Gene Expression in Two Hepatic Cell Lines, Cultured Primary Hepatocytes, and Liver Slices Compared to the in Vivo Liver Gene Expression in Rats: Possible Implications for Toxicogenomics Use of in Vitro Systems

Franziska Boess*,1, Markus Kamber{dagger}, Simona Romer*, Rodolfo Gasser*, Dieter Muller{ddagger}, Silvio Albertini* and Laura Suter*

* F. Hoffmann–La Roche Ltd., Pharmaceuticals Division, Nonclinical Development, CH-4070 Basel, Switzerland; {dagger} Basilea Pharmaceutica Ltd., CH-4002 Basel, Switzerland; and {ddagger} Institute for Pharmacology and Toxicology, University of Jena, Nonnenplan 4, D-07743 Jena, Germany

Received November 27, 2002; accepted February 13, 2003

ABSTRACT

Microarray technology allows the simultaneous analysis of mRNA expression levels of thousands of genes. In the field of toxicogenomics, this technology could help to identify potentially unsafe compounds based on the changes in mRNA expression patterns they induce. Rodent in vivo and in vitro systems are currently the experimental models of choice for predictive toxicology, especially in early phases of development. This study characterizes several hepatic in vitro systems based on mRNA expression profiles, comparing them to gene expression in liver tissue. The in vitro systems investigated comprise two rat liver cell lines (BRL3A and NRL clone 9), primary hepatocytes in conventional monolayer or in sandwich culture, and liver slices. The results demonstrate that liver slices exhibit the strongest similarity to liver tissue regarding mRNA expression, whereas the two cell lines are quite different from the whole liver. We were able to identify genes with strong changes in expression levels in all or at least one of the in vitro systems relative to whole liver. In particular, for some cytochrome P450s the differences observed on the mRNA expression level were paralleled by protein expression and enzymatic activity. In addition, the effect of time in culture was assessed. We were able to show a profound effect of the duration of culture. Expression patterns change most rapidly soon after cell isolation and culture initiation and stabilize with time in culture. The findings are discussed with respect to the usefulness of the various hepatic in vitro systems for microarray-based toxicological testing of compounds.

Key Words: microarray technology; hepatic in vitro system; rat liver cell line; sandwich culture; monolayer; liver slice.

In the past several years, novel systems (specifically microarrays) have quickly emerged, allowing genome-wide analysis of gene expression at the RNA level. These new technologies are not only heavily influencing drug discovery but have also raised high expectations regarding the prediction and mechanistic investigation of adverse events in preclinical drug safety (Lovett, 2000Go). On the one hand, toxicogenomics is hoped to help in drug candidate selection in very early phases of the development process by identifying compounds that may cause toxicity or adverse events in vivo. On the other hand, it should help elucidate the underlying mechanisms of toxicity in animals and in humans. This should facilitate the extrapolation of findings across species and the prediction of factors that might put some individuals at a higher risk. However, investigations with very limited amounts of test compound, as well as investigations with human tissue samples require in vitro experimentation methods using isolated tissues, cells, or cell lines. Therefore, the application of toxicogenomics is not only limited to experiments with whole animals but should also involve in vitro systems (Burczynski et al., 2000Go; Harries et al., 2001Go; Waring et al., 2001Go).

It was the aim of the current study to characterize various in vitro systems regarding their gene expression profiles in order to gain more insight into the potential usefulness and limitations of these systems. Special attention was given to the change of gene expression with time. For this purpose, three different primary cell culture systems (liver slices, monolayer, and sandwich cultures) were evaluated at different times after isolation (Fig. 1Go). Precision-cut rat liver slices were used as the system that is thought to better represent the in vivo situation, as they contain all the cell types present in whole liver and maintain their three-dimensional structure. In addition they are well accepted as an in vitro system to study the hepatotoxicity and biotransformation of compounds (Gandolfi et al., 1996Go; Goethals et al., 1992Go) and the in vitro induction of CYP-mRNAs (Glöckner et al., 1999Go; Müller et al., 1998Go, 2000Go). Even after cryopreservation and thawing, liver slices can be used for biotransformation and induction studies (Glöckner et al., 1998Go, 1999Go, 2001Go). Liver slices have been morphologically and immunohistochemically characterized for up to 48 h of incubation (Lupp et al., 2001Go). Rat primary hepatocytes were prepared by collagenase perfusion and cultured either as monolayers on collagen coated plates or between two layers of collagen (sandwich cultures). Cells for monolayer cultures were allowed to attach for 3 h and were then either preincubated overnight (standard protocol, MONO) or used immediately (monolayer after seeding, MONOAS). Sandwich cultures were allowed 3 h for attachment followed by collagen overlay and either an overnight (standard protocol, SAND) or a three day preculture period (sandwich after 4 days; SANDAD) before the start of the experiment. Hepatocytes represent the most abundant cell type and fulfill most of the main functions of the liver. Under sandwich configuration, hepatocytes have been shown to retain their polarity and cuboidal shape and to support an architecture similar to that found in the liver (George et al., 1996Go; LeCluyse et al., 2000Go). Also, hepatocytes in sandwich configuration can be cultured for a longer time period compared to hepatocytes in monolayer cultures. In addition, it has been reported that the sandwich configuration leads to better retention of cytochrome P450 inducibility in rat, but not human, cultured hepatocytes (LeCluyse, 2001Go). In addition to the mentioned primary culture systems, two stable rat liver cell lines—BRL 3A and NRL clone 9—were also investigated. The BRL 3A cell line is an epithelial cell line from buffalo rat liver, which is able to divide in the absence of serum. This cell line was found to produce a family of polypeptides termed MSA (multiplication stimulating activity), which can partially satisfy the serum requirement of other cells, for example, chick embryo fibroblasts (Nissley et al., 1977Go). The clone 9 cell line is an epithelial cell line isolated in 1968 from normal rat liver (NRL) taken from a young male rat and has mainly been used for studies of in vitro carcinogenesis (Weinstein et al., 1975Go).



View larger version (29K):
[in this window]
[in a new window]
 
FIG. 1. Overview of the primary sources of mRNA used in this study. Experimental procedures and the respective time intervals since necropsy are depicted schematically.

 
MATERIALS AND METHODS

Chemicals and reagents.
Rat gene expression microarrays (RG-U34A) were obtained from Affymetrix Inc (Santa Clara, CA). These arrays contain 8799 rat specific probe sets, representing approximately 5000 different genes. Unless stated otherwise, all chemicals were obtained from Sigma (Buchs, Switzerland). Fetal calf serum (FCS) and penicillin/streptomycin were from BioConcept (Basel, Switzerland); collagenase (type A, from Clostridium histolyticum) from Roche Diagnostics (Mannheim, Germany).

