Abstract
Peroxisome proliferator-activated receptor-α (PPARα) agonists such as fenofibrate are used to treat dyslipidemia. Although fenofibrate is considered safe in humans, it is known to cause hepatocarcinogenesis in rodents. To evaluate untargeted metabolic profiling as a tool for gaining insight into the underlying pharmacology and hepatotoxicology, Fischer 344 male rats were dosed with 300 mg/kg/day of fenofibrate for 14 days and the urine and plasma were analyzed on days 2 and 14. A combination of liquid and gas chromatography mass spectrometry returned the profiles of 486 plasma and 932 urinary metabolites. Aside from known pharmacological effects, such as accelerated fatty acid β-oxidation and reduced plasma cholesterol, new observations on the drug’s impact on cellular metabolism were generated. Reductions in TCA cycle intermediates and biochemical evidence of lactic acidosis demonstrated that energy metabolism homeostasis was altered. Perturbation of the glutathione biosynthesis and elevation of oxidative stress markers were observed. Furthermore, tryptophan metabolism was up-regulated, resulting in accumulation of tryptophan metabolites associated with reactive oxygen species generation, suggesting the possibility of oxidative stress as a mechanism of nongenotoxic carcinogenesis. Finally, several metabolites related to liver function, kidney function, cell damage, and cell proliferation were altered by fenofibrate-induced toxicity at this dose.
Introduction
Fenofibrate belongs to the fibrate class of drugs that have been widely used to treat patients with atherogenic dyslipidemia and is often the preferred drug in combination with statins (Backes et al. 2007; Fruchart and Duriez 2006). The mechanism of fibrate action is known to be mediated through binding and activation of peroxisome proliferator-activated receptor alpha (PPARα) (Fruchart, Duriez, and Staels 1999). PPARα is abundantly expressed in the liver (Auboeuf et al. 1997; Braissant et al. 1996) and is a member of the PPAR subfamily of the nuclear hormone receptors (Berger and Moller 2002; Staels et al. 1998). Upon agonist binding, PPARα forms a heterodimer with retinoid X receptor (RXR) and stimulates the expression of various genes involved in fatty acid β-oxidation and ω-oxidation, fatty acid intracellular and intercellular transport, and HDL cholesterol metabolism (Berger and Moller 2002; Lee, Olson, and Evans 2003; Staels et al. 1998). Hence, the net effects of fenofibrate treatment are to reduce serum triglyceride and cholesterol levels (Linton and Fazio 2000), decreasing cardiovascular risk factors.
Aside from the established attenuation of cardiovascular risk factors, fenofibrate and other PPARα agonists have been shown in rodents to cause significant peroxisome proliferation (hypertrophy and hyperplasia) and finally hepatocarcinogenesis (Peters, Cheung, and Gonzalez 2005). The precise mechanism of the PPARα agonist–induced liver tumor formation is not well understood, but there is a clear association between peroxisome proliferation and cancer (Lawrence and Eacho 1990). Various experimental evidences suggest that multiple factors may be involved, including oxidative stress and altered cell proliferation (Cariello et al. 2005; Nishimura et al. 2008; Oliver and Roberts 2002; Shah et al. 2007). Significantly, it is nongenotoxic in origin (Ashby et al. 1994), so prediction of tumor formation using standard (and more rapid) assays for genotoxic effects does not apply.
In humans, the clinical use of fenofibrate is generally regarded as safe. There are no reports of hepatocarcinogenesis (Ashby et al. 1994; Brown 2007), but there are reports in non-human primates of peroxisome proliferation (Hoivik et al. 2004). Hepatocarcinogenesis aside, various adverse effects (although low in frequency) have been reported, such as liver cirrhosis, abnormal liver function, muscle myopathy, and renal disorders (Davidson et al. 2007; Holoshitz, Alsheikh-Ali, and Karas 2008; Tahmaz et al. 2007). Thus, toxicological uncertainties of fenofibrate therapy, and PPARα agonists in general, remain a concern.
