Time- and Dose-based Gene Expression Profiles Produced by a Bile-duct–damaging Chemical,4,4′-methylene Dianiline,in Mouse Liver in an Acute Phase

Abstract

A toxicogenomics study was performed in the mouse liver after treatment of a bile-duct–damaging chemical, 4,4′-methylene dianiline (MDA), across multiple doses and sampling times in an acute phase using the AB Expression Array System. Imprinting control region (ICR) mice were given a single oral administration of a low (10 mg/kg b.w.) or high (100 mg/kg b.w.) dose of MDA. Mice were sacrificed six, twenty-four, and seventy-two hours after treatment for serum chemistry, histopathology, and mRNA preparation from liver samples. Treatment with MDA increased liver-toxicity–related enzymes in blood and induced bile-duct cell injury, followed by regeneration. To explore potential biomarker gene profiles, the altered genes were categorized into four expression patterns depending on dose and time. Numerous functionally defined and unclassified genes in each category were up- or down-regulated throughout the period from cellular injury to the recovery phase, verified by RT-PCR. Many genes associated with liver toxicity and diseases belonged to one of these categories. The chemokine-mediated Th1 pathway was implicated in the inflammatory process. The genes associated with oxidative stress, apoptosis, and cell-cycle regulation were also dynamically responsive to MDA treatment. The Wnt/β-catenin signaling pathway was likely responsible for the reconstitution process of the MDA-injured liver.

Introduction

The intra- and intercellular microenvironmental changes induced by a variety of toxicants are characterized by their unique gene expression profiles, which makes it possible to identify the expression signatures of altered genes following treatment with different chemicals (Hamadeh et al. 2002; Huang et al. 2004). The unique gene expression profiles produced by chemicals can therefore provide practical information to not only predict their potential toxicity in specific tissues, but also to understand the molecular mechanisms involved (Bartosiewicz et al. 2001; Walker et al. 2006).

The liver is vulnerable to a variety of metabolic, toxic, microbial, circulating, and neoplastic insults. Therefore, microenvironmental changes in liver tissues will reflect the biochemical characteristics and potential toxicity of certain chemicals and compounds. Understanding such changes is important for risk assessment in humans and animals. Toxicogenomics studies have focused on identifying gene expression changes in the liver following treatment with hepatotoxicants, which induce hepatocytic cell injury (Bartosiewicz et al. 2001; Huang et al. 2004). However, very few genomics studies have examined altered gene expression profiles induced by bile-duct-cell–damaging chemicals.

4,4′-methylene dianiline (MDA) is used in the synthesis of isocyanates and polyurethane polymers and in the production of dyes, polycondensation products, copolymers, curatives for epoxy resins, and other compounds (Kanz et al. 1992). Bile-duct cells are the main target cells of this chemical, which induces cholangitis with cellular necrosis and inflammation, resulting in “epping jaundice” in humans (Kanz et al. 1992; Kopelman et al. 1966).

In the present study, we performed a toxicogenomics study to identify the distinct gene expression profiles altered by MDA in an acute phase. The aim of this study is to extend our understanding of the possible mechanisms by which MDA induces bile-duct damage, followed by cellular recovery, at the level of global genes and to identify the potential target gene profiles for risk assessment and classification of the chemicals or compounds that cause liver toxicity, especially those like MDA that cause toxicity toward bile-duct cells.

Materials and Methods

Materials

MDA (CAS #101-77-9), chloroform, RNase-free 75% ethanol, RNase-free distilled water (DEPC water), and iso-propyl alcohol were all purchased from Aldrich Sigma (St. Louis, MO, USA). RNAlater RNA stabilization solution was obtained from Ambion Inc. (Austin, TX, USA). TRIzol and RNeasy Mini kits for RNA isolation were purchased from Invitrogen (Carlsbad, CA, USA) and Qiazen (Valencia, CA, USA), respectively.