Animals and treatment.
Permission for animal studies was obtained from the local regulatory agencies, and all study protocols were in compliance with the federal guidelines. Male HanBrl:WIST rats (SPF; 10–12 weeks of age) were obtained from Biological Research Laboratories (BRL) Ltd., Füllinsdorf, Switzerland. The animals were housed individually in Macrolone cages with wood shavings as bedding at 20°C and 50% relative humidity in a 12-h light/dark rhythm with free access to water and Kliba 3433 rodent pellets (Provimi Kliba AG, Kaiseraugst, Switzerland). For liver slice experiments 40-day-old male Han:Wist rats from the breeding colony of the Institute of Pharmacology and Toxicology at the University of Jena were used. They were house under similar conditions as described above with the exception of food, which consisted of Altromin 1316 rodent pellets (Altromin GmbH, Lage, Germany).

Animals for the in vivo studies were treated with vehicle solutions. Used vehicle solutions and the route of administration were chosen according to the investigated test compound (test compound treated animals are not included in this study) and do not have a major impact in liver gene expression. The following vehicles were used: saline (n = 6), corn oil (n = 6), 7.5% gelatine (n = 3), or 1% DMSO (n = 6) were administered ip. Saline (n = 8) and a 10% intralipid/1:10 EtOH solution (n = 3) were administered iv. Water (n = 3) and corn oil (n = 3) were used as vehicles for po treatment via gavage. The number (n) of animals treated with the respective vehicle and sacrificed 6 h later are indicated in parenthesis. Naive animals were used for the preparation of hepatocytes or slices.

In the in vivo studies, liver samples from the left medial lobe were immediately placed in RNALater (Ambion, TX) for RNA extraction and gene expression analysis. The samples were stored at -20°C until further processing. In addition, a portion of the liver of an untreated rat was perfused with saline solution to eliminate excess blood contamination and immediately snap-frozen for protein extraction and subsequent Western blot analysis.

Preparation, culture, and treatment of rat liver slices.
In two independent experiments, precision-cut liver slices were prepared from randomly selected liver cores (diameter 8 mm) with a Krumdieck slicer filled with carbogen (95% O2/5% CO2)-saturated Krebs-Henseleit HEPES buffer, pH 7.4 (slice thickness 0.20–0.25 mm). Six slices were incubated in 25 ml Erlenmeyer flasks containing 10 ml William’s medium E with 1.3 IU insulin, 2.9 mg glutamine, 0.5 mg gentamicin, and 0.1 mg ampicillin at 37°C and gassed with carbogen. After preincubation for 2 h the medium was changed. After incubation for 6 or 24 h the mean gene expression value was calculated (Fig. 1Go). Since the influence of test compounds on liver slices was studied in parallel experiments (not shown), some slices were also exposed to 1% DMSO as a vehicle (n = 3 of out of 8 slices per time point).

Preparation, culture, and treatment of primary rat hepatocytes.
Hepatocytes were isolated from 10- to 14-week-old male HanBrl:WIST rats by a two-step collagenase liver perfusion method (Berry and Friend, 1969Go) as previously described (Göldlin and Boelsterli, 1991Go). Briefly, the rats were anaesthetized with sodium pentobarbital (120 mg/kg, ip). The perfusate tubing was inserted via the portal vein, then the v. cava caudalis was cut, and the perfusion was started. The liver was first perfused for 5 min with a preperfusing solution consisting of calcium-free, EGTA (0.5 mM)-supplemented, HEPES (20 mM)-buffered Hank’s balanced salt solution (5.36 mM KCl, 0.44 mM KH2PO4, 137 mM NaCl, 4.2 mM NaHCO3, 0.34 mM Na2HPO4, 5.55 mM D-glucose). This was followed by a 12-min perfusion with NaHCO3 (25 mM)-supplemented Hank’s solution containing CaCl2 (5 mM) and collagenase (0.2 U/ml). Flow rate was maintained at 28 ml/min and all solutions were kept at 37°C. After in situ perfusion the liver was excised and the liver capsule was mechanically disrupted. The cells were suspended in William’s Medium E without phenol red (WME) and filtered through a set of tissue sieves (30-, 50-, and 80-mesh). Dead cells were removed by a sedimentation step (1 x g, for 15 min at 4°C) followed by a Percoll (Sigma) centrifugation step (Percoll density: 1.06 g/ml, 50 x g, 10 min) and an additional centrifugation in WME (50 x g, 3 min). Typically, 100–300 x 106 cells were obtained from one rat liver. Hepatocyte viability was assessed by trypan blue exclusion and typically ranged between 85 and 95%.

Monolayer cultures.
The cells were seeded into collagen-coated six-well Falcon Primaria® plates (Fisher Scientific AG, Wohlen, Switzerland), at a density of 9 x 105 cells/well in 2 ml WME supplemented with 10% fetal calf serum, penicillin (100 U/ml), streptomycin (0.1 mg/ml), insulin (100 nM), and dexamethasone (100 nM). After an attachment period of 3 h, the medium was replaced by 1.5 ml/well serum-free WME, supplemented with antibiotics and hormones, and further kept at 37°C in an atmosphere of 5% CO2/95% air. Usually, the cells were precultured overnight prior to experiments (MONO; Fig. 1Go). For some experiments the cells were treated directly after the 3 h attachment period (treatment after seeding = MONOAS; Fig. 1Go). At the beginning of the experiments, cells were incubated either with medium alone (n = 14 per time point for monolayers; n = 3 per time point for MONOAS) or medium containing 0.1% DMSO (n = 19 per time point for monolayers; n = 1 per time point for MONOAS) or 0.1% ethanol (n = 3 per time point for monolayers) as vehicle. Cells were harvested at several time points after the experiment had started either for gene expression analysis or for protein extraction. Generally, the cells from three wells were pooled to obtain sufficient amounts of RNA and protein.

Sandwich cultures.
Isolated cells were seeded into collagen-coated (rat tail collagen, 15 µg/cm2) NunclonTM six-well plates (Fisher Scientific, Wohlen, Switzerland), at a density of 1.125 x 106 cells/well in 2.5 ml WME supplemented with 10% fetal calf serum, penicillin (100 U/ml), streptomycin (0.1 mg/ml), insulin (100 nM), and dexamethasone (100 nM). After an attachment period of 3 h, the medium was removed and the cells were covered with a layer of collagen (rat tail collagen, 15 µg/cm2), which was allowed to set for about 1 h. Afterwards 1.5 ml/well serum-free WME, supplemented with 2% fetal calf serum, antibiotics, and hormones was added, and the cells were incubated at 37°C in an atmosphere of 5% CO2/95% air. Usually, the cells were precultured overnight prior to experiments (= SAND; Fig. 1Go). For some experiments the sandwiched hepatocytes were cultured for three days before treatment (treatment after three days = SANDAD; Fig. 1Go). At the beginning of the experiments cells were incubated with medium alone (no other vehicles used). Cells were harvested at several time points after the experiment had started for gene expression analysis or for protein extraction. Generally, the cells from three wells were pooled to obtain sufficient amounts of RNA and protein.

