Effects of drugs in subtoxic concentrations on the metabolic fluxes in human hepatoma cell line Hep G2

Toxicology and Applied Pharmacology 240 (2009) 327–336

Contents lists available at ScienceDirect

Toxicology and Applied Pharmacology j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / y t a a p

Effects of drugs in subtoxic concentrations on the metabolic fluxes in human hepatoma cell line Hep G2 Jens Niklas, Fozia Noor ⁎, Elmar Heinzle Biochemical Engineering Institute, Saarland University, Campus A1.5, D-66123 Saarbrücken, Germany

a r t i c l e

i n f o

Article history: Received 13 May 2009 Revised 17 June 2009 Accepted 3 July 2009 Available online 14 July 2009 Keywords: Toxicity screening Subtoxic effects Preclinical drug development In-vitro assays HepG2 cells Metabolic flux analysis Respiration

a b s t r a c t Commonly used cytotoxicity assays assess the toxicity of a compound by measuring certain parameters which directly or indirectly correlate to the viability of the cells. However, the effects of a given compound at concentrations considerably below EC50 values are usually not evaluated. These subtoxic effects are difficult to identify but may eventually cause severe and costly long term problems such as idiosyncratic hepatotoxicity. We determined the toxicity of three hepatotoxic compounds, namely amiodarone, diclofenac and tacrine on the human hepatoma cell line Hep G2 using an online kinetic respiration assay and analysed the effects of subtoxic concentrations of these drugs on the cellular metabolism by using metabolic flux analysis. Several changes in the metabolism could be detected upon exposure to subtoxic concentrations of the test compounds. Upon exposure to diclofenac and tacrine an increase in the TCA-cycle activity was observed which could be a signature of an uncoupling of the oxidative phosphorylation. The results indicate that metabolic flux analysis could serve as an invaluable novel tool for the investigation of the effects of drugs. The described methodology enables tracking the toxicity of compounds dynamically using the respiration assay in a range of concentrations and the metabolic flux analysis permits interesting insights into the changes in the central metabolism of the cell upon exposure to drugs. © 2009 Elsevier Inc. All rights reserved.

Introduction The development of drugs is a very expensive and time-consuming process. The average clinical success rate in all therapeutic areas is approximately 11%. This means that just one in nine compounds makes it through the clinical development and reaches the market. The costs for the development of a drug are about $900 million and most of these incur later in the pipeline where major attrition occurs (Kola and Landis, 2004). One of the leading causes of attrition at all stages of drug development is toxicity (Kramer et al., 2007). In particular, hepato- and cardiovascular toxicities were accounted for two out of three post-market withdrawals of drugs (Schuster et al., 2005). Drug induced liver impairment is one of the major causes of acute liver failure in western countries and the mortality rate is about 80% in patients with acute liver failure (Ostapowicz and Lee, 2000). There is, therefore, an urgent need to identify better models to predict human hepatotoxicity (Sahu, 2007). While animal based toxicity assays could predict 70% of the toxicity in human beings in a retrospective analysis, hepatotoxicity has a very bad concordance between human beings and animals. Only half of the new compounds that caused clinical hepatotoxicity had a concordance with animal hepatotoxicity (Olson et al., 2000; O'Brien et al., 2006).

⁎ Corresponding author. Fax: +49 681 302 4572. E-mail address: [email protected] (F. Noor). 0041-008X/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.taap.2009.07.005

Primary hepatocytes represent the gold-standard for the analysis of the hepatotoxic potential of new drug entities but other cellular models can also be used for similar predictions (Davila et al., 1998; Castell et al., 2006). Several cell lines are available for the analysis of toxicity mechanisms. The human hepatoma cell line Hep G2, established in 1979, is the best-characterized and most frequently used cell line with respect to hepatotoxic endpoints and has been used to examine various mechanisms of hepatotoxicity (Aden et al., 1979; Brandon et al., 2003; Wilkening et al., 2003; Hewitt and Hewitt, 2004; Vermeir et al., 2005; Noor et al., 2009). The evaluation of multiple endpoints on Hep G2 cells allows the prediction of human hepatotoxicity with more than 80% sensitivity and 90% specificity (O'Brien et al., 2006). Therefore, the idea that Hep G2 cells can predict overall hepatotoxicity using hepatospecific endpoints is becoming more and more accepted (Sahu, 2007). Various mechanisms can be involved in the onset and progression of hepatotoxicity. There are different cellular targets of drug-related toxicity. Moreover, liver injury can be initiated by mechanisms specific to the toxicant (Chang and Schiano, 2007; Sahu, 2007). Biological systems are quite robust concerning their ability to adapt to environmental changes. For example alternative pathways can be activated, isoenzymes can undertake the task of an enzyme or alternative substrates can be used. The metabolic system may adapt without visible changes in the phenotype or even before the manifestation of toxicity. These metabolic changes could, therefore, be indications for the hepatotoxic potential of a test compound.

328

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

It is highly probable that the effects of subtoxic concentrations contribute to the overall toxicity of the compound which may later lead to the failure of the drug. These effects may be difficult to identify but could later cause severe long term problems. To our knowledge very little is known about the effects of drugs in subtoxic concentrations on the cellular metabolism. In the emerging field of systems biology, the analysis of metabolic flux data is highly relevant since intracellular reaction rates represent the functional endpoints of gene, protein and metabolic interactions (Sauer, 2006). Metabolic flux analysis aims at the quantitative analysis of in-vivo carbon fluxes across metabolic networks and has revealed fascinating insights into various biological systems (Wittmann, 2007). Such quantitative analysis has been applied to analyse and engineer the metabolism of microorganisms (Wittmann and Heinzle, 2002; Kiefer et al., 2004; Becker et al., 2005; Antoniewicz et al., 2007; Kim et al., 2008), and to study mammalian cells (Bonarius et al., 1996; Balcarcel and Clark, 2003; Vo and Palsson, 2006; Hofmann et al., 2008; Maier et al., 2008; Yoo et al., 2008; Deshpande et al., 2009) and plants (Heinzle et al., 2007; Libourel and Shachar-Hill, 2008). Assessment of the effects of drugs on mammalian metabolism requires methods allowing high throughput analysis. Relatively few studies focused on developing and applying such methods (Balcarcel and Clark, 2003; Sauer, 2004; Wittmann et al., 2004; Velagapudi et al., 2007). We used the Hep G2 cell line as a model system to determine whether metabolic flux analysis in a high throughput setup can be used to analyse changes in the metabolism of cells upon exposure to subtoxic concentrations of drugs. A number of compounds namely amiodarone, diclofenac and tacrine as well as the commonly used solvent dimethylsulphoxide (Busby et al., 1999; Easterbrook et al., 2001) were tested. For the determination of the EC50 and subtoxic concentrations of the test drugs, we used a time-resolved respiration assay which is reported as an invaluable non-invasive tool for the assessment of toxicity (Deshpande et al., 2005; Noor et al., 2009). The anabolic demand of the Hep G2 cells was determined, the exometabolome was analysed and the metabolic fluxes were estimated using metabolite balancing. In the presented work, the following questions were addressed: (i) can differences in metabolic fluxes of the central metabolism of HepG2 cells be observed upon exposure to subtoxic concentrations of the tested drugs? (ii) Which changes occur and iii) is it possible to reconciliate these changes with literature data according to functions of the test drugs and mechanisms of toxicity of these drugs? Materials and methods Cell line and culture conditions. The human hepatoma cell line HepG2 (DSMZ, Braunschweig, Germany) was maintained in Williams Medium E (WME, PAN Biotech, Aidenbach, Germany) supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin (penicillin/ streptomycin solution from C.C. Products, Oberdorla, Germany) and 3% fetal calf serum (FCS Gold, PAA Laboratories, Pasching, Austria). The cells were kept in culture flasks (Nunc, Wiesbaden, Germany) at 37 °C in a humidified 5% CO2 cell culture incubator and were subcultivated at 90% confluency. Cell number was determined using a hemocytometer and viability was assessed using the trypan blue exclusion method. Test compounds. Test drugs and DMSO were purchased from Sigma-Aldrich (Steinheim, Germany). Stock solutions of 50 mM were prepared in DMSO for all test compounds. The stock solutions were serially diluted with Dulbecco's phosphate buffered saline (PBS, PAA Laboratories, Pasching, Austria). An appropriate volume of the diluted stock solution was added in the medium to yield the desired final concentration of the test drug (n = 4). The highest final concentration of the solvent in toxicity assay did not exceed 2% (v/v). The DMSO concentration in subtoxic concentrations was always less than 0.1% (v/v).

