ToxSci Advance Access originally published online on January 25, 2007
Toxicological Sciences 2007 96(2):294-309; doi:10.1093/toxsci/kfm009
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Underlying Mechanisms of Pharmacology and Toxicity of a Novel PPAR Agonist Revealed Using Rodent and Canine Hepatocytes
* Department of Investigative Toxicology
Integrative Biology
Department of Pathology
Statistics and Information Science, Lilly Research Laboratories, Divisions of Eli Lilly and Company, Greenfield, Indiana 46140
1 To whom correspondence should be addressed. Fax: (317) 277-6770. E-mail: timryan{at}lilly.com.
Received December 22, 2006; accepted January 15, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Marked species-specific responses to agonists of the peroxisome proliferator-activated
receptor (PPAR) have been observed in rats and dogs, two species typically used to assess the potential human risk of pharmaceuticals in development. In this study, we used primary cultured rat and dog hepatocytes to investigate the underlying mechanisms of a novel PPAR and -
coagonist, LY465608, relative to fenofibrate, a prototypical PPAR agonist. As expected, rat hepatocytes incubated with these two agonists demonstrated an increase in peroxisome number as evaluated by electron microscopy, whereas the peroxisome number remained unchanged in dog hepatocytes. Biochemical analysis showed that rat hepatocytes responded to PPAR agonists with an induction of both peroxisomal and mitochondrial ß-oxidation (PBox and MBox) activities. Dog hepatocytes treated with both PPAR agonists, however, did not show increased PBox activity but did demonstrate increased MBox activity. Analysis of peroxisomal ß-oxidation gene expression markers by quantitative real-time PCR confirmed that PPAR agonists induced the peroxisomal enzymes, acyl-coenzyme A (CoA) oxidase (Acox), enoyl-CoA hydratase/L-3-hydroxyacyl–CoA dehydrogenase (Ehhadh), and 3-ketoacyl–CoA thiolase (Acaa1) at the transcriptional level in rat hepatocytes, but not dog hepatocytes. Expression of mRNA for the mitochondrial ß-oxidation gene hydroxyacyl-CoA dehydrogenase/3-ketoacyl–CoA thiolase (Hadhb), however, increased in both rat and dog hepatocytes, consistent with biochemical measurements of peroxisomal and mitochondrial ß-oxidation. Repeat-dose nonclinical safety studies of LY465608 revealed abnormities in mitochondrial morphology and evidence of single-cell necrosis following 30 days of dosing exclusively in dogs, but not in rats. Microarray analysis indicated that dog hepatocytes, but not rat hepatocytes, treated with LY465608 had an expression profile consistent with abnormalities in the regulation of cell renewal and death, oxidative stress, and mitochondrial bioenergetics, which may explain the canine-specific toxicity observed in vivo with this compound. This increased sensitivity to mitochondrial toxicity of canine hepatocytes relative to rat hepatocytes identified using gene expression was confirmed using the fluorescent indicator tetramethylrhodamine ethyl ester (TMRE) and flow cytometry. At doses of 0.1µM LY465608, canine hepatocytes showed a greater shift in fluorescence indicative of mitochondrial damage than observed with rat hepatocytes treated at 10µM. In summary, using rat and dog primary hepatocytes, we replicated the pharmacologic and toxicologic effects of LY465608 observed in vivo during preclinical development and propose an underlying mechanism for these species-specific effects. Key Words: PPAR agonist; hepatocyte; mitochondrial ß-oxidation; peroxisomal ß-oxidation; toxicogenomics; gene expression; LY465608; fenofibrate; Biomarker.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Peroxisome proliferators are a diverse group of chemicals that induce pleiotropic responses in the liver of rodents, including a dramatic increase in the number of peroxisomes in hepatocytes and an increase in liver weight (Lalwani et al., 1987
). After searching for receptors that modulate this response, a novel member of the nuclear hormone receptor superfamily, subsequently named the peroxisome proliferator-activated receptor (PPAR), was found to be responsible for these effects (Chen et al., 1993; Dreyer et al., 1992; Issemann and Green, 1990; Lalwani et al., 1987). The PPAR family includes three receptor subtypes, PPAR, -ß/
, and -. The PPAR subtype is highly expressed in liver, where it controls peroxisomal and mitochondrial fatty acid catabolism (Despres et al., 2004; Klaunig et al., 2003). PPAR functions as a ligand-activated transcription factor that regulates these and other metabolic processes largely via target gene transcription. Indeed, fibrates, such as ciprofibrate and fenofibrate, have been used clinically to treat hyperlipidemia through their ability to modulate the transcription of genes involved in fatty acid oxidation (FAO) (Despres et al., 2004; Graham et al., 1994). The thiazolidinediones, such as pioglitazone and rosiglitazone acting via PPAR, influence free fatty acid flux and reduce both insulin resistance and blood glucose levels making them useful drugs to treat type 2 diabetes mellitus (Vasudevan and Balasubramanyam, 2004). In an attempt to control the multiple disorders occurring in metabolic syndrome which include dyslipidemia and impaired glucose metabolism, the novel PPAR and - dual agonist, LY465608, was designed (Etgen et al., 2002; Zuckerman et al., 2002). In vitro, LY465608 exhibits potent activity at PPAR and moderate potency at PPAR, with EC50 values of 65 and 665nM, respectively (Etgen and Mantlo, 2003).