Culture and treatment of cell lines.
Two rat liver cell lines BRL 3A derived from Buffalo rats (Nissley et al., 1977Go) and NRL clone 9, derived from normal Sprague Dawley rats (Weinstein et al., 1975Go) were used. Cell lines were obtained from ATCC. Approximately 106 cells/well were plated onto six well plates and incubated under standard conditions (5% CO2, 37°C) in ISCOVE medium containing 5% FCS for 16 h. The medium was then replaced by fresh ISCOVE medium containing 5% FCS and, if appropriate, vehicle for test compound (either 0.02% DMSO [n = 2 for BRL 3A; n = 1 for NRL clone 9] or 0.1% ethanol [n = 2 for BRL 3A; n = 1 for NRL clone 9]). Cells were incubated for a further 6 or 24 h before harvesting for RNA or protein extraction. Generally, the cells from three wells were pooled to get sufficient amounts of RNA and protein.

Gene expression analysis.
Total RNA was extracted from liver pieces, liver slices, and harvested cells following standard laboratory protocols. Briefly, total RNA was extracted from liver tissue, liver slices or cultured cells (approx. 2.5 x 106 cells) using RNAzol (Tel-Tex Inc., TX) and a commercially available kit (Bio 101, CA). Total RNA was purified using RNeasy columns (Qiagen, Basel, Switzerland) before quantification and assessment of ribosomal RNA integrity on agarose gels. Double-stranded cDNA was synthesized from 20 µg of total RNA using an oligo dT-T7 promoter primer (Roche Molecular Biochemicals, Mannheim, Germany). The obtained cDNA was used as a template for in vitro transcription using the Megascript kit purchased from Ambion (TX) and biotinylated nucleotides (Bio-11-CTP and Bio-16-UTP) provided by Roche Molecular Biochemicals (Mannheim, Germany). Fragmented in vitro transcripts (cRNAs) were hybridized overnight onto commercially available rat microarrays containing 8799 rat specific probe sets (RG-U34A, Affymetrix, CA). The hybridized samples were stained with streptavidin-R-phycoerythrin (SAPE, Molecular Probes, CA) and the signal amplified using a biotinylated goat anti-streptavidin antibody (Vector Laboratories, CA) followed by a final staining with SAPE. Washing, staining, and amplification were carried out in a fluidics station provided by Affymetrix. Microarrays were scanned in an Affymetrix GeneArray scanner (gain setting: 18,000). The obtained image files were analyzed with the software Microarray Suite 4.0 (Affymetrix) while comparisons between treated and control groups and statistical analysis were performed with in-house developed software. The latter software calculates "change factors" (listed in Tables 2Go to 5Go), which report the change from condition 1 (in vivo) to condition 2 (in vitro). In case of an increase, it is the ratio (cond2/cond1) minus 1. In case of a decrease, it is -(cond1/cond2) plus 1 (the addition/subtraction of 1 avoids a gap in the scale between +1 and -1).


View this table:
[in this window]
[in a new window]
 
TABLE 2 Genes Up- or Downregulated in All Primary in Vitro Systems at 6 h
 

View this table:
[in this window]
[in a new window]
 
TABLE 5 Expression Changes in Cytochrome P450 Gene Probe Sets Comparing the 6 and 24 h Time Points
 
Western blot analysis.
Liver tissue was placed in PBS containing protease inhibitors and homogenized in a Potter-homogenizer. Approximately 2.5 x 106 cells from each in vitro experiment were harvested directly into Cytobuster reagent for protein extraction according to the manufacturer’s instructions. Twenty micrograms of total protein were denatured in load buffer and separated using SDS–PAGE on a Novex gradient gel (4–20%). After migration, proteins were transferred onto a nitrocellulose membrane. After transfer, nitrocellulose membranes were blocked overnight at 4°C in PBS containing 0.1% Tween-20 and 0.1 g/ml skimmed milk. After washing with phosphate buffered saline (PBS)/Tween-20, the membranes were incubated for an hour with the appropriate first antibody (either anti-cytochrome P450 2B1, anti-cytochrome P450 4A, or anti-cytochrome P450 2C11, generated in sheep) diluted 1:1000; followed by an incubation with the second antibody (donkey anti-sheep/Goat Immunoglobulin Horseradish Peroxidase Conjugated, Chemicon, Hofheim, Germany), diluted 1:1000 using the same incubation conditions. Chemiluminiscence was analyzed in a semiquantitative manner by densitometry in a Multimage Light Cabinet (Alpha Inotech Corporation, San Leandro, CA) using Lumilight solution.

P450 activity determination.
The formation of 6ß-hydroxytestosterone (indicative of P450 3A activity) and 2{alpha}-hydroxytestosterone (indicative of P450 2C11 activity) from radiolabeled testosterone was used to determine relative enzymatic activity in intact cultured hepatocytes (Arlotto et al., 1991Go). Cells were incubated for 20 min with medium containing 12.44 µM 14C-Testosterone (NEN Life Sciences, Boston, MA; 53.6 mCi/mmol). Supernatant was sampled at the beginning and end of incubation period and analyzed by radiometric reversed phase HPLC (Purdon and Lehmann-McKeeman, 1997Go). Peaks were identified by comparison to authentic 6ß- and 2{alpha}-hydroxytestosterone standards (Sigma, Buchs, Switzerland).

The relative enzymatic activity for P450 2B was determined using the benzyloxyresorufin O-dealkylation (BROD) assay (Burke et al., 1985Go). Formation of resorufin from benzyloxyresorufin was determined directly in the cultured cells using a fluorescence plate reader (Victor2 multilabel counter, Wallac) with excitation and emission wavelengths set to 544 nm and 590 nm, respectively. The cells were preincubated with 5 µM benzyloxyresorufin/10 µM dicoumarol in Hanks balanced salt solution for 20 min and then the formation of resorufin was monitored by fluorescence detection every 10 s over a time period of 3 min.

All P450 activity determinations were done in triplicates.