Evaluation of proliferation of Hep G2. For the determination of the proliferation of Hep G2 cells under serum free conditions, 5 × 104 cells/well were seeded in WME containing 3% FCS in 4-well plates (Nunc, Langenselbold, Germany). One well was always filled with medium as a control. After 24 h, the medium was aspirated, the cells were washed once with serum free WME and finally serum free WME was added. The cultivation volume was 500 μl. The proliferation was monitored using the Sulforhodamine B-assay (SRB-assay, SigmaAldrich). The SRB-assay is a rapid method for the quantification of the cellular protein content and is linear with the number of cells (Skehan et al., 1990). The assay was carried out according to the kit instructions. The absorption was measured at 540 nm against a reference wavelength at 660 nm. At each time point, one 4-well plate was used and the SRB-assay was performed. Growth rates (μ) were calculated using μ=

lnN − lnN0 t − t0

ð1Þ

where N represents the absorbance measured in the SRB-assay and t is the time. Doubling time (TD) was calculated using TD =

ln2 : μ

Determination of the biomass composition. Hep G2 cells were maintained in 175 cm2 culture flasks under standard culture conditions in a humidified 5% CO2 incubator at 37 °C. For the calculation of the anabolic demand of the cells, the biomass composition of Hep G2 cells was determined as follows: Sample preparation. Cells were harvested in the late exponential phase by trypsinisation, collected in a 50 ml centrifuge tube (Falcon) and spun down in a centrifuge (1500 U/min, 5 min, 25 °C, Function Line, Heraeus Instruments, Hanau, Germany). The cell pellet was washed twice with PBS. The total cell number was then determined using a hemocytometer. An appropriate amount of cells was used for each biomass-related analysis. Dry cell weight. Samples containing 5–15 × 106 cells were taken in preweighed 1.5 ml tubes (Eppendorf) and centrifuged (3000 U/ min, 5 min, 25 °C, Biofuge Pico, Heraeus Instruments, Hanau, Germany). The supernatant was carefully discarded and the pellets were frozen, lyophilized and finally weighed. Protein content. Samples containing 2–10 × 106 cells were taken in 1.5 ml Eppendorf tubes and centrifuged (3000 U/min, 5 min, 25 °C, Biofuge Pico, Heraeus Instruments, Hanau, Germany). The supernatant was carefully aspirated off and the cell pellets were resuspended in 200 μl CelLytic™ M solution (Sigma-Aldrich) and agitated for 15 min. After centrifugation (13,000 U/min, 10 min, 4 °C, Biofuge fresco, Heraeus Instruments, Hanau, Germany) the supernatant was transferred into fresh tubes. The cellular protein content was determined using the Bio-Rad Protein Assay which is based on the method described by Bradford (1976). The calibration was performed using different concentrations of bovine serum albumin (Sigma-Aldrich) and the absorbance was measured at 595 nm (iEMS Reader MF, Labsystems, Helsinki, Finland). Amino acid composition of cellular protein. To determine the amino acid composition of the total cellular protein, 2–5 × 106 cells were taken in 1.5 ml Eppendorf tubes. After centrifugation (3000 U/min, 5 min, 25 °C, Biofuge Pico, Heraeus Instruments) the supernatant was discarded and the pellets were resuspended in 450 μl PBS. For hydrolysis 50 μl of 6 M HCl was added and the tubes were incubated for 24 h at 110 °C. The solution was then neutralized using an appropriate volume of 6 M NaOH and the amino acid composition was determined by high performance

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

liquid chromatography (HPLC) as described later for the quantification of glucose, organic acids and amino acids. Nucleic acid content. The determination of the DNA and RNA content of Hep G2 cells was carried out using the TRI Reagent® Solution from Applied Biosystems/Ambion according to the manual. The cell pellet was obtained as described in the protein assay. The DNA and RNA content was determined spectrophotometrically by measuring the absorbance at 260 nm (Ultraspec, Pharmacia Biotech, Little Chalfont, England). Carbohydrate content. The carbohydrate assay is based on a method described by Xie and Wang (1994). The cell pellet was obtained as described in the protein assay. The supernatant was discarded and the cells were resuspended in 250 μl CelLytic™ M (Sigma-Aldrich) and agitated for 15 min. 250 μl distilled water and 500 μl phenol (50 g/l in distilled water) were added and mixed. 2.5 ml sulphuric acid (98%) was added to the samples. After incubation for 30 min at 20 °C, the absorbance was measured at 488 nm (iEMS Reader MF, Labsystems). A series of glucose standards was used for calibration. Lipid content. The method used for the extraction of lipids was described by Akopian and Medh (2006). Approximately 2.5 × 108 cells were collected in a 50 ml tube (Falcon) and centrifuged (1500 U/min, 5 min, 25 °C, Function Line). The pellet was resuspended in 10 ml cyclohexane/isopropanol solution (3:2 v/v) and incubated for 15 min on ice. After centrifugation (same conditions as before) the supernatant was transferred in a preweighed flask and the extraction step was repeated once. The supernatants were pooled and the solution was evaporated and dried under nitrogen atmosphere until a constant weight was achieved. Quantification of glucose, organic acids and amino acids. Quantification of glucose, lactate and pyruvate in supernatants was carried out by HPLC (Kontron Instruments, Neufahrn, Germany). The separation was performed on an Aminex HPX-87H ion exchange column (300 × 7.8 mm; Biorad, Hercules, CA, USA) at 55 °C isocratically, with 7 mM H2SO4 as the mobile phase and a flow rate of 0.8 ml/min. The peaks were detected via determination of the refractive index (glucose) or UV absorption at 210 nm (organic acids). Quantification of amino acids was performed as described by Kromer et al. (2005) in detail. Respiration measurement. Respiration measurements were carried out using 96-well flat bottom microtiter OxoPlates (OxoPlate96C, Presens, Precision Sensing GmbH, Regensburg, Germany). Each well in an OxoPlates contains an optical oxygen sensor immobilized in the centre. The sensor contains an oxygen-sensitive and a reference dye that acts as an internal standard. The measurements can be carried out in a fluorescence reader using the dual filter 544 nm/ 650 nm for the oxygen-sensitive dye and 544 nm/590 nm for the reference dye. Detailed procedure of the method and further information are described elsewhere (John et al., 2003; Deshpande and Heinzle, 2004; Deshpande et al., 2004, 2005; Noor et al., 2009).

