Application of microphysiological systems for nonclinical evaluation of cell therapies

Microphysiological systems (MPS) are gaining broader application in the pharmaceutical industry but have primarily been leveraged in early discovery toxicology and pharmacology studies with small molecules. The adoption of MPS offers a promising avenue to reduce animal use, improve in-vitro-to-in-vivo translation of pharmacokinetics/pharmacodynamics and toxicity correlation, and provide mechanistic understanding of model species suitability. While MPS have demonstrated utility in these areas with small molecules and biologics, MPS models in cell therapy development have not been fully explored, let alone validated. Distinguishing features of MPS, including long-term viability and physiologically relevant expression of functional enzymes, receptors, and pharmacological targets make them attractive tools for nonclinical characterization. However, there is currently limited published evidence of MPS being utilized to study the disposition, metabolism, pharmacology, and toxicity profiles of cell therapies. This review provides an industry perspective on the nonclinical application of MPS on cell therapies, first with a focus on oncology applications followed by examples in regenerative medicine.

Microphysiological systems (MPS) are advanced cell models, applied in the pharmaceutical industry to characterize novel therapies. While their application in studies of small molecule ther­apies has been very successful, the use of these models to study cell therapies has been limited. Cell therapies consist of cells and are living drugs, often with complex biological mechanisms of action, which can be very challenging to study. However, MPS have several features that make them attractive for studying cell therapies, including possibilities for longer-term studies and the ability to mimic physiologically relevant biological functions. MPS can mimic complex biological systems and processes, as such, the adoption of MPS offers a promising avenue to reduce the use of animals in the characterization of novel therapies. This review provides an industry perspective on current chal­lenges and highlights opportunities for using MPS in the development of cell therapies.

Evaluation of Erythromycin as a Tool to Assess CYP3A Contribution of Low Clearance Compounds in a Long-Term Hepatocyte Culture.

Long-term hepatocyte culture systems such as HepatoPac are well suited to evaluate the metabolic turnover of low clearance (CL) drugs because of their sustained metabolic capacity and longer-term viability. Erythromycin (ERY), a moderate, mechanism-based inhibitor of CYP3A, was evaluated as a tool in the HepatoPac model to assess contribution of CYP3A to the clearance of drug candidates. ERY inhibited CYP3A activity by 58% and 80% at 3 and 10 µM, respectively, for up to 72 hours. At 30 µM, ERY inhibited midazolam hydroxylation by >85% for the entire 144-hour duration of the incubation. Alprazolam CLint was inhibited 58% by 3 µM of ERY, 75% by 15 µM of ERY, 89% by 30 µM of ERY, and 94% by 60 µM of ERY. ERY (30 µM) did not markedly affect CLint of substrates for several other major cytochrome P450 isoforms evaluated and did not markedly inhibit uridine diphosphoglucuronosyl transferase (UGT) isoforms 1A1, 1A3, 1A4, 1A6, 1A9, 2B7, or 2B15 as assessed using recombinant UGTs. ERY only mildly increased CYP3A4 gene expression by 2.1-fold (14% of rifampicin induction) at 120 µM, indicating that at effective concentrations for inhibition of CYP3A activity (30-60 µM), arylhydrocarbon receptor, constitutive androstane receptor, and pregnane-X-receptor activation are not likely to markedly increase levels of other drug-metabolizing enzymes or transporters. ERY at concentrations up to 60 µM was not toxic for up to 6 days of incubation. Use of ERY to selectively inhibit CYP3A in high-functioning, long-term hepatocyte models such as HepatoPac can be a valuable strategy to evaluate the contribution of CYP3A metabolism to the overall clearance of slowly metabolized drug candidates. SIGNIFICANCE STATEMENT: This work describes the use of erythromycin as a selective inhibitor of CYP3A to assess the contribution of CYP3A in the metabolism of compounds using long-term hepatocyte cultures.

Upregulation of Krebs cycle and anaerobic glycolysis activity early after onset of liver ischemia.