Data analysis.
For the evaluation of gene expression data using microarrays, data obtained from the Affymetrix software (MAS 4.0) were further analyzed using in-house developed analysis tools (RACE-A, F. Hoffmann-La Roche AG, Basel, Switzerland). The expression level for each gene is expressed as the mean fluorescence intensity (average difference) and SD after Nalimov outlier removal (99% confidence) for n independent replicates as indicated in Table 1Go. Probe sets with an expression level (average difference) lower than 200 in all the systems (in vitro and in vivo) were considered below the detection limit and excluded from the analysis. Clustering was performed using a hierarchical clustering algorithm. Information about the function of the listed genes was obtained from public sources (e.g., Medline, Swissprot) but it might be incomplete because not all the functions of certain proteins are known or were included in the tables.


View this table:
[in this window]
[in a new window]
 
TABLE 1 Linear Correlation Coefficients for Gene Expression Compared to in Vivo and Numbers of Samples Included in the Analysis
 
Protein levels evaluated by Western blots were measured by densitometry and expressed in arbitrary units.

RESULTS

In Vitro versus in Vivo Comparison
In a first approach to make an overall comparison of the different in vitro systems, the gene expression of the three primary cell culture systems and the two studied cell lines was compared to the gene expression in whole livers. Linear regression was performed comparing the gene expression values for the 3984 probe sets above the defined detection limit (see data analysis section in Materials and Methods) on the Affymetrix GeneChips for whole liver in vivo to those from the different in vitro systems. A complete list of these genes with their respective gene expression values can be downloaded from the Internet (www.roche.com/science-download.htm). The calculated correlation coefficients are summarized in Table 1Go. Generally, at the 6 h time point all primary cell systems are similar to the liver tissue, showing a high correlation with the gene expression in the liver (r2 > 0.73). However, the data show that the liver slices are more similar to the whole liver in vivo (r2 = 0.95). It is also interesting to note that lower correlation coefficients, indicating higher divergences, are obtained at time points further away from the cell isolation. This holds especially true for the collagen sandwich cultures after 3 days in culture (SANDAD). The gene expression in the two cell lines BRL 3A and NRL clone 9, on the other hand, seem to be very different from that of the whole liver as reflected by the low correlation coefficients (approximately 0.10). This is reflected in the outcome of hierarchical clustering (Fig. 2Go). Six h after the beginning of the experiment, slices cluster nearest to the whole liver, while primary hepatocytes in monolayer or sandwich cultures are further away. Both hepatic cell lines appear to be similar to each other but quite different from the whole liver and the primary cell culture systems. Both the linear regressions and the hierarchical clustering strongly suggest that for the primary hepatocyte systems time in culture has a very strong influence on gene expression.



View larger version (13K):
[in this window]
[in a new window]
 
FIG. 2. Result of hierarchical cluster analysis using all genes expressed in at least one of the systems. All probe sets (n = 3984) with an expression value >=200 in at least one of the systems were included in the analysis.

 
Gene Expression Changes in the Various Systems over Time
The changes in gene expression over time were analyzed in the various systems. Gene expression was quantified 6 and 24 h after vehicle treatment and the differences in gene expression due to this time lapse were assessed by linear regressions. As the correlation coefficients would be strongly influenced by a few very highly expressed genes, only genes with an expression value between 200 (lower limit of detection) and 15,000 (upper limit to exclude to highly expressed genes) were included in this analysis. For the whole livers (ex vivo) the differences between the two sampling times are minor (Fig. 3A Gor,2 = 0.96), indicating little variation between the two time points chosen. As an exception, two genes, namely metallothionein 1 (MT-1) and metallothionein 2 (MT-2), showed very different expression levels at the measured times.



View larger version (51K):
[in this window]
[in a new window]
 
FIG. 3. Variation of gene expression between the 6 and 24 h time points. The expression for single genes at 6 h vs. 24 h is given for all the primary systems. All gene probes with an expression value >= 200 in at least one of the systems were included in the analysis. For each system the gene probes with expression values >=15,000 at either time point are not depicted and were excluded for the calculation of linear correlation (r2). (A) The encircled data points are gene probes for MT-1 and MT-2.

 
Primary cell cultures showed a greater divergence with time. The correlation between the 6 and 24 h time points after vehicle treatment was lower for hepatocytes in monolayer (MONO, r2 = 0.83) and sandwich (SAND, r2 = 0.83) configuration (Figs. 3BGo and 3CGo, respectively) than for liver tissue. The difference between the two time points is most pronounced for MONOAS and slices, as seen by the stronger scatter and the consequently lower correlation coefficients between the two time points (Fig. 3E Gor,2 = 0.79 and Fig. 3D Gor,2 = 0.71 for MONOAS and slices, respectively). These results indicate that changes in gene expression over time are more pronounced in the two systems with the shortest adaptation period to the culture conditions (see Fig. 1Go). Conventional sandwich (SAND) and monolayer (MONO) cultures that were precultured overnight before treatment showed less dramatic changes with time. Finally, the least marked changes in gene expression over time were observed in the SANDAD system (r2 = 0.90, Fig. 3FGo), which is the system with the longest preculture time before the beginning of the experiment.

Most Pronounced Alterations in the Primary in Vitro Systems Compared to Whole Liver
Comparison of expression profiles between the liver and primary culture systems is interesting and important. With this purpose in mind, the gene expression in collagen sandwich, monolayer, and slice cultures were compared with the gene expression levels in the liver. This comparison was performed using stringent analysis criteria: only genes showing a statistically significant (t-test p <= 0.05) sixfold up- or downregulation were considered modulated. Some of the identified genes were altered in all in vitro systems compared to whole liver (Table 2Go), most of them showing an upregulation in culture. Among these upregulated genes some are involved in apoptosis (p8, calpactin) and regulation of growth and differentiation (pro-neuregulin, nerve growth factor inducible anti-proliferative protein, insulin like growth factor). In addition, two genes, sm-20 and heme oxygenase, were significantly increased in all three in vitro systems at 6 h but returned to the control levels at 24 h in all three systems or at least (for heme oxygenase) in the two hepatocyte culture systems (data not shown).