These toxicological concerns, along with the relationship between PPARα and lipid homeostasis and its implications for new therapeutic development, have motivated intensive research. In addition to targeted biochemical and molecular analysis, transcriptomic and proteomic profiling have been increasingly used to study PPAR in recent years. These studies ranged from hepatocyte cell culture models to rodents and non-human primates. In addition to expected findings related to lipid metabolism (e.g., β-oxidation, lipid mobilization, fatty acid and cholesterol synthesis), many other insights about the broad impact of PPARα agonists were presented by these studies (Cariello et al. 2005; Guo et al. 2006; Leonard et al. 2006; Moffit et al. 2007; Tamura et al. 2006).
For example, new understanding on fibrate-induced hepatotoxicity and carcinogenesis provided by proteomic and transcriptomic studies have shown that fibrates elevate the level of proteins or gene transcripts that are involved in the production of reactive oxygen species (ROS) (Guo et al. 2006) and (possibly linked to ROS) DNA repair genes (Nishimura et al. 2007). More specifically, Nishumura et al. (2007) showed that in fenofibrate-treated rats, several DNA repair gene expression levels were increased. These increases were associated with an elevation in ROS, changes in enzymes such as superoxide dismutase, and elevations in 8-OHdG (oxidized deoxyguanosine). These findings strongly implicated oxidative stress–induced DNA damage as a plausible component of the cause of hepatocarcinogenesis. In this and other studies, genes related to apoptosis suppression and cell proliferation were altered by fibrate treatment, indicating that these activities are also likely involved (Cariello et al. 2005; Guo et al. 2006; Nishimura et al. 2007; Tamura et al. 2006).
Some of these studies have also uncovered additional insight into the impact of fibrates on energy metabolism. Genes involved in NAD and tryptophan metabolism were up-regulated (Cariello et al. 2005), which supports the notion that PPARα agonists affect energy homeostasis. Finally, closely related to energy metabolism, a proteomic analysis of clofibrate showed that all but two TCA cycle enzymes were increased in response to dosing with clofibrate (Leonard et al. 2006).
In general, these studies have provided rich data sets on fibrate action. However, mechanistic understanding on the full extent of fibrate impact on cellular energy homeostasis and oxidative stress still lacks clarity. Few studies have been reported that attempt to address these questions using a nontargeted metabolic profiling technology (one study performed a detailed targeted analysis on lipids, demonstrating many of the expected changes in lipid metabolites as well as many additional insights [Wheelock et al. 2007]). Hence, we sought to evaluate untargeted unbiased metabolic profiling as a means to gain further insight into the molecular mechanisms of observed fenofibrate pharmacology and toxicology. For this study, male rats were dosed with fenofibrate (300 mg/kg/day) or vehicle control. Plasma and urine samples were then collected at two time points for metabolic profile analysis.
Materials and Methods
Animal Studies:
Fischer 344 male rats of nine weeks of age used in this study were from Charles River Japan, Inc. (Atsugi Breeding Center, Tokyo, Japan). All the animal experiments were housed in stainless-steel cages in a room that was lit for 12 hr (7:00–19:00) daily, ventilated with an air-exchange rate of 10–20 times per hour, and maintained at 20–26°C with a relative humidity of 30–70%. The rats were housed individually in metabolic cages during the period when urine samples were collected and urine collection began after transfer to these cages. All animals were allowed free access to water and food (CRF-1, sterilized by radiation, Oriental Yeast Co., Tokyo, Japan) except for a 4-hour fasting period prior to plasma and tissue collection. After a 5-day acclimatization period, the rats were randomly assigned to four groups (n = 6 for each group) by body weight. Two groups received fenofibrate (300 mg/kg/ day for 14 days in 0.5% CMC-Na aqueous solution) by oral gavage, and the other two groups received a vehicle control (0.5% CMC-Na aqueous solution). The initial dosing day is designated as day 0. Urine was collected over 24 hours on days 1–2 and days 13–14 with the collection vessels surrounded by dry ice during the collection period. After urine collection, rats were maintained under fasting condition for 4 hr, and blood samples (anticoagulated with EDTA-2Na) were collected on day 2 or day 14. The animals were then euthanized to collect specimens of liver and kidney used in pathological examination. Plasma and urine samples were stored in a freezer set at −80°C. One control and one treatment group were dedicated for each time point. All procedures of animal studies were performed in accordance with the rule of the Institutional Animal Care and Use Committee at the study facility.