Animals and Chemical Treatment

Specific pathogen-free (SPF) slc:ICR mice were purchased from Charles River Laboratories of Japan. The mice were housed at five per polycarbonate cage in an approved animal facility of the College of Veterinary Medicine of Seoul National University, Korea. Mice were maintained on a twelve-hour light-dark cycle. A basic pellet diet and water were provided ad libitum throughout the study. Temperature and humidity inside the raising room were maintained automatically at 23°C ± 2°C and 50% ± 20%, respectively. The mice were acclimatized for one week prior to the beginning of the study. All animals used in this study were treated humanely according to the guidelines of the Institutional Animal Care and Use Committee (IACUC) at Seoul National University for compliance with the National Research Council’s Guide for the Care and Use of Laboratory Animals (National Academic Press 1996). Animals were randomly divided into three groups: a vehicle control and two experimental groups. Two different doses of MDA were determined based on histopathological data from preliminary experiments; 100 mg/kg b.w. and 10 mg/kg b.w. In the preliminary study, 100 mg/kg b.w. (selected as the high-dose group) was the lowest dose showing typical injurious morphological changes of the bile ducts. Because it is important that with low doses, cellular injury is difficult to detect by light microscopic examination, 10 mg/kg b.w. was also selected as the lower dose group where no morphological changes were observed; it was also the geometric ratio of the high dose.

After one week of acclimatization, the mice were subjected to experimental treatments. Food was withdrawn for four hours before treatment (from 8:00 AM to 1:00 pm ). Thereafter the mice of each dose group, including the vehicle group, were given a single oral administration of the selected dose of chemicals and/or vehicles. Based on the data of the preliminary study, which showed that twenty-four hours post-MDA single treatment was the peak time point of MDA-induced liver injury, three sampling time points were determined to observe the altered gene expression profiles at the early (six hours post-treatment) and the recovery (seventy-two hours post-treatment) phases, as well as the injury peak time point (twenty-four hours post-treatment). Animals were provided a normal diet (Purina Inc., Korea) and tap water ad libitum throughout the experimental period.

Histopathology and Serum Chemistry

The mice were sacrificed after collecting blood from the abdominal artery under ether anesthesia. Liver tissues were removed and then fixed in 10% neutral buffered formalin. After routine tissue processing, the tissues were embedded in low-melting-point paraffin. Tissue sections of 3 μm were stained with hematoxylin and eosin (H & E) for histological examination.

Serum was prepared from the collected blood and analyzed using chemical analyses for aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase, albumin, total bilirubin, blood urea nitrogen, creatinine, and triglyceride.

RNA Preparation and Microarray Analysis Using the Applied Biosystems Expression Array System

Three animals out of five in each group were selected as representative animals based on biochemical and histopathological review. Total RNA was extracted from the left lobe of the collected liver tissues from the three selected mice of each group using TRIzol and RNeasy Mini kits. The purity and quality of the isolated RNA samples were analyzed using an Agilent Bioanalyzer 2110 (Agilent Technologies, Santa Clara, CA, USA) to confirm that the 28S/18S ratio was between 1.8 and 2.0 and the 260/280 nm ratio was between 2.0 and 2.2, respectively. The quality of the isolated RNA samples was further determined by electrophoresis on denaturing formaldehyde–agarose gels. Applied Biosystems (AB, Foster City, CA, USA) Mouse Genome Survey Arrays were used to analyze gene expression profiles following the single administration of MDA and/or vehicle. The AB mouse genome survey array contains 33,315 probes, representing 32,381 curated genes targeting 44,498 transcripts. The usefulness and effectiveness of the AB microarray were validated in previous studies (Walker et al. 2006). Digoxigenin-UTP–labeled cRNA was generated and linearly amplified from 1 μg of total RNA from each sample using an AB Chemiluminescent RT-IVT Labeling Kit (version 1.5). Array hybridization chemiluminescence detection and image acquisition and analysis were performed using the AB Chemiluminescence Detection Kit and the AB 1700 Chemiluminescent Microarray Analyzer. Images were auto-gridded, then spotted and spatially normalized. Signals were quantified and corrected for background noise, and the final images and feature data were processed using AB 1700 Chemiluminescent Microarray Analyzer software (version 1.1). Data and images were collected through an automated process for each microarray using the 1700 analyzer. General information about the strategy for microarray data quality assurance in the ABI microarray system is available at http://www.appliedbiosystems.com/.

