Use of a Low-Density Microarray for Studying Gene Expression Patterns Induced by Hepatotoxicants on Primary Cultures of Rat Hepatocytes

Francoise de Longueville*,1, Franck A. Atienzar{dagger}, Laurence Marcq*, Simon Dufrane{dagger}, Stéphanie Evrard*, Lydia Wouters{dagger}, Florence Leroux{dagger}, Vincent Bertholet*, Brigitte Gerin{dagger}, Rhys Whomsley{dagger}, Thierry Arnould{ddagger}, José Remacle* and Mickael Canning{dagger}

* Eppendorf Array Technology (EAT), 20, Rue du Séminaire, 5000 Namur, Belgium; {dagger} UCB SA Pharma sector, Chemin du Foriest, B-1420 Braine-l’Alleud, Belgium; and {ddagger} Laboratory of Biochemistry and Cellular Biology, University of Namur, Belgium

Received May 6, 2003; accepted June 30, 2003

ABSTRACT

In the field of gene expression analysis, DNA microarray technology is having a major impact on many different areas including toxicology. For instance, a number of studies have shown that transcription profiling can generate the information needed to assign a compound to a mode-of-action class. In this study, we investigated whether compounds inducing similar toxicological endpoints produce similar changes in gene expression. In vitro primary rat hepatocytes were exposed to 11 different hepatotoxicants: acetaminophen, amiodarone, clofibrate, erythromycin estolate, isoniazid, {alpha}-naphtylylisothiocyanate, ß-naphtoflavone, 4-pentenoic acid, phenobarbital, tetracycline, and zileuton. These molecules were selected on the basis of their variety of hepatocellular effects observed such as necrosis, cholestasis, steatosis, and induction of CYP P450 enzymes. We used a low-density DNA microarray containing 59 genes chosen as relevant toxic and metabolic markers. The in vitro gene expression data generated in this study were generally in good agreement with the literature, which mainly concerns in vivo data. Furthermore, gene expression profiles observed in this study have been confirmed for several genes by real-time PCR assays. All the tested drugs generated a specific gene expression profile. Our results show that even with a relatively limited gene set, gene expression profiling allows a certain degree of classification of compounds with similar hepatocellular toxicities such as cholestasis, necrosis. The clustering analysis revealed that the compounds known to cause steatosis were linked, suggesting that they functionally regulate similar genes and possibly act through the same mechanisms of action. On the other hand, the drugs inducing necrosis and cholestasis were pooled in the same cluster. The drugs arbitrarily classified as the CYP450 inducers formed individual clusters. In conclusion, this study suggests that low-density microarrays could be useful in toxicological studies.

Key Words: hepatotoxicants; gene expression pattern; low-density microarray; toxicogenomics; drug metabolism.

With increasing costs of new drug development, there is a crucial need to conduct toxicity evaluation as early as possible and on as many potential chemical leads as feasible. During the drug developmental process, undesired toxicity accounts for about one-third of compound failures (Johnson and Wolfgang, 2000Go). Therefore, it is clear that new powerful technologies are needed as an alternative to classical toxicological tests for a rapid screening. Since some recent studies have shown the usefulness of DNA microarrays in toxicological studies, the scientific community is showing a growing interest for this kind of technology. The emerging field of "toxicogenomics" could be defined as the study of toxicological processes at the transcriptome level of a target organ or cell. It seems that DNA microarrays could be very helpful not only to predict drug induced toxicity but also to better understand mechanisms of actions of drugs (Fielden and Zacharewski, 2001Go; Storck et al., 2002Go). In this context, gene expression microarrays could help to prioritize lead compounds.

DNA microarrays consist of DNA fragments corresponding to genes. The use of high density microarrays containing thousands of DNA fragments has the main advantage that the expression level of a large number of genes can be studied simultaneously. However, the major drawbacks are related to the high cost and the time taken for analysis and interpretation of the data. Low-density microarrays, even though they contain fewer genes, can still offer the ability to rapidly study gene expression changes following chemical exposure (de Longueville et al., 2002Go). However, it is clear that with low-density microarrays, the effects on genes not selected will obviously be missed.

While it may take weeks, months, or even years before some traditional toxicological endpoints occur, specific changes in mRNA levels could occur within a few hours or days after exposure to chemical compounds. Toxicogenomics builds upon the fact that relevant toxicological outcomes are preceded by such changes in gene expression. A recent study revealed a strong correlation between the histopathology, clinical chemistry, and gene expression profiles induced by 15 different known hepatotoxicants (Waring et al., 2001bGo). In addition, comparison of gene expression profiles induced by new drugs with those induced by known toxicants obtained in a database could help identify and predict potential toxicities (Hamadeh et al., 2002aGo). Recent studies have also shown that not only gene expression analysis reveals chemical specific profiles (Hamadeh et al., 2002bGo) but also that compounds belonging to a same class of toxicant yield to similar gene expression patterns that are distinct from other profiles generated by other class of chemicals (Bartosiewicz et al., 2001Go; Morgan et al., 2002Go).

While the ultimate goal of toxicogenomics is to generate safe drugs for humans, the majority of studies are performed on rodents despite the fact that the human predictability of standard rodent tests shows only 45% concordance (Johnson and Wolfgang, 2000Go). However, primary hepatocytes are well suited for toxicogenomic studies because they display a certain level of metabolic activity and the liver is a major stage for toxic events (Waring et al., 2001aGo). Hepatotoxicity is a common reason for withdrawal of compounds from the market (Baker et al., 2003Go). In addition, the use of cell culture models reduces the animal utilization and need for the synthesis of new compounds on a large scale (Baker et al., 2001Go). However, it is also clear that there are a number of limitations in using in vitro approaches such as the functional differences observed in primary hepatocytes relative to the intact liver, the absence of interactions with biological entities (e.g., organs, blood) under in vitro conditions, and the difficulty to select doses and time points which are representative of an in vivo situation.

Compared to the input of drug developers in toxicogenomics, the number of published studies on toxicogenomic involving the analysis of several compounds is still limited and mainly restricted to high-density microarrays (Bulera et al., 2001Go; Burczynski et al., 2000Go; de Longueville et al., 2002Go; Gerhold et al., 2001Go; Hamadeh et al., 2002aGo,bGo; Waring et al., 2001aGo,bGo). In this study, we have used a low-density microarray containing 59 genes to analyze gene expression profiles generated in primary cultures of rat hepatocytes exposed to 11 different hepatotoxicants. These latter were pooled into four groups labeled: necrosis (isoniazid [ISN] and acetaminophen [APAP]), cholestasis(erythromycin estolate [ERY] and {alpha}-naphtylylisothiocyanate [ANIT]), steatosis (tetracycline, 4-pentenoic acid, and amiodarone [AM]), and induction of cytochromes P450 (CYP P450) subfamilies (clofibrate [CLO], ß-naphtoflavone [BNF], phenobarbital [PB], and zileuton).

The aims of this study were to analyze changes in gene expression levels induced by in vitro primary hepatocytes exposed to different xenobiotic treatments and to determine if gene expression profiles generated with a low-density microarray would permit a classification of compounds associated signatures.

MATERIALS AND METHODS

Rat hepatocyte isolation.
Wistar rats, 7–8 weeks old on the day of sacrifice, were obtained from Iffa Credo (L’Arbresles, France). Upon arrival and for the duration of the acclimatization period, animals had free access to UV-treated water and controlled rodent diet (Dietex, Witham, U.K.). The animal room temperature was maintained between 20 and 24°C with a relative humidity of 40 to 70%. The light cycle was 12 h of light and 12 h of darkness.