Metabolite Identification and Platform Technology General Overview:
Global unbiased metabolic profiling technology based on sample extraction and mass spectrometry was applied to urine and plasma samples as similarly described (Lawton et al. 2008). To efficiently recover metabolites with diverse chemical properties from plasma and urine samples, a four-step sequential extraction procedure using different solvents was used. The extracts were then combined and analyzed by both GC-MS and LC-MS. Chromatographic separation followed by full scan mass spectra was carried out to record all detectable ions presented in the samples. Metabolites with known chemical structure were identified by matching the ions’ chromatographic retention index and mass spectra fragmentation signatures with reference library entries created from authentic standard biochemicals. For ions that were not covered by the standards, additional library entries were added based on their unique ion signatures (chromatographic and mass spectral) and also by virtue of their recurrent nature among samples. Once the library entries were created for these metabolites with unknown structures, they can be routinely detected and quantified. In addition, the unknown biochemicals have the potential to be identified by future acquisition of matching purified standards or by classical structural analysis.
Sample Preparation:
100 μl of plasma or urine samples were extracted using an automated MicroLab STAR® system (Hamilton Company, Salt Lake City, UT, USA). The samples were extracted using a series of four solvent extraction steps: 400 μl tridecanoic acid (2.5 mg/mL) in ethyl acetate:ethyl alcohol (1:1), 200 μl methanol, 200 μl methanol:H2O (3:1), and 200 μl dichloromethane:methanol (1:1). Each solvent extraction step was performed by shaking for 2 min in the presence of glass beads using a Glen Mills Genogrinder 2000. After each extraction, the sample was centrifuged and the supernatant removed using the MicroLab STAR® robotics system, followed by reextraction of the pellet in subsequent steps. The multiple extract supernatants were pooled and then split into two equal aliquots, one for liquid chromatography/mass spectrometry (LC/MS) and one for gas chromatography/Mass Spectrometry (GC/MS). Aliquots were placed on a TurboVap® (Zymark) to remove the solvent, frozen, and dried under vacuum overnight. Samples were maintained at 40°C throughout the extraction process. For LC/MS analysis, extract aliquots were reconstituted in 10% methanol and 0.1% formic acid. GC/ MS aliquots were derivatized using equal parts bistrimethylsilyl-trifluoroacetamide and solvent mixture acetonitrile: dichloromethane:cyclohexane (5:4:1) with 5% triethylamine at 600°C for 1 hr.
Liquid Chromatography/Mass Spectrometry:
LC/MS was carried out using a Surveyor HPLC (ThermoElectron Corporation, San Jose, CA, USA) with an electrospray ionization (Katajamaa and Oresic 2005) source coupled to an LTQ mass spectrometer (MS) (ThermoElectron Corporation). The following solvents were used as mobile phase: 0.1% formic acid in H2O (solvent A) and 0.1% formic acid in methanol (solvent B). The extract was loaded onto a 100 × 2.1 mm, 3 μm particle, Aquasil column (ThermoElectron Corporation) via an CTC autosampler (LeapTechnologies, Carrboro, NC, USA) and gradient eluted (0% B, 4 min; 0–50% B, 2 min; 50–80% B, 5 min, 80–100% B, 1 min; maintain 100% B, 2 min) directly into the mass spectrometer at a flow rate of 200 uL/min. The LTQ took full scan mass spectra (99–1500 m/z) while switching polarity to monitor both negative and positive ions. An LTQ-FTICR hybrid MS (ThermoElectron Corporation) operated at 50,000 resolving power with a mass measurement error <+10 ppm using gradient conditions above was used to confirm reported biochemicals that were present above LOD of the instrument.