and enclosed in a chamber with constant 5% CO2 supply. After the test period the supernatants were collected for each tested concentration and pooled together (n = 4). The samples were analysed and resulting data was used for estimating yields for all measured components. These yields were further used for flux estimation as described below. EC50 estimation. EC50 value indicates the concentration of the test drugs at which 50% decrease in respiration is observed. Respiration is assessed by the measured dissolved oxygen concentration, given in % air saturation. EC50 values were determined using Matlab 7.2 (The MathWorks, Natick, MA, USA) applying a four-parameter sigmoidal concentration–response curve using the following equation;

Y = Bottom +

ðTop − BottomÞ

ð3Þ

1 + 10ð log EC50 − X Þ × ðHillslopeÞ

where X is the logarithm of concentration and Y is the response. Top is the maximum response value, and Bottom is the minimum response. The range of the EC50 was determined applying Monte-Carlo simulation considering the standard deviation of replicates (n = 4). Metabolic flux analysis. A metabolic network model (Fig. 1) was developed for the calculation of metabolic fluxes based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database (http://www.genome.jp/kegg) for Homo sapiens and biochemistry books (Michal, 1999; Berg et al., 2003). More than 230 reactions were considered. Assuming intracellular steady-state many reactions could be lumped together into single reactions. After simplification the resulting stoichiometric network contained 62 reactions in total, consisting of 34 intracellular reactions, 23 extracellular, i.e. transport reactions and 5 lumped reactions for the biomass synthesis. The main pathways of the central energy metabolism, glycolysis, oxidative decarboxylation and TCA-cycle are included. Since the P/ O ratio of Hep G2 cells is not known, oxidative phosphorylation is not included in the model but the excess of NAD(P)H, FADH2 and ATP from all catabolic and anabolic reactions considered in the model is calculated. The pentose phosphate pathway is represented only for the synthesis of nucleic acids, since the pentose phosphate pathway activity cannot be calculated by metabolite balancing alone. The anaplerotic fluxes represented by the enzymes pyruvate carboxylase, PEP carboxykinase and malic enzyme are lumped to one flux from oxaloacetate to pyruvate. Amino acid metabolism is represented by biochemical reactions that describe the degradation of amino acids to intermediates of the central pathways. The biomass production in the experiments for the analysis of the effects of test compounds was estimated using the glucose uptake of the cells. An excellent correlation was observed between glucose uptake and biomass production (R2 = 1.00) in a control experiment with different cell numbers

ΔBiomass½g  = 3:43 × ΔGlucose½μmol × 10 Assay procedure. Hep G2 cells were seeded in WME supplemented with 3% FCS, 100 U/ml penicillin and 100 μg/ml streptomycin in 96well OxoPlates at a cell density of 105 cells per well. In earlier experiments it was found that this cell density is most appropriate demonstrating measurable consumption of oxygen without oxygen limitation. The culture volume was 200 μl and the outer wells of the plate were filled with distilled water. The cells were allowed to adhere for 18 h in the incubator. Afterwards the medium was replaced by serum free WME after one washing and the test compound was added (n = 4). The respiration was recorded online for 48 h at intervals of 15 min in a BMG fluorescence reader (Fluostar, BMG Labtechnologies, Offenburg, Germany) set at 37 °C

329

−5

:

ð4Þ

Secretion and uptake rates were determined from the specific growth rate, μ, and yield coefficients

vm;i = μ

ΔMetabolitei : ΔBiomass

ð5Þ

ΔMetabolitei and ΔBiomass are the concentration differences of metabolite i and biomass between start of the experiment and 48 h incubation, respectively. The stoichiometric matrix has the dimensions

330

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

Fig. 1. Metabolic network model for the determination of metabolic fluxes using metabolite balancing in Hep G2 cells. ATP, NAD(P)H and FADH2 are not balanced but calculated from all catabolic and anabolic reactions included in the model. Oxidative phosphorylation in the respiratory chain is not calculated. The balance region is given in grey. The balance equations used in the model are given in the Supplementary material.

34 × 62 and has a rank of 34. The number of measurements and degrees of freedom is 28. The balance equations are shown in Table S1 of the Supplementary material. Assuming pseudo steady-state, fluxes were estimated using standard methods (Stephanopoulos et al., 1998) applying Matlab 7.2 using −1

vk = Sk

× ð−Sm × vm Þ

ð6Þ

where vk represents the vector and Sk the matrix of calculated rates, vm the vector and Sm the matrix of measured rates. Since flux estimations were based on the measurement of pooled samples of four parallel 96well cultivations, only the standard deviation of the HPLCmeasurement could be determined which was usually very low. To get a more realistic representation of the standard deviation of the calculated fluxes, the determined average standard deviation of 14 pairs of measurements of a separate set of experiments was used in Monte-Carlo simulation. Fluxes are given as molar % with respect to

glucose uptake flux. Flux differences are presented for easier comparison of drug effects (Eq. (7)). Error propagation is calculated by applying Gaussian error propagation. Δvi = vi;drug − vi;control :

ð7Þ

These values are also given as % of glucose uptake flux. Results Proliferation of Hep G2 cells A growth curve of Hep G2 cells in 3% serum containing WME and serum free WME is depicted in Fig. 2. Cells maintained in 3% serum containing WME had a growth rate of 0.026 h− 1 and a doubling time of 26.3 h. Although slightly slower, the cells in serum free conditions

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

331

Table 2 Amino acid composition of total cellular proteins of Hep G2 cells determined by HPLC analysis of cell hydrolysates.

Fig. 2. Growth curve for Hep G2 cells grown in WME supplemented with 3% FCS and without FCS (n = 3). For the cultivation under serum free conditions, the cells were seeded in WME containing 3% FCS and after 24 h (arrow) the medium was changed to WME without FCS. Seeded cell density was 5 × 104 cells/well.

were proliferating for up to 120 h with a growth rate of 0.025 h− 1 and a doubling time of 27.5 h. Determination of the biomass composition of Hep G2 cells Depending on the cell type and the growth conditions the biomass composition of mammalian cells can vary (Alberts et al., 1983; Xie and Wang, 1994; Zupke et al., 1995; Bonarius et al., 1996). For the calculation of the intracellular fluxes using metabolite balancing, the determination of the anabolic fluxes in the cells is essential. The macromolecular composition of Hep G2 cells as determined is shown in Table 1 whereas the amino acid composition of total cellular protein is given in Table 2. The values are comparable to literature data (Savinell and Palsson, 1992; Xie and Wang, 1994; Bonarius et al., 1996). Due to hydrolytic degradation, cysteine could not be measured. It was assumed to be equal to the tyrosine fraction. Toxicity assessment by respiration assay The selected drugs were screened for toxicity in a non-invasive kinetic respiration assay to obtain the concentration–response curves (Fig. 3) and EC50-values (Table 3). These were used to determine the subtoxic concentration range of the drugs. The assay was carried out for 48 h since during this time there was no limitation of oxygen and nutrients and therefore no effect on the growth of the cells. Effect of test compounds on cellular metabolism using metabolic flux analysis DMSO This is a commonly used solvent and was also used in the presented study for making drug stock solutions. DMSO may itself have effects on the cellular metabolism. It is therefore, important to

Table 1 Macromolecular composition of Hep G2 cells maintained in Williams Medium E containing 3% FCS in the late exponential phase.