Rats fasted for 24 h were anesthetized with an ip injection of sodium pentobarbital (Pharmacie du Val d’Hony, Esneux, Belgium) as a saline solution (80 mg/kg) before liver perfusion and hepatocytes were isolated using a modification of Seglen’s two step collagenase perfusion technique (Seglen, 1976Go). A laparotomy was performed and a catheter was introduced into the portal vein, allowing the perfusion of the liver in situ at 37°C, with Ca2+ and Mg2+ free Hank’s balanced salt solution (HBSS; BioWhittaker Inc., Walkersville, MD), supplemented with 0.47 mmol/l ethylene glycol-bis-(ß-aminoethyl ether)-N,N,N',N'-tetraacetic acid (EGTA; Sigma, St. Louis, MO) and 33.5 mmol/l NaHCO3 (J.T. Baker, Deventer, Holland). The pH of this solution was kept at 7.4 with permanent bubbling of sterile carbogen (5% CO2, 95% O2; Air Liquide Medical, Machelen, Belgium). A second catheter was introduced into the right atrium of the heart, permitting the recycling of the perfusion medium. After a few minutes perfusion, 90 U/ml collagenase "Hepatocytes qualified" (Invitrogen, Carlsbad, CA) and 1.5 mmol/l CaCl2 (Sigma, St. Louis, MO) were added to the perfusion medium. After 10 min of perfusion, the liver was removed and the cells were dissociated, filtered and washed in William’s E medium supplemented with 2 mmol/l L-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, and 10% v/v fetal bovine serum (WEC; Invitrogen, Carlsbad, CA). Hepatocytes number and viability were assessed by counting unstained and stained cells, after addition of trypan blue dye (Invitrogen, Carlsbad, CA), using a Burker haemocytometer. The cell suspension was considered to be valid and used when the cell viability was greater than 80%.

Cytotoxicity Assessment
To ensure that sublethal concentrations of test compounds were used in the experiment to establish the gene expression profile, the cytotoxicity was assessed in vitro, on freshly isolated male adult rat hepatocytes using the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] reduction method (Otoguro et al., 1991Go).

Compound solutions.
The different compounds were freshly dissolved at 100-fold the final concentrations either in dimethyl sulfoxide (DMSO; ICN Biomedical, Eschwege, Germany; all compounds except CLO and PB), in dimethyl formamide (DMF; Sigma, St. Louis, MO; CLO) or in H2O (PB). Each solution or corresponding vehicle was then diluted 100-fold in William’s E medium (WDI; Invitrogen, Carlsbad, CA) supplemented with 2 mmol/l L-glutamine (Invitrogen, Carlsbad, CA), 100 U/ml penicillin (Invitrogen, Carlsbad, CA), 100 µg/ml streptomycin (Invitrogen, Carlsbad, CA), 10 nmol/l insulin (Sigma, St. Louis, MO) and 10 mmol/l dexamethasone (ICN Biomedical, Eschwege, Germany) to obtain the desired final concentrations (Table 1Go).


View this table:
[in this window]
[in a new window]
 
TABLE 1 Range of Compound Concentrations and EC50 Values Determined in the MTT Assay as Well as Chemical Concentrations Used in the Experiment for Gene Expression Analysis
 
Hepatocyte incubations and MTT reduction assay.
Freshly isolated hepatocytes were seeded in collagen S-precoated 24-well plates (Becton Dickinson, Franklin Lakes, NJ) at a density of 105 viable cells/cm2 for 3 h at 37°C under a 5% CO2/95% humidified atmosphere in William’s E medium supplemented with 2 mmol/l L-glutamine, 100 U/ml penicillin, 100 µg/ml streptomycin, and 10% v/v fetal bovine serum (WEC; Invitrogen, Carlsbad, CA). After cell attachment, the medium was replaced with 1 ml of compound or vehicle solution, and the cells were incubated for a further 24 h before endpoint measurement. Compounds at the different concentrations and vehicles were tested in triplicates. Male rat hepatocytes were incubated with all the compounds while female rat hepatocytes were only incubated with acetaminophen.

The cytotoxicity was then assessed using the MTT reduction method. The test medium was discarded and fresh medium containing 1 mg/ml MTT (Sigma, St. Louis, MO) was added to the monolayers. After a 3 h incubation at 37°C in a humid atmosphere (5% C02:95% air), the medium was removed and the formazan dye formed by succinyl dehydrogenase-catalysed reaction solubilized with isopropranol. The absorbance was then measured at 550 nm using a microtiter plate reader SpectramaxPlus (Molecular Devices Corporation, Sunnyvale, CA).

The toxic effect of each compound at the different concentrations was expressed as the percentage of the absorbance determined for control cells incubated with the corresponding vehicle. The concentration that produces a change of 50% (EC50) in this endpoint assay was calculated by nonlinear iterative adjustment using the Levenburg Marquardt algorithm (XL fit Windows from Molecular Devices, Sunnyvale, CA).

Gene Expression Analysis
In order to evaluate the reliability of microarray experiments and to obtain an accurate gene expression profile for each compound, three independent hepatocyte preparations were carried out for each compound. A microarray experiment was performed on each hepatocyte preparation. The microarray experiment included the following steps; mRNA extraction, labeled cDNA synthesis, and microarray hybridization. In summary, three hybridization on microarray were performed for each compound (n = 3).

Cell treatments.
It has to be noted that for each drug and its control, the same hepatocyte preparation was used. In addition, the effect of each hepatotoxicant was tested on three independent hepatocyte preparations originating from three different rats.

Cells were seeded at 105 cells/cm2 in collagen S-precoated 75 cm2 flasks using WEC medium (15 ml/flask; Invitrogen, Carlsbad, CA). Hepatocytes were then incubated at 37°C under a 5% CO2/95% humidified atmosphere and allowed to attach for 3 h prior to the incubation with the reference compounds. The culture media was then replaced with WDI culture medium containing one of the test compounds or vehicle alone control (DMSO or DMF; Sigma, St. Louis, MO; Table 1Go). Fresh stock solutions of compounds were prepared at 100-fold the final concentration in DMSO (all compounds except CLO), at 400-fold the final concentration in DMF (CLO). The control cell medium was prepared by diluting the vehicle in WDI to reach a final concentration of 1% v/v for DMSO or 0.25 % v/v for DMF (Sigma, St. Louis, MO). The rat hepatocytes cultures were then incubated for a further 24 h. At the end of the treatment period, hepatocytes were washed twice with phosphate buffered saline (PBS) at 37°C. Cells were then stored at -80°C until mRNA extraction.

mRNA isolation.
mRNA was isolated using the KingFisherTM mRNA extraction kit according to the manufacturer’s protocol (Thermo Life Sciences, Brussels, Belgium). Cells were lysed at room temperature for 15 min. Cell lysates were centrifuged twice on Qiashredder columns (Westburg, Leusden, The Netherlands) at 14000 rpm for 2 min. mRNA extraction was then performed on non-viscous lysate with the KingFisher mlTM device. mRNA was resuspended in RNase free water and quantification was performed by spectrophotometry. Denaturing agarose gel electrophoresis was used to assess the integrity and relative contamination of mRNA with ribosomal RNA. Extracted mRNA was stored at -80°C until use.

Synthesis of labeled cDNA.
Labeled cDNA were prepared using 2 µg of mRNA. Three synthetic poly(A)+tailed RNA standards were spiked at three different amounts (10 ng, 1 ng, and 0.1 ng per reaction) into the purified mRNA as required by the microarray kit (Eppendorf, Hamburg, Germany). The RNA standards are used for quantification and estimation of experimental variation introduced during labeling and analysis. For more details concerning the cDNA preparation, please refer to de Longueville et al.(2002)Go.

Microarray design and hybridization.
The DualChip rat hepato (Eppendorf, Hamburg, Germany) contains two arrays per slide with a range of genes involved in basic cellular processes such as drug metabolism, stress responses, cell proliferation, cell cycle activation, transcription, inflammation, and apoptosis (de Longueville et al., 2002Go). To evaluate the reliability of the experimental data, several positive and negative hybridization and detection controls are included on the microarray. For normalization, three internal standard controls and eight housekeeping genes were arrayed on the slides.

The DualChip rat hepato hybridization was carried out according to the manufacturer’s instructions as reported in de Longueville et al.(2002)Go. The detection was performed by using a cyanin-3 streptavidin conjugate (Amersham Pharmacia Biotech, Buckinghamshire, England).