Gas Chromatography/Mass Spectrometry:
The derivatized samples were analyzed on a Thermo-Finnigan Trace DSQ fast-scanning single-quadrupole MS operated at unit mass resolving power with a 20 m × 0.18 mm (0.18 μm film phase consisting of 5% phenyldimethyl silicone) GC column. The initial oven temperature was 60°C ramped to 340°C and with helium as the carrier gas. GC/MS was operated using electron impact ionization with a 50–750 amu scan range and was tuned and calibrated daily for mass resolution and mass accuracy.
Biochemical Identification:
Biochemicals were identified by automated comparison to Metabolon’s reference library entries. The reference library was created using approximately 1,500 authentic standards that were analyzed in multiple concentrations and under the same conditions as the experimental samples. The combination of chromatographic retention index and mass spectra signatures gave an indication of a match to the specific biochemical.
Data Imputation and Statistical Analysis:
After the data were corrected for minor variation resulting from instrument inter-day tuning differences (Lawton et al. 2008), the missing values for a given biochemical were imputed with half the observed minimum on the assumption that they were below the limits of detection. For the convenience of data visualization, the raw area counts for each biochemical were rescaled by dividing each sample’s value by the median value for the specific biochemical. Statistical analysis of the data was performed using JMP (SAS, http://www.jmp.com), a commercial software package, and “R” (http://cran.r-project.org/), which is a freely available, open-source software package. ANOVA was performed between the vehicle control group and fenofibrate treatment group at each time point. A log transform was applied to the observed relative concentrations for each biochemical because, in general, the variance increased as a function of a biochemical’s average response.
Results
Dose Regimen and Histology Observations
A dose of 300 mg/kg/day of fenofibrate was used in this study to induce liver toxicology. Although lower doses (200 mg/kg/day) have been used in several longer-term studies (Nishimura et al. 2007), we chose a higher dose since this initial study aimed principally at eliciting short-term effects. Furthermore, gene expression studies have been performed at 400 mg/kg/day (Cornwell, De Souza, and Ulrich 2004). Thus, in this initial study aimed principally at evaluating this technology, we deemed 300 mg/kg/day as a suitable compromise. Under this dose, the liver was notably enlarged, and the relative weights for liver and kidney increased significantly with time, reaching 2.4- and 1.2-fold, respectively, at day 14 (Figure 1D-F). Grossly, the liver was abnormally brown on day 14. Consistent with these findings, hepatocellular hypertrophy and eosinophilia were histopathologically observed on both day 2 and day 14, with more apparent changes on day 14. Also, an increase in the number of mitotic figures was observed on day 2 and day 14 and hepatocyte proliferation was also detectable. Finally, possibly indicating liver injury was slight single-cell necrosis for hepatocytes (although only in 1 of 6 rats). Despite the elevated kidney weight, no noteworthy microscopic findings were observed in the kidneys at day 14. Therefore, the selected dose induced expected liver histological and pathological changes. The incidence of the pathological changes in the liver are presented in Table 1, and representative photomicrographs are provided in Figure 1A-C.
Metabolomic Profiling of Plasma and Urine and Changes Induced by Fenofibrate
The metabolomic profiling approach used in this study was a nonbiased, global analysis technology based on GC-MS and LC-MS. In short, plasma and urine samples were extracted, run over both MS platforms, ion peaks matched to standards in a reference library, and their relative levels quantitated.
A total of 486 biochemicals were detected in the plasma samples, of which 140 were biochemicals matching named structures in our library. The chemical composition for urine was found to be more complex with a total of 932 biochemicals detected (205 were biochemicals that matched named structures in our library). Many biochemicals were unique to either plasma or urine. Welch’s T-test was used to identify biochemicals that showed altered levels between the control and fenofibrate treated groups at both time points for both plasma and urine, with p < .05 and a q-value cutoff of .1 deemed to be significant (Table 2). After mapping the biochemicals with altered levels by fenofibrate into general biochemical pathways as illustrated in the Kyoto Encyclopedia of Genes and Genomes (KEGG), it was found that the most significant impact of the fenofibrate treatment was on cellular energy homeostasis and metabolism related to toxicology.