The liver is a highly vascularized organ receiving a dual input of oxygenated blood from the hepatic artery and portal vein. The impact of decreased blood flow on glucose metabolism and how hepatocytes could adapt to this restrictive environment are still unclear. Using the left portal vein ligation (LPVL) rat model, we found that cellular injury was delayed after the onset of liver ischemia. We hypothesized that a metabolic adaptation by hepatocytes to maintain energy homeostasis could account for this lag phase. Liver glucose metabolism was characterized by 13C- and 1H-NMR spectroscopy and analysis of high-energy metabolites. ALT levels and caspase 3 activity in LPVL animals remained normal during the first 12 h following surgery (P<0.05). Ischemia rapidly led to decreased intrahepatic tissue oxygen tension and blood flow (P<0.05) and increased expression of Hypoxia-inducible factor 1-alpha. Intrahepatic glucose uptake, ATP/ADP ratio and energy charge level remained stable for up to 12 h after ligation. Entry of glucose in the Krebs cycle was impaired with lowered incorporation of 13C from [U-13C]glucose into glutamate and succinate from 0.25 to 12 h after LPVL. However, total hepatic succinate and glutamate increased 6 and 12 h after ischemia (P<0.05). Glycolysis was initially reduced (P<0.05) but reached maximum 13C-lactate (P<0.001) and 13C-alanine (P<0.01) enrichments 12 h after LPVL. In conclusion, early liver homeostasis stems from an inherent ability of ischemic hepatocytes to metabolically adapt through increased Krebs cycle and glycolysis activity to preserve bioenergetics and cell viability. This metabolic plasticity of hepatocytes could be harnessed to develop novel metabolic strategies to prevent ischemic liver damage.

Bridging in vitro and in vivo metabolism and transport of faldaprevir in human using a novel cocultured human hepatocyte system, HepatoPac.

An increased appreciation of the importance of transporter and enzyme interplay in drug clearance and a desire to delineate these mechanisms necessitates the utilization of models that contain a full complement of enzymes and transporters at physiologically relevant activities. Additionally, the development of drugs with longer half-lives requires in vitro systems with extended incubation times that allow characterization of metabolic pathways for low-clearance drugs. A recently developed coculture hepatocyte model, HepatoPac, has been applied to meet these challenges. Faldaprevir is a drug in late-stage development for the treatment of hepatitis C. Faldaprevir is a low-clearance drug with the somewhat unique characteristic of being slowly metabolized, producing two abundant hydroxylated metabolites (M2a and M2b) in feces (∼40% of the dose) without exhibiting significant levels of circulating metabolites in humans. The human HepatoPac model was investigated to characterize the metabolism and transport of faldaprevir. In human HepatoPac cultures, M2a and M2b were the predominant metabolites formed, with extents of formation comparable to in vivo. Direct glucuronidation of faldaprevir was shown to be a minor metabolic pathway. HepatoPac studies also demonstrated that faldaprevir is concentrated in liver with active uptake by multiple transporters (including OATP1B1 and Na(+)-dependent transporters). Overall, human HepatoPac cultures provided valuable insights into the metabolism and disposition of faldaprevir in humans and demonstrated the importance of enzyme and transporter interplay in the clearance of the drug.

Hepatocellular apoptosis in mice is associated with early upregulation of mitochondrial glucose metabolism.

Hepatocyte death due to apoptosis is a hallmark of almost every liver disease. Manipulation of cell death regulatory steps during the apoptotic process is therefore an obvious goal of biomedical research. To clarify whether metabolic changes occur prior to the characteristic apoptotic events, we used ex vivo multinuclear NMR-spectroscopy to study metabolic pathways of [U-(13)C]glucose in mouse liver during Fas-induced apoptosis. We addressed whether these changes could be associated with protection against apoptosis afforded by Epidermal Growth Factor (EGF). Our results show that serum alanine and aspartate aminotransferase levels, caspase-3 activity, BID cleavage and changes in cellular energy stores were not observed before 3 h following anti-Fas injection. However, as early as 45 min after anti-Fas treatment, we observed upregulation of carbon entry (i.e. flux) from glucose into the Krebs-cycle via pyruvate dehydrogenase (PDH) and pyruvate carboxylase (PC) (up to 139% and 123% of controls, respectively, P < 0.001). This was associated with increased glutathione synthesis. EGF treatment significantly attenuated Fas-induced apoptosis, liver injury and the late decrease in energy stores, as well as the early fluxes through PDH and PC which were comparable to untreated controls. Using ex vivo multinuclear NMR-spectroscopic analysis, we have shown that Fas receptor activation in mouse liver time-dependently affects specific metabolic pathways of glucose. These early upregulations in glucose metabolic pathways occur prior to any visible signs of apoptosis and may have the potential to contribute to the initiation of apoptosis by maintaining mitochondrial energy production and cellular glutathione stores.