Imaging, statistical analysis, and clustering.
After hybridization, arrays were scanned using the GMS 418 laser confocal scanner (Genetic Microsystem, Woburn, MA) at a resolution of 10 µm. To maximize the dynamic range of microarrays, the same arrays were scanned using different photomultiplier settings (PMT). The use of different intensities allows the quantification of both the high and low copy expressed genes. After image acquisition, the scanned 16-bit image was used to quantify the signal intensities with the ImaGene 4.1 software (BioDiscovery, Los Angeles, CA). The fluorescent intensity of each DNA spot (average of intensity of each pixel present within the spot) was calculated using local mean background subtraction. A signal was accepted if the average intensity after background subtraction was at least 2.5-fold higher than its local background. The two intensity values of the duplicate DNA spots were averaged and used to determine the intensity ratio between the reference and the test samples. Very bright element intensities (saturated signals, highly expressed genes) were deemed unsuitable for accurate quantification because they underestimated the intensity ratios and were excluded from further analysis.

Several potential sources of experimental variation could occur during cDNA synthesis, labeling, hybridization, and indirect detection steps. To take into account these possible variations, the data were normalized in a two step procedure. The values were first corrected using a factor calculated from the intensity ratios of the internal standards in the references and test samples. The presence of the three internal standard probes at two different locations of the microarray allowed to measure a local background and to evaluate the microarray homogeneity, which is taken into account in the normalization (Schuchhardt et al., 2000Go). However, as the internal standard control does not take into account the purity and quality of the mRNA, a second step of normalization was performed based on the expression levels of the housekeeping genes. This process involved calculating the average intensity from a set of housekeeping genes. Among these housekeeping genes, only genes for which the expression was not changed after a particular treatment were taken into account for the normalization. Indeed, any drug may affect the expression of some of the housekeeping genes.

The variance of the normalized set of housekeeping genes (except those affected by the treatment) was used to generate an estimate of expected variance, leading to a predicted confidence interval (CI) to test the significance of the ratios obtained (Chen et al., 1997Go; de Longueville et al., 2002Go). Ratios outside the 99% confidence interval were determined to be significantly different. ANOVA was used to examine the data.

Before performing the cluster analysis, ratios falling inside the 99% confidence interval were replaced by the value 1. Clusters of hybridization profiles were created with the s-plus 2000 software (Insightful, Seattle, WA) using the classical agglomerative hierarchical with the single link. The distance computed between two hybridization profiles corresponds to the Manhattan distance (Van Custem et al., 1994Go).

Validation of relative gene expression by real-time PCR.
The single strand-cDNA (ss-cDNA) was synthesized from 0.5 µg mRNA according to the RNA labeling protocol described in de Longueville et al.(2002)Go with the following minor modifications: (1) a DNase treatment of mRNA was performed prior to cDNA synthesis; (2) the dNTP mixture contained dGTP, dATP, dTTP, and dCTP each at 500 µM but no biotinylated dCTP; (3) the second addition of reverse transcriptase was omitted.

Gene specific primers correspond to the gene sequence present on the DualChip rat hepato (Eppendorf, Hamburg, Germany). Forward and reverse primers for real-time PCR amplification were designed with the Primer Express Software (PE Applied Biosystem, Foster City, CA).

Real time PCR was performed on six genes, namely, CYP 2B1/2, CYP 3A, GST Ya, Smp30, GAPDH (house keeping gene), and ribosomal protein S9 (house keeping gene). mRNA extracted from hepatocytes exposed to PB and CLO was used in the real time PCR (n = 2) and each reaction was performed in triplicate.

PCR reaction mixtures contained of 12.5 µl SYBR green PCR Master Mix 2X (PE Applied Biosystems, Foster City, CA), 2.5 µl forward primer (3 mM), (PE Applied Biosystems, Foster City, CA), 2.5 µl reverse primer (3 mM) (PE Applied Biosystems, Foster City, CA), 5 µl cDNA and 2.5 µl distilled water. PCR reactions without cDNA were performed as template-free negative controls. All PCR reactions were made in duplicates with the following PCR conditions: 2 min at 50°C, 10 min at 95°C followed by 40 cycles of 15 s at 95°C and 1 min at 60°C in 96-well optical plates (PE Applied Biosystem, Foster City, CA) in the ABI 7000 Sequence Detection System (Perkin-Elmer Life Sciences, Boston, MA). The ABI PRISM 7700 sequence detection system software (version 1.6) was used for data analysis according to the manufacturer’s instructions (PE Applied Biosystem, Foster City, CA).

Fluorescence emission was detected for each PCR cycle and the threshold cycle (CT) values were determined. The CT value was defined as the actual PCR cycle when the fluorescence signal increased above the background threshold. Average CT values from duplicate PCR reactions were normalized to average CT values for housekeeping gene from the same cDNA preparations. The ratio of expression of each gene in hepatotoxicants treated vs. vehicle sample was calculated as 2-({Delta}{Delta}CT) of that treatment as recommended by Perkin-Elmer where CT is the threshold cycle and {Delta}{Delta}CT is the difference CT (test gene) – CT (housekeeping gene) for treated sample minus vehicle sample. Values were reported as an average of triplicate analyses.

RESULTS

Cytotoxicity Assessment
Incubation for 24 h with AM, ANIT, ERY, tetracycline, CLO, 4-pentenoic acid, and PB decreased the MTT reduction in freshly isolated male rat hepatocytes in a concentration dependent manner, with EC50 values of 14, 17, 64, 787, 2511, 6288, and 11253 µM, respectively (Table 1Go and Fig. 1Go). The BNF EC50 value of 30 µM was estimated by graph extrapolation. EC50 values could not be calculated for zileuton, ISN, and APAP in male hepatocytes because only mild effects were observed even for the highest tested concentrations. For these three compounds, the maximal reduction of the MTT end-point was 35, 12, and 18% respectively for 300 µM zileuton, 10 mM ISN, and APAP (male). Consequently, EC50 values for zileuton, ISN, and APAP (male) are higher than the highest concentrations tested. Finally, the female rat hepatocytes were more susceptible to APAP toxicity, with a decrease in the MTT metabolism of 67% compared to only 18% in male rat hepatocytes at a concentration of 10 mM.



View larger version (35K):
[in this window]
[in a new window]
 
FIG. 1. MTT reduction in rat hepatocytes treated for 24 h with various hepatotoxicants. The identities of the drugs are indicated in each figure. Male hepatocytes were used with all toxicants, but for APAP, female hepatocytes (F) were also used. Rat hepatocytes were incubated for 24 h with the different hepatotoxicants before assessing the cell viability by MTT reduction. Results (means ± SD for triplicate wells) were expressed as a percentage of the MTT determination for control cells incubated with control solvent. The arrows indicate the concentration of toxicant used in the gene expression experiment. For BNF, the concentration used was 2 µM. For the exact toxicant concentrations, please refer to Table 1Go. * and ** mean precipitation of the compound in the test medium and acidification of the test medium, respectively.

 
Analysis of Gene Expression Modifications Induced by the Hepatotoxicants
Vehicle-treated samples.
Hepatocellular gene expression changes induced by DMSO and DMF (the solvents used to dissolve the test compounds) are shown in Figure 2Go. Acyl-Co-oxidase, involved in peroxisome proliferation and HGPT (hypoxanthine guanine phosphoribosyl transferase), a housekeeping gene, were respectively up- and downregulated by DMSO (p < 0.01). On the other hand, two genes implicated in stress responses, namely GSH reductase and MDR-1b (multi-drug resistance) were downregulated by DMF (p < 0.01).



View larger version (20K):
[in this window]
[in a new window]
 
FIG. 2. Logarithmic scatter plots of normalized fluorescence intensity values from DualChip rat hepato hybridized with cDNA obtained from mRNA extracted from DMSO-treated hepatocytes (A) and from DMF-treated hepatocytes (B) versus reference (hepatocytes in WDI medium). The arrows indicate the genes that are significantly up- or downregulated by the solvent treatment (p < 0.01).