Statistical Analysis of Microarray Data

Microarray data were analyzed using the software Avadis 3.3 prophetic (Strand Genomics Pvt Ltd.). The local background was subtracted from the raw expression values for allspots. The ratios were then log-transformed (base 2) and normalized so that the median log-transformed ratio equaled zero. Gene expression ratios were centered on the median across all samples. The expression ratio of each gene was determined by dividing the normalized expression value of a gene in a chemical treatment group by the mean expression value in the vehicle control group for each sampling time. Genes indicating a change of more than 1.5-fold were usually included in the data analysis. We used the supervised analysis method for differential expression analysis (DEA). Differential expression analysis was done between the vehicle control and chemical treatment for each sampling time. DEA was performed using Avadis software. Permutation-based modified t tests were used to provide further confidence in these results. Differential gene expression was analyzed using a two-sample Welch Benjamin Hochberg t statistic. Thus, differential gene expression associated with each group was tested using the significance analysis of microarrays.

Gene expression values were manipulated and visualized using the R package (free software under the terms of the Free Software Foundation’s General Public License). For the analysis of data correlation, correspondence analysis was also performed using the AB1700 package in R. Hierarchical cluster analysis partitioned the data into discrete hierarchical groups based on data trends. The resulting gene lists were limited to genes with changes in proportions of 1.5 or more and p < .05.

Categorization of Microarray Gene Expression Profiles

Because our goal was to define gene expression profiles to predict liver toxicity, especially associated with bile duct cell injury, we categorized altered genes based on the expression patterns represented at the selected sampling times in the different dose groups (Figure 1) and further classified them by function.

Semiquantitative RT-PCR

To verify whether the genes identified from the microarray results were altered by MDA treatment, RT-PCR analyses were performed for twelve genes, as described previously (Park et al. 2004) using specific primers (Table 1). Equal amounts of total mRNA from the three samples selected from each group were pooled. First-strand cDNA was generated from 2 μg of total RNA by RT with 1 μL of 50 μM oligo (dT)₂₀ primer and 1 μL of SuperScript III Reverse Transcriptase (Life Technologies Inc., Gaithersburg, MD, USA) in a 20-μL reaction mixture, following the manufacturer’s instructions. From the 20 μL produced, PCR was performed using AccuPower PCR PreMix (Bioneer, Daejeon, Korea) and using a Mycycler (Bio-Rad, Hercules, CA, USA). The PCR products were analyzed by electrophoresis in 1.5% agarose gels containing 0.1 mg/mL ethidium bromide. Images of RT-PCR ethidium-bromide–stained agarose gels were acquired with ultraviolet illumination on a Gel Doc EQ System (Bio-Rad, Hercules, CA, USA).

Results

Histopathology and Serum Chemistry

In the low-dose group, no notable histopathological changes were observed for any of the sampling times (Figure 2). In the high-dose group, MDA-induced cellular toxicity was limited mainly to the bile duct cells (Figure 2). Six hours after treatment, bile duct epithelial cells showed hydropic degeneration and, sometimes, necrotic changes. Necrosis of bile duct cells, resulting in exfoliation of the epithelial cells from the basement membrane, was evident twenty-four hours after the high-dose treatment of MDA. Seventy-two hours after the high-dose treatment of MDA, bile duct epithelial cells were regenerative, which was characterized by large basophilic epithelial cells of tall cuboidal shape and relatively frequent mitotic figures. A periductular desmoplastic reaction was also prominent, with the infiltration of a number of mononuclear cells. Individual necrosis and multiple necrotic foci of hepatocytes were observed randomly in the liver.

