Mitochondrial dysfunction and apoptosis underlie the hepatotoxicity of perhexiline

Abstract

Perhexiline is an anti-anginal drug developed in the late 1960s. Despite its therapeutic success, it caused severe hepatoxicity in selective patients, which resulted in its withdrawal from the market. In the current study we explored the molecular mechanisms underlying the cytotoxicity of perhexiline. In primary human hepatocytes, HepaRG cells, and HepG2 cells, perhexiline induced cell death in a concentration- and time-dependent manner. Perhexiline treatment also caused a significant increase in caspase 3/7 activity at 2 h and 4 h. Pretreatment with specific caspase inhibitors suggested that both intrinsic and extrinsic apoptotic pathways contributed to perhexiline-induced cytotoxicity, which was confirmed by increased expression of TNF-α, cleavage of caspase 3 and 9 upon perhexiline treatment. Moreover, perhexiline caused mitochondrial dysfunction, demonstrated by the classic glucose-galactose assay at 4 h and 24 h. Results from JC-1 staining suggested perhexiline caused loss of mitochondrial potential. Blocking mitochondrial permeability transition pore using inhibitor bongkrekic acid attenuated the cytotoxicity of perhexiline. Western blotting analysis also showed decreased expression level of pro-survival proteins Bcl-2 and Mcl-1, and increased expression of pro-apoptotic protein Bad. Direct measurement of the activity of individual components of the mitochondrial respiratory complex demonstrated that perhexiline strongly inhibited Complex IV and Complex V and moderately inhibited Complex II and Complex II + III. Overall, our data demonstrated that both mitochondrial dysfunction and apoptosis underlies perhexiline-induced hepatotoxicity.

1. Introduction

Perhexiline is an anti-anginal drug developed in the late 1960s by Richardson-Merrell Pharmaceuticals, and later marketed worldwide, before β-blockers and calcium channel blockers became the mainstream therapies in the treatment of angina pectoris (Ashrafian et al., 2007). Although perhexiline was a great success therapeutically, it caused severe hepatotoxicity and neurotoxicity in a small population, especially over long-term use, which resulted in its withdrawal from most of the markets (Paliard et al., 1978; Poupon et al., 1980; Shah, 2006). Currently, perhexiline is used mainly in Australia and New Zealand, and patients using this drug are closely monitored to ensure that the plasma concentration of the drug is within a safe range (Inglis and Stewart, 2006). Perhexiline has been reported to cause fatty liver, inhibit mitochondrial β-oxidation, and may cause cholestasis (Deschamps et al., 1994; Fromenty and Pessayre, 1995; Oorts et al., 2016; Satapathy et al., 2015). However, there are still many gaps in our understanding of the mechanisms underlying perhexiline-induced liver injury.

Perhexiline is an inhibitor of the carnitine O-palmitoyltransferase (CPT) −1 and – 2, the enzymes responsible for transferring of free fatty acids from the cytosol to the mitochondria (Kennedy et al., 1996). Through this mechanism, perhexiline shifts the substrate utilization in myocardial cells from fatty acids to carbohydrates and carries out its therapeutic effects (Abozguia et al., 2010; Gehmlich et al., 2015; Horowitz and Chirkov, 2010). Mitochondria are critical organelles in the cells; they generate most of the ATP for the normal function of a cell and contribute to multiple cellular signaling pathways. Due to these vital roles, mitochondria are involved in various diseases and toxic effects of drugs and chemicals (Cereghetti and Scorrano, 2006; Chalah and Khosravi-Far, 2008; Nieminen, 2003; Vuda and Kamath, 2016; Wallace, 2015; Wallace and Starkov, 2000). Mitochondrial dysfunction has been recognized as one of the major mechanisms underlying drug-induced liver injury, a major cause of premature termination of drug development and drug withdrawal from the market (Boelsterli and Lim, 2007; Kass, 2006; Labbe et al., 2008; Pessayre et al., 2012). Disruption of mitochondrial function could lead to the inhibition or uncoupling of mitochondrial oxidative phosphorylation system and the loss of mitochondrial membrane potential. These primary toxicities could further result in the activation of downstream signaling pathways and the changes in gene expression and protein location. In addition, mitochondrial dysfunction has been recognized as one of common events leading to both apoptosis and necrosis (Malhi et al., 2006; Pradelli et al., 2010).