Copper-catalyzed ascorbate oxidation results in glyoxal/AGE formation and cytotoxicity.

Previously we showed that 10 muM glyoxal compromised hepatocyte resistance to hydrogen peroxide (H(2)O(2)) by increasing glutathione (GSH) and NADPH oxidation and decreasing mitochondrial membrane potential (MMP) before cytotoxicity ensued. Since transition metal-catalyzed oxidation of ascorbate (Asc) has been shown to result in the generation of both glyoxal and H(2)O(2), we hypothesized that glyoxal formation during this process compromises hepatocyte resistance to H(2)O(2). We used isolated rat hepatocytes and incubated them with Asc/copper and measured cytotoxicity, glyoxal levels, H(2)O(2), GSH levels, and MMP. To investigate the role of Asc/copper on glyoxal-BSA adducts, we measured the appearance of advanced glycation end-products (AGE) in the presence and absence of catalase or aminoguanidine (AG). Asc/copper increased glyoxal and H(2)O(2) formation. Hepatocyte GSH levels were decreased and cytotoxicity ensued after a collapse of the hepatocyte MMP. Glyoxal traps protected hepatocytes against Asc/copper-induced cytotoxicity. In cell-free studies with BSA, incubation with Asc and copper resulted in glyoxal-hydroimidazolone formation, which was decreased by both AG and catalase. To the best of our knowledge, this is the first study that illustrates the importance of glyoxal production by transition metal-catalyzed Asc autoxidation. Understanding this mechanism of toxicity could lead to the development of novel copper chelating drug therapies to treat diabetic complications.

Glyoxal markedly compromises hepatocyte resistance to hydrogen peroxide.

Glyoxal is an interesting endogenous alpha-oxoaldehyde as it originates from pathways that have been linked to various pathologies, including lipid peroxidation, DNA oxidation and glucose autoxidation. In our previous study we showed that the LD(50) of glyoxal towards isolated rat hepatocytes was 5mM. However, 10microM glyoxal was sufficient to overcome hepatocyte resistance to H(2)O(2)-mediated cytotoxicity. Hepatocyte GSH oxidation, NADPH oxidation, reactive oxygen species formation, DNA oxidation, protein carbonylation and loss of mitochondrial potential were also markedly increased before cytotoxicity ensued. Cytotoxicity was prevented by glyoxal traps, the ferric chelator, desferoxamine, and antioxidants such as quercetin and propyl gallate. These results suggest there is a powerful relationship between H(2)O(2)-induced oxidative stress and glyoxal which involves an inhibition of the NADPH supply by glyoxal resulting in cytotoxicity caused by H(2)O(2)-induced mitochondrial oxidative stress.

The biosynthesis of ascorbate protects isolated rat hepatocytes from cumene hydroperoxide-mediated oxidative stress.

Most animals synthesize ascorbate. It is an essential enzymatic cofactor for the synthesis of a variety of biological molecules and also a powerful antioxidant. There is, however, little direct evidence supporting an antioxidant role for endogenously produced ascorbate. Recently, we demonstrated that incubation of rat hepatocytes with 1-bromoheptane or phorone simultaneously depleted glutathione (GSH) and triggered rapid ascorbate synthesis. The present study investigates the hypothesis that endogenous ascorbate synthesis can confer protection against oxidative stress. Rat and guinea pig hepatocytes were depleted of GSH with 1-bromoheptane and subsequently treated with the oxidative stressor cumene hydroperoxide (CHP) in the presence or absence of the ascorbate synthesis inhibitor sorbinil. In rat hepatocytes, ascorbate content increased linearly (from 15.1 to 35.8 nmol/10(6) cells) over a 105-min incubation. Prior depletion of GSH increased CHP-induced cellular reactive oxygen species (ROS) production, lipid peroxidation, and cell death in rat and guinea pig hepatocytes. Inhibiting ascorbate synthesis, however, further elevated ROS production (2-fold), lipid peroxidation (1.5-fold), and cell death (2-fold) in rat hepatocytes only. This is the first time that endogenous ascorbate synthesis has been shown to decrease cellular susceptibility to oxidative stress. Protection by endogenously produced ascorbate may therefore need to be addressed when extrapolating data to humans from experiments using rodents capable of synthesizing ascorbate.

Application of quantitative structure-toxicity relationships for the comparison of the cytotoxicity of 14 p-benzoquinone congeners in primary cultured rat hepatocytes versus PC12 cells.