 
Hepatotoxicant treated samples.
A global view of the different gene expression profiles induced by the various treatments is presented in Figure 3Go as well as an example of the microarray pictures obtained after hybridization.




View larger version (97K):
[in this window]
[in a new window]
 
FIG. 3. Gene expression profiles of DualChip rat hepato hybridized with cDNA obtained from mRNA extracted from control and compound treated rat hepatocytes (n = 3). (A) The data are expressed as mean ratio (treatment vs. reference as well as solvent vs. WDI) outside the 99% confidence interval. The range of changes is represented by a code of colors at the bottom of the chart. House keeping genes appear in red. Individual genes are grouped into functional classes. Genbank accession numbers are also given under the column heading "Acc #." All compounds were dissolved in DMSO except CLO in DMF. (B) Dualchip rat hepato hybridized with cDNA obtained from mRNA extracted from reference and BNF treated sample. Fluorescence is represented in pseudocolor scale and corresponds to the expression levels of genes. The arrows show two examples of changes in gene expression levels (induction of CYP 1A1 and GST Ya in BNF treated hepatocytes). For the description of each spot, please refer de Longueville et al.(2002)Go.

 
Cytochrome P450s Inducers
Different subfamilies of CYP P450 1A, 2B, 3A, and 4A were significantly induced by BNF, PB, zileuton, and CLO (p < 0.01, Figs. 3Go and 4AGo). Expression of the major CYP P450 genes was increased by a factor of at least 25. For example BNF, PB, and CLO induced CYP1A1, 2B, and 4A1 by a factor 25.41, 45.01, and 33.03 respectively. In addition, PB decreased the expression of CYP4A1 (Figs. 3Go and 4AGo). An overall view of the data is presented in Figure 4BGo. CLO treatment significantly changed the expression of 12 genes included on the microarray versus 7 genes for zileuton and 6 genes for PB and BNF.



View larger version (49K):
[in this window]
[in a new window]
 
FIG. 4. Gene expression profiles obtained from rat hepatocytes exposed to (A) BNF, PB, CLO, or zileuton, (B) ISN and APAP, (C) ERY and ANIT, and (D) tetracycline, pentenoic acid, and AM. Expression data are presented as a mean ratio (treatment vs. reference) (n = 3, p < 0.01). Only genes having a significant ratio are presented in the table. The red and green codes correspond to up and down regulated genes, respectively. The gray code means no significant change between treatment and reference (p < 0.01). Genbank accession numbers are also given under the column heading "Acc #."

 
Drugs Inducing Hepatocellular Necrosis
The data revealed significant gene expression changes for 10 and 11 genes after a treatment with ISN and APAP, respectively (Figs. 3Go and 4BGo). A comparison among the drugs inducing necrosis shows that four genes implicated in phase I metabolism (CYP3A1 and Acyl-coA-oxidase), phase II metabolism (glutathione S-transferase Ya; GST Ya), and growth arrest and DNA damage response (GADD153) followed the same tendency (Fig. 4Go). In addition, the gene expression profiles were not identical between female and male hepatocytes exposed to APAP. Seven genes were differentially expressed in response to APAP. APOJ, albumin, and fibronectin were induced to a greater extent in female hepatocytes whereas MDR-1b and Ornithine decarboxylase (ODC) were only repressed in male hepatocytes treated with APAP, respectively. In addition, the UDP-glucuronosyltransferase gene (UDPGT 1A6) implicated in phase II metabolism was differentially regulated in male and female hepatocytes exposed to APAP. Finally, seven genes followed the same tendency after treatment to APAP in male and female hepatocytes (Fig. 4BGo).

Drugs Inducing Hepatocellular Cholestasis
The expression of three genes involved in apoptosis (Bcl-2), phases I (CYP 3A1) and II (UDPGT 1A) metabolism were significantly upregulated in rat hepatocytes after a treatment with ERY whereas Bax was downregulated (Figs. 3Go and 4CGo). On the other hand, ANIT induced the expression of 12 genes involved in phases I (CYP3A1) and II (GST Ya and theta5) metabolism, apoptosis (Bax, Bcl-2), oncogenesis (c-Myc), stress response (Hsp 70, MnSOD; manganese superoxide dismutase), cell proliferation (PCNA; proliferation cellular nuclear antigen), cellular markers such as alpha-2-macroglobulin, structural element like fibronectin and arginine synthesis (ODC). The gene expression comparison between both treatments (ERY and ANIT) revealed that only two genes had the same gene expression profile (CYP 3A1 and Bcl-2) whereas the expression of Bax was either down- or upregulated in hepatocytes exposed to ERY or ANIT.

Tetracycline, Pentenoic Acid, and Amiodarone
After tetracycline treatment, the expression of only three genes was changed; CYP4A1 and GADD153 were upregulated while senescence marker protein-30 (Smp30) was downregulated (Figs. 3Go and 4DGo). Three genes were also found to be differentially expressed after pentenoic acid and AM treatments. CYP4A1 and C-Jun were upregulated and Smp30 was downregulated by pentenoic acid. On the other hand, AM significantly induced the expression of cytochrome-c-oxidase and CYP4A1 and repressed the expression of Smp30. Both CYP 4A1 and Smp30 transcript levels were significantly up- and downregulated by tetracycline, pentenoic acid and AM (Figs. 3Go and 4DGo).

Gene Expression Validation
Based on data collected from the microarray technique and presented in Fig. 4AGo, the expression profile was validated for six genes responsive to CLO and PB treatment (CYP 2B1/2, CYP 3A, GST Ya, Smp-30, GAPDH, and ribosomal protein S29; Fig. 5Go).



View larger version (55K):
[in this window]
[in a new window]
 
FIG. 5. Comparison of gene expression data determined by DNA microarray and real-time PCR. (A) Real-time PCR SYBR Green I Fluorescence versus cycle number of CYP 2B1/2 gene and reference gene (ribosomal protein S29) in PB treated sample and vehicle treated sample. In the amplification plot, A refers to ribosomal protein S29 in vehicle and PB treated samples, B to CYP2B1/2 in PB treated sample. Closed arrowhead shows PCR with no template. Striped arrow indicates the position of the noise band. (B) Comparison of microarray and real time PCR gene expression data. Microarray data derived from Figure 4AGo are shown as mean ratios (treatment vs. reference, n = 3, p < 0.01). The red and green codes correspond to up and down regulated genes, respectively. The gray code means no significant change between treatment and reference (p < 0.01). House keeping genes appear in red. Genbank accession numbers are also given under the column heading "Acc #."

 
Real time PCR data revealed that CYP 2B1/2, CYP 3A1, and GST-Ya were induced 189.2, 14.66, and 3.95 times in PB treated hepatocytes (Fig. 5BGo). On the other hand, the expression of smp30 was repressed by a factor 0.56 in the cells exposed to PB. The expression of GADPH and ribosomal S29 was not affected by PB as the ratios were quite close to 1. Figure 5BGo shows as well that CLO changed the expression of CYP 2B1/2, CYP 3A1, and GST-Ya by a factor 100.33, 1.97, and 2.07. The expression of smp30 is not affected by CLO as the ratios were quite close to 1.

The gene expression measured by the means of DNA microarrays followed the same tendency for the six genes measured.

Clustering Analysis
The comparison of all gene expression profiles generated by the 11 reference compounds revealed that six different clusters were observed (Fig. 6Go). The first cluster contains tetracycline, pentanoic acid, and AM. The second one includes APAP, ANIT, ERY, and ISN. Four other clusters were formed by individual drugs that are BNF, PB, zileuton, and CLO.



View larger version (14K):
[in this window]
[in a new window]
 
FIG. 6. Clustering analysis of the gene expression patterns induced by the 11 hepatotoxicants. A cluster was formed for tetracycline, pentenoic acid, and AM (red color). The drugs APAP, ISN, ANIT, and ERY were pooled in the same cluster which can be divided in two sub-clusters (firstly: ANIT and APAP and secondly: ISN and ERY) (blue color). BNF, PB, zileuton and CLO formed four individual clusters (in pink, brown, green, and black colors, respectively). The drugs inducing cholestasis, necrosis and steatosis are represented by the asterisk, filled circle, and filled diamond, respectively.