Apoptosis is programmed cell death; increasing research work has shown that apoptosis underlies drug-induced liver injury (Chen et al., 2014; Elmore, 2007; Iorga et al., 2017; Wang, 2014). Different from necrosis, apoptosis is strictly regulated and energy dependent. Apoptosis signaling pathway is composed of two branches, an intrinsic pathway and an extrinsic pathway. The B-cell lymphoma 2 (Bcl-2) protein family, which is generally bound to the membrane of mitochondria, tightly regulates the activation of the intrinsic pathway. The Bcl-2 family has anti-apoptotic members, such as Bcl-2 and Myeloid cell leukemia-1 (Mcl-1), and pro-apoptotic members, such as Bcl-2 associated agonist of cell death (Bad) and Bcl-2-associated X (Bax). Activation of the pro-apoptotic Bcl-2 family proteins can lead to the activation of caspase 9, which further activates caspase 3, the executer of apoptosis (Adams and Cory, 2007; Antonsson, 2004; Gross et al., 1999; Hatok and Racay, 2016). In the extrinsic pathway, the signal transduction is initiated by the binding of tumor necrosis factor α (TNF-α), the main mediator in the extrinsic pathway, to its receptors. This results in the recruitment and activation of downstream factors including caspase 8, which eventually activates caspase 3 as well (Elmore, 2007; Nair et al., 2014).

In this study, using multiple hepatic cell models, we explored the mechanisms underlying the cytotoxicity of perhexiline. We established that perhexiline induced apoptosis, through both the intrinsic and extrinsic pathways. In addition, exposure to perhexiline caused mitochondrial dysfunction in HepG2 cells.

2. Materials and methods

2.1. Chemicals and reagents

Williams' Medium E, perhexiline maleate salt, D-(+)-glucose, D-(+)-galactose, cyclosporine A, bongkrekic acid, and dimethysulfoxide (DMSO) were purchased from Sigma-Aldrich (St. Louis, MO). Fetal bovine serum (FBS) was obtained from Atlanta Biologicals (Lawrenceville, GA). Dulbecco's modified Eagle's medium (DMEM), DMEM deprived of glucose, sodium pyruvate, N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), and antibiotic-antimycotic were obtained from Life Technologies (Carlsbad, CA).

2.2. Cell culture and drug treatment

The human hepatoma HepG2 cells were routinely cultured in Williams' Medium E supplemented with 10% FBS and 1 × antibiotic-antimycotic, as described previously (Ren et al., 2016). The passage number did not exceed 10 for all the experiments. Cells were either seeded at a density of 2.5 × 10⁵ cells/ml in volumes of 100 μl in the wells of 96-well tissue culture plates for toxicity assays, or in volumes of 5 ml in 60 mm plates for biochemical assays. Unless otherwise specified, cells were maintained in growth medium for approximately 24 h before treatment with indicated concentrations of perhexiline and/or inhibitors, or the vehicle DMSO control. The final concentration of DMSO did not exceed 0.1%.

Cell culture conditions for the comparison of glucose- and galactose-containing medium comparison followed the description of Marroquin et al. (Marroquin et al., 2007). HepG2 cells were switched to glucose- or galactose-containing medium upon seeding, approximately 24 h prior to drug treatment. Glucose-containing medium was DMEM, supplemented with 25 mM glucose, 5 mM HEPES, 10% FBS, 1 mM sodium pyruvate, and 1× antibiotic-antimycotic. Galactose-containing medium was DMEM, deprived of glucose and supplemented with 10 mM galactose, 5 mM HEPES, 10% FBS, 1 mM sodium pyruvate, and 1× antibiotic-antimycotic.

Primary human hepatocytes, pooled from 10 donors, were purchased from In Vitro ADMET Laboratories (Columbia, MD). Plates used for primary human hepatocytes were pre-coated with PureCol^® (Advanced BioMatrix, Carlsbad, CA) following the manufacturer's protocol. Cells were thawed and seeded at a density of 4 × 10⁵ cells/ml in volumes of 100 μl in the wells of 96-well tissue culture plates for toxicity assays or in volumes of 5 ml in 60 mm plates for RNA isolation or protein extraction approximately 24 h before drug treatment, according to supplier's protocol. Cells were maintained in Universal Primary Cell Plating Medium (UPCM™) provided by the supplier.