DNA RNA Carbohydrates Lipids Proteins Rest/Ash Dry weight

Content [pg/cell]

Mass percentage [%]

10.1 16.2 14.1 75.5 257.5 45.7 419.1

3.9 2.4 3.4 18.0 61.4 10.9

Amino acid

Fraction [%]

Alanine Arginine Aspartate/asparagine Cysteinea Glutamate/glutamine Glycine Histidine Isoleucine Leucine Lysine Methionine Phenylalanine Proline Serine Threonine Tryptophane Tyrosine Valine

8.5 4.7 10.6 2.6 12.3 12.7 1.4 2.5 7.2 12 1.3 2.8 4.6 6.6 3.7 0.8 2.6 3.3

a

The cysteine fraction was assumed to be equal to the tyrosine content.

analyse the effects of different concentrations of DMSO. The glucose uptake was obviously not influenced at DMSO concentrations up to about 0.5% (Fig. 4A). With DMSO concentrations higher than 1% the glucose uptake decreased clearly with increasing solvent concentrations. The respiration measurement did not show any significant change up to 2% DMSO as can be deduced from a nearly constant oxygen concentration after 48 h incubation (Fig. 4B). 3% DMSO, corresponding to the EC50 value (Table 3), caused a remarkable decrease in respiration as well as in glucose uptake (Figs. 4A and B) indicating a loss of cell viability. TCA-cycle activity, represented by the flux through the α-ketoglutarate dehydrogenase (v7, Fig. 1), was around 15% in the null reference and at 0.001% DMSO, remained at about 33% up to 2% DMSO and then dropped to about 0% (Table S2 of Supplementary material). Flux differences are depicted in Fig. 4C. The observed average lactate production (v10), was 155 ± 8% and did not change significantly (Fig. 4D and Table S2 of Supplementary material). Therefore, anaerobic energy production remained constant. However, other fluxes were more visibly influenced. Up to 0.1% DMSO alanine was secreted (v21) but at DMSO concentrations higher than 1% alanine was consumed which resulted in a change of the flux from alanine to pyruvate (v22) of about 7% (Fig. 4E). Exposure to DMSO concentrations higher than 0.5% resulted in a decrease in glutamine uptake from about 11% in the untreated control to about 3% at 3% DMSO (Table S2) that caused at the same time a decrease of the flux from glutamine to glutamate (v26) of about 7% (Fig. 4F). The flux from glutamate to αketoglutarate (v24) also changed, particularly at 3% DMSO where a reverse flux to glutamate was calculated (Table S2). The pyruvate production (v12) slightly increased with increasing concentrations of DMSO (Table S2). Amiodarone For the analysis of the effects of amiodarone at subtoxic range, a series of concentrations from 0.5 μM to 10 μM was analysed. 10 μM represents the EC50 calculated from the respiration assay (Table 3). The glucose uptake (Fig. 5A1) and respiration (Fig. 5A2) were not significantly influenced upon exposure to amiodarone concentrations less than 10 μM. However, 10 μM amiodarone caused a clear decrease in glucose uptake (Fig. 5A1) and respiration indicated by the higher pO2 value after 48 h of treatment (Fig. 5A2). The calculated flux distribution indicated that there were no significant concentration dependent changes in metabolism at amiodarone concentrations from 0.5 μM to 10 μM. Treatment with 10 μM amiodarone resulted in a decrease in the activity of the TCA-cycle which is represented by the flux through the α-ketoglutarate dehydrogenase (v7) by about 25%

332

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

Fig. 3. Concentration response curves for DMSO, amiodarone, diclofenac and tacrine tested on Hep G2 cells in the 96-well respiration assay (n = 4). pO2 is the dissolved oxygen concentration given as % air saturation after 48 h incubation. Seeded cell density was 105 cells/well.

as well as a slight increase in lactate production (v10) to about 15% (Figs. 5A3 and A4). These changes lead to a decreased production of NAD(P)H, FADH2 and ATP (Table S3). Compared with the untreated control, less glutamine was taken up (v25) upon exposure to amiodarone (Fig. 5A5), whereas the glutamate production (v23) was relatively uninfluenced (Fig. 5A6). The flux from glutamate to αketoglutarate (v7) was about 6% in untreated control and decreased to about 0% during drug treatment (Fig. 5A7). Similarly, the flux from oxaloacetate to pyruvate (v62), decreased from about 7% to about 2% (Fig. 5A8). In addition the flux from serine to glycine (v14) was slightly increased by about 3% in the cells treated with the drug (Table S3 of Supplementary material). Diclofenac An EC50 of 146 μM was calculated for diclofenac in the respiration assay (Table 3). Therefore, the exometabolome of the cells treated with diclofenac concentrations ranging from 0.5 μM to 100 μM was analysed (Table S4 of Supplementary material). Treatment with diclofenac concentrations up to 10 μM did not result in any change in glucose uptake or on the respiration of Hep G2 cells (Figs. 5B1 and B2). At concentrations higher than 10 μM glucose uptake decreased proportionally whereas the respiration decreased slightly (Figs. 5B1 Table 3 Test compounds used in the study and their EC50 values determined by the respiration method after 48 h of treatment (n = 4). Drugs

EC50 [μM] (range)

SC [μM]

Amiodarone-HCl Diclofenac-Na Tacrine-HCl DMSO (solvent)

10 (4–28) 146 (90–237) 137 (62–302) 3% (3–4)

0.5–10 0.5–100 1–50 0.0001–3%

The range of the EC50 was determined by Monte-Carlo simulation. The range of subtoxic concentrations (SC) used for exometabolome analysis is also shown.

and B2). The flux distribution showed an increase in TCA-cycle activity upon treatment with 10 μM and higher concentrations of diclofenac; the flux through the α-ketoglutarate dehydrogenase (v7) increased by approximately 40% (Fig. 5B3). At the same time, lactate production (v10) and therefore anaerobic energy production were decreased by up to 30% at 10 μM diclofenac exposure or even by 45% at 100 μM exposure (Fig. 5B4). Therefore, the calculated NAD(P)H and FADH2 production per cell increased with increasing concentration of the drug (Table S4 of Supplementary material). In addition, a concentration dependent decrease in the serine production (v15) and therefore an increase in the flux through the serine dehydratase (v16) were observed (Fig. 5B5). Glutamine uptake (v25) increased proportionally to the glutamate production (v2) (Figs. 5B6 and B7). The flux through glutaminase (v26) changed accordingly (Fig. 5B8).