Changes in Metabolites Associated with Fatty Acid β-Oxidation
Several metabolites involved in lipid metabolism were found to be significantly altered by fenofibrate treatment. Figure 2 shows a condensed scheme of fatty acid metabolism and β-oxidation and the observed metabolites using box plots. The urinary level of carnitine, a metabolite that is essential for the transportation of long chain fatty acid into mitochondria, was decreased approximately fivefold by fenofibrate at both time points (Figure 2, Table 2). Also, both plasma and urinary acetylcarnitine were reduced at both doses (Tables 2 and 3). Plasma 3-hydroxybutanoic acid, a ketone body and the end product of fatty acid β-oxidation, was significantly increased (Figure 2, Table 2). In addition, the levels of two urinary dicarboxylic acids, suberic acid and adipic acid, typically derived from fatty acid ω-oxidation pathway (Draye et al. 1988; Mortensen 1986), were significantly decreased by fenofibrate (Figure 2, Table 3). Collectively, these changes are consistent with the expected action of fenofibrate in induction of β-oxidation.
TCA Cycle and Lactic Acidosis
One of the most dramatic series of changes observed in this study was that the levels of urinary TCA cycle metabolites were significantly reduced by fenofibrate treatment (Figure 3). At both day 2 and day 14, α-ketoglutarate was decreased to 10% of the control level. Fumarate and malate were reduced greater than 50%. Citrate and aconitate also significantly declined at day 14 (although the changes were less apparent at day 2 [Figure 3]). These results indicated that fenofibrate may cause alterations in the TCA cycle. Consistent with less biochemical flux through the TCA cycle, pyruvate (Figure 3) and lactate (Figure 4) were 2.5-fold and 1.5-fold greater in the fenofibrate treated group at day 2 compared to control, respectively.
In addition to lactate and pyruvate, elevations of p-hydroxyphenyllactic acid (HPLA) in the urine and 2-hydroxybutyric acid in the plasma are observed in the fenofibrate-dosed rats (Figure 4). The levels of these metabolites are often used as indicators of lactic acidosis (Kumps, Duez, and Mardens 2002).
Tryptophan Metabolism and Metabolites Associated with Oxidative Stress
The main tryptophan metabolic pathway is the kynurenine pathway. It involves the conversion of tryptophan to kynurenine and ultimately precursors for NAD+ synthesis. Tryptophan can also be degraded to serotonin (Figure 5). Figure 5 shows that fenofibrate treatment up-regulates the kynurenine pathway. The levels of kynurenine and kynurenate in urine were significantly increased by fenofibrate treatment as early as at day 2. The quinolinate change was not obvious at day 2 but increased as much as twelvefold at day 14. The downstream metabolites of this pathway, nicotinamide, methyl nicotinamide, and 6-hydroxynicotinate were also significantly elevated at day 14.
Figure 6 shows that several metabolites associated with oxidative stress were elevated in the fenofibrate-treated group. In particular, components of the glutathione biosynthetic pathways were increased, including γ-glutamylleucine, γ-glutamyltyrosine, and 5-oxoproline. Furthermore, allantoin is increased at both day 2 and day 14. Allantoin is produced from uric acid oxidation and proposed to be a marker for oxidative stress (Hellsten et al. 2001; Zitnanova et al. 2004). Finally, N,N-dimethylarginine in urine is elevated at both day 2 and 14. This is a posttranslational modification of arginine and is considered a marker of oxidative stress (because the enzyme that catabolizes it, dimethylarginine dimethylaminohydrolase, is sensitive to oxidizing conditions) (Sydow and Munzel 2003).
Metabolites Associated with Fenofibrate-Induced Toxicology
A panel of metabolites related to liver function and cell proliferation were significantly altered by fenofibrate treatment as early as the day 2 time point and preceded the changes in liver weight and histopathology at day 14. Thus, these metabolites can be potentially used as early biomarkers for liver toxicology.