The human hepatoma HepaRG cells were purchased from Biopredic International (Saint Grégoire, France) and cultured following the manufacturer's instruction. Briefly, the cells were thawed, counted, and plated at a density of 0.7 × 10⁶ cells in a 100 mm cell culture dish and maintained in 10 ml Williams' Medium E supplemented with a growth supplement (Lonza, Walkersville, MD), 1% GlutaMax (ThermoFisher Scientific, Waltham, MA), and 100 μg/ml primocin (Invivogen, San Diego, CA) at 37 °C for 14 days. Differentiation of the cells was induced by adding differentiation supplement (Lonza) to the culture medium for an additional 14 days. The differentiated cells were then seeded at a density of 2.5 × 10⁵ cells/ml in volumes of 100 μl in the wells of 96-well cell culture plates for 72 h before drug treatment.

2.3. Cellular ATP level measurement

Cellular ATP levels were measured using a CellTiter-Glo Luminescent Cell Viability Assay Kit (Promega Corporation, Madison, WI), as described previously (Ren et al., 2017). The results in treated cells were normalized to the intensity of luminescent signals in DMSO controls.

2.4. MTS cell viability assay

Cell viability was measured by MTS assay using a Cell Titer 96 Aqueous Non-Radioactive Cell Proliferation Assay Kit (Promega Corporation), as described previously (Chen et al., 2018). The results in treated cells were normalized to the intensity of luminescent signals in DMSO treated controls.

2.5. Lactate dehydrogenase assay

Lactate dehydrogenase (LDH) release upon drug treatment was measured as described previously (Chen et al., 2018).

2.6. Measurement of caspase 3/7 and caspase 9 activity

Caspase 3/7 and caspase 9 activities were measured using Caspase-Glo^® 3/7 Assay System and Caspase-Glo^® 9 Assay System (Promega Corporation), respectively, following the manufacturer's instruction. The results in treated cells were normalized to the luminescent signals in DMSO controls.

2.7. Isolation of RNA and quantitative real-time PCR assay

Total RNA was isolated using RNeasy Mini kits (Qiagen, Germantown, MD). The yield of the extracted RNA was determined by measurement of absorption at 260 nm using a NanoDrop 2000 spectrometer (ThermoFisher Scientific). cDNAs were generated through reverse transcription of 2 μg of total RNA using a High-Capacity cDNA Reverse Transcription kit (Applied Biosystem, Foster City, CA). Quantitative real-time PCR assay for TNFα, NQO1, HMOX1, GSS, and GCLC was conducted as described previously (Chen et al., 2014) and GAPDH was used as an internal control. All the primers were obtained from ThermoFisher Scientific.

2.8. Western blot analysis

Upon completion of drug treatments, cells were harvested using RIPA buffer supplemented with Halt Protease Inhibitor Cocktail (ThermoFisher Scientific). The concentrations of the samples were determined by Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA). Standard Western blot analysis was performed using antibodies against caspase 3, caspase 9, Bcl-2, Mcl-1, Bad, Bax, TNF-α, SOD1, SOD2, glutathione reductase, γ-GCS, HO1 and NQO1 (Cell Signaling Technology, Danvers, Massachusetts). GAPDH (antibody from Santa Cruz Biotechnology, Santa Cruz, CA) was used as the internal control. The bands were detected using FluorChem E and M Imager (ProteinSimple, San Jose, CA) and were quantified using ImageJ software (NIH, Bethesda, MD).

2.9. Measurement of mitochondrial membrane potential

Mitochondrial membrane potential (MMP) was measured using the dual fluorescence dye JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolyl-carbocyanine iodide). Cells seeded in 96-well black wall/clear bottom plates were treated with various concentrations of perhexiline for 4 h or 24 h. Upon completion of the treatment, the supernatant was aspirated, and cells were incubated with 100 μl of 2.5 μg/ml JC-1 in medium in each well for 20 min at 37 °C. Cells were then washed twice with PBS, and the fluorescence intensities of JC-1 monomers and aggregates were measured using a Cytation 5 Multi-Mode Microplate Reader (BioTek, Winooski, VT). JC-1 monomers were detected at wavelength of 485 nm for excitation and 535 nm for emission, whereas JC-1 aggregates were detected at wavelength of 530 nm for excitation and 590 nm for emission. MMP was measured as the ratio of JC-1 aggregate to monomer, and the results in drug treated cells were expressed as percentage of the results in DMSO controls.