As shown in the cluster analysis (see previous section), gene expression in liver slices is more similar to that in intact liver than is the gene expression from the other cell culture systems. Hence, many genes were differentially expressed in hepatocyte cultures (monolayer or sandwich culture) but remained unchanged in the slices when compared to their expression in whole livers (Table 3Go). Among these for example, alpha-2-macroglobulin, P450 1A1 (see also Figure 4Go), p21/waf, guanylate cyclase, and hepatocyte nuclear factor 4-alpha were overexpressed in the sandwich and monolayer cultures. Besides these genes induced by certain culture conditions, some genes, including MHC II, hemoglobin beta-globin, and structural proteins (e.g., collagen) were downregulated in the sandwich and monolayer cultures. Although the gene expression in liver slices appears to be closest to the whole liver, this similarity decreases with time in culture. This is suggested by a subset of genes whose expression in liver slices at 6 h does not differ from the expression in the liver applying the mentioned stringent criteria for induction/repression but whose expression at 24 h shows marked induction or repression (Table 3Go). For these genes, the expression level in liver slices at 24 h is more similar to that of hepatocyte cultures than to the whole liver. Some of these genes such as annexin II, acyl-coA desaturase, dig-1, and carbonyl reductase are induced while sulfotransferases and several genes involved in fatty acid metabolism, gluconeogenesis, growth regulation, and cellular transport are downregulated.


View this table:
[in this window]
[in a new window]
 
TABLE 3 Genes Up- or Downregulated in Hepatocyte Cultures but Not in Slices at 6 h
 


View larger version (102K):
[in this window]
[in a new window]
 
FIG. 4. Expression of cytochrome P450 genes at the 6 h time point. Expression values at the 6 h time point of the various P450 gene probes on the Affymetrix RG-U34A chip are represented as shadowed bars. Each row corresponds to one P450 gene probe on the chip. The darker the color, the higher the expression of the corresponding gene probe at the 6 h time point in the respective experimental system.

 
Even though hepatic slices appear to be the most similar system to the whole liver, there are some genes whose expression, when compared to whole liver, were altered only in slices but not in hepatocyte cultures (Table 4Go). Most of these genes were upregulated and among them were many cytokines, the inducible nitric oxide synthase (iNOS), as well as some members of the GADD family and several transcriptional regulators. Some genes were only transiently up- or downregulated, showing a significant up- or downregulation at the 6 h time point only but not after 24 h.


View this table:
[in this window]
[in a new window]
 
TABLE 4 Genes Up- or Downregulated in Slices but Not in Hepatocyte Cultures at 6 h
 
Most Pronounced Alterations in the Rat Liver Cell Lines Compared to Whole Liver
As Table 1Go and Figure 2Go indicate, there are many differences in gene expression not only between whole liver and the two investigated cell lines, but also between the primary in vitro systems and the cell lines. Compared to whole liver 781 and 714 genes are up- or downregulated by more than a factor of 10 in BRL 3A and NRL clone 9 cells, respectively. The majority of these genes show concordant regulation in both cell lines. The messenger RNA of many structural proteins (e.g., proliferin, elastin, crystallin, and collagen) or of proteins involved in growth regulation and differentiation (e.g., galectine, insulin-like growth factor [igf] binding protein and igf receptor) appear overexpressed in the cell lines. Among the repressed genes, there are many metabolic phase I and phase II enzymes (e.g., cytochrome P450 isoenzymes [see also Fig. 4Go], sulfotransferases, and glucuronosyltransferases) and many plasma proteins such as serum albumin, fibrinogen, and complement factors.

Expression Levels of Cytochrome P450s: mRNA Expression, Protein Expression, and Activity
In order to investigate whether there are marked differences between the various in vitro systems with respect to maintenance of P450 expression, particular attention was given to the mRNA levels of various cytochrome P450 isoenzymes (Fig. 4Go). When assessing the P450 gene expression levels across the various systems, slices and MONOAS, which have the shortest adaptation time before harvesting, appear to have very similar expression profiles to the whole liver in vivo. A clear reduction in expression can be seen for monolayers and sandwich hepatocytes that have been kept in culture overnight. Consistent with this, the most pronounced differences were observed in sandwich cultures kept for several days (up to eight days) before harvesting. In the two cell lines BRL 3A and NRL clone 9 the overall expression of P450 mRNA was weak. CYP1A1 was only found to be expressed in hepatocyte cultures, with no expression observed in whole liver, liver slices, or the two cell lines tested.

While in vivo the expression of the P450 genes appeared to be quite stable, showing similar levels in animals sacrificed 6 or 24 h after dosing with a vehicle, there were many significant changes in the in vitro systems. For the majority of the studied P450s, the gene expression level in culture declined over time, although some exceptions were observed (Table 5Go).

Some cytochromes P450 were further investigated in cultured hepatocytes in order to verify whether the observed decreases in mRNA-levels are also reflected in protein amount and enzymatic activities. In an additional experiment, several P450 activites were measured and cells were harvested for protein extraction at several times in culture. Western blot analysis was performed for the P450 isoforms CYP2C11, CYP2B, and CYP4A1 and the metabolic activities of CYP2C11, CYP2B, and CYP3A were measured (Fig. 5Go). In general, the results obtained show a tendency towards lower protein expression and enzymatic activity with increased time after isolation, consistent with the gene expression findings. Some loss in protein level was already seen after liver perfusion and cell isolation. When compared to the liver decreases in protein levels of 37, 55, and 60% were seen for CYP2B, CYP2C11, and CYP4A1, respectively, after seeding (corresponding to ~ 5 h after necropsy). On average the decrease in the mRNA level was less pronounced. However, this depended on the different probes for the respective gene on the Affymetrix chip. For enzymatic activity, the direct comparison between in vivo and early in vitro time frames could not be determined, as the earliest measurement of activity in intact cells could only be done after cell isolation. After longer time in culture (>1 day) the mRNA levels of the determined CYPs became markedly reduced. This seemed to be even more pronounced in sandwich culture than in conventional monolayer cultures. After the first 48 h in culture the remaining activity, protein, and mRNA levels of all CYPs remained constant or even recovered again in sandwich cultured hepatocytes. For CYP 2C11 there seems to be a discrepancy between the time course of mRNA, protein expression, and enzyme activity, especially after 48 h in monolayer culture. At this time point almost no protein could be detected although enzyme activity was not reduced compared to that in freshly isolated cells.



View larger version (64K):
[in this window]
[in a new window]
 
FIG. 5. Time course of mRNA, protein expression and enzymatic activity for CYPs 2B, 2C11, 3A4, and 4A. The expression values for the probe sets representing the respective CYP were expressed as percentage of the expression in whole liver. Protein levels evaluated by Western blots were measured by densitometry and are depicted as percentage of the value obtained from whole liver. Enzymatic activities were determined as described under Materials and Methods and are represented as percentage of the activity in cells in suspension shortly after isolation from the liver by a two step perfusion procedure.

 
Western blot analysis of the two cell lines showed that BRL 3A and NRL clone 9 cells do not express the proteins for CYPs 2B, 4A, and 2C11 in detectable amounts (data not shown), which corresponds well with the respective low mRNA expression of the respective probe sets (data not shown, but see Fig. 4Go).

