![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Business Unit Biosciences, TNO Quality of Life (Netherlands Organization for Applied Scientific Research), Zeist, The Netherlands (A.S.K., H.M.W., B.v.O., R.H.S.); Department of Health Risk Analysis and Toxicology, University of Maastricht, Maastricht, The Netherlands (A.S.K., E.J.M., J.C.S.K., J.H.M.v.D.); Independent author (J.C.H.)
(Received June 1, 2006; accepted September 7, 2006)
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
; Waring and Ulrich, 2000). Simple, well established in vitro assays, such as primary hepatocytes, precision-cut liver slices, and hepatic cell lines, are increasingly in demand for identifying potential hepatotoxicity in early stages of investigative toxicology and for decreasing attrition rates of drugs during lead optimization (Dambach et al., 2005). However, extrapolation of in vitro results to the in vivo situation remains a scientific challenge (Guillouzo, 1998).
Toxicogenomics, the application of the genomics technologies in toxicology, would be particularly useful in the extrapolation from in vitro experiments to the in vivo situation. Extrapolations can be made at the molecular level, comparing similar mechanistic endpoints, namely compound-induced changes in multiple genes and signaling pathways (Hamadeh et al., 2002; Stierum et al., 2005). Several toxicogenomics-based studies have already been performed comparing rat hepatic in vitro models with the situation in vivo (Waring et al., 2001; Boess et al., 2003; Jessen et al., 2003). These studies concluded that, to date, no toxicogenomics-based in vitro system allowed for prediction of hepatotoxic responses in vivo. The main limitation of hepatic in vitro assays used in these toxicogenomics-based studies is the loss of liver-specific functions, in particular, cytochrome P450 (P450) monooxygenase activities (Balls et al., 2002; Boess et al., 2003). Extrapolation from these in vitro models to the in vivo situation is therefore hampered when examining compounds for which toxicity depends on bioactivation by the P450 enzyme system.
In the aim to develop an in vitro toxicogenomics-based system, the relevance of a hepatocyte sandwich culture with improved metabolic competence toward prediction of in vivo toxicity was assessed. Sandwich-cultured hepatocytes were used because, compared with other hepatocyte-based in vitro models, hepatocyte longevity is increased and a polarized cell and membrane architecture resembling that in vivo is maintained for several weeks (Dunn et al., 1989; LeCluyse et al., 1996). Furthermore, rat hepatocytes maintained in sandwich configuration display a more optimal P450 inducibility (LeCluyse et al., 1999; Richert et al., 2002). In a separate study performed by this laboratory, it was shown that in sandwich-cultured hepatocytes maintained in modified medium, enriched with low concentrations of the known P450 inducers phenobarbital (PB), dexamethasone (DEX), and ß-naphthoflavone (ß-NF), the metabolic competence, as reflected by retention of P450 enzyme activities and gene expression levels of several phase I and phase II enzymes, was enhanced (A. S. Kienhuis, H. M. Wortelboer, W. J. Maas, M. van Herwijnen, J. C. S. Kleinjans, J. H. M. Delft, and R. H. Stierum, manuscript submitted for publication). In the present study, the effect of coumarin (cis-o-coumaric acid lactone), a compound for which toxicity depends on metabolic activation by the P450 enzyme system (Born et al., 2002; Lake et al., 2002), on toxicity and gene expression profiles was studied in sandwich-cultured hepatocytes maintained in standard medium, the standard model (without inducers), and in sandwich-cultured hepatocytes maintained in modified medium, the modified model (containing enzyme inducers). Coumarin-induced cytotoxicity data and gene expression profiles generated in both in vitro models were compared with toxicity and gene expression profiles in rat liver in the in vivo study in which rats were exposed to coumarin, since in vivo verification of in vitro toxicogenomics data has been proposed to be a necessity to judge the value of the in vitro model for in vivo toxicity prediction (Jaeschke, 2003). The expected differences of bioactivation of coumarin in the standard model versus the modified model were verified by measurement of coumarin and one of its metabolites in medium.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Animals. Male Wistar rats (Crl:(WI)WU BR), 9 to 12 weeks of age, 180 to 250 g, were obtained from Charles River GmbH (Sulzfeld, Germany). During the acclimatization period and until sacrifice, animals were housed individually in Macrolon cages with wire tops and sawdust bedding at 22°C and 50 to 60% humidity. The light cycle was 12-h light/12-h dark. Feed and tap water were available ad libitum.