Tacrine The subtoxic concentrations of tacrine ranged from 1 μM to 50 μM (Table 3). Exposure to these concentrations did not result in a change in glucose uptake and respiration (Figs. 5C1 and C2) indicating unchanged cell viability. However, with increasing tacrine concentration the calculated flux distribution revealed a clear decrease in the lactate production (v10), by about 50% i.e. from 157% to 103%, upon exposure to 50 μM tacrine (Fig. 5C4). An increase of the flux through the TCA-cycle was observed. The flux through the α-ketoglutarate dehydrogenase (v7), increased from 30% in the reference value to about 70% at 50 μM tacrine (Fig. 5C3). The alanine secretion (v21) and therefore also the flux through the alanine transaminase (v22) (Fig. 5C5), and the pyruvate production (v12) (Fig. 5C6), decreased with the increasing concentration of the drug. The glutamate production (v23) was slightly increased (Fig. 5C7) whereas the glutamine uptake (v25) was reduced upon exposure to 1 μM and 50 μM tacrine (Fig. 5C8).

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

Discussion Hep G2 cells have demonstrated usefulness in the detection of parent compound toxicity (Noor et al., 2009; Beckers et al., 2009). The tested drugs, amiodarone, diclofenac and tacrine, are known to have basal toxicity. It has been reported that in-vitro assays correlate well with in-vivo parent compound or basal toxicity (Ekwall, 1999) Although, few studies are reported on primary human hepatocytes, it has been suggested that for the tested drugs primary rat hepatocytes and primary human hepatocytes yield similar results for parent compound toxicity (Monteith et al., 1998; Kaufmann et al., 2005; Bort et al., 1999). Hep G2 cells could therefore, be used for the detection of parent compound toxicity and prediction of human liver toxicity potential. The EC50 values obtained in this study with the respiration method after 48 h of treatment are slightly lower than EC50 values in the literature for the tested drugs. Wang et al. (2002) observed an EC50 of 79 μM for amiodarone and 399 μM for diclofenac in Hep G2 cells after 24 h exposure. For tacrine an EC50 of 230 μM is reported using the same cell line and exposure time (Viau et al., 1993). In the presented study the cells were treated for 48 h with the test compounds under serum free conditions. It has been reported that a longer exposure to the drug and a lower amount of serum usually decreases the EC50 (Halle, 2003). The time period of 48 h was chosen as there was no oxygen or nutrient limitation during this period. In addition, there was no inhibition of growth (Fig. 2) and changes in the metabolites could be observed in subtoxic concentrations. The concentrations used in in-vitro assays are usually higher than the concentration in-vivo. Nevertheless, on a high content platform, it has been reported that in-vitro concentrations which are about 30 times higher than the Cmax could predict toxicity potential of 80% of the test compounds (O'Brien et al., 2006). Therefore, subtoxic concentrations in our in-vitro setup may correspond to the in-vivo free plasma concentrations. At these concentrations there is no manifestation of toxicity in conventional endpoint assays. We also did not observe a toxic effect in the respiration assay. However,

333

metabolic flux analyses did indicate changes at the metabolite level as discussed below. The analysis of the effects of DMSO on cellular metabolism showed that solvent concentrations lower than 0.5% should be used for the investigation of drug effects since higher DMSO concentrations had a clear influence on the growth and viability as well as on the metabolism in Hep G2 cells. In addition it is known that DMSO has multiple effects on the metabolism and toxicity of xenobiotics. DMSO has a strong effect on the expression and activity of several CYPs and can therefore influence the biotransformation of a given drug (Busby et al., 1999; Easterbrook et al., 2001; Kim et al., 2008). The drugs used in this study were dissolved in DMSO but the final concentration of DMSO did not exceed 0.1% in the samples used for the analysis of the cellular metabolism. However, it cannot be excluded that the combination of both drug and solvent caused the effects on cellular metabolism. The class III antiarrhythmic drug amiodarone is a well-studied hepatic mitochondrial toxin. It inhibits the mitochondrial β-oxidation, exerts a protonophoric uncoupling effect and inhibits enzyme complexes of the respiratory chain (Fromenty et al., 1990a, 1990b; Spaniol et al., 2001; Kaufmann et al., 2005; Waldhauser et al., 2006). The presented results indirectly indicate an inhibitory effect of amiodarone on the mitochondria. The increase in lactate production and therefore increase in anaerobic energy production support this assumption. However, this effect could only be observed by treatment with 10 μM amiodarone which represents the EC50 value. An uncoupling of the oxidative phosphorylation could not be observed at the analysed subtoxic concentrations. A partial uncoupling of the oxidative phosphorylation should lead to an increased activity of the TCA-cycle which could then compensate the lower energy production caused by the uncoupling effect. However, the flux distribution did not indicate any increase in TCA-cycle at subtoxic levels of amiodarone. Non steroidal anti-inflammatory drugs (NSAIDs), clinically prescribed for pain relief, such as diphenylamine derivative diclofenac, are associated with idiosyncratic hepatotoxicity. As reported, diclofenac treatment in hepatocytes leads to an inhibition of the ATP-

Fig. 4. Glucose uptake (A); respiration, indicated by the dissolved oxygen concentration, pO2 [% air saturation] (n = 4; B); and flux differences (Δv = vdrug − vcontrol) of the fluxes of the α-ketoglutarate dehydrogenase, v7 (C); lactate dehydrogenase, v10 (D); alanine aminotransferase, v22 (E); and glutaminase, v26 (F); after 48 h of exposure to different DMSO concentrations. Molar fluxes are normalized to the molar glucose uptake (= 100%). Mean values and standard deviations of fluxes were determined by Monte-Carlo simulation. Seeded cell density was 105 cells/well. All measured and calculated fluxes, defined in Fig. 1 and Table S1 are listed in Table S2 of the Supplementary material.

334

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

Fig. 5. Glucose uptake (A1, B1, C1); respiration, indicated by the measured dissolved oxygen concentration, pO2 [% air saturation] (A2, B2, C2, n = 4); and flux differences (Δv = vdrug − vcontrol) of the fluxes through the α-ketoglutarate dehydrogenase, v7 (A3, B3, C3); lactate dehydrogenase, v10 (A4, B4, C4); glutamate dehydrogenase, v24 (A7); malic enzyme, v62 (A8); serine dehydratase, v16 (B5); glutaminase, v26 (B8); alanine aminotransferase, v22 (C5); and of the secretion/production rates of glutamine, v25 (A5, B6, C8); glutamate, v23 (A6, B7, C7); pyruvate, v12 (C6); after 48 h of treatment with different subtoxic concentrations of drugs namely amiodarone (A), diclofenac (B) and tacrine (C). The molar fluxes are normalized to the molar glucose uptake (= 100%). Mean values and standard deviations of fluxes were determined by Monte-Carlo simulation. Seeded cell density was 105 cells/well. All measured and calculated fluxes, defined in Fig. 1 and Table S1, are listed in Tables S3–S5 of the Supplementary material.

synthesis, a decrease of the mitochondrial membrane potential and a change in the respiration indicating an uncoupling of the oxidative phosphorylation (Bort et al., 1999; Masubuchi et al., 1999). In experiments on perfused liver and isolated hepatic mitochondria of rats it was shown that diclofenac increases glycogenolysis, glycolysis and oxygen consumption (Petrescu and Tarba, 1997). However, this study was carried out for a brief period of some minutes whereas in our experimental setup a period of two days is examined. The results indicate clearly an influence of diclofenac on the mitochondria. The increase in TCA-cycle activity could be an adaption of the cellular

metabolism to an uncoupling of the oxidative phosphorylation caused by the drug. The observed increase in glutamine uptake and glutamate production at higher concentrations of diclofenac could also be caused by an increased apoptosis of the cells and therefore increased extracellular glutaminase levels. This would lead to increased extracellular hydrolytic desamination of glutamine to glutamate and would misleadingly indicate a higher uptake of glutamine and production of glutamate. However this effect would not influence the intracellular flux distribution since it would only increase the glutaminase flux from glutamine to glutamate.