The majority of dietary fatty acids are oxidized in the mitochondria; however, very long-chain fatty acids of more than 20 carbons in length are shortened exclusively in peroxisomes and then transported into the mitochondria for complete oxidation (Wanders et al., 2001). Marked species differences in these oxidation processes are well documented. For example, using the enzyme activity of acyl-coenzyme A (CoA) oxidase (Acox) as a biomarker for peroxisomal ß-oxidation, the induction observed in rodent livers (Reddy and Mannaerts, 1994) far exceeded that observed in non-rodent livers from animals treated with PPAR agonists (Graham et al., 2006; Hoivik et al., 2004; Morimura et al., 2006; Schafer et al., 2004). This difference was not due to altered pharmacokinetics or disposition of peroxisome proliferators between species (Pacot et al., 1996). Instead, data from several investigations suggest that the relatively weak non-rodent response to peroxisome proliferators results from fundamental species differences in PPAR transcriptional activation (Bosgra et al., 2005). These fundamental differences include observations that non-rodents have less constitutive hepatic PPAR expression or intrinsically different activity than rodents (Cheung et al., 2004; Gervois et al., 1999; Morimura et al., 2006; Palmer et al., 1998), that non-rodent PPREs are not as efficient as rodent PPREs (Lambe et al., 1999; Woodyatt et al., 1999), and that the presence of a dominant-negative form of PPAR, as has been observed in human hepatocytes, results in less overall transcriptional activity in liver (Gervois et al., 1999; Palmer et al., 1998).
A single cause for the weak response of PPAR agonists in non-rodent species is unlikely. For example, although humans have lower constitutive PPAR receptor levels in liver than rodents (Gervois et al. 1999; Palmer et al. 1998), overexpression of PPAR in human hepatocytes to levels comparable to those observed in rat primary hepatocytes does not increase the induction of peroxisomal activity, suggesting that differences in receptor levels alone cannot account for a lack of peroxisome proliferation (Ammerschlaeger et al., 2004; Hsu et al., 2001; Lawrence et al., 2001). Recent work also suggests that there are inherent differences in the PPAR receptor with regard to tumor formation using PPAR-humanized mice. These transgenic mice express human PPAR mainly in liver and do not exhibit carcinogenic responses observed in wild-type mice after treatment with potent PPAR agonists, suggesting that structural differences in PPAR itself may underlie species-specific susceptibility to hepatocarcinogenesis (Cheung et al., 2004; Morimura et al., 2006). In addition, it has been postulated that PPAR-regulated responses are proportional to the efficiency of the PPAR response elements (PPREs) located within the promoter region of target genes. Indeed, evidence suggests that human PPREs are not as efficient as rat PPREs (Lambe et al., 1999; Woodyatt et al., 1999). For example, Lambe et al. (1999) substituted a human Acox PPRE for the rat Acox PPRE and abolished rat PPAR-induced Acox activity (Ammerschlaeger et al., 2004; Lambe et al., 1999). In contrast, human PPAR exhibited activity at the rat Acox promoter (Hasmall et al., 2000; Lambe et al., 1999) and induced rat genes involved in fatty acid metabolism (Lawrence et al., 2001). Finally, although a dominant-negative form of PPAR may contribute to the lack of peroxisome proliferation effects observed in human hepatocytes, evidence for this effect is equivocal (Gervois et al., 1999; Palmer et al., 1998). Thus, observations suggest that these species differences result in differential receptor activation, cell proliferation, or apoptosis and that these responses are likely ligand dependent (Peraza et al., 2006; Peters et al., 2005). Regardless of the above mechanisms, PPAR agonist administration in humans is clearly sufficient to provide therapeutic benefit while avoiding the undesirable side effects observed in rodents.
Preclinical safety data must be obtained from at least one rodent and one non-rodent species in drug development, and the dog is often used as the non-rodent species as it is considered a good predictor of human toxicity (Olson et al., 1998). Typically, the dose selected for governing human exposure in clinical trials is based on safety data from the most sensitive preclinical species used in preclinical toxicology testing U.S. Food and Drug Administration. During preclinical safety assessment for LY465608, the dog proved to be very sensitive to hepatotoxicity as assessed by elevated (alanine aminotransferase/aspartate aminotransferase) and hepatocellular necrosis. Rat ALT/AST levels and hepatic morphology remained normal through 30 days of testing. Ultrastructural evaluation using electron microscopy of liver from dogs revealed enlarged and megamitochondria with abnormal cristae, without increases in peroxisome number or volume. In contrast, hepatic ultrastructure was normal in LY465608-treated rats, with the exception of expected increases in peroxisomal number (Carfagna et al., 2006; Reynolds et al., 2006). Because the toxicity in dogs included mitochondrial abnormalities and suggested bioenergetic changes, it was decided to address the role of peroxisomes and mitochondria in the hepatoxocity observed with LY465608.
We chose cultured primary hepatocyes as an in vitro model to explore mechanisms of toxicity with LY465608 (Baker et al., 2001; Farkas and Tannenbaum, 2005). A number of investigators have studied peroxisome proliferation in cultured rat hepatocytes (Foxworthy and Eacho, 1986); however, there is only one documented study in dog hepatocytes, and this study presented a limited biochemical description of the canine response to PPAR agonists (Foxworthy et al., 1990). Herein, we explored more deeply metabolic regulation via PPAR by correlating gene expression profiles with biochemical activities using both dog and rat hepatocytes in vitro, providing data that differentiates LY465608 from the prototypical PPAR agonist, fenofibrate. As a result of these biochemical and expression-based studies, we propose mechanisms underlying the differential pharmacology and toxicity of rat and dog PPAR function that may be applicable to developing future molecules to treat metabolic syndrome.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Chemicals.
LY465608 was synthesized at Eli Lilly and Company (Indianapolis, IN). Fenofibrate was purchased from Sigma (St Louis, MO). William's medium E (WE) was purchased from Gibco BRL (Gaithersburg, MD), and fetal bovine serum (FBS) was purchased from Invitrogen (Carlsbad, CA).
Animals.
Male Fischer 344 rats (160–230 g) were purchased from Harlan Sprague-Dawley (Indianapolis, IN), and male or female beagle dogs were purchased from Marshall Farms, North Rose, NY. All the animals were housed individually in stainless steel cages and exposed to a 12-h light/dark cycle.
