frozen in liquid nitrogen. All samples were stored at –80°C.

RNA Isolation and Microarray

Total RNA was isolated using the Clontech Atlas Pure Total RNA Labeling System (Palo Alto, CA) which is a modified guanidinium thiocyanate/phenol/chloroform method of Chomczynski and Sacchi (1987). Isolation was performed as described by the manufacturer. In brief, equal portions of liver from the same chemical/dose groups were pooled and placed in denaturing solution. The livers were homogenized in a Duall tissue grinder, then the homogenate was passed through a 30-gauge needle and the cellular debris pelleted. Several rounds of phenol/chloroform extraction were performed on the supernatant then the RNA was precipitated and resuspended. Next the RNA samples were DNase I treated followed by phenol/chloroform and then chloroform extraction. The DNase-treated RNA was precipitated, quantitated, and aliquoted for future use. The absence of DNA contamination and RNA integrity was checked by electrophoresing the RNA samples through a denaturing agarose gel. Poly A⁺ RNA was isolated by incubating biotinylated oligo(dT) with isolated total RNA and allowing the poly A⁺ RNA to hybridize forming a biotinylated oligo(dT)–poly A⁺ RNA complex. Following this, magnetic beads containing streptavidin were added and allowed to bind with the biotin on the oligo(dT). Finally, using a magnet, the biotinylated oligo(dT)–poly A⁺ RNA–magnetic bead complex was separated from the rest of the RNA and washed. cDNA was made by incubating the poly A⁺ RNA with specific primers, against the genes spotted on the array, to incorporate ³²P using reverse transcriptase. Atlas Rat Toxicology II (Clontech) arrays were prehybridized for 30 min, then the ³²P-labeled cDNA was denatured, added to the prehybridization solution, and allowed to hybridize overnight to the array. The arrays were washed, exposed to a phosphorimaging plate, and then scanned using a Molecular Dynamics (Sunnyvale, CA) phosphorimager. For each experiment, RNA samples from each pooled treatment were labeled and hybridized to a microarray membrane on two separate occasions. Each experiment utilized arrays with the same lot number (i.e., from the same manufacturing batch).

Microarray Analysis

Each phosphorimager hybridization spot was aligned using Atlas Imager (Clontech) with the background being defined as the open area around the gene panels. Data were then analyzed using GeneSpring (Agilent Technologies, Palo Alto, CA) software. For a gene to be considered expressed, the phosphorimager value after background subtraction had to be of an intensity ≥20 (approximately ≥2 times the background) on one of the blots in the dose curve. Only these genes were analyzed further for altered expression. To determine fold change, each array was normalized to the median intensity measurements for that array; this is intended to eliminate differences in exposure/specific activity among arrays. Each gene was further normalized to its solvent control (which was considered to be 1.0). Genes determined to be induced or repressed had normalized values greater than 1.5 or less than 0.67, respectively, on one of the dose-response arrays. Genes that were induced or repressed on both sets of dose curve microarrays (each RNA analyzed twice) were considered to have been altered by treatment. For graphing purposes, fold-change values were converted to log₂.

Enzyme Assays

Frozen livers were thawed on ice and homogenized with a Potter-Elvehjem apparatus. Cytosol and microsomes were isolated using the method described by Guengerich (1982). Briefly, tissue was pooled and homogenized using a Tris-buffered KCl solution with EDTA and an antioxidant. The homogenate was centrifuged at 9000 × g to remove cellular debris, large organelles, and blood cells. The supernatant was removed and recentrifuged at 100,000 × g for 60 min. The cytosolic fraction (supernatant) was flash frozen in liquid nitrogen. The resulting microsome pellet was washed with pyrophosphate solution and recentrifuged. Washed microsomes were resuspended in a Tris-buffered 20% glycerol solution, frozen on dry ice, and stored at –80°C. Protein concentrations were determined using the Pierce BCA method (Pierce, Rockford, IL). The activity of several phase I and II enzymes were measured in vitro. Microsomal cytochrome P450 CYP1A1, CYP1A2, and CYP2B1/2 activities were measured using ethoxyresorufin (EROD), methoxyresorufin (MROD), and pentoxyresorufin (PROD) O-dealkylase activities, respectively, using a microplate fluorometric method outlined by Kennedy and Jones (1994). CYP2E1 activity was measured by detecting the hydroxylation of p-nitrophenol (pNPH) using a spectrophotometric method (Reinke and Moyer 1985). The phase II enzyme activities, UDP-glucuronosyl transferase (UGT) and glutathione-S-transferase (GST) were both assayed using a 96-well format. Microsomal UGT activity, which detects several isoforms, was measured using a modified spectrophotometeric method of Stewart and McCrary (1987) and 1-naphthol as a substrate. Cytosolic GST activity, which also included several isoforms, was measured using a spectrophotometric method described by Habig and Jakoby (1981), with CDNB (1-chloro-2,4-dinitrobenzene) as a substrate. Enzyme assay data was evaluated using analysis of variance (ANOVA). Comparisons of individual dose groups to the control group were made using Dunnett’s test (α = 0.05, two sided) only when a statistically significant dose effect exists.