The plasma levels of three bile acids, glycocholic acid, taurocholic acid, and cholic acid, increased significantly by fenofibrate treatment at both time points (Figure 7). In a healthy state, circulating bile acid levels are generally low (for each cycle of enterohepatic circulation, about 95% of bile acids are actively reabsorbed [Hofmann 1999]). However, when liver function is compromised, more bile acids can arise in the circulation due to inadequate removal during the enterohepatic circulatory cycle. Fasting serum bile acid levels have been reported as indicators to examine early liver function. Thus, they are considered by some to be more sensitive than many traditional liver enzyme assays (Barnes et al. 1975; de Caestecker et al. 1995). Hence, the elevated levels of bile acids may be an indication that fenofibrate treatment at 300 mg/kg/day resulted in some level of liver malfunction as early as day 2.
A 30% reduction was observed in cholesterol levels at day 2 by fenofibrate (Figure 7). Fenofibrate is known to affect cholesterol metabolism (Berger and Moller 2002; Lee, Olson, and Evans 2003; Staels et al. 1998). However, at day 14, fenofibrate increased the cholesterol level by 70% (Figure 7). One of the liver’s major functions is to capture and metabolize dietary cholesterol in circulation (Dietschy 1998). The increase of cholesterol by fenofibrate at the later time point could likely be due to altered liver function induced by fenofibrate at the high dose of 300 mg/kg/day.
Methyl p-hydroxyphenyllactate (MeHPLA) is an important cell growth–regulating agent that inhibits cell growth and proliferation (Markaverich et al. 1988) and is easily converted to the free acid, p-hydroxyphenyllactic acid (HPLA), by an esterase in various organs. The level of urinary HPLA was increased by fenofibrate treatment at both day 2 and day 14 (Figure 4). Although MeHPLA was not measured by this approach, the levels of HPLA may indicate that MeHPLA is also reduced. In the event that MeHPLA is reduced in cells, a loss of regulatory control for cell growth and proliferation would be predicted to ensue. However, MeHPLA may also have increased because of a more abundant substrate supply. This ultimately can be determined in a further study.
Decrease of urinary putrescine was observed in the fenofibrate treated group at both day 2 and day 14 (Table 3). Because polyamines such as putrescine are generally known to have cancer-promoting effects, the evidence that fenofibrate reduced the level of putrescine was unexpected. However, preliminary targeted metabolite analysis revealed an increase in urinary N 1, N 12-diacethyspermine and N 1, N 8-diacethyspemidine (data not shown). The two diacetyl polyamine derivatives recently have attracted attention as promising tumor markers for diagnosis of cancer (Kawakita and Hiramatsu 2006). Thus, the decreased putrescine noted in this study may be related to cell proliferation (i.e., the reduction being due to conversion to their diacetyl forms).
Discussion
Here, using metabolic profiling, a global “snapshot” of a number of diverse metabolic pathways has been captured for the effects of fenofibrate treatment in rodents. The simultaneous monitoring of this extensive panel of metabolites produced connections between global metabolic changes and fenofibrate pharmacology and toxicology. In addition to recapitulating some of the canonical activities of fenofibrate, we have identified metabolic changes that were previously unknown or unclear. One caveat to the effects described in this study may be present because we employed a dose that is higher than what is used in many studies (300 mg/kg/day compared to the more frequently chosen dose of 200 mg/kg/day). The basis for this was that we were particularly interested in short-term effects (i.e., as early as 2 days). Despite many of the changes being consistent with what might be predicted both for fenofibrate pharmacology and hepatotoxicology, the fact that a higher dose was used in this study can not allow us to exclude a nonspecific generalized toxicity. Thus, some of the observations could possibly be confounded by a generalized toxicity.
Among these changes are alterations to energy homeostasis that are possibly linked to the established central activity of fenofibrate—β-oxidation. Supporting this are reduced urinary carnitine and acetyl-carnitine (urine and plasma) levels (a general indication for fatty acid β-oxidation enhancement). For long-chain fatty acids (LCFAs) to be moved into the mitochondrial matrix for oxidation, the fatty acid CoA esters must first exchange with carnitine to form acyl-carnitine. The reduced levels of free carnitine and acetyl-carnitine following drug treatment may indicate that these pools of carnitine were shifted to LCFA–carnitine (and the intermediates of the oxidation of LCFAs [i.e., medium- and short-chain acyl-carnitines]). Another indication of a high rate of β-oxidation is the elevation in plasma of 3-hydroxybutanoic acid in the fenofibrate dosed animals (Figure 2). In addition, the decreases of suberic acid and adipic acid in the urine are consistent with the decreased ω-oxidation due to the greater flux of fatty acid β-oxidation.