Serum chemical analyses indicated significant increases in AST, ALT, and total bilirubin twenty-four hours after treatment with the high dose of MDA, followed by a return toward control values at seventy-two hours (Table 2). With the low dose of MDA, no significant changes in the parameters examined were observed for any sampling time.

Gene Expression Changes

After quantile-normalization of the raw data, statistical microarray analyses were conducted. First, we verified a good correlation (> 95%) of the gene expression profile between individuals in the same group for each sampling time. Whole microarray data are found at http://www.snubi.org/publication/TGRC_MDA. The low and high doses of MDA produced differential gene expression profiles for each selected sampling time. In the hierarchical clustering, the gene expression profiles were likely to be affected by the sampling time, rather than the dose of MDA, until twenty-four hours after treatment. However, at seventy-two hours, the gene expression profiles were distinctly comparable between the low- and high-dose groups (Figure 3). Two-way ANOVA of the dose- and time-based microarray data identified 4,257 altered genes in the low-dose group and 4,534 genes in the high-dose group, all with statistically significant changes (p < .05). These microarray data were further filtered into a list of genes with changes more than 1.5 times the control level. The numbers of altered genes (p < .05, with a change of more than two times) for each sampling time in the low- and high-dose groups are shown in Figure 4. Categorization of the altered genes based on the PANTHER biological process indicated that a variety of genes were implicated in the biological processes of the liver after MDA treatment, associated with cell cycle and proliferation; cell structure and motility; developmental processes; immunity and defense; intra-cellular protein traffic; lipid, fatty acid, and steroid metabolism; protein metabolism and modification; signal transduction; transport; and so on (Figure 5). The global gene expression profiles were classified further based on the comparative gene expression patterns between the low- and high-dose groups (Figure 1). Semiquantitative RT-PCR was performed using specific primers (Table 1) to verify the gene expression profiles detected using the oligonucleotide microarray. The semiquantitative RT-PCR and microarray results were quite compatible for each sampling time: ten of twelve genes (83.3%) at six hours, nine of twelve genes (75.0%) at twenty-four hours, and ten of twelve genes (83.3%) at seventy-two hours showed the same patterns of change (Figure 6).

Class I: Genes That Showed Similar Expression Patterns at the Low and High Dose

This class contains genes that were up- or down-regulated in both the low- and high-dose groups with similar patterns of expression (Supplementary Table 1: http://www.snubi.org/publication/TGRC_MDA). Various genes associated with amino acid metabolism; apoptosis; carbohydrate metabolism; the cell cycle; cell structure and motility; developmental processes; electron transport; immunity and defense; lipid, fatty acid, and steroid metabolism; nucleoside, nucleotide, and nucleic acid metabolism; oncogenesis; signal transduction; and other functions were placed in this category. In particular, the genes that were altered concomitantly with both the low and high doses were found most frequently at the peak injury time twenty-four hours after treatment with MDA.

The genes associated with the induction of apoptosis, for example, Chordc1, Tnfrsf6, and Casp3, were up-regulated irrespective of dose. Several notable genes were associated with carbohydrate metabolism: Gbe1, Galk2, and Galnt11. Gbe1 is up-regulated in the hypoxic environmental conditions of the mouse liver (Zhao et al. 2004). In association with the cell cycle, cyclin D1 was concomitantly down-regulated at the late phase, whereas other cell cycle-related genes, including Cks2, Nudc, Mobk1b, and Rback, were up-regulated at twenty-four hours after MDA treatment. The alteration of cortactin, which is involved in cytoskeletal reorganization (Lua et al. 2005), was notable because it was highly up-regulated by more than ten times at twenty-four hours after MDA treatment for both dose levels. Genes associated with developmental processes and electron transport tended to be down-regulated by MDA treatment, especially at the time of peak injury at twenty-four hours. In contrast, many genes associated with intracellular protein traffic and lipid, fatty acid, and steroid metabolism were up-regulated. In particular, Copg and Sgpl1 (apoptosis-associated genes under stress) were prominently elevated and Gpx6 was affected at the peracute phase.