RESULTS

Analysis of ³²P signal from the Clontech Rat Toxicology II nylon array identified 15 and 18 genes with altered regulation from AAF- and PB-treated rat livers, respectively (Table 1, Figures 1 and 2). Many of these genes are known to be altered by either AAF or PB treatment. Many of the genes identified as having altered gene expression show an atypical dose-response relationship. These genes do not plateau at the higher doses, instead many start to return towards expression levels found in untreated animals. Several P450 enzymes (EROD, MROD, PROD, and pNPH) were analyzed showing a relatively good correlation with the microarray data (Table 2) except PB induction of PROD/CYP2B1. Finally, UGT and GST enzyme activity was also measured showing statistically significant induction of both in the livers of AAF- and PB-treated animals (Figure 3).

DISCUSSION

Microarray analysis of rat livers treated with various doses of AAF or PB show gene expression and enzyme activity changes, many previously reported by others in the literature (Gerbal-Choloin et al. 2001; Sidhu, Farin, and Omiecinski 1993; Srivastava et al. 1990). Dose levels employed ranged from nontumorigenic to a level above that reportedly induces liver tumors (Gold and Zeiger 1997). However, no increase in liver cII mutation levels in AAF-treated animals was observed (data not shown), suggesting effective repair of DNA damage at the dose levels employed. Nonetheless, several genes, from an array containing 465 genes, displayed altered regulation over the doses of AAF administered (Figure 1, Table 1). Interestingly, the gene with the sharpest and greatest induction was O-6-methylguanine-DNA methyltransferase (MGMT) (Figure 1A), a DNA repair (dealkylating) enzyme. AAF has previously been shown to induce MGMT activity (Chinnasamy et al. 1997) via a p53 pathway (Grombacher, Eichhorn, and Kaina 1998) and is a general response to a variety of DNA damage. Finally, osteopontin, another AAF–up-regulated gene, has been shown to be directly induced by p53 in response to DNA damage (Morimoto et al. 2002). These findings indicate that AAF produced DNA damage in the present study despite lack of detected induction of Big Blue cII mutations.

The 40S ribosomal protein S19 (RPS19) was highly induced in AAF-treated rats. RPS19 is one of approximately 80 proteins that comprise the mammalian ribosome that previous studies have shown to be produced and released by apoptotic cells (Yamamoto 2000). Another apoptosis related gene up-regulated by AAF is defender against cell death 1 (DAD1), which reportedly has antiapoptotic activity (Nakashima et al. 1993). In addition to apoptosis-related genes was the up-regulation of nucleoside diphosphate kinase A (NDHA), which is linked to cell proliferation (Keim et al. 1992). Consistent with these findings has been a report that male Wistar rats fed a diet with 0.04% AAF reportedly had increased hepatocellular apoptosis in areas of the liver but also cell proliferation in other areas (Bitsch et al. 2000; Hadjiolov and Bitsch 1997). Thus, both apoptosis and antiapoptosis signaling may be occurring simultaneously in AAF-treated rat livers.

Several genes were also down-regulated by AAF (Figure 1B, Table 1). Two of these genes, complement component 3 (C3) and complement component 4–binding protein alpha (C4BPA), are an important part of the immune systems complement pathway. The three pathways of the complement response are dependent on the cleavage of C3. This cleavage is performed by C3 convertase, which is comprised of C2 and C4 (which consists of seven C4BPA subunits and one beta subunit) (Volanakis 1998). Therefore, the down-regulation of both C3 and C4BPA appear to be indicative of reported AAF-immunosuppressive activity (Lee and Yang 2000). Another AAF down-regulated gene is liver fatty acid–binding protein (L-FABP), presumably related to its reported binding of AAF and decrease in the concentration of this protein in the liver of treated rats (Blackburn et al. 1980).

Enzyme assays were performed on livers from AAF treated animals (Table 2, Figure 3A ) to detect CYP1A1, CYP1A2, CYP2B1/2, CYP2E1, UGT, and GST activities. Both UGT and GST displayed a modest increase in activity over the doses administered; however, no UGT or GST genes were shown to have altered expression. This could result from the fact that the enzyme assays detect several isoforms of these enzymes; therefore, if multiple isoforms were slightly increased, the cumulative effect could be significant where the expression of the individual isoforms would not. Another possibility is that secondary modification to the enzyme, and not direct effect on gene regulation, caused the increase in enzyme activity. Finally, the array utilized has a very limited number of genes represented and the particular isoform responsible for the increase may not be represented. The P450s show fairly good correlation with the microarray data (Table 2).