Animal Treatment in Vivo. Wistar rats, housed under conditions as described above, were injected i.p. with 17.5 mg/kg b.wt. (low dose), 75 mg/kg b.wt. (mid dose), and 200 mg/kg b.wt. (high dose) coumarin dissolved in corn oil. Doses were defined in a range finding study. As a solvent control, only corn oil was injected (vehicle-treated control). In each dose group and vehicle-treated group, n = 5 rats were included. Injection volume for each treatment was 10 ml/kg b.wt. Body weight was recorded on day 0 and at 24 h, just before sacrifice. Rats were anesthetized by inhalation of CO2/O2. Twenty-four hours after i.p. injection, animals were sacrificed by bleeding through the aorta abdominalis; the blood was collected in a heparin tube, from which plasma was isolated for clinical chemistry. Thereafter, livers were immediately dissected, frozen in liquid nitrogen, and stored at –80°C until further processing. A section of the liver was kept aside in formalin for pathological examination.
Clinical Chemistry. Serum alanine aminotransferase (ALAT) activity, serum aspartate aminotransferase (ASAT) activity, lactate dehydrogenase (LDH) leakage, serum alkaline phosphatase activity, glucose, cholesterol, 
-glutamyl transferase (GGT) activity, and phospholipid levels were analyzed on a Hitachi 911 centrifugal analyzer using Roche reagent kits. Differences between plasma levels in treated and nontreated animals were defined as statistically significant at a p value below 0.01, determined by one-way analysis of variance followed by Dunnett's test.
Culture of Rat Hepatocytes. Male Wistar rats similar to those used in the in vivo study and housed under identical conditions were used for hepatocyte isolation. Hepatocytes were isolated from livers from three individual rats according to a two-step collagenase perfusion technique as described by Seglen (Seglen, 1976; Paine et al., 1979) with minor modifications (Paine et al., 1979). Hepatocyte preparations with viability greater than 85% as determined by trypan blue exclusion were used and cultured on collagen gelprecoated six-well plates at a density of 1.25 x 106 cells per well. Sandwich cultures were essentially prepared according to the method of Beken et al. (2004). Hepatocytes were allowed to attach for 4 h in DMEM supplemented with 10% fetal calf serum, insulin (0.5 U/ml), glucagon (7 ng/ml), and gentamicin (50 µg/ml). After attachment, dead cells were removed by washing, and the upper collagen layer was applied. Cells were kept in standard medium consisting of DMEM containing 25 mM HEPES and 4.5 g/l D-glucose supplemented with insulin (0.5 U/ml), glucagon (7 ng/ml), hydrocortisone (7.5 µg/ml), and gentamicin (50 µg/ml). In the modified model, standard culture medium was modified by supplementation of an inducer mix that consisted of 1 mM PB, 10 µM DEX, and 5 µM ß-NF (A. S. Kienhuis, H. M. Wortelboer, W. J. Maas, M. van Herwijnen, J. C. S. Kleinjans, J. H. M. Delft, and R. H. Stierum, manuscript submitted for publication). PB was added as a concentrated stock solution in phosphate-buffered saline. DEX and ß-NF were added as concentrated stock solutions in dimethyl sulfoxide (DMSO). The final concentration of DMSO was equalized in all culture media and did not exceed 0.2% (v/v). Standard and modified media were applied 4 h after seeding the cells. Cultures were incubated at 37°C in a humidified incubator, O2 95% and CO2 5%. Medium was changed on a daily basis during a period of 72 h.