Hepatocyte isolation.
Independent isolations of rat and dog hepatocytes were obtained from three animals (n = 3 isolations from each species) using a modified two-stage portal vein perfusion technique (Berry and Friend, 1969; Seglen, 1972). Hepatocyte viability was assessed by trypan blue exclusion, and only cell preparations with viability greater than 85% were used. Cells were seeded in a density at 2.75 x 106/ml approximately 70–80% confluent. In general, rat hepatocytes appeared larger than dog hepatocytes, and dog hepatocytes were slower to spread to confluence as judged by light microscopy. The cell pellets were plated with WE containing 10% FBS, 5.0 µg/ml insulin, 5.0 µg/ml transferrin, 5.0 g/ml sodium selenite, 10 ng/ml dexamathasone, 2mM L-glutamine, and 50 µg/ml getamicin (WE complete). Cells were incubated in a 37°C, 100% humidified environment (5% CO2, 95% air). Rat cells were plated on Falcon Primaria TM 100-mm culture dishes, and dog cells were plated on Collagen I–coated 100-mm culture dishes (Becton Dickinson, Lincoln Park, NJ).
Hepatocyte culture.
Following an attachment period of 4 h, the culture medium was replaced with fresh WE complete. After a subsequent 20 h in culture, medium was replaced with WE complete without serum, with or without test compounds prepared in dimethyl sulfoxide (DMSO, 1% v/v). Fresh media and test compounds were added every 24 h thereafter for up to 48 h. Each treatment included three doses, with 0.1, 1, and 10µM for LY465608 and 1, 10, 100µM for fenofibrate. Doses used in these studies were all less than an established cytotoxic dose determined previously in dose-ranging studies. At the end of treatment, cells were washed and collected for electron microscopy, biochemical evaluation, and gene expression. All biochemical and genomic experiments were repeated on three independent hepatocyte isolations, performed at different times, and combined using the statistical tools described below.
Electron microscopy.
Plated hepatocytes were fixed in phosphate-buffered modified Karnovsky's fixative, pH 7.4, at 4°C. Cell pellets were processed and embedded in epoxy resin. Multiple sections were cut and stained with toluidine blue for light microscopic evaluation of area selection. Ultrathin sections were produced from selected areas, mounted on copper grids, and evaluated with a transmission electron microscope.
Biochemical determinations.
For peroxisomal ß-oxidation (ACOX assay), cells from each dish were washed and harvested with a rubber policeman in 1 ml phosphate-buffered saline (PBS) and homogenized with 0.2 ml of tissue solubilization buffer (10% sucrose [w/v], 3mM imidazole [Sigma], pH 7.4. w/KOH). The oxidation of palmitoyl-CoA was quantified spectrophotometrically by measuring the production of H2O2 coupled to oxidation of leuco-DCF (Sigma Chemical Co.) at 502 nm (Small et al., 1985).
Mitochondria were isolated from hepatocytes in the following manner. Cells were harvested and placed in 10-ml buffer containing 220mM mannitol, 70mM sucrose, 2mM HEPES, and 5% bovine serum albumin (BSA), pH 7.0. Cells were homogenized with a Polytron homogenizer with a 12-mm generator for 10 s at a power setting of 19. The homogenate was centrifuged at 400 g for 10 min, and the resulting supernatant was centrifuged at 5700 g for 10 min at 4°C. The resulting mitochondrial fractions were resuspended in the same buffer. Mitochondria (200–250 µg) were evaluated for carnitine palmitoyl transferase 1 (CPT-1) activity according to the methods of Bieber et al. (1972), using a Cobas MiraS Clinical Chemistry Analyzer (Roche Diagnostic System, Somerville, NJ.). Briefly, by subtracting the formation of TNB in the D(+)-carnitine reaction from the L(–)-carnitine reaction, 1,3,5-trinitrobenzene (TNB) formed through CPT-1 activity was determined using 13,600/cm as the molar extinction coefficient of TNB. Protein was determined by Bradford method by using reagent purchased from Pierce (Rockford, IL) with BSA as a standard.
For the biochemical assays, a one-way ANOVA followed by two-sided Dunnett's multiple comparison test with Graphpad Prism software 3.1 was employed (Graphpad Software, Inc., San Diego, CA).
RNA isolation and microarray hybridization.
Total RNA from each plate was isolated using RNA STAT-60 (Tel-Test, Friendswood, TX) and subsequently purified using RNeasy columns (Qiagen, Valencia, CA). Total RNA (10 mg) was converted to double-stranded cDNA, followed by cRNA synthesis, hybridization, washing, and scanning using standard microarray processing protocols; GeneChip analysis was performed according to the Affymetrix manufacturer's protocol (Affymetrix GeneChip Expression Analysis, Technical Manual). Rat hepatocyte samples were hybridized to rat genome 230 2.0 GeneChips, and canine genome GeneChips were utilized to analyze RNA obtained from dog hepatocytes. Microarray fluorescence was normalized by scaling the 2% trimmed average signal of each array to 1500. The fluorometric data were then analyzed using Affymetrix Software (MicroArray Suite 5.0). Data filtering was performed using Data Mining Tool software (Affymetrix, Santa Clara, CA).
Microarray data analysis.