DISCUSSION

As gene expression is a dynamic process, it is not surprising that the expression levels of genes are not stable over time. Under standardized laboratory conditions marked alterations in the expression of certain genes can be found in vivo (Fig. 3AGo). For example we found a marked difference in the in vivo expression of metallothionein (MT-1 and MT-2), which had approximately three times lower expression at 24 h than at 6 h after vehicle treatment. Metallothioneins are cysteine-rich, ubiquitous metal binding proteins involved in the homeostasis of essential metals and metal detoxification and are thought to play an important role in tissue protection against various stresses. Possible reasons for the observed changes in MT expression, which is thought to be an overexpression at the 6 h time point rather than a depression at 24 h, might be treatment-related stress or reduced food intake of the animals during the day light phase. Both stress (Miles et al., 2000Go) and food-deprivation (Shinogi et al., 1999Go) have been reported to induce MT expression. Circadian expression of MT might serve as another explanation as circadian variation in MT levels has been observed in the kidney of rats, although this variation with time was not statistically significant in the liver (Cahill et al., 1983Go).

In the primary in vitro systems used in this study, the gene expression differences between the two time points investigated were much more pronounced than those found in vivo (Figs. 3BGo–3FGo). The main objective of this study was to learn more about the gene expression differences in the experimental liver systems used.

In order to gain statistical power we maximized the number of replicates for gene expression analysis (Table 1Go) by including control samples obtained from several independent in vivo and in vitro toxicogenomics studies. All animals and cell cultures were treated with a vehicle solution that was chosen in each case in accordance to the compound under investigation. However, compound-related changes in gene expression are neither presented nor discussed in this article. Using principal component analysis (data not shown) we found that the exposure to different vehicles had only a minor impact on gene expression. However, the combination of different vehicle treatments might have led to a higher variability of certain genes, possibly diminishing the discrimination power of the statistical test used.

Using stringent statistical criteria (sixfold up or downregulation and a significance level of p < 0.05) we found several genes induced or repressed in all in vitro conditions (Table 2Go) when compared to the expression in vivo. Most of the upregulated genes, such as Sm-20 and heme oxygenase 1, seem to be involved in cellular growth, death, or differentiation. Sm-20 is only transiently upregulated in all primary in vitro systems. It is a member of the immediate early gene family that has been reported to be involved in the development and differentiation of muscle cells (Moschella et al., 1999Go) and in apoptosis in neuronal cells (Lipscomb et al., 1999Go, 2001Go). Its role in the liver, however, has not yet been elucidated. Heme oxygenase 1 (HO-1), also termed heat shock protein 32, is induced in liver by various stresses, e.g., cytokines, hypoxia, and reactive oxygen species including nitric oxide (Suematsu and Ishimura, 2000Go). Increased HO protein expression and activity has been hypothesized to contribute to the initial decrease in P450 activity and content observed in cultured rat hepatocytes and liver slices in culture (Kutty et al., 1998Go).

When comparing the various in vitro liver systems to whole liver by linear regression (Table 1Go) or hierarchical clustering analysis (Fig. 2Go), gene expression in liver slices is most similar to the in vivo gene expression. The divergence between in vitro and in vivo gene expression seems to increase with the total time elapsed between cell isolation from the liver and cell harvest for gene expression analysis (Table 1Go). Finally, the two investigated cell lines were very different from whole liver but they also differed from all primary in vitro systems.

There are two reasons that might explain why gene expression in liver slices is most similar to that in whole liver:

First, slices still contain all the different cell types present in the liver in vivo, whereas the hepatocytes in our culture systems were Percoll© purified to remove nonparenchymal cells. MHC II for example is normally not expressed on hepatocytes (Eisenberger et al., 1994Go; Senaldi et al., 1991Go) but only on antigen presenting cells. Thus, the apparent lower expression of this gene in monolayers and sandwich cultures compared to in vivo or slices is believed to be due to the removal of nonparenchymal cells—e.g., Kupffer cells, blood macrophages, and lymphocytes. In addition, the three-dimensional architecture and the extracellular matrix are to a large extent maintained in slices but not in hepatocyte cultures. mRNA expression for collagen and other structural proteins is markedly decreased in hepatocytes in culture but maintained in liver slices. This might be indicative of a strongly altered cellular arrangement in the hepatocyte culture system compared to the three-dimensional arrangement in intact tissue.

Second, the time elapsed between excision of the liver and gene expression analysis from slices at the 6 h time point was approximately 10 h whereas it was roughly 30 h for monolayers and sandwich hepatocytes due to an overnight preculture period (Fig. 1Go). If the time of mRNA isolation relative to necropsy is matched, e.g., if the 24 h time point for slices is compared to the 6 h time point for monolayers and sandwich hepatocytes, the systems show similar correlation coefficients to liver tissue (Table 1Go). This effect of culture time on gene expression is in agreement with published gene expression data from primary hepatocytes (Baker et al., 2001Go).

The difference in total culture time between slices at the 6 h time point and primary hepatocytes 6 h after treatment is also responsible for some of the differences between these two systems with respect to the expression of single genes (Table 3Go). Many of the genes that show different expression levels at the 6 h time point in hepatocytes compared to whole liver and to slices at 6 h, show comparable differential expression in slices at the 24 h time point. Among these genes are many sulfotransferases. The marked downregulation of sulfotransferases is in good agreement with previously reported gene expression data from primary hepatocytes (Baker et al., 2001Go) and might be related to the absence of appropriate sulfate, hormone, and substrate (Seo et al., 1999Go).

A very interesting finding when comparing the gene expression in slices with that in whole liver was the upregulation of genes involved in inflammatory reaction (cytokines, iNOS). This upregulation of cytokines or iNOS was not observed in any of the hepatocyte cultures systems (Table 4Go). The inflammatory reaction observed in the liver slices on a gene expression level is possibly mediated by nonparenchymal cells and is especially pronounced at early time points after slicing. This finding is in agreement with those reported by other authors, who detected TNF{alpha}, IL-1ß, IL-10, and nitric oxide in the incubation media of slices (Neyrinck et al., 1999Go; Olinga et al., 2001Go).

Metabolism of endogenous and exogenous compounds is one of the key functions of the liver. Therefore special attention was paid to the expression of the various P450 isoenzymes in the different in vitro systems used in this study (Figs. 4Go and 5Go, Table 5Go). It is well known that the expression of P450 diminishes in vitro depending on culture time (Wright and Paine, 1992Go). Our results presented in Figures 4Go and 5Go confirm these findings. The overall decrease in cytochrome P450 mRNA expression over time was comparable in all primary in vitro systems; slices, conventionally cultured hepatocytes or hepatocytes in sandwich configuration (Table 5Go).