2.10. Measurement of mitochondrial complex activity

The activity of individual mitochondrial complex was measured using MitoTox™ Complete OXPHOS Activity Assay kit (Abcam, Cambridge, MA) and a Cytation 5 Microplate Reader (BioTek). All the chemicals used, except perhexiline, were provided in the kit.

Complex I (NADH ubiquinone oxidoreductase) activity measurement: 320 μl of detergent-solubilized bovine heart mitochondria was diluted with 5 ml of Mito Buffer, and 50 μl of the dilution was added to each well of a 96-well plate pre-coated with an anti-Complex I monoclonal antibody. After 2 h of incubation at room temperature, the solutions were removed by aspiration and 40 μl of phospholipids were added to each well. Following another 45 min of incubation at room temperature, perhexiline or DMSO was diluted in Complex I activity solution to their final concentrations, and 200 μl of the compound-containing solutions was added to each well. Absorbance was measured at 340 nm at 30 °C for 2 h at 1-min intervals using Cytation 5 Microplate Reader. The activity of Complex I was calculated as a decrease in absorbance at the opical density (OD) 340 nm, and the calculation was made using the time window between 20 and 60 min, where the decrease was most linear.

Complex II (succinate-coenzyme Q reductase) activity measurement: 513 μl of detergent-solubilized mitochondria were diluted with 20 ml of Mito Buffer, and 200 μl of the dilution was added to each well in a 96-well plate pre-coated with an anti-Complex II monoclonal antibody. After 2 h of incubation at room temperature, the solutions were removed by aspiration and the plates were rinsed with Buffer solution. Perhexiline or DMSO was added to the Complex II activity solution, and 200 μl of the compound-containing solution was added to each well of the 96-well plate. Absorbance was measured at 600 nm at room temperature for 1 h at 1-min intervals using the Microplate Reader. The activity of Complex II was calculated as a decrease in absorbance at OD 600 nm, and the calculation was made using the time window between 5 and 15 min, where the decrease was most linear.

Complex II + III (ubiquinolcytochrome c oxidoreductase) activity measurement: perhexiline or DMSO was diluted to the final concentration with the Complex III activity solution, and 100 μl of the solution was added to each well of two rows in a plain 96-well plate, which served as controls. Mitochondria (120 μl) was diluted with 880 μl of Complex III Mito Buffer, and then 5 μl of the diluted mitochondria was added to each remaining cells of the 96-well plate together with 95 μl of the compound-containing Complex III activity solution. Absorbance was measured immediately at 550 nm for 5 min at 30-s intervals at room temperature. The activity of Complex III was calculated as the rate of increase in absorbance at OD 550 nm, and the calculation was made using the time window between 1 and 4 min, where the increase was most linear.

Complex IV (cytochrome c oxidase) activity measurement: 300 μl of the Blocking Solution was added to each well in a 96-well plate pre-coated with an anti-Complex IV monoclonal antibody and incubated for 1 h at room temperature. The solutions were removed by aspiration and the plates were rinsed. Detergent-solubilized mitochondria (4.4 μl) was diluted with 22 ml of Mito Buffer, and 200 μl of the dilution was added to each well in the 96-well plate, followed by 3 h incubation at room temperature. At the end of the incubation, the solution was removed and the plates were rinsed again. Perhexiline or DMSO was diluted to their final concentrations with the Complex IV activity solution, and 200 μl of the compound-containing solutions was added to each well. Absorbance was measured at 550 nm at room temperature for 1 h at 1-min intervals. The activity of Complex IV was calculated as a decrease in absorbance at OD 550 nm, and the calculation was made using the time window between 15 and 30 min, where the decline was most linear.