To determine whether a gene was differentially expressed, an ANOVA model was fitted to each probe set on the array (Jolly et al., 2005). Fold-change (FC) calculations were used to indicate the relative change in expression levels between the experimental and baseline (control) targets. FC is expressed as a positive number when the expression level in the experiment increased and a negative number when the expression level in the experiment decreased relative to the control. False discovery rate (FDR) of Benjamini and Hochberg (1995) was used to adjust the p values derived from the above ANOVA model to estimate the number of false positives. Expression data were filtered using the following criteria: a minimum signal of 250 for either the vehicle or treated samples, at least one sample determined present or marginal in each treatment group, and FDR 
0.2. Subsequently, differentially expressed genes were mapped onto a network using the Ingenuity Pathway Analysis application which were assembled into biological networks and were ranked by score (Ingenuity Systems; http://www.ingenuity.com, Redwood City, CA). The score was the likelihood of a set of genes being found in the networks due to random chance calculated by Fischer's exact test. For example, a score of 3 indicates that there is a 1/1000 chance that the focus gene in the network changed by random chance. Therefore, a score of 1.3 or higher has a 95% confidence of not being generated by random chance alone and was chosen as a threshold for this analysis. Hierarchical clustering (HC) was performed using the Pearson correlation as the distance metric, and average linkage clustering was visualized in Spotfire Decision Suite.
Quantitative PCR.
Genes chosen for quantitative PCR analysis are listed in Table 1. Gene-specific primers were designed using Primer Express 1.5 software (Applied Biosystems, Foster City, CA). Gene-specific primer sequences used in this study are presented in Table 2. Briefly, total RNA was DNase treated (DNA-free, Ambion) and reverse transcribed using the SuperScript First-Strand Synthesis system for reverse transcription (RT)–PCR kit (Invitrogen). PCR products were electrophoresed on a 4% agarose gel to confirm primer specificity and amplicon size prior to quantitative fluorescence-based detection using the Applied Biosystem 7700 Sequence Detector System. Each assay was performed in triplicate for each independent hepatocyte isolation. Data were analyzed by the comparative CT method of relative quantitation, according to the calculations described in the Applied Biosystems User Bulletin #2. All values were expressed as FC relative to the vehicle treatment group.
|
|
Associations between canine and rat microarray probe sets and genes.
For both microarray types utilized in these studies, the target sequences for probe set were downloaded from the Affymetrix Web site and used to identify all matches where at least 70% of the target sequence matched at the 90% identity level or better, on the correct strand, against the following sequence Ensembl and the National Center for Biotechnology Information (NCBI) data sets: Canine genomic build one and Ensembl canine transcripts (downloaded from Ensembl) (Birney et al., 2004
), Canine genomic build two, rat genome build three, and canine and rat RefSeq transcripts (downloaded from the NCBI ftp site). Matches were made using Blat version 26 (Kent, 2002) and performed against sequences from the appropriate organism. All matches were converted into EntrezGene IDs or Ensembl Gene IDs using the appropriate resources (seq_gene.md and gene2refseq for NCBI, EnsMart and GenBank formatted genome sequence for Ensembl). These gene associations were supplemented by using UniGene (Pontius et al., 2003) associations with the transcript ID in the NetAffyx annotation file for the appropriate chip. Resulting associations were categorized as high, medium, or low reliability for unique associations depending on the level of agreement between different sources of information. When a single probe set was associated with more than one locus, all associations were recorded. Functional information for the rat genes was retrieved from the Proteome database licensed from BioBase (Hodges et al., 2002). Canine genes were associated with orthologous rat genes by consulting HomoloGene (Wheeler et al., 2006) and by consulting EnsMart (Kasprzyk et al., 2004). Where these agreed on unique associations, the association was considered high reliability. In some cases, we were unable to make these associations directly and instead made joint associations to mouse or, rarely, human loci. Functional annotation for the canine probe sets was then retrieved from the orthologous loci in Proteome (which covers only human, mouse, and rat loci).
Measurement of mitochondrial membrane potential by flow cytometry.
Mitochondrial membrane potential 
m was assessed using a fluorescent indicator, tetramethylrhodamine ethyl ester (TMRE, Molecular Probes, Leiden, the Netherlands) coupled with flow cytometry (Ehrenberg et al., 1988). A TMRE stock was prepared at a concentration of 10mM in DMSO and stored at – 20°C. Working stocks of 100µM were prepared fresh in PBS on the day of assay. After 48 h of treatment with test compounds, cells were collected and incubated for 15 min with 100nM TMRE at room temperature in the dark. The cell fluorescence was measured by flow cytometry. After gating out small-sized debris, 10,000 events were collected for each analysis using CELLQuest software for data analysis (Becton-Dickson Immunocytochemistry Systems, Mountain View, CA, USA). Data shown are representative of two independent hepatocyte preparations, which differed by less than 15% at every dose level.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Microscopic Evaluation of Cultured Primary Rat and Dog Hepatocytes Following PPAR Agonist Treatment
Hepatocytes were analyzed for morphologic changes 48 h after treatment with test PPAR agonists. No signs of cell death were observed at the doses tested, as judged by morphologic observation and lactate dehydrogenase (LDH) release (data not shown). There were no visible compound-related differences between control hepatocytes and hepatocytes incubated with LY465608 or fenofibrate. Hepatocytes in all groups maintained polygonal cell shapes and established extensive cell-to-cell contacts, typical morphological correlates of adaptive dedifferentiation (Figs. 1A–D). However, ultrastructural examination of LY465608-treated rat and dog hepatocytes using electron microscopy revealed compound-related increases in the number of peroxisomes in rat (Figs. 1E and 1F), but not dog, hepatocytes (Figs. 1G and 1H). Mitochondrial morphology was normal in hepatocytes from both species (Figs. 1E–H). Similar ultrastructural observations were made with fenofibrate-treated rat and dog hepatocytes (data not shown).