As shown in Figure 5Go, the decrease in cytochrome P450 mRNA expression is more or less paralleled by a corresponding decrease in protein expression and enzymatic activity, especially at later time points. For CYP2B and CYP4A however, protein expression seemed to decrease more rapidly than mRNA expression within the first hour after isolation. This might indicate that shortly after isolation increased protein degradation might at least partially be responsible for the loss of P450 enzymes. Whether such an enhancement of P450 protein degradation might be related, as already discussed, to the transient increase in heme oxygenase gene expression observed in this study remains speculative. Another possible mediator of P450 degradation could be nitric oxide (NO; Lopez-Garcia, 1998Go). Although no induction of NOS, neither the constitutive form cNOS nor the inducible form iNOS, was seen in cultured hepatocytes in our study, the involvement of increased NO production cannot be ruled out since the production of NO can become increased without a change in mRNA expression (Lopez-Garcia and Sanz-Gonzalez, 2000Go).

Regarding the investigated cell lines (BRL 3A and NRL clone 9), only very low mRNA levels for P450 enzymes were found (Fig. 4Go), which agrees with lack of immunoreactive protein observed on the Western blots.

In conclusion, pronounced gene expression changes took place in all the investigated primary in vitro systems (Fig. 3Go). These changes are thought to be part of an adaptation process to the new in vitro environment at earlier time points after isolation and a de-differentiation process later on. These dynamics in gene expression presents the toxicogenomics scientist with a "dilemma." On the one hand, the longer the time period between isolation/preparation and treatment, the less comparable the initial gene expressions to the in vivo situation. On the other hand, our data indicate that the ongoing adaptation changes are more pronounced early after isolation/preparation than after an initial period of preculture (Fig. 3Go). These ongoing adaptation processes might mask or counteract changes induced by compound treatment. Therefore, no ideal in vitro liver system seems to exist for toxicogenomics investigations.

From our experiences, in vitro toxicogenomics results—regardless of the system used—do not directly compare to the results obtained in vivo, at least not on a gene to gene comparison basis. Therefore, whatever system is used, knowledge about system-related differences, for example an ongoing inflammatory reaction in slices, and ongoing adaptation changes, is needed for better understanding and interpretation of genomics data.

ACKNOWLEDGMENTS

We thank E. Durr and N. Schaub for hepatocyte preparation and H. Guder for slice preparation and culture; A.-C. Boscato, M.-C. Boy, N. Flint, M. Haiker, and K. Rupp for mRNA preparation and chip work; M. Brecheisen and C. Sarron for in vivo experimentation; Detlef Wolf and the bioinformatics group of Roche Basel for providing excellent bioinformatics tools; and S. Fowler for help with revising the manuscript.

NOTES

1 To whom correspondence should be addressed at F. Hoffmann-La Roche Ltd., Pharmaceuticals Division, Nonclinical Development, PRNS, Bldg. 068/110, CH-4070 Basel, Switzerland. Fax: +41-61-6882908. E-mail: franziska.boess{at}roche.com. Back

REFERENCES

Arlotto, M. P., Trant, J. M., and Estabrook, R. W. (1991). Measurement of steroid hydroxylation reactions by high performance liquid chromatography as indicator of P450 identity and function. In Methods in Enzymology (M. R. Waterman and E. F. Johnson, Eds.), Vol. 206, pp. 454–462. Academic Press, San Diego.

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]

Berry, M. N., and Friend, D. S. (1969). High-yield preparation of isolated rat liver parenchymal cells. A biochemical and fine structural study. J. Cell. Biol. 43, 506–520.[Abstract/Free Full Text]

Burczynski, M. E., McMillian, M., Ciervo, J., Parker, J. B., Dunn II, R. T., Hicken, S., Farr, S., and Johnson, M. D. (2000). Toxicogenomics-based discrimination of toxic mechanisms in HepG2 human hepatoma cells. Toxicol. Sci. 58, 300–415.

Burke, M. D., Thompson, S., Elcomb, C. R., Halpert, J., Haaparanta, T., and Mayer, R. T. (1985). Ethoxy-, pentoxy- and benzyloxyphenoxazones and homologues: A series of substrates to distinguish between different induced cytochromes P-450. Biochem. Pharmacol. 34, 3337–3345.[CrossRef][ISI][Medline]

Cahill, A. L., Nyberg, D., and Ehret, C. F. (1983). Tissue distribution of cadmium and metallothionein as a function of time of day and dosage. Environ. Res. 31, 54–65.[ISI][Medline]

Eisenberger, C. F., Viebahn, R., Lauchchart, W., de Groot, H., and Becker, H. D. (1994). MHC antigen presentation on the surface of hepatocytes: Modulation during and after hypoxic stress. Transpl. Int. 7, S163–S166.[CrossRef][Medline]

Gandolfi, A. J., Wijeweera, J., and Brendel, K. (1996). Use of precision-cut liver slices as an in vitro tool for evaluating liver function. Toxicol. Pathol. 24, 58–61.[ISI][Medline]

George, E., Hamilton, G., and Westmoreland, C. (1996). The use of in vitro models in hepatotoxicity testing. Toxicol. Ecotoxicol. News 3, 142–152.

Glöckner, R., Rost, M., Pissowotzki, K., and Müller, D. (2001). Monooxygenation, conjugation and other functions in cryopreserved rat liver slices until 24 h after thawing. Toxicology 161, 103–109.[CrossRef][ISI][Medline]

Glöckner, R., Steinmetzer, P., Drobner, C., and Müller, D. (1998). Application of cryopreserved precision-cut liver slices in pharmacotoxicology - principles, literature data and own investigations with special reference to CYP1A1-mRNA induction. Exp. Toxic. Pathol. 50, 440–449.[ISI][Medline]

Glöckner, R., Steinmetzer, P., Drobner, C., and Müller, D. (1999). Use of fresh and cryopreserved human liver slices in toxicology with special reference to in vitro induction of cytochrome P450. Toxicol. in Vitro 13, 531–535.[CrossRef][ISI]

Goethals, F., Deboyser, D., Cailliau, E., and Roberroid, M. (1992). Liver slices for the in vitro determination of hepatotoxicity. In In Vitro Methods in Toxicology (G. Jolles and A. Cordier, Eds.), pp. 197–209. Academic Press, London.