 
DISCUSSION

In the present study, we used a low-density microarray containing 59 genes to analyze the gene expression profiles generated in primary cultures of rat hepatocytes exposed to 11 different known hepatotoxicants. The drugs were pooled into four groups labeled necrosis (ISN and APAP), cholestasis (ERY and ANIT), steatosis (tetracycline, 4-pentenoic acid, and AM), and induction of CYP P450 subfamilies (CLO, BNF, PB, and zileuton). Our main goal was to analyze changes in gene expression levels induced by these drugs and to determine if the transcription profiles would permit a classification of compounds associated signatures.

Male hepatocytes were used for all drugs except for APAP for which female hepatocytes were also used. Indeed, APAP is known to produce sex-dependent hepatotoxicity in young adult rats (Tarloff et al., 1996Go) and consequently we expected to see some modifications in gene expression between male and female hepatocytes after a treatment with APAP. Incubation conditions with compounds can vary considerably between toxicological studies leading to various results.

Based on a previous study dealing with the kinetic of gene expression (hepatocytes exposed to PB for 24, 48, or 72 h), 24 h was selected because this time point provided for PB a good gene expression response. Twenty-four h was also selected for the other drugs used in this study because it would most likely provide a complete gene response in hepatocytes without the interference of significant secondary responses that could be encountered at later timepoints. In addition, a time of 24 h after treatment has been frequently selected to analyze gene expression modifications under in vitro conditions (Baker et al., 2001Go; Waring et al., 2001aGo). The range of concentration tested in the cytotoxicity assay was based on preliminary assays using a wide range of concentrations and concentrations of the different compounds selected for the gene expression experiment were chosen based on cell viability assays (MTT curve). For all compounds, the selected concentrations did not induce any cell death (even for female hepatocytes exposed to APAP after curve correction (see Fig. 1Go) but were generally close to the lowest concentrations inducing cell mortality. This approach was used since literature data were not available for many compounds regarding to the concentrations required to elicit toxic effects on hepatocytes either under in vitro nor in vivo conditions. Based on mechanisms of action and differential induced-toxicity, it is almost impossible to test the different classes of molecules in the same range of concentrations. Indeed, it is clear that toxic effects of similar order are observed at different doses for diverse drugs.

Although the mode of action of BNF, CLO, PB, and zileuton are different, we arbitrarily pooled these drugs because they are inducers of cytochrome P450 isoforms (Gerhold et al., 2001Go; Sundseth and Waxman, 1992Go). BNF is an aromatic hydrocarbon that induces the expression of CYP1A family by activating the Aryl hydrocarbon (Ah) receptor (Denison and Heath-Pagliuso, 1998Go). In the present study, induction of gene encoding CYP1A and several phase II enzymes namely UDPGT1A6 and GST Ya was also observed in rat hepatocytes after a BNF treatment as reported elsewhere (Maheo et al., 1997Go; Saarikoski et al., 1998Go).

The barbiturate PB induces the transcription of the rat gene CYP2B and CYP3A (Frueh et al., 1997Go; Meyer and Hoffmann, 1999Go) through the constitutive androstane receptor (CAR; (Honkakoski et al., 1998Go; Masahiko and Honkakoski, 2000Go). In our study, PB induced CYP2B, CYP3A and repressed CYP4A1. CYP4A1 is involved in the {omega}-hydroxylation of fatty acids (Gibson et al., 1982Go) and therefore a down regulation could lead to cellular dysfunction. Induction of genes encoding phase II metabolism enzymes (GST Ya and UDGT1A) was also observed in our study. In addition, the mRNA level of the senescence marker protein-30 (Smp30) was significantly reduced following PB treatment as reported in other studies (Fujita et al., 1999Go).

CLO, a lipid lowering agent, triggers peroxisome proliferation in rodents and induces genes involved in the ß-oxidation of fatty acids by the activation of the peroxisome proliferator-activated receptor alpha (PPAR{alpha}; Corton et al., 2000Go; Lindquist et al., 1998Go; Simpson, 1997Go). CLO is known to induce members of the CYP4A subfamily genes (Surry et al., 2000Go and present study). CLO significantly induced the expression of CYP 2B, CYP 3A, GST ya, theta5, and UDPGT1A as reported elsewhere (Jemnitz et al., 2000Go; Ritter and Franklin, 1987Go; Ronis et al., 1994Go). In our study, increased lipid ß-oxidation in response to CLO is supported by the induction of genes encoding peroxisomal enzymes such as acyl-CoA oxidase and peroxisomal enoyl-CoA-hydratase.

Zileuton, a 5-lipoxygenase inhibitor, is considered to be a moderate inducer of CYP 450 (Rodrigues and Machinist, 1996Go) and it induced CYP2B in our study. However, zileuton also significantly modified the expression of GADD153, Erk1, Histone deacetylase (Hdac), Ferritin subunit H, fibronectin and HMG-CoA-synthetase (3-hydroxy-3methylglutaryl-CoA-synthetase). A comparison of our data with other studies is difficult since, to our knowledge, the effect of zileuton on gene expression has not been published yet.

The next compounds studied were ISN and APAP, which are known to induce necrosis. ISN is a first-line drug in the prophylaxis and treatment of tuberculosis (Sadaphal et al., 2001Go). Little is known about the effect of ISN on gene expression. In our study, ISN induced CYP3A, GST Ya, GADD153, Hsp70, HO-2 (heme oxygenase 2), transferin, and cytochrome c-oxidase.

APAP is known to induce the depletion of glutathione and cell death (Ray and Jena, 2000Go), impair the mitochondrial respiration (Burcham and Harman, 1991Go), and interfere with Ca++ homeostasis (Salas and Corcoran, 1997Go). However so far, the exact mechanism of action of acetominophen has not been completely elucidated although recent reports have identified the constitutive androstane receptor as a regulator of APAP hepatotoxicity (Zhang et al., 2002Go). APAP is mainly metabolized by cytochromes P450 (CYP 3A), by glucuronidation and sulforination pathways (Tygstrup et al., 2002Go). Our data show that APAP changed the expression of genes implicated in drug metabolism (induction of CYP 3A1, GST Ya, UDGT1a, UDGT1a6), stress response (repression of MDR-1b, induction of Hsp70), DNA repair (induction of GADD153 and repression of MGMT), and oncogenesis (induction of c-myc). It is noteworthy that growth arrest and DNA damage (GADD) 153 and Hsp70 have been associated with the induction of apoptosis (Fontanier-Razzaq et al., 1999Go; Reilly et al., 2001Go). APAP is also known to produce sex-dependent hepatotoxicity in young adult rats (Tarloff et al., 1996Go). Our data reveal some important gene expression differences between male and female rat hepatocytes. For instance, APOJ, albumin and fibronectin were only induced in female hepatocytes, whereas repression of MDR-1b and induction of ODC only occurred in male hepatocytes.

Intrahepatic cholestasis has been reported to occur during ERY and ANIT therapy (Orsler et al., 1999Go). In our study, CYP3A and Bcl-2 were induced by both drugs as reported elsewhere (Celli et al., 1998Go; Que et al., 1997Go) and Bax was repressed and induced by ERY and ANIT, respectively. Aoshiba and co-workers (1995)Go also reported the effects of ERY on apoptotic genes. The increased of Bcl-2 expression is protective against apoptosis due to its intracellular antioxidant action (Gottlieb et al., 2000Go). In addition to these apoptotic genes, ANIT upregulated GST-Ya and theta 5 as already reported by other studies (Lesage et al., 2001Go; Ohta et al., 2001Go), PCNA (Ranganna et al., 2000Go), c-myc, Hsp70, fibronectin, alpha-2-macroglobulin, and ODC. However, the correlation between ANIT toxicity and these genes is not yet established.