|
Mitochondrial and Peroxisomal Enzyme Activity in Cultured Primary Rat and Dog Hepatocytes Following PPAR Agonist Treatment
To evaluate functional differences in fatty acid ß-oxidation in rats and dogs, we examined the ability of PPAR agonists to regulate the activity of enzymes that are rate limiting for peroxisomal-mediated (Acox) and mitochondrial-mediated (CPT-1) ß-oxidation activity in rat and dog hepatocytes. In rat hepatocytes, a dose-dependent increase in Acox activity was apparent with LY465608 and fenofibrate treatment. The highest dose of LY465608 caused a threefold increase in Acox activity over control rat hepatocytes (Fig. 2A). In contrast, there were no notable increases in Acox activity for PPAR agonist–treated dog hepatocytes (Fig. 2B). The levels of Acox measured in our experiments were lower than those previously reported; however, these studies showed a linear increase in Acox activity beyond the highest dose of 100µM used in our studies (Gray et al., 1983
; Mitchell et al., 1985). We observed a dose-related increase in CPT-1 activity upon treatment of rat hepatocytes with LY465608 and fenofibrate (Fig. 2C). In dogs, there was also an increase in mitochondrial CPT-1 activity upon treatment with LY465608, but not fenofibrate (Fig. 2D), although basal CPT-1 activity in dog hepatocytes was consistently lower than that measured in rat hepatocytes (scale difference, Figs. 2C and 2D). These results are consistent with species differences noted for peroxisomal ß-oxidation and demonstrate in vitro compound-mediated differences in sensitivity to the induction of mitochondrial ß-oxidation.
|
Comparison of Mitochondrial and Peroxisomal ß-Oxidation Gene Expression Between Rat and Dog Hepatocytes
To further investigate the mechanism of PPAR agonist–mediated mitochondrial and peroxisomal ß-oxidation, microarray analyses were undertaken. In the peroxisomal fatty acid ß-oxidation pathway, genes encoding three enzymes are necessary for metabolizing acyl-CoA to acetyl-CoA: Acox, Ehhadh, and Acaa1 (Fig. 3B). Comparatively, a trifunctional enzyme in mitochondria encoded by two genes, Hadha and Hadhb, catalyzes all three steps in mitochondrial ß-oxidation. In rat hepatocytes, transcripts for all three enzymes for peroxisomal ß-oxidation were increased upon treatment with LY465608 or fenofibrate (numbers indicated change from control, upward arrow indicates statistical significance). These changes did not occur in dog hepatocytes (Fig. 3C). In contrast, the expression of the mitochondrial ß-oxidation genes increased in both rat and dog hepatocytes upon treatment with LY465608 and fenofibrate, with a larger increase being noted in rat than in dog hepatocytes.
|
To confirm the microarray results, we measured Hadhb, Acox Ehhadh, and Acaa1 expression in cultures using quantitative real-time PCR (Fig. 4). Consistent with microarray data, RT-PCR analysis indicated that transcription of all four genes in rat hepatocytes was induced upon agonist treatment. Ehhadh demonstrated the largest response among measured transcripts, with upregulation the greatest magnitude of induction at the medium and high doses (Fig. 4A). In dog hepatocytes, there was little change in the expression of the peroxisomal genes, Acox Ehhadh, and Acaa1, but the mitochondrial gene, Hadhb, increased about fourfold relative to control (Fig. 4B). In summary, both microarray and RT-PCR showed that rat hepatocytes induced both peroxisomal and mitochondrial gene expression upon treatment with PPAR agonists, while dog hepatocytes responded by increasing the expression of RNAs mitochondrial ß-oxidation genes only.
|
Pharmacology: PPAR Agonist Regulated Induction of Genes with Functional PPRE-Containing Promoters
PPAR
agonists not only control peroxisomal and mitochondrial FAO but also regulate microsomal fatty acid hydroxylation, ketogenesis, lipoprotein metabolism, bile and amino acid metabolism, and other metabolic pathways (Mandard et al., 2004). Since the pharmacology of PPAR is driven by the transcriptional activation of target genes, we compared the transcriptional response in rat and dog hepatocytes for a panel of genes previously shown to be regulated by PPAR in humans and mice (Mandard et al., 2004). Table 3 shows a list of genes for which functional PPREs have been documented within the promoter and the magnitude of change for these genes in dog and rat hepatocytes upon PPAR agonist treatment. Overall, there was a trend toward upregulation of genes peripherally associated with fatty acid ß-oxidation in both rat and dog hepatocytes with several notable differences between species. For example, the expression level of the fatty acid–binding protein was elevated in rat hepatocytes but not in dog hepatocytes. Microsomal 
-hydroxylation catalyzed largely by CYP 4A family members is an alternative means by which fatty acids are catabolized in liver. A dramatic increase in CYP 4A1 expression was measured in rat hepatocytes, but in the dog hepatocytes it did not change upon treatment. Consistent with biochemical results, mRNA levels for CPT-1, the enzyme catalyzing the rate-limiting step in mitochondrial FAO, and CPT-2, which performs a similar function at the inner membrane of mitochondria, were elevated in both rat and dog hepatocytes. Mitochondrial ketogenesis acts in concert with fatty acid metabolism during prolonged fasting. This pathway was also upregulated in both rat and dog hepatocytes as indicated by the expression of the rate-limiting gene of this pathway, Hmgcs. The induction of Hmgcs was again in line with biochemical measurements of mitochondrial ß-oxidation, reinforcing the observation that metabolism in mitochondria is increased in both rat and dog hepatocytes. Overall, the gene expression pattern in Table 3 showed that PPAR agonists regulated fatty acid and ketone metabolism in a coordinated manner, with the key observation being that rat hepatocytes responded more robustly than dog hepatocytes, clarifying species-dependent responses in PPAR pharmacology at the level of gene expression.