Aside from the effects on lipid metabolism, we observed a marked reduction in TCA-cycle intermediates in urine as early as day 2 (Figure 3), suggesting a diminished flux in this pathway. Fatty acid oxidation is a nicotinamide adenine dinucleotide (NAD+)–intensive process where each reduction of two carbon units from the fatty acid chain reduces one mole of NAD+ to NADH. Plausibly, the decline in TCA cycle intermediates is a barometer for high rates of fenofibrate-induced β-oxidation and the commensurate change in the overall redox status (an increase in the NADH/NAD+ and/or ATP/ADP ratios). Several enzymes in the TCA cycle are inhibited by high levels of NADH (more specifically a large ratio of NADH/ NAD+), such as isocitrate dehydrogenase, α-ketoglutarate dehydrogenase, citrate synthase, and malate dehydrogenase. High quantities of NADH generated by high rates of β-oxidation might be expected to down-regulate several TCA cycle enzymes. Interestingly, our observation of reduced TCA cycle intermediates is similar to that observed in lipid-induced insulin resistance in skeletal muscle (Koves et al. 2008; Muoio and Koves 2007). This is a condition where PPAR-target genes are upregulated and there are high rates of β-oxidation. Similar to what is reported in the present study, a decrease in TCA cycle intermediates was also reported (Koves et al. 2008). The authors comment that in a state where energy expenditure remains the same with a chronic activation of β-oxidation, the NADH/NAD ratio rises, and this elevated ratio serves to inhibit enzymes within the TCA cycle (Koves et al. 2008). They further elaborate on the concept of “mitochondrial overload” (the idea that high rates of fatty acid oxidation and products of incomplete oxidation of those fatty acids lead to mitochondrial dysfunction). In addition to the incomplete oxidation of fatty acids, when β-oxidation is high, a general redox imbalance can follow, facilitating mitochondrial exposure to oxidative stress. Other plausible explanations for the reduction of urinary TCA metabolites could be inappetence of the animals in response to toxins (Connor et al. 2004) or hepatocyte injury. However, during the course of the study, there were no abnormal feeding behaviors associated with fenofibrate treatment (i.e., they exhibited normal dietary intake). The body weights between the control and the treatment groups were also not significantly different (Figure 1D). In addition, single-cell necrosis for hepatocytes and abnormal brownish liver color were only detected at day 14.
The major metabolic fate of the pyruvate that fails to enter the TCA cycle is reduction to lactate. Consistent with this, increased lactate levels were observed in this study (Figure 4). Perhaps this observation can be linked to reports of muscle myopathy as a rare fibrate-induced side effect (Holoshitz, Alsheikh-Ali, and Karas 2008). If high lactate levels translate into myocytes, perhaps some form of myopathy may be induced (from chronically present mild lactic acidosis). In this study, urinary 3-methylhistidine (a metabolite related to muscle breakdown) increases in the fenofibrate treated group (Table 3). Whether the increase in 3-methylhistidine can be attributed to lactic acidosis is unclear, but 3-methylhistidine is a marker of muscle protein breakdown (Long et al. 1988). In fact, we have observed muscle toxicity (such as necrosis of femoral and soleus muscle fiber) under the same experimental conditions used in this study (300 mg/kg for 14 consecutive days) in a separate rat study.