Class II: Genes That Showed Similar Expression Patterns at Any Sampling Time(s) at Both Dose Levels, but Showed Dose-dependent Expression at Any Other Sampling Time(s)

This class contains genes that were dose-independently up- or down-regulated at any sampling time, but also showed dose-dependent expression at any other sampling time (Supplementary Table 2: http://www.snubi.org/publication/TGRC_MDA). The genes in this class generally showed up- or down-regulation at twenty-four hours for both the low and high doses and reverse alteration at seventy-two hours for the high dose. For instance, arginase, which is involved in nitric oxide (NO) and reactive oxygen species (ROS) metabolism, was down-regulated at twenty-four hours for both doses, but was up-regulated at seventy-two hours in the high-dose treatment. Many genes postulated to be associated with liver diseases, including biliary diseases, belonged to this class (Table 3). Anapc5, which plays an important role in the regulation of unscheduled hepatocyte cell proliferation (Wirth et al. 2004), was prominently overexpressed at twenty-four hours for both doses (12.3- and 9.3-fold changes in low and high doses, respectively) but was dramatically down-regulated at the regenerating phase for the high dose. Raf1 showed an expression pattern similar to that of Anapc5. These genes appeared to play important roles in reconstitution processes after MDA-induced liver injury. Emid2 and Lepr were noteworthy as genes with rapid responses to MDA treatment.

Class III: Genes That Showed Reverse Expression Patterns Between the Low and High Dose

This class contains genes that were down-regulated for the low dose and up-regulated for the high dose, and vice versa (Supplementary Table 3: http://www.snubi.org/publication/TGRC_MDA). These genes may reflect differential responses of the liver depending on intensity and type of cellular injury. Catalase was included in this category; it was highly overexpressed at twenty-four hours for the low dose, but it significantly down-regulated at the active regenerating phase for the high dose, suggesting the importance of oxidative stress in association with toxic mechanisms and the preventive function of catalase at a low dose of MDA. Ehpb1, Map3k4, Mipep, Hsd3b7, Slc25a25, and Slc25a15 showed the same expression patterns as catalase. In contrast, Galnt7, B4galt4, and Mt1 had expression patterns opposite to that of catalase. They were down-regulated for the low dose, but they were highly over-expressed at the regenerating phase for the high dose.

Class IV: Genes That Were Separately Affected in an On/Off Pattern Dose-dependently

The genes in this class may represent differential and separate genetic responses in proportion to the dose of MDA (Supplementary Table 4: http://www.snubi.org/publication/TGRC_MDA. In association with apoptosis, at the low dose, Bcl-2 and related genes appeared to function in maintaining cellular homeostasis, whereas at the high dose, the alteration of the apoptosis-inhibiting genes Angptl4 and Birc5, and the apoptosis-induction–associated genes Lgals4, Foxo1, Foxo3, Tnfrsf23, Ing1l, and Pdcd4 was noted, with the down-regulation of caspase 3 at the regenerating phase. The expression of cell-cycle–related genes was very different depending on the dose of MDA. At the low dose, the cell cycle checkpoint genes cyclin G1 and G2 were up-regulated, but at the high dose, cyclin B2, Cdc20, Cks1, Mad2l1, and Prc1 were up-regulated at the regenerating phase. However, cyclin D3 was rather depressed by the high dose of MDA at the regenerating phase. At the peak injury time of the high-dose group, marked up-regulation of Pak4 (p21-activated kinase 4) was noted. In association with immunity and defense, the chemokine ligand series genes were usually up-regulated at both the low and high doses. In particular, marked up-regulation of Cxcl1 by 11.1 times at the peracute phase was notable in the low-dose group. Histocompatibility-2–related genes were activated in the low-dose group but down-regulated in the high-dose group. Lymphocyte-related genes were usually activated at the late phase in the high-dose group, possibly reflecting lymphocyte-mediated chronic inflammation, which was also noted in histopathological findings (Figure 2). In addition, Sod1, an endogenous antioxidant, was significantly activated in the high-dose group. Numerous genes associated with nucleoside, nucleotide, and nucleic acid metabolism belonged in this class. These genes were usually found at the early and peak injury times in the low-dose group and at the late phase in the high-dose group. In particular, in the low-dose group, Egr1 (early growth response 1) was dramatically up-regulated by 44.5 times at the peracute phase, possibly associated with liver regeneration and cholestasis (Table 3) (Liao et al. 2004; Kim et al. 2006). Many ribosomal proteins responded differently to MDA, depending on the dose. They were usually down-regulated at the early phase at the low dose, whereas they were down-regulated at the peak injury time followed by up-regulation in the regenerating phase at the high dose.