Hepatocyte Treatment and Cytotoxicity Analysis. Hepatocytes cultured in either standard or modified medium for 72 h were exposed to coumarin in a concentration range of 0 to 1000 µM dissolved in DMSO for 24 h. The final concentration of DMSO was 0.5% (v/v). Cytotoxicity was determined using the MTT reduction method (Mosmann, 1983). LDH leakage was determined spectrophotometrically in pooled medium obtained from three wells per rat per dose group, on a Hitachi 911 centrifugal analyzer using Roche reagent kits. The final coumarin concentrations for the gene expression study selected were 0 µM (control), 70 µM (low dose), 200 µM (mid dose), and 600 µM (high dose). Doses corresponded to 0%, 10%, 20%, and 50% cytotoxicity as determined by the MTT reduction assay, respectively (see results).
Gas Chromatography to Study Metabolism of Coumarin. Coumarin and the o-hydroxyphenylacetic acid (o-HPAA) metabolite were measured in standard and modified culture media of hepatocytes exposed to coumarin for 24 h using gas chromatography with flame ionization detection as described previously (Born et al., 2000a, 2003). Samples (0.1 ml) were diluted with water to a final volume of 1 ml. To increase the extraction efficiency of o-HPAA, 0.5 g of sodium chloride was added to each sample, after extraction with ethyl acetate. The extraction efficiencies of coumarin and o-HPAA were 30% and 65%, respectively. Coumarin and o-HPAA were separated and quantitated using a Fisons HRGC 8650 gas chromatograph (Fisons, Loughborough, UK) and a Varian VG-5ms column, 50 m x 0.25 mm (Varian, Inc., Palo Alto, CA). Quantification was accomplished by calculation of peak-area ratios relative to the n-tridecane internal standard, and use of the external standard curves. The same extraction procedure was applied to the external standards.
Total RNA Isolation. RNA was extracted using TRIzol (Invitrogen) according to the manufacturer's protocol. TRIzol was added to frozen liver samples obtained from the in vivo study which subsequently were pulverized with mortar and pestle under liquid nitrogen before extraction. To obtain RNA from sandwich cultures, TRIzol was added on the upper collagen layer, and cells were collected. RNA was purified using the RNeasy mini kit including an additional DNase digestion step. RNA concentration was calculated from the absorbance at 260 nm as measured spectrophotometrically. RNA quality was assessed by agarose gel electrophoresis.
Microarray Design. A local reference design was used for microarray hybridization (Table 1). For the in vivo samples, tester RNA labeled with Cy5 from individual rat livers was combined with Cy3-labeled reference RNA consisting of a pool of RNA extracted from livers obtained from five vehicle-treated control rats. For the in vitro studies, Cy5-labeled tester RNA from one experiment was combined with Cy3-labeled reference RNA from the same experiment; hepatocytes in one experiment were obtained from one rat.
|
Microarray Labeling. RNA samples were indirectly labeled according to the amino-allyl labeling procedure for microarrays from The Institute for Genomic Research (TIGR) (http://www.tigr.org/tdb/microarray/protocolsTIGR.shtml). For each labeling reaction, 25 µg of total RNA was used as the starting amount for reverse transcription of mRNA. In brief, for reverse transcription, mRNA was selected by oligo(dT) priming with primer p(DT)15. The reverse transcription reaction, in which amino-allyl dUTP is incorporated, was conducted for 3 h at 42°C. Nontranscribed RNA was degraded by alkaline hydrolysis in a final concentration of 0.25 M NaOH for 30 min at 37°C. Thereafter, the mixture was neutralized with an equimolar amount of acetic acid. The cDNA was purified using the QIAGEN QIAquick PCR purification kit. Columns were washed with 10 mM sodium-borate in 80% ethanol (pH 8.5). Column-bound cDNA was eluted two times in 30 µl of MilliQ H2O. Samples were labeled with either Cy3 (reference) or Cy5 (tester) monofunctional reactive dyes and afterward cleaned from unincorporated Cy dyes using AutoseqG-50 Sephadex chromatography columns.