Possibly also related to fenofibrate’s impact on energy homeostasis by β-oxidation are the increased levels of metabolites of the kynurenine pathway. As discussed above, many metabolites in this pathway were increased (Figure 5). Interestingly, nicotinamide is a product of this pathway and a precursor to NAD+. Since the precursor to NAD+ is nicotinamide, the enhanced utilization of the kynurenine pathway may be a reflection that there are high demands for NAD+ (in fact, nicotinamide is elevated thirteenfold at day 14 by fenofibrate). Generally, the combined level of NAD+ and NADH is maintained fairly closely, but given the enhanced demand for NAD+ from β-oxidation induced by an exogenous agonist, this seems to be a plausible idea for why this pathway is activated. (However, the changes occurring more substantially at day 14 [when many signs of toxicology are present] suggests that this response is a failed strategy to maintain metabolic function.) Fibrate activation of several kynurenine pathway enzymes has been reported previously (Delaney et al. 2005; Shin et al. 2006; Shin et al. 1999), and metabolites within these pathways have been proposed as urinary markers for fibrate-induced peroxisome proliferation (Delaney et al. 2005). But despite these observations, little discussion about the consequences of accumulation of certain metabolites within this pathway is present in the literature for fibrates.
One of the least resolved questions with fenofibrate toxicology and hepatocarcinogenesis in rodents is the underlying mechanism. Although still an open question, several intriguing possibilities are raised in this profiling study—chiefly related to oxidative stress markers. Figure 7 shows several metabolites that are indicative of perturbation of glutathione biosynthesis and elevated oxidative stress conditions in the fenofibrate-treated group. The accumulation of deleterious kynurenine pathway intermediates is consistent with the proposed oxidative stress connection to rodent hepatocarcinogenesis. Related to the toxicity and hepatocarcinogenesis induced by fenofibrate, oxidative stress is attributed to peroxisome proliferation and hepatocarcinogenesis in rodents. The dramatic increase of quinolinate at day 14 (Figure 5) is one of the most evident observations in this study. It is suggested that quinolinate could (at least in part) exert deleterious effects by oxidative stress since it can generate free radicals through iron-catalyzed fenton reactions (Stone et al. 2007). Although there are few comments in the literature regarding the possible impact of quinolinate and fibrates, the effect of quinolinate in many cell types is well established—in particular, in neurological tissues it is considered a potent neurotoxin (Sas et al. 2007). This association with cell death, oxidative stress, and a possible link to a core activity of fenofibrate (β-oxidation/NAD+ use) warrant consideration that fenofibrate-induced hepatotoxicity could be mediated, at least in part, by quinolinate.
In addition to the question of hepatocarcinogenesis, the changes of a panel of metabolites were indicative of liver dysfunction as a result of fenofibrate treatment (although this indication is not supported by clinical chemistry data since these assays were not performed). These metabolites could potentially serve as toxicological markers. As described in Figure 6, bile acids were elevated in the fenofibrate treated group. Bile acids represent sensitive markers of hepatic function and are used as indicators of liver toxicology (Barnes et al. 1975; de Caestecker et al. 1995).
Another deleterious effect reported for fenofibrate is kidney dysfunction. Indeed, some signs of this were clearly identified. The levels of 1-methylguanidine was increased in serum and decreased in urine (Tables 2 and 3) by fenofibrate treatment. The level of xanthosine was also decreased in urine (Table 3). The change with the levels of the above nucleosides could be due to fenofibrate-induced renal stress. In fact, our observations are supported by accumulation of nucleosides in serum and decrease in urine due to renal disorder (Niwa, Takeda, and Yoshizumi 1998; Seidel et al. 2006). Furthermore, distinct indicators of cell damage, such as sphinganine and glycerophosphorylcholine in the plasma, were elevated in the fenofibrate treated group. Sphinganine is a component of phospholipids and may be an indication of breakdown of sphingolipids from cell damage. Finally, glycerophosphorylcholine is also elevated in the fenofibrate group, possibly to function as an osmolyte and/or in response to cell damage (Feng et al. 2001; Waldegger et al. 1998).
Collectively, global metabolic-profiling technology produced a simultaneous overview of a number of effects induced by fenofibrate treatment in rodents. Beyond the established effects on lipid metabolites, many additional metabolic changes may serve to enlighten current understanding of fenofibrate action and toxicology. However, it is worth noting that the present study used a high dose, 300 mg/kg/day, to elicit effects in a short time period. It is possible that this high dose caused nonspecific generalized toxic effects that could have confounded the study. Along these lines, it may be a promising approach to evaluate fenofibrate with this technology using a lower dose in a long-term carcinogenicity study or a nontoxic dose to elaborate on mechanism of action-induced effects.