Tetracycline, pentenoic acid, and AM are known to induce hepatocellular steatosis (Fromenty and Pessayre, 1995Go; Loscher et al., 1993Go). In the present study, these three compounds up- and downregulated CYP4A1 and Smp30, respectively. Smp30 seems to play a critical role in the highly differentiated functions of the liver and its down-regulation may contribute to hepatic deterioration of cellular functions induced by steatosis (Fujita et al., 1999Go; Ishigami et al., 2002Go). CYP4A induction always accompanies any substantial drug-dependent increases in beta-oxidation (Amacher and Martin, 1997Go; Tang et al., 1995Go). Robertson and co-workers (2001)Go suggested that the induction of CYP4A could be used as a good marker to assess steatosis injury. Other genes were differently expressed after tetracycline, pentenoic acid, and AM treatments. For instance, tetracycline induced the expression of GADD153, a growth arrest and DNA damage gene. On the other hand, pentenoic acid induced C-Jun, a nuclear transcription factor and such events may lead to toxic events (Chung et al., 2001Go; Kovary and Bravo, 1991Go).

Microarray measurements are usually semi-quantitative, with compression of values occurring at high-fold changes (Gerhold et al., 2001Go; Rajeevan et al., 2001Go; Yuen et al., 2002Go) but generally the data generated by microarray are in agreement with real time PCR results. In the present study, it was shown that the six genes measured with both technologies followed the same tendency for PB and CLO treated hepatocytes. Indeed, a compression of the values occurs at high-fold changes in expression but as observed, the quantifications made by the two methods are well correlated.

Even with a relatively limited gene set, all the 11 compounds gave rise to discernable gene expression profiles as already obtained with high-density microarrays (Hamadeh et al., 2002bGo; Morgan et al., 2002Go). When clustering analysis is performed, it has to be noted that drugs inducing similar endpoints (e.g., cholestasis) may trigger different mechanism of actions. Thus, such drugs will not necessarily change the expression of the same set of genes. BNF, PB, CLO, and zileuton arbitrarily pooled in the CYP450 inducers formed four individual clusters, which confirmed that they act through different mechanisms of action. The present study also shows that some compounds belonging to the same class of toxicant were linked, suggesting that they target similar genes and possibly through the same mechanism of action. For instance, a cluster was formed for tetracycline, pentenoic acid, and amiodarone, drugs that are known to induce steatosis. On the other hand, the drugs inducing necrosis (APAP and ISN) and cholestasis (ANIT and ERY) were pooled in the same cluster which can be divided in two sub-clusters (firstly: ANIT and APAP and secondly: ISN and ERY). Interestingly, APAP and ANIT, which belong to the cholestasis and necrosis groups, respectively, are also known to be inducers of apoptosis. Thus, this may explain why both drugs were clustered together. However, it has to be said that the low number of genes studied may also diminish the power of the clustering analysis.

The use of cultured hepatocytes to model hepatotoxicity has proven to be a valuable tool despite some limitations (see Introduction for more details). Some reports have shown that gene expression data showed a good correlation between in vitro and in vivo models. For instance, hepatocytes treated with PPAR{alpha} agonist fenofibrate produced gene expression changes characteristic of the in vivo response in rat liver (Baker et al., 2003Go). The data revealed remarkable similarities in both the affected biological pathways and the rank-order magnitude of the response. The present study shows also a good correlation with regard to induction and repression of gene expression obtained in primary rat hepatocytes when compared to in vivo data.

Low-density microarrays seem to represent a useful tool to select drug candidates early in the development in conjunction with other data (e.g., toxicokinetic and pharmacological studies). For instance, drugs that do not change the expression of genes implicated in phase-1 and -2 metabolisms could be of particular interest. In addition, another attractive application could be to compare gene expression patterns generated by a key compound and its analogs. This would allow the selection of the best analogs based on gene expression comparison.

In conclusion, the in vitro gene expression data generated in this study were in good agreement with the literature, which mainly concerns in vivo data. Furthermore, gene expression profiles observed in this study have been confirmed for several genes by real-time PCR assays. This confirmation validates our results and supports the use of microarray technology in toxicogenomics. Each drug gave unique gene expression profile. Despite the low number of genes studied, the gene expression patterns allowed a certain degree of classification of compounds with similar hepatocellular injuries. Finally, low-density microarrays represent a powerful tool to investigate mechanistic toxicology issues and to help in the selection of the best drug candidates in conjunction with other data.

ACKNOWLEDGMENTS

This work was supported by the Region Wallonne, Belgium. Furthermore, we would like to thank Anne-France Dabee for her technical assistance. Thierry Arnould is a Research Associate of FNRS (Fonds National de la Recherche Scientifique, Brussels, Belgium).

NOTES

1 To whom correspondence should be addressed. Fax: +32-81-72-56-23. E-mail: delongueville.f{at}eppendorf.be. Back

REFERENCES

Amacher, D. E., and Martin, B. A. (1997). Tetracycline-induced steatosis in primary canine hepatocyte cultures. Fundam. Appl. Toxicol. 40, 256–263.[CrossRef][ISI][Medline]

Aoshiba, K., Nagai, A., and Konno, K. (1995). Erythromycin shortens neutrophil survival by accelerating apoptosis. Antimicrob. Agents Chemother. 39, 872–877.[Abstract]

Baker, T. K., Carfagna, M. A., Gao, H., Dow, E. R., Li, Q., Searfoss, G. H., and Ryan, T. P. (2001). Temporal gene expression analysis of monolayer cultured rat hepatocytes. Chem. Res. Toxicol. 14, 1218–1231.[CrossRef][ISI][Medline]

Baker, T. K., Higgins, M. A., Carfagna, M. A., and Ryan, T. P. (2003). Characterization of hepatocytes and their use as a model system in toxicogenomics. In An Introduction to Toxicogenomics (M. E. Burczynski, Ed.), pp. 117–143. CRC Press, Boca Raton, FL.

Bartosiewicz, M. J., Jenkins, D., Penn, S., Emery, J., and Buckpitt, A. (2001). Unique gene expression patterns in liver and kidney associated with exposure to chemical toxicants. J. Pharmacol. Exp. Ther. 297, 895–905.[Abstract/Free Full Text]

Bulera, S. J., Eddy S. M., Ferguson, E., Jatkoe, T. A., Reindel, J. F., Bleavins, M. R., and De La Iglesia, F. A. (2001). RNA expression in the early characterization of hepatotoxicants in Wistar rats by high-density DNA microarrays. Hepatology 33, 1239–1258.[CrossRef][ISI][Medline]

Burcham, P. C., and Harman, A. W. (1991). Acetaminophen toxicity results in site-specific mitochondrial damage in isolated mouse hepatocytes. J. Biol. Chem. 266, 5049–5054.[Abstract/Free Full Text]

Burczynski, M. E., McMillian, M., Ciervo, J., Li, L., Parker, J. B., Dunn, R. T., Hicken, S., Farr, S., and Johnson, M. D. (2000). Toxicogenomics-based discrimination of toxic mechanism in HepG2 human hepatoma cells. Toxicol. Sci. 58, 399–415.[Abstract/Free Full Text]

Celli, A., Que F. G., Gores, G. J., and LaRusso, N. F. (1998). Glutathione depletion is associated with decreased Bcl-2 expression and increased apoptosis in cholangiocytes. Am. J. Physiol. 275, G749–G757.[ISI][Medline]

Chen, Y., Dougherty, E. R., and Bittner, M. L. (1997). Ratio-based decisions and the quantitative analysis of cDNA microarray images. J. Biomed. Optics 2, 364–374.[CrossRef]

Chung, W. H., Bennett B. M., Racz, W. J., Brien, J. F., and Massey, T. E. (2001). Induction of c-jun and TGF-beta 1 in Fischer 344 rats during amiodarone-induced pulmonary fibrosis. Am. J. Physiol. Lung Cell Mol. Physiol. 281, L1180–L1188.[Abstract/Free Full Text]