Summary and Discussion

MDA is a selective bile duct toxicant (Kanz et al. 1992), as also indicated by our histopathological observations in an association with serum chemistry. However, the MDA-induced toxic effect is achieved not only by targeting biliary cells, but also by breaking the functional relationships with other liver cell components. In the genomic circumstances, host cell responses against chemical-inducing effects are even more dynamic and active, but very much conserved. Thus, our microarray data represented complicated and understandable expression changes in a variety of genes maintaining cellular microenvironmental homeostasis across MDA-induced toxic effects and adaptation in the liver, including the genes altered in response to bile duct cell injury by MDA.

Exploration of Potential Biomarker Gene Profiles to Discern the Toxic Endpoints of Bile Duct Cells

The gene expression profiles produced by oral treatment with MDA differed greatly depending on the dose and sampling time after treatment. This finding reflects the highly dynamic genetics that maintain cellular homeostasis despite acute cellular injury. Based on the comparative gene expression patterns between the low and high doses, MDA-affected genes were classified into four categories, Classes I to IV. Classes I and II contained genes affected by MDA independent of the dose, whereas Classes III and IV contained entirely dose-dependent genes.

The genes belonging to Classes I and II could be affected by chemicals similar to MDA independent of dose. In fact, several genes that are postulated to be associated with liver toxicity, including bile duct damage, were frequently found in these classes (Table 3). Class I and II genes may also represent the microenvironmental conditions associated with the toxic mechanism of MDA. Cortactin (Cttn) and catenin beta (Catnb) are functionally involved in cell mortility (Chuma et al. 2004; Lua and Low 2005) and Wnt signaling (Janssens et al. 2006), respectively, and will contribute to the remodeling of the MDA-injured liver together with cell-cycle– and apoptosis–regulating genes such as cyclin D, Rback, caspase3, sphingosine phosphate lyase 1, and so on (Supplementary Table 3: http://www.snubi.org/publication/TGRC_MDA). An iron homeostatic gene, Alas1, is also activated by MDA treatment, which is associated with heme synthesis regulated by bile acids in the liver (Peyer et al. 2007). In contrast, the genes belonging to Classes III and IV could be potential biomarker genes in predicting the level of exposure to MDA-like chemicals because their expression was dose dependent. As shown in Class III, many genes that were altered at the time of peak injury (i.e., twenty-four hours) in the low-dose group had reversed expression patterns during the recovery phase (i.e., seventy-two hours) in the high-dose group. Metallothionein-1 and catalase have been shown to be altered by some liver toxicants such as furan (Bartosiewicz et al. 2001; Huang et al. 2004). In contrast, genes belonging to Class IV were affected in an on/off dose-dependent expression pattern. Among the genes affected only by the low dose of MDA, some genes such as Sdh1 (2.5-times increase), Gck (4.1-times increase), and Hyal (2.1-times decrease followed by 3.0-times increase) have already been postulated as biochemical markers associated with liver function and toxicity, including biliary diseases (Table 3). In addition, Caveolin 2 (Cav2), which is highly expressed in the proliferating bile ductules in primary biliary cirrhosis (Yokomori et al. 2005), was significantly up-regulated at the time of peak injury at the high dose of MDA. In association with liver fibrosis, overexpression of smad 1 was noteworthy at the late stage of severely MDA-injured liver (Table 3) (Fan et al. 2006). Glypican 3, a cell-adhesion–related gene, is a useful marker for hepatocellular carcinoma (Libbrecht et al. 2006), but MDA, a bile-duct-damaging chemical, induced glypican 1 at the low dose and glypicans 4 and 5 at the high dose.