Microarray Hybridization. Samples were hybridized on rat oligonucleotide microarrays containing approximately 5800 different oligonucleotide fragments of 70 nucleotides in length (QIAGEN Operon, Westburg B.V., Leusden, The Netherlands), spotted in duplicate at the Frank Holstege group (Utrecht Genomics Laboratory, Utrecht, The Netherlands) as described previously (van de Peppel et al., 2003; Heijne et al., 2005). In brief, Cy3- and Cy5-labeled samples were combined according to the experimental design. To avoid nonspecific hybridization, yeast tRNA, poly(dA) · poly(dT), and human Cot-1 DNA were added. Samples were dissolved in 110 µl of EasyHyb hybridization buffer. After cDNA denaturation at 100°C, human Cot-1 DNA was allowed to anneal for half an hour at 42°C. The hybridization mixture was pipetted directly in the center of the hybridization chamber (Corning Life Sciences B.V., Schiphol-Rijk, The Netherlands). Slides were prehybridized according to the TIGR protocol and placed carefully on top. To keep the chamber moisturized during hybridization, a precut paper drenched in MilliQ H2O was put on top. Samples were hybridized overnight at 42°C in a water bath. Slides were washed with sodium chloride/sodium citrate buffers decreasing in stringency. Microarrays were scanned using a Packard Scanarray confocal laser scanner (PerkinElmer, Boston, MA). Resulting TIFF images were loaded into Imagene 5.0 (Biodiscovery Inc., El Segundo, CA) and saved to further process and analyze the data.
Microarray Quality Criteria. Criteria for microarrays to be further analyzed considered the homogeneity of the spot signal intensities, the effect of bleaching of the fluorescence, the number of manually flagged (excluded) spots, the spatial distribution of the signals over the slide surface, the balance between Cy3 and Cy5 signal intensity, the number of saturated spots, and the quality of the slide with respect to other slides of the experiment, as described by Heijne et al. (2005).
Microarray Data Analysis. Flagged spots and controls were excluded from further analyses. Only spots on qualified microarrays with intensities higher than 1.5 times the intensity of their local background were included in data analysis. Ratios of the background-corrected intensities of tester over reference were calculated for each slide. To account for technical variations introduced during labeling or hybridization, data were normalized using the LOWESS algorithm (Yang et al., 2002). Normalized ratios were log-transformed with base two in SAS Enterprise guide V2 (SAS Institute Inc., Cary, NC). The resulting data set was loaded into Excel (Microsoft Corporation, Redmond, WA). Replicate genes per array were averaged. For a gene to be included in the analysis for both in vivo and in vitro studies, a maximum of 30% missing values per gene was accepted. Two-sample t tests (with random variance model) were performed using BRB ArrayTools Version 3.3.1, developed by Dr. R. Simon and A. Peng Lam (Simon and Peng, 2002). To reduce false positives discovery, the nominal significance level of every univariate test was set at 0.001. To study whether similarities between biological processes could be observed between the in vitro models and in vivo, pathway analysis was performed using genes significantly expressed as determined by BRB ArrayTools, and Genelists were uploaded on the website of the Database for Annotation, Visualization and Integrated Discovery (DAVID) (http://apps1.niaid.nih.gov/david/). Pathway analyses were performed using the Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.ad.jp/kegg). A pathway was considered "triggered" when at least two genes within the pathway were significantly modulated by coumarin as assessed by means of Fisher's exact test. Therefore, for each pathway and per system (in vitro and in vivo), the total number of genes that gave a signal on the microarray was divided into four groups: the number of genes not significantly modulated by coumarin and representing the pathway, those not significantly modulated by coumarin and not representing the pathway, those significantly modulated by coumarin and representing the pathway, and finally, those significantly modulated by coumarin and not representing the pathway. Pathways were considered significant at a p value below 0.05. Furthermore, principal component analysis (PCA) was performed using Matlab software Version 6.5 (The MathWorks Inc., Natick, MA).