|
Toxicity: Functional Categorization of Rat and Dog Hepatocytes Treated with PPAR Agonists
To assess the effects of PPAR agonists on global gene expression, we broadly categorized expression data from treated rat and dog hepatocytes, mining microarray data sets for classified biological functions using gene ontology and functional annotation (see Materials and Methods). Using this approach, the top four biological functions affected by LY465608 in rat and dog hepatocytes were ranked by calculated significance (– log) and presented in Table 4. In rat hepatocytes, gene expression enrichment upon treatement with LY465608 indicated that lipid metabolism was the biological function most strongly perturbed, consistent with the demonstrated pharmacology of this molecule. In contrast, the highest ranking biological function enriched with differentially expressed genes in dog hepatocytes treated with LY465608 was cell death, not lipid metabolism. Recalling the liver toxicity observed in LY465608-treated dogs, we further investigated the specific genes responsible for enrichment of this cell death gene ontology grouping. The expression patterns of 65 genes identified from the dog hepatocyte profile, involved in cell death, cancer, or the cell cycle, were grouped using an HC algorithm based upon expression changes (Fig. 5). When this list of 65 genes was compared between species, the changes unique to the dog expression profile became evident and could be grouped into three functional categories: proapoptosis, anti-apoptosis, with the pro- and anti-apoptotic signals being nearly equal. Thus, our data suggest that the balance between cell death and cell survival is equally perturbed in LY465608-treated dog hepatocytes.
|
|
As an alternative data mining approach, an internal statistical analysis of Gene Ontology was performed with PPAR agonist–treated gene expression data (http://www.geneontology.org/). This analysis was based upon biological processes, molecular functions, and cellular compartments that were overrepresented with differentially expressed genes relative to the entire gene distribution on the Affymetrix chips. Using this approach, we found that mitochondrial inner membrane genes and genes involved in oxidoreduction processes were highly enriched from hepatocytes treated with LY456608 in dogs, but not in rats. Furthermore, this group of genes was not perturbed in fenofibrate-treated samples. To investigate this unique group of genes induced by LY465608, gene expression changes were identified and clustered (Fig. 6). Selected genes were subsequently grouped into three functional categories—group I: electron transport, respiratory chain, and ATP synthesis; group II: mitochondrial membrane transporter; and group III: oxidative stress. Genes in group I were differentially regulated exclusively in LY465608-treated hepatocytes, suggesting the capacity for increased energy production and ATP content.
|
To more directly probe mitochondrial changes unveiled using gene expression analysis in PPAR agonist–treated rat and dog hepatocytes, we used flow cytometry and the potential-dependent probe TMRE. Mitochondrial membrane potential is a sensitive indicator of the activity of the mitochondrial proton pumps, electrogenic transport systems, and a key monitor for depolarization initiated cell death (Brand et al., 1994
; Zoratti and Szabo, 1995). We adapted a flow cytometry method for measurement of m in primary hepatocytes, measuring TMRE accumulation in the negatively charged mitochondrial matrix (Ehrenberg et al., 1988). Upon mitochondrial depolarization, a shift in cellular fluorescence was measured by measuring the leftward distribution in the flow cytometry trace. As can be seen in Figure 7, LY465608 caused a decrease of the number of cells with high TMRE fluorescence in a concentration-dependent manner, indicating a depolarization of mitochondria. This depolarization is illustrated by the increased number of leftward shifted cells in the flow cytometry traces of LY465608-treated hepatocytes (Figs. 7A and 7B, insets). LY465608-induced changes in fluorescence were more pronounced in dog hepatocytes than in rat hepatocytes, producing equivalent TMRE responses at nearly 100-fold lower doses. Fenofibrate caused similar changes in mitochondrial depolarization, including an increased sensitivity in dog hepatocytes, although doses much greater than those obtained with LY465608 were needed to elicit effects (data not shown). To confirm that the TMRE dye was sensitive to mitochondrial membrane depolarization, hepatocytes were treated with the mitochondrial membrane-specific K+ ionophore vanilomycin (1–100µM). After 2 h, a significant leftward shift in a concentration-dependent manner indicated the reduction of m. (data not shown).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The goal of this study was to better understand the mechanisms underlying observed species differences in response to the drug candidate, LY465608, and to aid in future development of candidate drugs to treat metabolic syndrome. In general, both biochemical and genomic data supported the notion that mitochondria were a central component of the LY465608 response in our model systems, with a very robust and potentially adverse mitochondrial response being measured in the dog hepatocytes relative to rat hepatocytes. The responses measured in the current study focused on identifying species-specific changes in fatty acid ß-oxidation and mitochondrial bioenergetics. Viewing these data in total, a possible explanation for the LY465608-induced species-specific toxicity can be proposed. The rat exhibited increased capacity to oxidize fatty acids by increasing the number of peroxisomes. This may ease the mitochondrial workload and lessen oxidative demands and damage to this highly important organelle. The dog, however, may be more susceptible to hepatocellular damage upon treatment with LY465608 due to an increased mitochondrial workload, which is supported by our data showing an increased bioenergetic demand and stress response in LY465608-treated hepatocytes.
As mentioned above, peroxisomal activity was induced in rat hepatocytes but not in dog hepatocytes treated with LY465608. For example, Acox mRNA and corresponding enzyme activities were significantly increased in rat hepatocytes upon treatment with LY465608, while no significant changes in Acox mRNA levels or enzyme activities were observed in similarly treated dog hepatocytes. An interesting observation in this study was that, although mitochondrial ß-oxidation was induced in rat and dog hepatocytes treated with LY465608, fenofibrate, the molecule we chose as a PPAR comparator, did not stimulate appreciable mitochondrial ß-oxidation in dog hepatocytes. A possible explanation for this discrepancy may be that LY465608, with its increased potency at the PPAR receptor, produced toxicity through hyperpharmacology or an altered activity at the PPAR promoter in dog. Indeed, LY465608 is more potent at the PPAR receptor than is fenofibrate and carries additional activity at PPAR (EC50 of LY465608—PPAR: 65nM, PPAR: 665nM; EC50 of fenofibrate—PPAR: 30mM, PPAR: 300mM) (Etgen and Mantlo, 2003). This explanation is supported by one in vivo study describing long-term effects of fenofibrate in dogs in which giant mitochondria and crystalline inclusions were observed (Sameshima et al., 1995). These effects were also observed in vivo with LY465608 (Baker et al., in preparation), suggesting hyperpharmacology as one means of driving the mitochondrial effects observed herein. Because LY465608 is not of the same structural class as fenofibrate and carries activity of at least one other receptor, other off-target effects could also be responsible for the effects observed in peroxisomal and mitochondrial biology. Furthermore, how the fundamental changes in peroxisomal and mitochondrial biology observed in vitro manifest in vivo are not known, and the toxicity that we observed in long-term studies may instead be due to parameters not identified herein. We assume that the canine hepatotoxicity of LY465608 is driven largely by PPAR modulation because of its increased potency at this receptor because there is little PPAR expressed in liver and because the data showed subtle, yet comprehensive, induction of metabolic gene expression when analyzed in total. Although LY465608 has PPAR activity, the overall response that we observed matched known PPAR agonists when compared to gene expression of hundreds of drugs and prototypical toxicants contained within an internal reference database (data not shown).