Quinones are believed to induce their toxicity by two main mechanisms: oxygen activation by redox cycling and alkylation of essential macromolecules. The physicochemical parameters that underlie this activity have not been elucidated, although redox potential is believed to play a significant role. In this study, we have evaluated the cytotoxicity, formation of reactive oxygen species (ROS), and the glutathione (GSH) depleting ability of 14 p-benzoquinone congeners in primary rat hepatocyte and PC12 cell cultures. All experiments were performed under identical conditions (37 degrees C, 5% CO2/air) in 96-well plates. The most cytotoxic quinone was found to be tetrachloro-p-benzoquinone (chloranil), and the least toxic was duroquinone or 2,6-di-tert-butyl-p-benzoquinone. The cytotoxic order varied between the cell types, and in particular, the di-substituted methoxy or methyl p-benzoquinones were particularly more cytotoxic towards PC12 cells. We have derived one- and two-parameter quantitative structure-toxicity relationships (QSTRs) which revealed that the most cytotoxic quinones had the highest electron affinity and the smallest volume. Cytotoxicity did not correlate with the lipophilicity of the quinone. Furthermore, we found that p-benzoquinone cytotoxicity correlated well with hepatocyte ROS formation and GSH depletion, whereas in PC12 cells, cytotoxicity did not correlate with ROS formation and somewhat correlated with GSH depletion. Hepatocytes had far greater hydrogen peroxide detoxifying capacity than PC12 cells, but PC12 cells contained more GSH/mg protein. Thus, p-benzoquinone-induced ROS formation was greater towards PC12 cells than with hepatocytes. To our knowledge, this is the first QSTR derived for p-benzoquinone cytotoxicity in these cell types and could form the basis for distinguishing certain cell-specific cytotoxic mechanisms.

Glycogenolysis is directed towards ascorbate synthesis by glutathione conjugation.

Using isolated rat hepatocytes we have shown that glutathione (GSH) depletion by glutathione-S-transferase (GST)-catalyzed conjugation with 1-bromoheptane or phorone was accompanied by a significant elevation in ascorbate synthesis. Glycogenolysis was also stimulated without a significant rise in glucose synthesis. Furthermore, when glycogenolysis was stimulated in control hepatocytes by increasing intracellular cAMP levels (with glucagon or dibutyryl cAMP), cellular glucose levels, but not ascorbate levels, increased. These data suggest that GSH depletion can stimulate ascorbate synthesis independently of glucose synthesis and that hepatocytes can direct glycogenolysis towards ascorbate synthesis during GSH conjugation.

Effects of phosphodiesterase 3,4,5 inhibitors on hepatocyte cAMP levels, glycogenolysis, gluconeogenesis and susceptibility to a mitochondrial toxin.

Various phosphodiesterase (PDE) 3,4 and 5 inhibitors have been compared with glucagon for their effectiveness at increasing hepatocyte cAMP, glycogenolysis and gluconeogenesis. Preincubation of isolated hepatocytes with PDE 3 and 4 inhibitors (50 microM) for 2 h induced significant increases in cellular cAMP level. The order of effectiveness was: glucagon (78%), V11294A (42%), rolipram (40%), milrinone (36%), CDP-840 (33%), R(0) 20-1724 (31%), papaverine (27%), isobutylmethylxanthine (28%), isoliquiritigenin (25%), theophylline (22%), and amrinone (22%). The PDE 5 inhibitors dipyridamol and sildenafil had only a slight effect on cAMP levels. Glucose formation was increased as a result of increased glycogenolysis in the following order of effectiveness: glucagon (89%), V11294A (63%), rolipram (61%), milrinone (50%), CDP-840 (46%), R(0) 20-1724 (45%), sildenafil (34%), dipyridamol (31%), papaverine (30%), isobutylmethylxanthine (29%), theophylline (20%), amrinone (20%), and isoliquiritigenin (20%). Rolipram and milrinone, selective PDE 4 and PDE 3 inhibitors respectively, stimulated the gluconeogenesis of alanine, lactate + pyruvate, or fructose in hepatocytes isolated from fasted rats. On the other hand, selective cGMP specific phospodiesterase inhibitors, sildenafil and dipyridamol inhibited alanine-induced gluconeogenesis. All PDE inhibitors increased hepatocyte susceptibility to cyanide toxicity (3-4 fold) which was prevented by fructose whereas PDE 5 inhibitors did not significantly increase hepatocyte susceptibility.