|
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
Coumarin Metabolism. To verify that the improved metabolic competence of the modified model indeed affects the metabolism of coumarin and thus is responsible for differences in cytotoxicity of coumarin in standard culture medium and modified culture medium, both coumarin and the major coumarin metabolite reported to be detected in rat urine (Lake, 1999; Born et al., 2000a,b) were measured in both standard and modified culture media 24 h after exposure. A dose-dependent increase in the formation of o-HPAA was exclusively measured in the modified culture medium, but no o-HPAA was present in the standard culture medium. Furthermore, for all dose groups, coumarin concentration was significantly lower in the modified medium compared with the standard medium (the reduction varied from 15% for the high dose group to 59% for the low dose group; data not presented).
Gene Expression Profiling. Analysis of significantly modulated genes. In total, the criteria settings for microarray analysis used in this study resulted in data sets containing gene expression ratios of 1621 genes in vivo, 2536 genes in the standard system, and 2368 genes in the modified system. The large volume of data in the data set was reduced by performing robust statistical analysis that separated genes in which expression was actually changed by coumarin from genes that remained unchanged. Resulting data sets contained, regardless of dose dependence, 321, 13, and 92 genes that were significantly changed by coumarin in vivo, in the standard system, and in the modified system, respectively. Overlap of the changed genes between both in vitro models and in vivo is presented by the Venn diagram in Fig. 2. Twenty-three genes were altered in both the modified model and in vivo, and only three genes in both the standard system and in vivo. One gene was altered in all systems, in vitro and in vivo. These 27 genes form the subset of genes significantly changed by coumarin which are listed in Table 3. Both in vivo and in the in vitro models, most genes showed a dose-dependent down-regulation (for 52–85% of the genes, the Pearson correlation analysis coefficient is less than –0.8; data not shown). When comparing the different systems with each other, the direction of modulation was always comparable between the standard and the modified in vitro models and, in most of the cases, also between the in vivo model and one of the in vitro models.
|
Pathway analysis. Pathways triggered by coumarin in at least two systems and significant in at least one are presented in Table 4. In the standard model, the modified model, and in vivo, one, eight, and six pathways, respectively, were significantly triggered by coumarin. Four pathways, i.e., methionine metabolism, fatty acid metabolism, -hexachlorocyclohexane degradation, and complement and coagulation cascades, were significant in both the modified system and in vivo. Only one pathway was significantly triggered in the standard system, but not in vivo.
|
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
; Lake et al., 2002). CE rearranges spontaneously to the more stable o-hydroxyphenylacetaldehyde (o-HPA). Both CE and o-HPA are assumed to contribute to the hepatotoxicity of coumarin in rats as they conjugate with critical cellular macromolecules (Lake et al., 1989, 1994; Fentem and Fry, 1993; Born et al., 1997,2000b, 2003). Before engaging in gene expression studies, it was hypothesized that an increased metabolic competence of sandwich cultures would improve metabolic conversion of coumarin to toxic metabolites, thereby increasing the resemblance with aspects of in vivo coumarin-induced hepatotoxicity. Therefore, sandwich-cultured hepatocytes were maintained either in a standard medium or in a modified medium that was enriched with a mixture of low concentrations of known P450 inducers to enhance metabolic capacity (A. S. Kienhuis, H. M. Wortelboer, W. J. Maas, M. van Herwijnen, J. C. S. Kleinjans, J. H. M. Delft, and R. H. Stierum, manuscript submitted for publication). Findings in this study showed that both traditional measures of toxicity and coumarin-induced gene expression in the modified model were more pronounced and closer to in vivo compared with the standard model.