Complex V (ATP synthase complex) activity measurement: 320 μl of detergent-solubilized mitochondria was diluted with 5 ml of Mito Buffer, and 50 μl of the dilution was added to each well in a 96-well plate pre-coated with an anti-Complex V monoclonal antibody and incubated for 2 h at room temperature. After the incubation, the solution was removed and the plate was rinsed before 40 μl of phospholipids was added to each well, followed by 45 min incubation at room temperature. Perhexiline or DMSO was diluted with Complex V activity solution, and 200 μl of the compound-containing solutions was added to each well. Absorbance was measured at 340 nm at 30 °C for 1 h at 1-min intervals. The activity of Complex V was calculated as a decrease in absorbance at OD 340 nm, and the calculation was made using the time window between 20 and 40 min, where the decrease was most linear.

2.11. Measurement of ROS generation

Cellular ROS levels were assessed using DCFDA/H2DCFDA Cellular ROS Assay Kit (Abcam) following manufacturer's instruction. The results in treated cells were normalized to the luminescent signals in DMSO controls.

2.12. Statistical analysis

Data are presented as the mean ± standard deviation (SD) of at least three independent experiments. Analyses were conducted using GraphPad Prism 5 (GraphPad Software, San Diego, CA). Statistical significance was determined by one-way analysis of variance (ANOVA) followed by the Dunnett's tests for pairwise-comparisons or two-way ANOVA followed by the Bonferroni post-test. The difference was considered statistically significant when p was less than 0.05.

3. Results

3.1. Perhexiline induces cellular damage in hepatic cells

Initially, we exposed HepG2 cells from 5 to 25 μM perhexiline for 2, 4, and 6 h to obtain a cytotoxicity profile of the drug. Three cytotoxicity assays were employed to monitor different parameters: cellular ATP content, cell viability by MTS viability assay, and cell membrane integrity by LDH release assay. As demonstrated in Fig. 1A, the cellular ATP content showed the greatest sensitivity to perhexiline exposure. A reduction of ATP content was observed at a concentration as low as 5 μM at 4 h and 6 h. An increase in concentration drastically decreased the cellular ATP content, especially at longer exposure time.

Fig. 1.

Perhexiline induces cellular damage in hepatic cells. HepG2 cells, HepaRG cells, and primary human hepatocytes were exposed to 5, 10, 15, 20, 25 μM perhexiline. DMSO treated cells served as the vehicle control. The cytotoxicity of perhexiline in HepG2 cells was measured using cellular ATP levels (A), MTS assay (B) and LDH assay (C) at 2, 4, and 6 h. Cellular ATP levels in HepaRG cells and primary human hepatocytes upon perhexiline treatment were measured at 4 h (D). The results shown are mean ± S.D. from 3 to 4 independent experiments. *p < .05 compared to DMSO control.

MTS assay displayed a similar trend (Fig. 1B); 10 μM perhexiline exposure significantly reduced cell viability as early as 4 h. At a concentration of ≥15 μM, a 6 h exposure to perhexiline reduced cell viability to only 5% of control, suggesting strong toxicity. This observation was further confirmed by an LDH release assay (Fig. 1C). Treatment of 10 μM perhexiline for 6 h significantly increased LDH release to 42%, suggesting excessive cell death. Higher concentrations resulted in an increase in LDH release at earlier time points.

To ensure that the toxic effect observed in HepG2 cells was not dependent on a specific cell-line, we exposed HepaRG cells and primary human hepatocytes to perhexiline from 5 to 25 μM for 4 h to measure cellular ATP contents. Similar to the results of HepG2 cells, both HepaRG cells and primary human hepatocytes demonstrated ATP reduction upon perhexiline treatment and the response was also time- and concentration-dependent (Fig. 1D).

Overall, our results show that perhexiline induced time- and concentration-dependent cytotoxicity in different hepatic cells. Since HepG2 cells have value in liver toxicity studies (O'Brien et al., 2006; Ren et al., 2018), show good reproducibility in experiments, and are ease to use, we chose HepG2 cells for most of our following experiments.