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

Another drug tested in presented study is tacrine. It was reported that treatment with the acetylcholinesterase inhibitor tacrine used in the treatment of Alzheimer's disease leads to a decrease in intracellular glutathione concentration and alteration in the membrane fluidity (Galisteo et al., 2000). At higher concentrations tacrine induces a decrease in protein synthesis, intracellular ATP and glycogen content (Lagadic-Gossmann et al., 1998). The mitochondrial membrane potential is decreased while the respiration is stimulated. Tacrine exerts a protonophoric effect causing an uncoupling of the oxidative phosphorylation and therefore a decrease in the aerobic energy production (Berson et al., 1996). Our results indicate an influence of tacrine on the mitochondria in accordance to literature. The increased activity of the TCA-cycle could be caused by a partial uncoupling of the oxidative phosphorylation. This effect is compensated by an increase of the flux of pyruvate into the TCA-cycle. For all the drugs used in this study it was reported that they can cause an uncoupling of the oxidative phosphorylation. The observed increase in TCA-cycle activity could be a signature of this uncoupling effect. But interestingly this increase of the TCA-cycle was not observed in case of amiodarone at the concentrations applied in this study. This could indicate that an uncoupling of the oxidative phosphorylation is not a subtoxic effect of amiodarone and occurs only upon treatment with higher concentrations of this compound. The plasma concentrations of amiodarone, diclofenac and tacrine, depending on the dose and exposure time, range from 0.05 to 3 μM (Mukherjee et al., 2006; Johansson et al., 1996; Kashima et al., 2005). Moreover, for some drugs initially loading doses are given followed by maintenance doses as in the case of amiodarone. The plasma half lives are about 58 days for amiodarone, 1–2 h for diclofenac and 3–4 h for tacrine. In the respiration assay, low concentrations corresponding to the plasma concentrations of the test drugs did not show a toxic effect as respiration and glucose uptake was similar to the untreated control. However, metabolic flux analysis indicated changes at the metabolite level upon exposure to concentrations ranging from 0.5 to 50 μM. If a compound impairs or disrupts key processes of cell homeostasis, it has a high probability of producing the same event in-vivo. These changes may eventually in the long term lead to manifestation of the toxicity. The present study demonstrates that metabolic flux analysis is a useful tool for the investigation of the effects of drugs on cellular metabolism. The toxicity of the compound can be tracked directly using the online respiration assay and the exometabolome analysis permits the estimation of metabolic fluxes and therefore the analysis of effects of the compound on cellular metabolism. It is also possible to carry out further endpoint assays at the end of the treatment period which allows multiplexing of assays. Hence, complementary and additional information can be obtained from just one experiment in a microtiter plate. The described methodology could also be adapted to high throughput screening. The method could be improved further using 13C metabolic flux analysis which allows a deeper insight into the cellular metabolism. Modulations of fluxes could serve as toxicity indicators and specific flux patterns as signatures of toxicity. The presented systems biology approach could be applied in the future for the elucidation of side effects of compounds during the drug development process and could lead to an overall better understanding of toxicity at the metabolome level.

Acknowledgments This study was funded by the FP6 EU-STREP project “Vitrocellomics”: reducing animal experimentation in preclinical predictive drug testing by human hepatic in-vitro models derived from embryonic stem cells (LFHB-CT-2006 018940). We thank all the collaborators of this project for the joint work during the selection of test drugs. We thank Michel Fritz and Veronika Witte for their most valuable assistance for carrying out the experiments.

335

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.taap.2009.07.005.

References Aden, D.P., Fogel, A., Plotkin, S., Damjanov, I., Knowles, B.B., 1979. Controlled synthesis of HBsAg in a differentiated human liver carcinoma-derived cell line. Nature 282, 615–616. Akopia, D.P., Medh, J.D., 2006. Simultaneous isolation of total cellular lipids and RNA from cultured cells. Biotechniques 41, 426, 428, 430. Alberts, B., Bray, D., Lewis, J., Raff, M., Roberts, K., Watson, J.D., 1983. Molecular Biology of the Cell. Garland publishing Inc., New York. Antoniewicz, M.R., Kraynie, D.F., Laffend, L.A., Gonzalez-Lergier, J., Kelleher, J.K., Stephanopoulos, G., 2007. Metabolic flux analysis in a nonstationary system: fedbatch fermentation of a high yielding strain of E. coli producing 1,3-propanediol. Metab./Eng. 9, 277–292. Balcarcel, R.R., Clark, L.M., 2003. Metabolic screening of mammalian cell cultures using well-plates. Biotechnol. Prog. 19, 98–108. Becker, J., Klopprogge, C., Zelder, O., Heinzle, E., Wittmann, C., 2005. Amplified expression of fructose 1,6-bisphosphatase in Corynebacterium glutamicum increases in vivo flux through the pentose phosphate pathway and lysine production on different carbon sources. Appl. Environ. Microbiol. 71, 8587–8596. Beckers, S., Noor, F., Müller-Vieira, U., Mayer, M., Strigun, A., Heinzle, E., 2009. High throughput, non-invasive and dynamic toxicity screening on adherent cells using respiratory measurements. Toxicol. In vitro [Electronic publication ahead of print]. Berg, J.M., Tymoczko, J.L., Stryer, L., 2003. Biochemie. Spektrum Akademischer Verlag, Berlin. Berson, A., Renault, S., Letteron, P., Robin, M.A., Fromenty, B., Fau, D., Le Bot, M.A., Riche, C., Durand-Schneider, A.M., Feldmann, G., Pessayre, D., 1996. Uncoupling of rat and human mitochondria: a possible explanation for tacrine-induced liver dysfunction. Gastroenterology 110, 1878–1890. Bonarius, H.P., Hatzimanikatis, V., Meesters, K.P., de Gooijer, C.D., Schmid, G., Tramper, J., 1996. Metabolic flux analysis of hybridoma cells in different culture media using mass balances. Biotechnol. Bioeng. 50, 299–318. Bort, R., Ponsoda, X., Jover, R., Gomez-Lechon, M.J., Castell, J.V., 1999. Diclofenac toxicity to hepatocytes: a role for drug metabolism in cell toxicity. J. Pharmacol. Exp. Ther. 288, 65–72. Bradford, M.M., 1976. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254. Brandon, E.F., Raap, C.D., Meijerman, I., Beijnen, J.H., Schellens, J.H., 2003. An update on in vitro test methods in human hepatic drug biotransformation research: pros and cons. Toxicol. Appl. Pharmacol. 189, 233–246. Busby Jr., W.F., Ackermann, J.M., Crespi, C.L., 1999. Effect of methanol, ethanol, dimethyl sulfoxide, and acetonitrile on in vitro activities of cDNA-expressed human cytochromes P-450. Drug Metab. Dispos. 27, 246–249. Castell, J.V., Jover, R., Martinez-Jimenez, C.P., Gomez-Lechon, M.J., 2006. Hepatocyte cell lines: their use, scope and limitations in drug metabolism studies. Expert Opin. Drug Metab. Toxicol. 2, 183–212. Chang, C.Y., Schiano, T.D., 2007. Review article: drug hepatotoxicity. Aliment Pharmacol. Ther. 25, 1135–1151. Davila, J.C., Rodriguez, R.J., Melchert, R.B., Acosta Jr., D., 1998. Predictive value of in vitro model systems in toxicology. Annu. Rev. Pharmacol. Toxicol. 38, 63–96. Deshpande, R.R., Heinzle, E., 2004. On-line oxygen uptake rate and culture viability measurement of animal cell culture using microplates with integrated oxygen sensors. Biotechnol. Lett. 26, 763–767. Deshpande, R.R., Wittmann, C., Heinzle, E., 2004. Microplates with integrated oxygen sensing for medium optimization in animal cell culture. Cytotechnology 46, 1–8. Deshpande, R.R., Koch-Kirsch, Y., Maas, R., John, G.T., Krause, C., Heinzle, E., 2005. Microplates with integrated oxygen sensors for kinetic cell respiration measurement and cytotoxicity testing in primary and secondary cell lines. Assay Drug Dev. Technol. 3, 299–307. Deshpande, R., Yang, T.H., Heinzle, E., 2009. Towards a metabolic and isotopic steady state in CHO batch cultures for reliable isotope-based metabolic profiling. Biotechnol. J. 4, 247–263. Easterbrook, J., Lu, C., Sakai, Y., Li, A.P., 2001. Effects of organic solvents on the activities of cytochrome P450 isoforms, UDP-dependent glucuronyl transferase, and phenol sulfotransferase in human hepatocytes. Drug Metab. Dispos. 29, 141–144. Ekwall, B., 1999. Overview of the final MEIC results: II. The in vitro–in vivo evaluation, including the selection of a practical battery of cell tests for prediction of acute lethal blood concentrations in humans. Toxicol. in vitro 13, 665–673. Fromenty, B., Fisch, C., Berson, A., Letteron, P., Larrey, D., Pessayre, D., 1990a. Dual effect of amiodarone on mitochondrial respiration. Initial protonophoric uncoupling effect followed by inhibition of the respiratory chain at the levels of complex I and complex II. J. Pharmacol. Exp. Ther. 255, 1377–1384. Fromenty, B., Fisch, C., Labbe, G., Degott, C., Deschamps, D., Berson, A., Letteron, P., Pessayre, D., 1990b. Amiodarone inhibits the mitochondrial beta-oxidation of fatty acids and produces microvesicular steatosis of the liver in mice. J. Pharmacol. Exp. Ther. 255, 1371–1376. Galisteo, M., Rissel, M., Sergent, O., Chevanne, M., Cillard, J., Guillouzo, A., LagadicGossmann, D., 2000. Hepatotoxicity of tacrine: occurrence of membrane fluidity