Corton, J. C., Lapinskas, P. J., and Gonzalez, F. J. (2000). Central role of PPARalpha in the mechanism of action of hepatocarcinogenic peroxisome proliferators. Mutat. Res. 448, 139–151.[ISI][Medline]

de Longueville, F., Surry, D., Meneses-Lorente, G., Bertholet, V., Talbot, V., Evrard, S., Chandelier, N., Pike, A., Worboys, P., Rasson, J. P., et al. (2002). Gene expression profiling of drug metabolism and toxicology markers using a low-density DNA microarray. Biochem. Pharmacol. 64, 137–149.[CrossRef][ISI][Medline]

Denison, M. S., and Heath-Pagliuso, S. (1998). The Ah receptor: A regulator of the biochemical and toxicological actions of structurally diverse chemicals. Bull. Environ. Contam. Toxicol. 61, 557–568.[ISI][Medline]

Fielden, M. R., and Zacharewski, T. R. (2001). Challenges and limitations of gene expression profiling in mechanistic and predictive toxicology. Toxicol. Sci. 60, 6–10.[Abstract/Free Full Text]

Fontanier-Razzaq, N. C., Hay, S. M., and Rees, W. D. (1999). Upregulation of CHOP-10 (gadd153) expression in the mouse blastocyst as a response to stress. Mol. Reprod. Dev. 54, 326–332.[CrossRef][ISI][Medline]

Fromenty, B., and Pessayre, D. (1995). Inhibition of mitochondrial beta-oxidation as a mechanism of hepatotoxicity. Pharmacol. Ther. 67, 101–154.[CrossRef][ISI][Medline]

Frueh, F. W., Zanger, U. M., and Meyer, U. A. (1997). Extent and character of phenobarbital-mediated changes in gene expression in the liver. Mol. Pharmacol. 51, 363–369.[Abstract/Free Full Text]

Fujita, T., Shirasawa, T., and Maruyama, N. (1999). Expression and structure of senescence marker protein-30 (SMP30) and its biological significance. Mech. Ageing Dev. 107, 271–280.[CrossRef][ISI][Medline]

Gerhold, D., Lu, M., Xu, J., Austin, C., Caskey, C. T., and Rushmore, T. (2001). Monitoring expression of genes involved in drug metabolism and toxicology using DNA microarrays. Physiol. Genomics 5, 161–170.[Abstract/Free Full Text]

Gibson, G. G., Orton, T. C., and Tamburini, P. P. (1982). Cytochrome P-450 induction by clofibrate. Purification and properties of a hepatic cytochrome P-450 relatively specific for the 12- and 11-hydroxylation of dodecanoic acid (lauric acid). Biochem. J. 203, 161–168.[ISI][Medline]

Gottlieb, E., Vander Heiden, M. G., and Thompson, C. B. (2000). Bcl-x(L) prevents the initial decrease in mitochondrial membrane potential and subsequent reactive oxygen species production during tumor necrosis factor alpha-induced apoptosis. Mol. Cell Biol. 20, 5680–5689.[Abstract/Free Full Text]

Hamadeh, H. K., Bushel, P. R., Jayadev, S., DiSorbo, O., Bennett, L., Li, L., Tennant, R., Stoll, R., Barrett, J. C., Paules, R. S., et al. (2002a). Prediction of compound signature using high density gene expression profiling. Toxicol. Sci. 67, 232–240.[Abstract/Free Full Text]

Hamadeh, H. K., Bushel, P. R., Jayadev, S., Martin, K., DiSorbo, O., Sieber, S., Bennett, L., Tennant, R., Stoll, R., Barrett, J. C., et al. (2002b). Gene expression analysis reveals chemical-specific profiles. Toxicol. Sci. 67, 219–231.[Abstract/Free Full Text]

Honkakoski, P., Zelko, I., Sueyoshi, T., and Negishi, M. (1998). The nuclear orphan receptor CAR-retinoid X receptor heterodimer activates the phenobarbital-responsive enhancer module of the CYP2B gene. Mol. Cell. Biol. 18, 5652–5658.[Abstract/Free Full Text]

Ishigami, A., Fujita, T., Handa, S., Shirasawa, T., Koseki, H., Kitamura, T., Enomoto, N., Sato, N., Shimosawa, T., and Maruyama, N. (2002). Senescence marker protein-30 knockout mouse liver is highly susceptible to tumor necrosis factor-alpha- and Fas-mediated apoptosis. Am. J. Pathol. 161, 1273–1281.[Abstract/Free Full Text]

Jemnitz, K., Veres, Z., Monostory, K., and Vereczkey, L. (2000). Glucuronidation of thyroxine in primary monolayer cultures of rat hepatocytes: In vitro induction of UDP-glucuronosyltranferases by methylcholanthrene, clofibrate, and dexamethasone alone and in combination. Drug Metab. Dispos. 28, 34–37.[Abstract/Free Full Text]

Johnson, D. E., and Wolfgang, G. H. (2000). Predicting human safety: Screening and computational approaches. Drug Discovery Today 5, 445–454.[CrossRef][ISI][Medline]

Kovary, K., and Bravo, R. (1991). Expression of different Jun and Fos proteins during the G0-to-G1 transition in mouse fibroblasts: in vitro and in vivo associations. Mol. Cell Biol. 11, 2451–2459.[ISI][Medline]

Lesage, G., Glaser, S., Ueno, Y., Alvaro, D., Baiocchi, L., Kanno, N., Phinizy, J. L., Francis, H., and Alpini, G. (2001). Regression of cholangiocyte proliferation after cessation of ANIT feeding is coupled with increased apoptosis. Am. J. Physiol. Gastrointest. Liver Physiol. 281, G182–190.[Abstract/Free Full Text]

Lindquist, P. J., Svensson, L. T., and Alexson, S. E. (1998). Molecular cloning of the peroxisome proliferator-induced 46-kDa cytosolic acyl-CoA thioesterase from mouse and rat liver-recombinant expression in Escherichia coli, tissue expression, and nutritional regulation. Eur. J. Biochem. 251, 631–640.[Abstract]

Loscher, W., Nau, H., Wahnschaffe, U., Honack, D., Rundfeldt, C., Wittfoht, W., and Bojic, U. (1993). Effects of valproate and E-2-en-valproate on functional and morphological parameters of rat liver. II. Influence of phenobarbital comedication. Epilepsy Res. 15, 113–131.[CrossRef][ISI][Medline]

Maheo, K., J. Antras-Ferry, J., Morel, F., Langouet, S., and Guillouzo, A. (1997). Modulation of glutathione S-transferase subunits A2, M1, and P1 expression by interleukin-1beta in rat hepatocytes in primary culture. J. Biol. Chem. 272, 16125–16132.[Abstract/Free Full Text]

Masahiko, N., and Honkakoski, P. (2000). Induction of drug metabolism by nuclear receptor CAR: Molecular mechanisms and implications for drug research. Eur. J. Pharm. Sci. 11, 259–264.[CrossRef][ISI][Medline]

Meyer, U. A., and Hoffmann, K. (1999). Phenobarbital-mediated changes in gene expression in the liver. Drug Metab. Rev. 31, 365–373.[CrossRef][ISI][Medline]

Morgan, K. T., Ni, H., Brown, H. R., Yoon, L., Qualls, C. W., Crosby, L. M., Reynolds, R., Gaskill, B., Anderson, S. P., Kepler, T. B., et al. (2002). Application of cDNA microarray technology to in vitro toxicology and the selection of genes for a real-time RT-PCR-based screen for oxidative stress in Hep-G2 cells. Toxicol. Pathol. 30, 435–451.[CrossRef][ISI][Medline]

Ohta, Y., Kongo, M., and Kishikawa, T. (2001). Effect of melatonin on changes in hepatic antioxidant enzyme activities in rats treated with alpha-naphthylisothiocyanate. J. Pineal. Res. 31, 370–377.[CrossRef][ISI][Medline]

Orsler, D. J., Ahmed-Choudhury, J., Chipman, J. K., Hammond, T., and Coleman, R. (1999). ANIT-induced disruption of biliary function in rat hepatocyte couplets. Toxicol. Sci. 47, 203–210.[Abstract]