Simultaneous detection of the antioxidant and pro-oxidant activity of dietary polyphenolics in a peroxidase system.

The ability to reduce the peroxidase (myeloglobin/H2O2)-generated ABTS*+ [2,2'-azinobis-(3-ethylbenzthiazoline-6-sulfonic acid) radical cation] has been used to rank the antioxidant activity of various agents including dietary flavonoids and chalcones. Surprisingly, we found that in the presence of catalytic concentrations of the phenol B-ring containing flavonoids, apigenin, naringenin and the chalcone phloretin, the formation of the ABTS*+ was initially increased. The enhanced formation of the ABTS*+ was attributed to the peroxidase/H2O2 mediated generation of polyphenolic phenoxyl radicals that were able to co-oxidize ABTS. The relative ABTS*+ generating ability of these dietary polyphenolics correlated with their ability to co-oxidize NADH to the NAD* radical with the resultant generation of superoxide. This pro-oxidant activity was not observed for either luteolin or eriodyctiol, which are B-ring catecholic analogues of apigenin and naringenin, respectively, suggesting that these antioxidants are incapable of the transition metal-independent generation of reactive oxygen species. This pro-oxidant activity of the polyphenolics therefore needs to be taken into account when quantifying antioxidant activity.

Idiosyncratic NSAID drug induced oxidative stress.

Many idiosyncratic non-steroidal anti-inflammatory drugs (NSAIDs) cause GI, liver and bone marrow toxicity in some patients which results in GI bleeding/ulceration/fulminant hepatic failure/hepatitis or agranulocytosis/aplastic anemia. The toxic mechanisms proposed have been reviewed. Evidence is presented showing that idiosyncratic NSAID drugs form prooxidant radicals when metabolised by peroxidases known to be present in these tissues. Thus GSH, NADH and/or ascorbate were cooxidised by catalytic amounts of NSAIDs and hydrogen peroxide in the presence of peroxidase. During GSH and NADH cooxidation, oxygen uptake and activation occurred. Furthermore the formation of NSAID oxidation products was prevented during the cooxidation indicating that the cooxidation involved redox cycling of the first formed NSAID radical product. The order of prooxidant catalytic effectiveness of fenamate and arylacetic acid NSAIDs was mefenamic acid>tolfenamic acid>flufenamic acid, meclofenamic acid or diclofenac. Diphenylamine, a common moiety to all of these NSAIDs was a more active prooxidant for NADH and ascorbate cooxidation than these NSAIDs which suggests that oxidation of the NSAID diphenylamine moiety to a cation and/or nitroxide radical was responsible for the NSAID prooxidant activity. The order of catalytic effectiveness found for sulfonamide derivatives was sulfaphenazole>sulfisoxazolez.Gt;dapsone>sulfanilic acid>procainamide>sulfamethoxazole>sulfadiazine>sulfadimethoxine whereas sulfanilamide, sulfapyridine or nimesulide had no prooxidant activity. Although indomethacin had little prooxidant activity, its major in vivo metabolite, N-deschlorobenzoyl indomethacin had significant prooxidant activity. Aminoantipyrine the major in vivo metabolite of aminopyrine or dipyrone was also more prooxidant than the parent drugs. It is hypothesized that the NSAID radicals and/or the resulting oxidative stress initiates the cytotoxic processes leading to idiosyncratic toxicity.

N-oxidation of aromatic amines by intracellular oxidases.