336

J. Niklas et al. / Toxicology and Applied Pharmacology 240 (2009) 327–336

alterations without involvement of lipid peroxidation. J. Pharmacol. Exp. Ther. 294, 160–167. Halle, W., 2003. The Registry of Cytotoxicity: toxicity testing in cell cultures to predict acute toxicity (LD50) and to reduce testing in animals. Altern. Lab. Anim. 31, 89–198. Heinzle, E., Matsuda, F., Miyagawa, H., Wakasa, K., Nishioka, T., 2007. Estimation of metabolic fluxes, expression levels and metabolite dynamics of a secondary metabolic pathway in potato using label pulse-feeding experiments combined with kinetic network modelling and simulation. Plant J. 50, 176–187. Hewitt, N.J., Hewitt, P., 2004. Phase I and II enzyme characterization of two sources of HepG2 cell lines. Xenobiotica 34, 243–256. Hofmann, U., Maier, K., Niebel, A., Vacun, G., Reuss, M., Mauch, K., 2008. Identification of metabolic fluxes in hepatic cells from transient 13C-labeling experiments: Part I. Experimental observations. Biotechnol. Bioeng. 100, 344–354. Johansson, M., Hellström-Lindahl, Ewa, Nordberg, Agneta, 1996. Steady-state pharmacokinetics of tacrine in long-term treatment of Alzheimer patients. Dementia Geria. Cognit. Disorders 7 (2), 111–117. John, G.T., Klimant, I., Wittmann, C., Heinzle, E., 2003. Integrated optical sensing of dissolved oxygen in microtiter plates: a novel tool for microbial cultivation. Biotechnol. Bioeng. 81, 829–836. Kashima, A., Funahashi, M., Fukumoto, K., Komamura, K., Kamakura, S., Kitakaze, M., Uneo, K., 2005. Pharmacokinetic characteristics of amiodarone in long-term oral therapy in Japanese population. Biol. Pharm. bull. 28 (10), 1934–1938. Kaufmann, P., Török, M., Hänni, A., Roberts, P., Gasser, R., Krähenbühl, S., 2005. Mechanisms of benzarone and benzbromarone-induced hepatic toxicity. Hepatology 41, 925–935. Kiefer, P., Heinzle, E., Zelder, O., Wittmann, C., 2004. Comparative metabolic flux analysis of lysine-producing Corynebacterium glutamicum cultured on glucose or fructose. Appl. Environ. Microbiol. 70, 229–239. Kim, H.U., Kim, T.Y., Lee, S.Y., 2008. Metabolic flux analysis and metabolic engineering of microorganisms. Mol. Biosyst. 4, 113–120. Kola, I., Landis, J., 2004. Can the pharmaceutical industry reduce attrition rates? Nat. Rev. Drug Discov. 3, 711–715. Kramer, J.A., Sagartz, J.E., Morris, D.L., 2007. The application of discovery toxicology and pathology towards the design of safer pharmaceutical lead candidates. Nat. Rev. Drug Discov. 6, 636–649. Kromer, J.O., Fritz, M., Heinzle, E., Wittmann, C., 2005. In vivo quantification of intracellular amino acids and intermediates of the methionine pathway in Corynebacterium glutamicum. Anal. Biochem. 340, 171–173. Lagadic-Gossmann, D., Rissel, M., Le Bot, M.A., Guillouzo, A., 1998. Toxic effects of tacrine on primary hepatocytes and liver epithelial cells in culture. Cell Biol. Toxicol. 14, 361–373. Libourel, I.G., Shachar-Hill, Y., 2008. Metabolic flux analysis in plants: from intelligent design to rational engineering. Annu. Rev. Plant Biol. 59, 625–650. Maier, K., Hofmann, U., Reuss, M., Mauch, K., 2008. Identification of metabolic fluxes in hepatic cells from transient 13C-labeling experiments: Part II. Flux estimation. Biotechnol. Bioeng. 100, 355–370. Masubuchi, Y., Yamada, S., Horie, T., 1999. Diphenylamine as an important structure of nonsteroidal anti-inflammatory drugs to uncouple mitochondrial oxidative phosphorylation. Biochem. Pharmacol. 58, 861–865. Michal, G., 1999. Biochemical Pathways. Spektrum Akademischer Verlag, Heidelberg. Monteith, D.K., Theiss, J.C., Haskins, J.R., dela Iglesia, F.A., 1998. Functional and subcellular organelle changes in isolated rat and human hepatocytes induced by tetrahydroaminoacridine. Arch. Toxicol. 72, 147–156. Mukherjee, B., Mahapatra, S., Das, S., Roy, G., Dey, S., 2006. HPLC detection of plasma concentrations of diclofenac in human volunteers administered with povidoneethylcellulose-based experimental transdermal matrix-type patches. Methods Find Exp. Clin. Pharmacol. 28 (5), 301–306. Noor, F., Niklas, J., Müller-Viera, U., Heinzle, E., 2009. An integrated approach to improved toxicity prediction for the safety assessment during preclinical drug development using Hep G2 cells. Toxicol. Appl. Pharmacol. 237, 221–231. O'Brien, P.J., Irwin, W., Diaz, D., Howard-Cofield, E., Krejsa, C.M., Slaughter, M.R., Gao, B., Kaludercic, N., Angeline, A., Bernardi, P., Brain, P., Hougham, C., 2006. High concordance of drug-induced human hepatotoxicity with in vitro cytotoxicity measured in a novel cell-based model using high content screening. Arch. Toxicol. 80, 580–604. Olson, H., Betton, G., Robinson, D., Thomas, K., Monro, A., Kolaja, G., Lilly, P., Sanders, J., Sipes, G., Bracken, W., Dorato, M., Van Deun, K., Smith, P., Berger, B., Heller, A., 2000. Concordance of the toxicity of pharmaceuticals in humans and in animals. Regul. Toxicol. Pharmacol. 32, 56–67. Ostapowicz, G., Lee, W.M., 2000. Acute hepatic failure: a Western perspective. J. Gastroenterol. Hepatol. 15, 480–488. Petrescu, I., Tarba, C., 1997. Uncoupling effects of diclofenac and aspirin in the perfused liver and isolated hepatic mitochondria of rat. Biochim. Biophys. Acta 1318, 385–394. Sahu, S.C., 2007. Hepatotoxicity: From Genomics to In Vitro and In Vivo Models. John Wiley & Sons Ltd. Sauer, U., 2004. High-throughput phenomics: experimental methods for mapping fluxomes. Curr. Opin. Biotechnol. 15, 58–63.