In vivo, coumarin toxicity was indicated by increased plasma ALAT, ASAT, and GGT levels, primarily at the highest dose level of 200 mg/kg b.wt. Moreover, hepatotoxicity did manifest itself as severe single cell necrosis and minimal centrilobular necrosis, as observed by histopathological observations in livers of rats treated with 200 mg/kg coumarin. These findings agree with other studies which report that administration of coumarin in doses ranging from 125 to 500 mg/kg results in severe centrilobular necrosis after 24 h (Lake et al., 1989, 1994; Lake, 1999).
In vitro, coumarin appeared to be cytotoxic in hepatocytes cultured in modified medium, whereas no cytotoxicity was measured in the standard model. In our laboratory, it has been shown that gene expression of CYP1A2 and CYP2E1 in the modified model is closer to in vivo compared with the standard model, in which gene expression of these genes is highly down-regulated. Furthermore, upon addition of the inducer mixture, enzyme activities of the 1A subfamily were increased to levels comparable to in vivo (A. S. Kienhuis, H. M. Wortelboer, W. J. Maas, M. van Herwijnen, J. C. S. Kleinjans, J. H. M. Delft, and R. H. Stierum, manuscript submitted for publication). Since coumarin is converted by CYP1A and CYP2E to o-HPA (Born et al., 2002; Lake et al., 2002), this metabolite possibly causes the cytotoxicity in the modified model. It was shown earlier that the compounds in the inducer mixture, PB, DEX, and ß-NF, increase the formation of o-HPA and other coumarin metabolites, enhancing coumarin toxicity (Peters et al., 1991; Fentem and Fry, 1992). In the present study, these findings were confirmed by measurement of the coumarin metabolite o-HPAA, which is formed after oxidation of o-HPA. After 24 h of coumarin exposure, a dose-dependent formation of o-HPAA was exclusively observed in hepatocytes cultured in the modified model, whereas no o-HPAA was formed in the standard model. Furthermore, coumarin clearance was significantly higher in the modified culture medium compared with the standard culture medium in all dose groups.
Analysis of gene expression revealed that both in vivo and in vitro, genes are affected by coumarin at dose levels at which no toxicity occurs as determined by traditional toxicology measures. This finding suggests that gene expression changes may very well be more sensitive indicators of potential adverse effects, as was indicated before by Heinloth et al. (2004). Considerably more genes were significantly altered and similarly affected by coumarin in both the modified system and in vivo compared with the standard system. In this study, not all affected genes are discussed in detail. It is of more relevance to investigate analogies in the biological pathways these genes trigger. The four pathways that were statistically significantly triggered both in vivo and in the modified model, and the significant differentially expressed genes within these pathways will be discussed.
Betaine-homocysteine methyltransferase (Bhmt) and CTL target antigen (Cth), a putative cystathione lyase enzyme, both down-regulated in vivo and in the modified system, appear in the methionine metabolism pathway. It has been shown that down-regulation of Bhmt and Cth impairs conversion of homocysteine to methionine and cysteine, respectively, resulting in increased homocysteine levels in the cell (Torres et al., 1999; Forestier et al., 2003). Homocysteine accumulation is associated with impaired liver function, necrosis, liver cirrhosis, and fibrogenesis (Torres et al., 1999; Finkelstein, 2003).
Expression of CYP1A2, a gene which represents both the -hexachlorocyclohexane degradation pathway and the fatty acid metabolism pathway, is significantly down-regulated in vivo and in the modified model. This P450 belongs to the CYP1A subfamily that converts coumarin to toxic metabolites. Therefore, down-regulation of gene expression of this enzyme could be interpreted as a classical negative feedback loop, turning out in the present case to prevent additional liver injury.
Fibrinogen -polypeptide (fgg) is significantly up-regulated in vivo and significantly down-regulated in the modified model. This gene is part of the complement and coagulation cascade, to which coumarin is associated since coumarin derivatives, such as warfarin, are well known for their anticoagulant properties (Egan et al., 1990). One possible explanation for the significant, but different response of particular genes to coumarin in vivo and in vitro can be illustrated by fgg. fgg is mainly expressed in hepatocytes. Up-regulation of fgg, as occurred in vivo, would be expected in terms of toxicity since fibrinogen levels are increased after liver injury (Redman and Xia, 2001). Fibrinogen synthesis has been shown to be stimulated in hepatocytes by factors excreted by extrahepatic tissues or nonparenchymal cells (Otto et al., 1987). In vitro, these factors are absent. Therefore, stimulation of fibrinogen synthesis in hepatocyte cultures in vitro may only occur when these factors are added to the culture medium.