We chose four transcripts to be used as surrogate biomarkers for mitochondrial and peroxisomal ß-oxidation activity in this study, three enzymes of the classic peroxisomal ß-oxidation cycle (Acox, Ehhadh, and Acaa1), and one trifunctional enzyme of the classic mitochondrial ß-oxidation chain (Hadhb). Quantitative PCR analysis of these markers successfully verified expression data obtained from the rat and canine arrays and were consistent with the measured biochemical changes, indicating perturbations of these metabolic pathways upon agonist treatment. Thus, these transcripts may be employed as markers to differentiate organelle-specific FAO and predict pharmacological responses among species via activation of PPAR. The PCR is a simple and sensitive assay compared to biochemical assays and highly amenable to multiplex screening alongside other biomarkers. Therefore, we propose that these markers can be used to disseminate fatty acid ß-oxidation between peroxisomes or mitochondria. Given that there are many reasons for using cell-based systems as surrogate models to detect toxicities, a panel of assays which include peroxisomal and mitochondrial assessment alongside other toxicity endpoints holds promise as a comprehensive approach for compound toxicity screening.
Novel findings in the present studies that may begin to explain the toxicity of LY465608 in dogs include effects on several biological processes measured using a global analysis of gene expression. When expression data were grouped based on overall biological function, opposing biological processes of cell proliferation and cell death were uniquely perturbed in dog hepatocytes treated with LY465608. For example, CAPNS2, a large subunit of the calpain family which induces apoptosis, was upregulated. Interestingly, it has been reported that treating rat hepatocytes with an agent that induces oxidative stress leads to apoptosis by activation of the calpain pathway (Ding and Nam Ong, 2003). Bcl-2, a classical apoptosis inhibitor, was upregulated in canine hepatocytes treated with LY465608. CDKN24, which negatively regulates proliferation through arresting G1-phase checkpoint, was downregulated after treatment with LY465608. Despite these mixed signals of death and proliferation, dog hepatocytes retained their integrity as assessed by measuring LDH release and morphology. We hypothesize that an imbalance between these two processes would affect cellular integrity but may not manifest morphologically within the period of time in which hepatocytes can be kept viable. Further insight into the underlying mechanisms responsible for LY465608-induced toxicity include changes in the mRNA-encoding matrix-localized TCA cycle, the inner membrane–associated FAO pathway, the respiratory complexes (I–V), and the proton-transporting uncoupling proteins. Since FAO is the energy source of NADH for the TCA cycle and electron transport chain function, it is critical in driving bioenergetics. Defective FAO disturbs this balance of electron transport and can subsequently stimulate reactive oxygen species (ROS) generation and cell demise (Zhou and Wallace, 1999). Changes in the redox status of the intracellular environment can also alter signaling pathways, including the cell cycle and apoptosis. These findings are consistent with altered oxidative phosphorylation and mitochondrial ROS production, which might suggest an underlying mechanism by which cellular integrity is lost with toxicity in LY465608-treated dogs.
One difference between in vivo and in vitro data obtained with LY465608 was that altered mitochondrial morphology was observed in the in vivo toxicology study but not the in vitro studies conducted herein. Our in vivo data on liver from LY465608-treated dogs demonstrated a noticeable deterioration in mitochondria with altered function and morphology, including enlarged and pleomorphic mitochondria in liver sections as observed with fenofibrate in long-term studies (Sameshima et al., 1995). This phenotype was not replicated in the dog hepatocytes after 48 h of LY465608 treatment. Assays for mitochondrial apoptosis and oxidative stress were not sensitive enough to detect the subtle changes that LY465608 produced in our primary hepatocyte systems, and these subtle changes may underlie the chronic toxicity that we observed in vivo. Indeed, the gene expression changes reported herein for any one gene measured were subtle, but when viewed collectively using gene ontology, overwhelmingly pointed to a mitochondrially driven toxicity. Flow cytometry measuring mitochondrial potential confirmed these results, indicating that gene expression and mitochondrial potential are more sensitive indicators of hepatocellular damage than are traditionally employed assays for toxicity, such as caspase and cell proliferation reagent WST-1 (tetrazolium salt) measurements. Additional parameters such as treatment duration, loss of cytokine cell interactions with nonhepatocyte cells present in the in situ liver, and artificial environment culture medium may explain the differences between in vitro and in vivo observations in mitochondrial morphology in response to LY465608.