Early-response genes have been an important topic in toxi-cogenomics studies. In the microarray, Egr1, Nmyc1, Cxcl1, Clrf, Hmgcs1, P2rx2, Gng2, and Dnajb4 were considered potential early biomarkers associated with biliary cell damage in liver exposed to a low dose of MDA because they were dramatically up-regulated at the early phase (six hours) in the low-dose group. In particular, Egr1 was up-regulated by approximately forty-five times in the early phase after the low dose of MDA. This gene changes rapidly at the mRNA level in the livers of mice with cholestasis (Kim et al. 2006). Galk2 (galactokinase 2), which is regulated by Egr1 (Yang et al. 2004), was subsequently up-regulated by low and high doses of MDA. More importantly, Gpx6 (+2.5 times, +2.9 times), Emid2 (+2.8, +2.8), Nr0b2 (+4.1, +3.2), Slc23a3 (+3.2, +2.3), Dio1 (+5.5, +5.0), Ard1 (−2.4, −3.0), Prss22 (−5.3, −5.8), And (−2.2, −2.5), Wbscr16 (−2.6, −3.3), Havcr1 (−4.8, −4.5), Cox10 (−2.2, −2.9), and other genes showed marked alterations at both the low and high doses in the early phase after MDA treatment. The altered expression of these genes was further verified using RT-PCR in cases (Figure 6).

Comprehensive Action Mechanisms of Altered Genes Following MDA Treatment

Wnt Signaling Pathway

The Wnt genes are involved in cell differentiation, proliferation, maturation, and cell fate determination by regulating gene expression (Janssens et al. 2006). Thus, they play important roles in homeostatic mechanisms in adult tissues. Based on the microarray, the Wnt/β-catenin signaling pathway is most likely to play important roles in the reconstitution of MDA-injured liver (Figure 7). The first genes activated by MDA treatment were Nmyc1, Tle1 (transducin-like enhancer of split 1), Gng2 (guanine nucleotide binding protein gamma 2), and Pcdha8 (protocadherin alpha 8). Following the activation of these was the activation of β-catenin and its downstream genes, such as the protein phos-phatase family (Ppp2ca, Ppp3ca, and calcineurin B), cadherin, calcineurin, SWI/SNF, Tle3 (transducin-like enhancer of split 3), Btaf1 (BTAF1 RNA polymerase II), Acvr1 (activin A receptor), and Pcdhga12 (protocadherin gamma subfamily A 12). Myconcogenes were activated only by high doses of MDA. In contrast, Wnt2, cyclin D1, and Prkcz (protein kincase C zeta) were down-regulated in the late recovery phase after MDA treatment, presumably representing a feedback response to regulate cell proliferation.

Immunity and Inflammation

In the high-dose group, periductal inflammation characterized by lymphocyte infiltration was histopathologically evident in the acute late phase. This finding was supported by the up-regulation of numerous immunity-associated genes. Lymphocyte-related genes such as Lsp1, Ly6a, and Ly96, and chemokine-related genes such as Ccl2, Ccl9, Cxcl9, and Xcl1 were up-regulated, implying that the Th1 pathway was involved in MDA-induced inflammation. The activation of these types of proinflammatory cytokines also occurs in mice suffering from biliary atresia (Leonhardt et al. 2006). Further, regulators of cytokine signaling such as Jak1 and Map3k4 were also activated by MDA treatment.