Sauer, U., 2006. Metabolic networks in motion: 13C-based flux analysis. Mol. Syst. Biol. 2, 62. Savinell, J.M., Palsson, B.O., 1992. Network analysis of intermediary metabolism using linear optimization. I. Development of mathematical formalism. J. Theor. Biol. 154, 421–454. Schuster, D., Laggner, C., Langer, T., 2005. Why drugs fail—a study on side effects in new chemical entities. Curr. Pharm. Des. 11, 3545–3559. Skehan, P., Storeng, R., Scudiero, D., Monks, A., McMahon, J., Vistica, D., Warren, J.T., Bokesch, H., Kenney, S., Boyd, M.R., 1990. New colorimetric cytotoxicity assay for anticancer-drug screening. J. Natl. Cancer Inst. 82, 1107–1112. Spaniol, M., Bracher, R., Ha, H.R., Follath, F., Krahenbuhl, S., 2001. Toxicity of amiodarone and amiodarone analogues on isolated rat liver mitochondria. J. Hepatol. 35, 628–636. Stephanopoulos, G.N., Aristidou, A.A., Nielsen, J., 1998. Metabolic Engineering: Principles and Methodologies. Academic Press. Velagapudi, V.R., Wittmann, C., Schneider, K., Heinzle, E., 2007. Metabolic flux screening of Saccharomyces cerevisiae single knockout strains on glucose and galactose supports elucidation of gene function. J. Biotechnol. 132, 395–404. Vermeir, M., Annaert, P., Mamidi, R.N., Roymans, D., Meuldermans, W., Mannens, G., 2005. Cell-based models to study hepatic drug metabolism and enzyme induction in humans. Expert Opin. Drug Metab. Toxicol. 1, 75–90. Viau, C.J., Curren, R.D., Wallace, K., 1993. Cytotoxicity of tacrine and velnacrine metabolites in cultured rat, dog and human hepatocytes. Drug Chem. Toxicol. 16, 227–239. Vo, T.D., Palsson, B.O., 2006. Isotopomer analysis of myocardial substrate metabolism: a systems biology approach. Biotechnol. Bioeng. 95, 972–983. Waldhauser, K.M., Torok, M., Ha, H.R., Thomet, U., Konrad, D., Brecht, K., Follath, F., Krahenbuhl, S., 2006. Hepatocellular toxicity and pharmacological effect of amiodarone and amiodarone derivatives. J. Pharmacol. Exp. Ther. 319, 1413–1423. Wang, K., Shindoh, H., Inoue, T., Horii, I., 2002. Advantages of in vitro cytotoxicity testing by using primary rat hepatocytes in comparison with established cell lines. J. Toxicol. Sci. 27, 229–237. Wilkening, S., Stahl, F., Bader, A., 2003. Comparison of primary human hepatocytes and hepatoma cell line Hepg2 with regard to their biotransformation properties. Drug Metab. Dispos. 31, 1035–1042. Wittmann, C., 2007. Fluxome analysis using GC-MS. Microb. Cell Fact 6, 6. Wittmann, C., Heinzle, E., 2002. Genealogy profiling through strain improvement by using metabolic network analysis: metabolic flux genealogy of several generations of lysine-producing corynebacteria. Appl. Environ. Microbiol. 68, 5843–5859. Wittmann, C., Kim, H.M., Heinzle, E., 2004. Metabolic network analysis of lysine producing Corynebacterium glutamicum at a miniaturized scale. Biotechnol. Bioeng. 87, 1–6. Xie, L., Wang, D.I., 1994. Applications of improved stoichiometric model in medium design and fed-batch cultivation of animal cells in bioreactor. Cytotechnology 15, 17–29. Yoo, H., Antoniewicz, M.R., Stephanopoulos, G., Kelleher, J.K., 2008. Quantifying reductive carboxylation flux of glutamine to lipid in a brown adipocyte cell line. J. Biol. Chem. 283, 20621–20627. Zupke, C., Sinskey, A.J., Stephanopoulos, G., 1995. Intracellular flux analysis applied to the effect of dissolved oxygen on hybridomas. Appl. Microbiol. Biotechnol. 44, 27–36.

Glossary1 AcCoA: acetyl coenzyme A AKG: α-ketoglutarate Carbo: carbohydrates CYP: cytochrome P 450 DMSO: dimethylsulphoxide EC50: effective concentration where 50% effect is observed F6P: fructose-6-phosphate FCS: fetal calf serum Fum: fumarate G6P: glucose-6-phosphate GAP: glyceraldehyde-3-phosphate Glc: glucose Lac: lactate NAD+/NADH: nicotinamide adenine dinucleotide ox/red NADP+/NADPH: nicotinamide adenine dinucleotide phosphate ox/red OAA: oxaloacetate PBS: phosphate buffered saline PPP: pentose phosphate pathway Pyr: pyruvate SC: subtoxic concentrations SuCoA: succinyl coenzyme A TCA: tricarboxylic acid WME: Williams Medium E

1

Common abbreviations, e.g. of amino acids are not listed.