The introduction includes a literature review of DNA reactive species and DNA adduct formation that results from aromatic amine N-oxidation catalyzed by hepatic cytochrome P450 vs. that catalyzed by nonhepatic peroxidases. Experimental evidence is then described for a novel oxidative stress mechanism involving prooxidant N-cation radical formation by both oxidases, which is proposed as a contributing mechanism for aromatic amine induced cytotoxicity and carcinogenesis. Aromatic amine N-cation radicals formed by peroxidases were found to cooxidize GSH or NADH and form reactive oxygen species. The latter could explain the reported DNA oxidative damage found in vivo following methylaminoazobenzene administration [Hirano et al. Analyses of Oxidative DNA Damage and Its Repair Activity in the Livers of 3'-Methyl-4-dimethylaminoazobenzene-Treated Rodents. Jpn. J. Cancer Res. 2000, 91, 681-685]. It was also found that the prooxidant activity of the aromatic amine increased as its redox potential, i.e., ease of oxidation decreased with o-anisidine and aminofluorene being the most effective at forming reactive oxygen species. This suggests that the rate-limiting step in the cooxidation is the rate of arylamine oxidation by the peroxidase. Incubation of hepatocytes with aromatic amines caused a decrease in the mitochondrial membrane potential before cytotoxicity ensued. The CYP1A2-induced hepatocytes isolated from 3-methylcholanthrene administered rats were much more susceptible to some arylamines and were protected by CYP1A2 inhibitors. Hepatocyte GSH was also depleted by all arylamines tested and extensive GSH oxidation occurred with o-anisidine and aminofluorene, which was prevented by CYP1A2 inhibitors. This suggests that in intact hepatocytes CYP1A2 may also catalyze a one-electron oxidation of some arylamines to form prooxidant cation radicals, which cooxidize GSH to form the reactive oxygen species.

Coenzyme Q cytoprotective mechanisms for mitochondrial complex I cytopathies involves NAD(P)H: quinone oxidoreductase 1(NQO1).

The commonest mitochondrial diseases are probably those impairing the function of complex I of the respiratory electron transport chain. Such complex I impairment may contribute to various neurodegenerative disorders e.g. Parkinson's disease. In the following, using hepatocytes as a model cell, we have shown for the first time that the cytotoxicity caused by complex I inhibition by rotenone but not that caused by complex III inhibition by antimycin can be prevented by coenzyme Q (CoQ1) or menadione. Furthermore, complex I inhibitor cytotoxicity was associated with the collapse of the mitochondrial membrane potential and reactive oxygen species (ROS) formation. ROS scavengers or inhibitors of the mitochondrial permeability transition prevented cytotoxicity. The CoQ1 cytoprotective mechanism required CoQ1 reduction by DT-diaphorase (NQO1). Furthermore, the mitochondrial membrane potential and ATP levels were restored at low CoQ1 concentrations (5 microM). This suggests that the CoQ1H2 formed by NQO1 reduced complex III and acted as an electron bypass of the rotenone block. However cytoprotection still occurred at higher CoQ1 concentrations (>10 microM), which were less effective at restoring ATP levels but readily restored the cellular cytosolic redox potential (i.e. lactate: pyruvate ratio) and prevented ROS formation. This suggests that CoQ1 or menadione cytoprotection also involves the NQO1 catalysed reoxidation of NADH that accumulates as a result of complex I inhibition. The CoQ1H2 formed would then also act as a ROS scavenger.

Endogenous and endobiotic induced reactive oxygen species formation by isolated hepatocytes.

The rat hepatocyte catalyzed oxidation of 2',7'-dichlorofluorescin to form the fluorescent 2,7'-dichlorofluorescein was used to measure endogenous and xenobiotic-induced reactive oxygen species (ROS) formation by intact isolated rat hepatocytes. Various oxidase substrates and inhibitors were then used to identify the intracellular oxidases responsible. Endogenous ROS formation was markedly increased in catalase-inhibited or GSH-depleted hepatocytes, and was inhibited by ROS scavengers or desferoxamine. Endogenous ROS formation was also inhibited by cytochrome P450 inhibitors, but was not affected by oxypurinol, a xanthine oxidase inhibitor, or phenelzine, a monoamine oxidase inhibitor. Mitochondrial respiratory chain inhibitors or hypoxia, on the other hand, markedly increased ROS formation before cytotoxicity ensued. Furthermore, uncouplers of oxidative phosphorylation inhibited endogenous ROS formation. This suggests endogenous ROS formation can largely be attributed to oxygen reduction by reduced mitochondrial electron transport components and reduced cytochrome P450 isozymes. Addition of monoamine oxidase substrates increased antimycin A-resistant respiration and ROS formation before cytotoxicity ensued. Addition of peroxisomal substrates also increased antimycin A-resistant respiration but they were less effective at inducing ROS formation and were not cytotoxic. However, peroxisomal substrates readily induced ROS formation and were cytotoxic towards catalase-inhibited hepatocytes, which suggests that peroxisomal catalase removes endogenous H(2)O(2) formed in the peroxisomes. Hepatocyte catalyzed dichlorofluorescin oxidation induced by oxidase substrates, e.g., benzylamine, was correlated with the cytotoxicity induced in catalase-inhibited hepatocytes.