Cell Cycle and Proliferation

Cell proliferation after acute injury by MDA was achieved via pathways associated with insulin-like growth factor and epidermal growth factor. The p53-gene–mediated pathways seemed to play an important role in cell-cycle regulation during the reconstitution of liver with bile duct cell damage. In the recovery phase, which was characterized histopathologically by the regenerative proliferation of bile duct epithelial cells and periductal fibroplasia, Cyclin B2, Cdc20, and Mad2 were highly up-regulated, whereas Cdkn2b (a cyclin-dependent kinase inhibitor), Anapc5 (a regulator of unscheduled cell proliferation), and Gadd45g were down-regulated, indicating cell cycle progression in agreement with histopathological observations (Figure 2). Cyclin B2, as a regulatory subunit of the maturation-promoting factor, is considered to be associated with the formation of the bipolar spindle at metaphase during cell meiosis (Kotani et al. 2001). However, the G1 progression genes cyclin D1 and D3 were down-regulated during the recovery phase. In particular, cyclin D1 was inhibited by MDA at both the low and high doses. Cell-cycle regulation also appeared to be achieved by RB and its related genes, as supported by the high expression of Rbak (RB-associated KRAB repressor) and Rbbp9. The altered expression of Cul4a (cullin 4A), which is involved in the MDM2-mediated proteolysis of p53 (Nag et al. 2004), also suggests a regulatory role in cell proliferation in MDA-injured liver. In contrast, Cyclin G1 and G2, which are cell-cycle arrest genes mediated by p53 (Harris and Levine 2005), were activated twenty-four hours after a low dose of MDA, which might be associated with conserved cell-cycle regulation by feedback loops in cases of reversible injury without serious cellular damage.

Apoptosis

Programmed cell death is a process that precedes tissue reconstruction and causes irreversibly damaged cells to be removed. A single dose of MDA induced the activation of the apoptosis-inducing genes Chordc1 and Tnfrsf6 and the up-regulation of caspase 3 at twenty-four hours in both the low- and high-dose groups. In the low-dose group, Bcl-2–related genes seemed to play key roles in the regulation of cell death. In the high-dose group, apoptosis-inducing and -inhibiting genes were counterbalanced (Lgals4, Foxo1, and Ing1l vs. Angptl4 and Birc5).

Oxidative Stress

Several endogenous antioxidant genes and redox-regulating genes showed altered expression after MDA treatment. Gpx6 (glutathione peroxidase 6), catalase, and Txndc1 (thioredoxin domain-containing 1) were significantly activated in the low-dose group. The increased expression of these antioxidant enzymes suggests that treatment with MDA may induce oxidative stress in the liver, particularly at a low dose. In contrast, at the high dose, catalase was markedly down-regulated, which agreed with toxicogenomic results for other biliary toxic chemicals such as methapyrilene and furan in the liver (Huang et al. 2004). A decrease in catalase activity occurs in liver exposed to oxidative-stress–inducible chemicals and toxins such as 2-nitropropane and heptapeptide toxin microcystins (Borges et al. 2006; Jayaraj et al. 2006). The importance of catalase in chemical-induced liver toxicity is strongly supported by its preventive effect against carbon tetrachloride (CCl₄) toxicity (Ma et al. 2006). At the high dose of MDA, the redox-regulating system seemed to be disturbed, because Gpx1, Gpx5, and thioredoxin reductase 1 (Txbrd1), as well as catalase, were significantly suppressed. Thus, oxidative stress production and redox system disruption may be other important ways in which MDA induces liver toxicity.

In summary, a single oral administration of MDA produced differential gene expression profiles in mouse liver tissues that were dependent on or independent of sampling time and dose. With regard to potential biomarker genes, we divided the altered genes into four categories, Classes I to IV. Numerous functionally defined and unclassified genes in each category were differentially up- or down-regulated throughout the period from cellular injury to the recovery phase at the examined doses, providing potential biomarker gene profiles. These gene expression profiles were also comprehensive with regard to the mechanistic responses maintaining cellular homeostasis or recovery from MDA-induced liver toxicity. Thus, our results are valuable not only for understanding the mechanisms by which MDA-like chemicals induce liver toxicity, in particular bile duct cell injury, but also for risk assessment and classification of the chemicals using biotoxchips.