Göldlin, C. R., and Boelsterli, U. A. (1991). Reactive oxygen species and non-peroxidative mechanisms of cocaine-induced cytotoxicity in rat hepatocyte cultures. Toxicology 69, 79–91.[CrossRef][ISI][Medline]

Harries, H. M., Fletcher, S. T., Diggan, C. M., and Baker, V. A. (2001). The use of genomics technology to investigate gene expression changes in cultured human liver cells. Toxicol. in Vitro 15, 399–405.[CrossRef][ISI][Medline]

Kutty, R. K., Daniel, R. F., Ryan, D. E., Levin, W., and Maines, M. D. (1998). Rat liver cytochromes P-450b, P-420b, and P-420c are degraded to biliverdin by heme oxygenase. Arch. Biochem. Biophys. 260, 638–644.

LeCluyse, E. L. (2001). Human hepatocyte culture systems for the in vitro evaluation of cytochrome P450 expression and regulation. Eur. J. Pharm. Sci. 13, 343–368.[CrossRef][ISI][Medline]

LeCluyse, E. L., Fix, J. A., Audus, K. L., and Hochmann, J. H. (2000). Regeneration and maintenance of bile canilicular networks in collagen-sandwiched hepatocytes. Toxicol. in Vitro 14, 117–132.[CrossRef][ISI][Medline]

Lipscomb, E. A., Sarmiere, P. D., Crowder, R. J., and Freeman, R. S. (1999). Expression of SM-20 gene promotes death in nerve growth factor-dependent sympathetic neurons. J. Neurochem. 73, 429–432.[CrossRef][ISI][Medline]

Lipscomb, E. A., Sarmiere, P. D., and Freeman, R. S. (2001). SM-20 is a novel mitochondrial protein that causes caspase-dependent cell death in nerve growth factor-dependent neutrons. J. Biol. Chem. 276, 5085–5092.[Abstract/Free Full Text]

Lopez-Garcia, M. P. (1998). Endogenous nitric oxide is responsible for the early loss of P450 in cultured rat hepatocytes. FEBS Lett. 438, 145–149.[CrossRef][ISI][Medline]

Lopez-Garcia, M. P., and Sanz-Gonzalez, S. M. (2000). Peroxynitrite generated from constitutive nitric oxide synthase mediates the early biochemical injury in short-term cultured hepatocytes. FEBS Lett. 466, 187–191.[CrossRef][ISI][Medline]

Lovett, R. A. (2000). Toxicologists brace for genomics revolution. Science 289, 536–537.[Free Full Text]

Lupp, A., Danz, M., and Müller, D. (2001). Morphology and cytochrome P450 isoforms expression in precision-cut rat liver slices. Toxicology 161, 53–66.[CrossRef][ISI][Medline]

Miles, A. T., Hawksworth, G. M., Beattie, J. H., and Rodilla, V. (2000). Induction, regulation, degradation, and biological significance of mammalian metallothioneins. Crit. Rev. Biochem. Mol. Biol. 35, 35–70.[Abstract/Free Full Text]

Moschella, M. C., Menzies, K., Tsao, L., Lieb, M. A., Kohtz, J. D., and Taubman, M. B. (1999). SM-20 is a novel growth factor-responsive gene regulated during skeletal muscle development and differentiation. Gene. Expr. 8, 59–66.[ISI][Medline]

Müller, D., Glöckner, R., Rost, M., and Steinmetzer, P. (1998). Monooxygenation, cytochrome P450-mRNA expression and other functions in precision-cut rat liver slices. Exp. Toxic. Pathol. 50, 507–513.[ISI][Medline]

Müller, D., Steinmetzer, P., Pissowotzki, K., and Glöckner, R. (2000). Induction of cytochrome P450 2B1-mRNA and pentoxyresorufin O-depentylation after exposure of precision-cut rat liver slices to phenobarbital. Toxicology 144, 93–97.[CrossRef][ISI][Medline]

Neyrinck, A., Eeckhoudt, S. L., Meunier, C. J., Pampfer, S., Taper, H. S., Verbeek, R. K., and Delzenne, N. (1999). Modulation of paracetamol metabolism by Kupffer cells: A study on rat liver slices. Life Sci. 65, 2851–2859.[CrossRef][ISI][Medline]

Nissley, S. P., Short, P. A., Rechler, M. M., Podskalny, J. M., and Coon, H. G. (1977). Proliferation of buffalo rat liver cells in serum-free medium does not depend upon multiplication-stimulating activity (MSA). Cell 11, 441–446.[CrossRef][ISI][Medline]

Olinga, P., Merema, M. T., de Jager, M. H., Derks, F., Melgert, B. N., Moshage, H., Slooff, M. J. H., Meijer, D. K. F., Poelstra, K., and Groothuis, G. M. M. (2001). Rat liver slices as a tool to study LPS-induced inflammatory response in the liver. J. Hepatol. 35, 187–194.[CrossRef][ISI][Medline]

Purdon, M. P., and Lehman-McKeeman, D. (1997). Improved high-performance liquid chromatographic procedure for the separation and quantification of hydroxytestosterone metabolites. J. Pharmacol. Toxicol. Methods 37, 67–73.[CrossRef][ISI][Medline]

Senaldi, G., Lobo-Yeo, A., Mowat, A. P., Mieli-Vergani, G., and Vergani, D. (1991). Class I and class II major histocompatibility complex antigens on hepatocytes: Importance of the method of detection and expression in histologically normal and diseased livers. J. Clin. Pathol. 44, 107–114.[Abstract]

Seo, K. W., Oh, M. H., Choung, S. Y., Kim, S. J., and Kim, H. J. (1999). Alteration of acetaminophen metabolism by sulfate and steroids in primary monolayer hepatocyte cultures of rat and mice. Biol. Pharm. Bull. 22, 261–264.[ISI][Medline]

Shinogi, M., Sakaridani, M., and Yokoyama, I. (1999). Metallothionein induction in rat liver by dietary restriction or exercise and reduction of exercise-induced hepatic lipid peroxidation. Bio. Pharm. Bull. 22, 132–136.[ISI]

Suematsu, M., and Ishimura, Y. (2000). The heme oxygenase-carbon monoxide system: A regulator of hepatobiliary function. Hepatology 31, 3–5.[ISI][Medline]

Waring, J. F., Ciurlionis, R., Jolly, R. A., Heindel, M., and Ulrich, R. G. (2001). 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]

Weinstein, I. B., Orenstein, J. M., Gebert, R., Kaighn, E., and Stadler, U. C. (1975). Growth and structural properties of epithelial cell cultures established from normal rat liver and chemically induced hepatomas. Cancer Res. 35, 253–263.[Abstract]

Wright, M. C., and Paine, A. J. (1992). Evidence that the loss of rat liver cytochrome P450 in vitro is not solely associated with the use of collagenase, the loss of cell–cell contacts and/or the absence of an extracellular matrix. Biochem. Pharmacol. 43, 237–243.[CrossRef][ISI][Medline]