In summary, these studies were conducted to characterize the hepatic effects of LY465608 using rat and dog hepatocytes. We found that in vitro studies recapitulated most pharmacological and toxicological observations of this PPAR coagonist and that robust gene expression changes highlighted bioenergetic changes caused by LY465608. The impaired mitochondrial respiratory complex activities, possibly associated with oxidant/antioxidant imbalance, were thought to underlie defects in energy metabolism and may be considered as the mechanism responsible for LY465608-induced liver toxicity. Therefore, the canine transcription data from this in vitro model has proven to be a powerful and sensitive tool that can be employed to gain insight into the mechanisms underlying toxicity observed in LY465608-treated dogs relative to those observed in the rat.
|
ACKNOWLEDGMENTS |
|---|
Electron microscopy was performed by Jeffrey Horn of the Special Microscopy Laboratory at Lilly Research Laboratories. We thank Wei Tao for his assistance on software and data mining. Special thanks to Dr Mark Carfagna and Dr Phil Solter for critical review of the manuscript.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
U.S. Food and Drug Administration. (2005) Estimating the maximum safe starting dose in initial clinical trials for therapeutics in adult healthy volunteers (Center for Drug Evaluation and Research). Available at: http://www.fda.gov/cder/guidance/5541fnl.htm.
Ammerschlaeger M, Beigel J, Klein KU, Mueller SO. (2004) Characterization of the species-specificity of peroxisome proliferators in rat and human hepatocytes. Toxicol. Sci. 78:229–240.
Baker TK, Carfagna MA, Gao H, Dow ER, Li Q, Searfoss GH, Ryan TP. (2001) Temporal gene expression analysis of monolayer cultured rat hepatocytes. Chem. Res. Toxicol. 14:1218–1231.[CrossRef][ISI][Medline]
Benjamini Y and Hochberg Y. (1995) Controlling the false discovery rate: A practical and powerful approach to multiple testing. J. R. Stat. Soc. B (Methodological) 57:289–300.
Berry MN and Friend DS. (1969) High-yield preparation of isolated rat liver parenchymal cells: A biochemical and fine structural study. J. Cell Biol. 43:506–520.
Bieber LL, Abraham T, Helmrath T. (1972) A rapid spectrophotometric assay for carnitine palmitoyltransferase. Anal. Biochem. 50:509–518.[CrossRef][ISI][Medline]
Birney E, Andrews TD, Bevan P, Caccamo M, Chen Y, Clarke L, Coates G, Cuff J, Curwen V, Cutts T, et al. (2004) An overview of Ensembl. Genome Res. 14:925–928.
Bosgra S, Mennes W, Seinen W. (2005) Proceedings in uncovering the mechanism behind peroxisome proliferator-induced hepatocarcinogenesis. Toxicology 206:309–323.[CrossRef][ISI][Medline]
Brand MD, Chien LF, Ainscow EK, Rolfe DF, Porter RK. (1994) The causes and functions of mitochondrial proton leak. Biochim. Biophys. Acta 1187:132–139.[Medline]
Carfagna MA, Boone LI, Buenger DA, Donnelly KB, Fitzsimmons M, Reynolds VL, Sullivan JM, Williams GD. (2006) A toxicity study in beagle dogs treated for 1 month with LY465608, a PPAR alpha and gamma dual agonist. (SOT 2006 Annual Meeting, San Diego, CA)107.
Chen F, Law SW, O'Malley BW. (1993) Identification of two mPPAR related receptors and evidence for the existence of five subfamily members. Biochem. Biophys. Res. Commun. 196:671–677.[CrossRef][ISI][Medline]
Cheung C, Akiyama TE, Ward JM, Nicol CJ, Feigenbaum L, Vinson C, Gonzalez FJ. (2004) Diminished hepatocellular proliferation in mice humanized for the nuclear receptor peroxisome proliferator-activated receptor alpha. Cancer Res. 64:3849–3854.
Despres JP, Lemieux I, Robins SJ. (2004) Role of fibric acid derivatives in the management of risk factors for coronary heart disease. Drugs 64:2177–2198.[CrossRef][ISI][Medline]
Ding WX and Nam Ong C. (2003) Role of oxidative stress and mitochondrial changes in cyanobacteria-induced apoptosis and hepatotoxicity. FEMS Microbiol. Lett. 220:1–7.[CrossRef][ISI][Medline]
Dreyer C, Krey G, Keller H, Givel F, Helftenbein G, Wahli W. (1992) Control of the peroxisomal beta-oxidation pathway by a novel family of nuclear hormone receptors. Cell 68:879–887.[CrossRef][ISI][Medline]
Ehrenberg B, Montana V, Wei MD, Wuskell JP, Loew LM. (1988) Membrane potential can be determined in individual cells from the nernstian distribution of cationic dyes. Biophys. J. 53:785–794.
Etgen GJ and Mantlo N. (2003) PPAR ligands for metabolic disorders. Curr. Top. Med. Chem. 3:1649–1661.[CrossRef][ISI][Medline]
Etgen GJ, Oldham BA, Johnson WT, Broderick CL, Montrose CR, Brozinick JT, Misener EA, Bean JS, Bensch WR, Brooks DA, et al. (2002) A tailored therapy for the metabolic syndrome: The dual peroxisome proliferator-activated receptor-alpha/gamma agonist LY465608 ameliorates insulin resistance and diabetic hyperglycemia while improving cardiovascular risk factors in preclinical models. Diabetes 51:1083–1087.
Farkas D and Tannenbaum SR. (2005) In vitro methods to study chemically-induced hepatotoxicity: A literature review. Curr. Drug Metab. 6:111–125.[CrossRef][ISI][Medline]
Foxworthy PS and Eacho PI. (1986) Conditions influencing the induction of peroxisomal beta-oxidation in cultured rat hepatocytes. Toxicol. Lett. 30:189–196.[CrossRef][ISI][Medline]
Foxworthy PS, White SL, Hoover DM, Eacho PI. (1990) Effect of ciprofibrate, bezafibrate, and LY171883 on peroxisomal beta-oxidation in cultured rat, dog, and rhesus monkey hepatocytes. Toxicol. Appl. Pharmacol. 104:386–394.[CrossRef][ISI][Medline]
Gervois P, Torra IP, Chinetti G, Grotzinger T, Dubois G, Fruchart JC, Fruchart-Najib J, Leitersd