Nongenotoxin phenobarbital (PB), a nongenotoxic tumorigen, is known to alter numerous genes regulated through the nuclear receptors CAR and PXR (Honkakoski et al. 1998; Lehmann et al. 1998). Five of the genes identified as being induced on the arrays (GST Ya, CYP2B1, CYP2C9, CYP3A1, and UDPGT 2B) (Figure 2A and B ) are previously known to be affected after PB treatment (Gerbal-Choloin et al. 2001; Sidhu, Farin and Omiecinski 1993; Srivastava et al. 1990). Liver phase I and II metabolic enzyme activities were measured from this study (Figure 3B , Table 2). PROD activity showed a strong induction which correlates with that seen with the microarray analysis (CYP2B1/2) (Figure 2A , Tables 1 and 2). The lower values seen with the microarray is most likely caused by value compression which occurs with cDNA probes on microarrays (Yuen et al. 2002). The discrepancy of pNPH activity with the microarray results for CYP2E1 is caused by the ability of CYP3As, of which CYP3A1 is induced, to metabolize pNPH (Zerilli et al. 1998). Also UDPGT and GST activities were also induced though this activity does not distinguish the different isoforms of UGT and GST. However, an induction of GST Ya and UDPGT 2B was detected which correlates with that of PB-induced enzyme activities. Other GST and UGT isoform genes may have been altered by PB but were not included on the microarray used. Two genes, heat shock 27-kDa protein (HSP27) and hepatic triacylglycerol lipase precursor (HTLP), which were down-regulated in this study, have previously been identified as being affected by PB exposure (Figure 2C , Table 1). HSP27 has been shown to have increased expression in the hippocampus and cerebral cortex after withdrawal from PB (Tanaka et al. 2001), therefore, it is plausible that exposure to PB could repress HSP27. PB treatment has also been reported to affect HTLP, causing a redistribution of enzyme activity from intracellular to extracellular, however, the researchers did not evaluate gene expression (Parkes, Chan, and Goldberg 1986).

In general, most of the genes altered by treatment displayed an atypical dose-response. Their expressions were more similar to the basal levels at the high dose relative to the response observed at the low and/or intermediate dose levels for both the up- and down-regulated genes. This phenomenon is likely not an artifact of the array analysis given that the atypical response for induced and repressed genes are not parallel. This altered dose response may represent an adaptive change by the animal to 28 days of chemical treatment. This response may not have occurred if the treatment time had been shorter. This type of effect was also reported by Ellinger-Ziegelbauer et al. (2004) during a 14-day time course with rats treated with several other genotoxins. When genes were classified into genotoxic relevant processes and the cumulative gene change determined by chemical, some chemicals within the processed groups determined had less overall change at day 14 versus other time points. Another possibility is increased toxicity with higher doses causes a decrease in the ability of the liver to regulate gene expression and genes revert to their basal levels. Finally, liver remodeling may be occurring in which damaged cells have been removed and proliferation of existing cells is taking place. This new population of cells represents a different cell population from the original cells and therefore exhibits a different gene expression pattern.

It is interesting to note that four genes (organic cation transporter 1, DAD1, serotransferrin precursor, and phosphoenolpyruvate carboxykinase) were induced/repressed in common between PB and AAF and have similar expression profiles (Figures 1 and 2). These genes may represent common markers of early tumorigenesis or stress of the liver exposed to PB or AAF despite the induction of different pathways. Increased expression of DAD1 has an inhibitory affect on apoptosis (Nakashima et al. 1993) and thus prevents the elimination of unwanted/harmful cells. Phosphoenolpyruvate carboxykinase (PC), which was down-regulated by both chemicals, is a key enzyme in the regulation of gluconeogenesis. With this and the observed induction of phosphoglyceride kinase (PK), a major enzyme in glycolysis, by PB suggests an increase in glucose utilization. The latter has been observed during carcinogenesis (Bannasch 1986; Gatenby and Gawlinski 2003). It remains to be seen if changes in these genes could represent markers of early tumorigenesis.

This study evaluated the dose-response changes in gene expression after continuous exposure to relatively low doses of a genotoxic or nongenotoxic carcinogen. We have demonstrated that many gene expression changes showed atypical dose-responses most likely due to either an adaptive response and/or remodeling of the liver cell population.