Otoguro, K., Komiyama, K., Omura, S., and Tyson, C. A. (1991). An in vitro cytotoxicity assay using rat hepatocytes and MTT and Coomassie blue dye as indicators. ATLA 19, 352–360.[ISI]

Que, F. G., Gores, G. J., and LaRusso, N. F. (1997). Development and initial application of an in vitro model of apoptosis in rodent cholangiocytes. Am. J. Physiol. 272, G106–G115.[ISI][Medline]

Rajeevan, M. S., Vernon, S. D., Taysavang, N., and Unger, E. R. (2001). Validation of array-based gene expression profiles by real-time (kinetic) RT-PCR. J. Mol. Diagnostics 3, 26–31[ISI]

Ranganna, K., Yatsu, F. M., Hayes, B. E., Milton, S. G., and Jayakumar, A. (2000). Butyrate inhibits proliferation-induced proliferating cell nuclear antigen expression (PCNA) in rat vascular smooth muscle cells. Mol. Cell. Biochem. 205, 149–161.[CrossRef][ISI][Medline]

Ray, S. D., and Jena, N. (2000). A hepatotoxic dose of acetaminophen modulates expression of BCL-2, BCL-X(L), and BCL-X(S) during apoptotic and necrotic death of mouse liver cells in vivo. Arch. Toxicol. 73, 594–606.[CrossRef][ISI][Medline]

Reilly, T., Bourdi, M., Brady, J. N., Pise-Masison, C. A., Radonovich, M. F., George, J. W., and Pohl, L. R. (2001). Expression profiling of acetaminophen liver toxicity in mice using microarray technology. Biochem. Biophys. Res. Commun. 282, 321–328.[CrossRef][ISI][Medline]

Ritter, J. K., and Franklin, M. R. (1987). Clotrimazole induction of cytochrome P-450: Dose-differentiated isozyme induction. Mol. Pharmacol. 31, 135–139.[Abstract]

Robertson, G., Leclercq, I., and Farrell, G. C. (2001). Nonalcoholic steatosis and steatohepatitis. II. Cytochrome P-450 enzymes and oxidative stress. Am. J. Physiol. Gastrointest. Liver Physiol. 281, G1135–G1139.[Abstract/Free Full Text]

Rodrigues, A. D., and Machinist, J. M. (1996). Hepatic peroxisomal and drug metabolizing activity in CD-1 mice after oral treatment with a novel 5-lipoxygenase inhibitor. Toxicol. Appl. Pharmacol. 137, 193–201.[CrossRef][ISI][Medline]

Ronis, M. J., Ingelman-Sundberg, M., and Badger, T. M. (1994). Induction, suppression and inhibition of multiple hepatic cytochrome P450 isozymes in the male rat and bobwhite quail (Colinus virginianus) by ergosterol biosynthesis inhibiting fungicides (EBIFs). Biochem. Pharmacol. 48, 1953–1965.[CrossRef][ISI][Medline]

Saarikoski, S. T., Ikonen, T. S., Oinonen, T., Lindros, K. O., Ulmanen, I., and Husgafvel-Pursiainen, K. (1998). Induction of UDP-glycosyltransferase family 1 genes in rat liver: Different patterns of mRNA expression with two inducers, 3-methylcholanthrene and beta-naphthoflavone. Biochem. Pharmacol. 56, 569–575.[CrossRef][ISI][Medline]

Sadaphal, P., Astemborski, J., Graham, N. M., Sheely, L., Bonds, M., Madison, A., Vlahov, D., Thomas, D. L., and Sterling, T. R. (2001). Isoniazid preventive therapy, hepatitis C virus infection, and hepatotoxicity among injection drug users infected with Mycobacterium tuberculosis. Clin. Infect. Dis. 33, 1687–1691.[CrossRef][ISI][Medline]

Salas, V. M., and Corcoran, G. B. (1997). Calcium-dependent DNA damage and adenosine 3',5'-cyclic monophosphate-independent glycogen phosphorylase activation in an in vitro model of acetaminophen-induced liver injury. Hepatology 25(6), 1432–1438.[ISI][Medline]

Schuchhardt, J., Beule, D., Malik, A., Wolski, E., Eickhoff, H., Lehrach, H., and Herzel, H. (2000). Normalization strategies for cDNA microarrays. Nucleic Acids Res. 28, E47.[Medline]

Seglen, P. O. (1976). Preparation of isolated rat liver cells. Methods Cell. Biol. 13, 29–83.[Medline]

Simpson, A. E. (1997). The cytochrome P450 4 (CYP4) family. Gen. Pharmacol. 28, 351–359.[CrossRef][Medline]

Storck, T., von Brevern, M. C., Behrens, C. K., Scheel, J., and Bach, A. (2002). Transcriptomics in predictive toxicology. Curr. Opin. Drug Discov. Devel. 5, 90–97.[ISI][Medline]

Sundseth, S. S., and Waxman, D. J. (1992). Sex-dependent expression and clofibrate inducibility of cytochrome P450 4A fatty acid omega-hydroxylases. Male specificity of liver and kidney CYP4A2 mRNA and tissue-specific regulation by growth hormone and testosterone. J. Biol. Chem. 267, 3915–3921.[Abstract/Free Full Text]

Surry, D. D., Meneses-Lorente, G., Heavens, R., Jack, A., and Evans, D. C. (2000). Rapid determination of rat hepatocyte mRNA induction potential using oligonucleotide probes for CYP1A1, 1A2, 3A and 4A1. Xenobiotica 30, 441–456.[CrossRef][ISI][Medline]

Tang, W., Borel, A. G., Fujimiya, T., and Abbott, F. S. (1995). Fluorinated analogues as mechanistic probes in valproic acid hepatotoxicity: Hepatic microvesicular steatosis and glutathione status. Chem. Res. Toxicol. 8, 671–82.[ISI][Medline]

Tarloff, J. B., Khairallah, E. A., Cohen, S. D., and Goldstein, R. S. (1996). Sex- and age-dependent acetaminophen hepato- and nephrotoxicity in Sprague-Dawley rats: Role of tissue accumulation, nonprotein sulfhydryl depletion, and covalent binding. Fundam. Appl. Toxicol. 30, 13–22.[CrossRef][ISI][Medline]

Tygstrup, N., Bangert, K., Ott, P., and Bisgaard, H. C. (2002). Messenger RNA profiles in liver injury and stress: A comparison of lethal and nonlethal rat models. Biochem. Biophys. Res. Commun. 290, 518–525.[CrossRef][ISI][Medline]

Van Custem, B. (1994). Classification and dissimilarity analysis. In Lecture Notes in Statistics (P. Bickel, P. Diggle, S. Fienberg, K. Krickeberg, I. Olkin, N. Wermuth, and S. Zeger, Eds.), Vol. 93, pp. 1–238. Springer-Verlag, Heidelberg, Germany.

Waring, J. F., Ciurlionis, R., Jolly, R. A., Heindel, M., and Ulrich, R. G. (2001a). Microarray analysis of hepatotoxins in vitro reveals a correlation between gene expression profiles and mechanisms of toxicity. Toxicol. Lett. 120, 359–368.[CrossRef][ISI][Medline]

Waring, J. F., Jolly, R. A., Ciurlionis, R., Lum, P. Y., Praestgaard, J. T., Morfitt, D. C., Buratto, B., Roberts, C., Schadt, E., and Ulrich, R. G. (2001b). Clustering of hepatotoxins based on mechanism of toxicity using gene expression profiles. Toxicol. Appl. Pharmacol. 175, 28–42.[CrossRef][ISI][Medline]

Yuen, T., Wurmbach, E., Pfeffer, R., Ebersole, B. J., and Sealfon, S. C. (2002). Accurary and calibration of commercial oligonucleotide and custom cDNA microarrays. Nucleid Acid Res. 30(10), e48.

Zhang, J., W. Huang, W., Chua, S. S., Wei, P., and Moore, D. D. (2002). Modulation of acetaminophen-induced hepatotoxicity by the xenobiotic receptor CAR. Science 298, 422–424.[Abstract/Free Full Text]