In addition to the statistical analysis of microarray data which resulted in a list of significantly modulated genes, PCA was performed. PCA allows inclusion of all genes into analysis, preserving their interrelationships. Results showed that even though differences between in vitro and in vivo data remain evident, the dose-response effect with respect to gene expression of coumarin in the modified system is more similar to in vivo data compared with the standard system.
Dose levels in vivo were compared with the coumarin concentrations applied in vitro. Using pharmacokinetic data from literature, a distribution coefficient of coumarin of 3.33 l/kg was estimated (Ritschel and Hussain, 1988; Lake, 1999; Born et al., 2003). Assuming that in vivo bioavailability of coumarin is 100% after i.p. injection, one can estimate a maximum in vivo plasma concentration of 36 µM, 154 µM, and 411 µM for the low, mid, and high in vivo doses, respectively. These concentrations are in the same range as the in vitro concentrations of 70 µM, 200 µM, and 600 µM. Although plasma levels of coumarin were not determined in vivo, the range of estimated maximum plasma levels in vivo and, therefore, potential target organ concentrations were comparable to the dose level range used in the culture media in vitro.
It is important to note that in addition to coumarin, the inducer mixture may also influence gene expression and that these effects interact with each other. In the design of the present study, however, the effect of the inducer mixture on gene expression was filtered out because matching controls were used; i.e., both test and control samples are from hepatocytes cultured in the modified system. Nonetheless, it remains possible that the inducer mixture affected the expression of some genes in the modified system in such a way that no additional effect of coumarin could have been detected.
Furthermore, effects on gene expression in the highest dose group of the modified model may not be solely attributed to coumarin. The highest coumarin dose group in this model resulted in a 50% loss of cell viability as measured by the MTT reduction assay. In a sandwich configuration, dead cells remain between the two layers of collagen. As a consequence, necrosis caused by coumarin exposure in one cell can affect gene expression in viable neighbor cells (Fielden and Zacharewski, 2001). Excluding the high dose group of the modified model from analysis, still 80% of the genes in Table 2 are retained. Moreover, the methionine metabolism and -hexachlorocyclohexane degradation pathways are still significantly triggered by coumarin.
In summary, our experiments have shown that the metabolism of coumarin toward active metabolites, coumarin-induced toxicity, gene expression profiles, and, consequently, biological pathways in the modified system containing sandwich-cultured hepatocytes with enhanced metabolic capacity better represent the situation in vivo compared with conventional sandwich-cultured hepatocytes. This finding highlights the need for a metabolically competent, toxicogenomics-based, hepatocyte in vitro system.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
ABBREVIATIONS: P450, cytochrome P450; PB, phenobarbital; DEX, dexamethasone; ß-NF, ß-naphthoflavone; DMEM, Dulbecco's modified Eagle's medium; MTT, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide; ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; LDH, lactate dehydrogenase; GGT, -glutamyl transferase; DMSO, dimethyl sulfoxide; o-HPAA, o-hydroxyphenylacetic acid; TIGR, The Institute for Genomic Research; BRB, Biometric Research Branch; PCA, principal component analysis; PC, principal component; CE, coumarin 3,4-epoxide; o-HPA, o-hydroxyphenylacetaldehyde.
Address correspondence to: Rob H. Stierum, TNO Quality of Life, Business Unit Biosciences, Physiological Genomics, P.O. Box 360 (PP 8), 3700 AJ, Zeist, The Netherlands. E-mail: rob.stierum{at}voeding.tno.nl.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
 |
 |
|
 |
 |
 |