3.2. Perhexiline activates caspase 3/7 through both intrinsic and extrinsic pathways

To explore whether or not apoptosis is involved in the cell death observed upon perhexiline treatment, we measured caspase 3/7 activity in HepG2 cells exposed to 5 to 25 μM of perhexiline for 2 h and 4 h. As demonstrated in Fig. 2A, exposure to 20 μM of perhexiline significantly increased the caspase 3/7 activity at both 2 h and 4 h time point. For the 2 h treatment, a further increase in caspase 3/7 activity was observed at 25 μM; however, the caspase 3/7 activity dropped after the 4 h treatment at the same concentration. It is very likely that under this condition, excessive cell death already occurred, as the results in Fig. 1 showed, which caused the decrease in caspase 3/7 activity.

Fig. 2.

Perhexiline induces apoptosis in hepatic cells. HepG2 cells were exposed to 5, 10, 15, 20, and 25 μM perhexiline for 1, 2, or 4 h. DMSO treated cells served as the vehicle control. Cellular caspase 3/7 activity (A), caspase 9 activity (B), and relative TNF-α mRNA level (C) in HepG2 cells upon perhexiline treatment were measured and expressed as fold change to DMSO vehicle control. Total cellular proteins from HepG2 cells were extracted after 1 and 2 h perhexiline exposure, and cleavage of caspase 3 and 9 were assessed using Western blotting with GAPDH as the internal control (D). The results shown are the mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

Apoptosis can be triggered through an intrinsic pathway mediated by caspase 9, and an extrinsic pathway mediated by TNF-α and caspase 8. The activation of both pathways could lead to the activation of downstream executor caspase 3/7. Therefore, we next explored which pathway contributed to perhexiline-induced apoptosis. Exposing HepG2 cells to 5 to 25 μM of perhexiline for 2 h resulted in a small but significant increase in caspase 9 activity (Fig. 2B). Furthermore, using qRT-PCR, we found that mRNA level of TNF-α displayed a concentration-dependent increase at as early as 1 h (Fig. 2C). Western blot analysis showed that both caspase 3 and caspase 9 were cleaved, which further confirmed their activation (Fig. 2D).

3.3. Suppression of apoptosis and necrosis attenuates perhexiline-induced cytotoxicity

To confirm further the contribution of intrinsic and extrinsic pathways of apoptosis in the cytotoxicity of perhexiline, we applied a series of caspase inhibitors, with different targets, to the HepG2 cells before exposing the cells to various concentrations of perhexiline. The inhibitors used were: general caspase (pan-caspase) inhibitor Z-VAD-FMK, caspase 3 inhibitor Z-DEVD-FMK, caspase 8 inhibitor Z-IETD-FMK, and caspase 9 inhibitor Z-LEHD-FMK. First, we measured caspase 3/7 activity levels to test the inhibitory effects of these inhibitors. HepG2 cells were pretreated with 10 μM of the individual inhibitors for 2 h before exposure to 20 or 25 μM perhexiline or the DMSO control. As demonstrated in Fig. 3A, pretreatment with the pan-caspase inhibitor (Z-VAD-FKM) or the caspase 3 inhibitor (Z-DEVD-FKM) almost completely abolished the perhexiline-induced increase in caspase 3/7 activity. Application of caspase 8 (Z-IETD-FMK) or caspase 9 (Z-LEHD-FMK) inhibitors significantly decreased caspase 3/7 activity, suggesting that both intrinsic and extrinsic pathways contributed to perhexiline-induced activation of caspase 3/7. LDH release assay after 4 h perhexiline exposure also demonstrated that suppression of caspase 3/7, caspase 8, or caspase 9 attenuated the cytotoxicity of perhexiline (Fig. 3B).

Fig. 3.

Inhibition of apoptosis or necrosis attenuates perhexiline-induced cell death. (A, B) HepG2 cells were pretreated with 10 μM caspase general inhibitor Z-VAD-FMK, caspase 3 inhibitor Z-DEVD-FMK, caspase 8 inhibitor Z-IETD-FMK, or caspase 9 inhibitor Z-LEHD-FMK 2 h prior to exposure to perhexiline. Caspase 3/7 activity (A) and LDH release (B) were measured. (C) HepG2 cells were pretreated with 10 μM necrosis inhibitor NecroX 5 for 2 h prior to exposure to perhexiline. Cytotoxicity was assess using LDH release. The results shown are the mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

It is worth noting that although caspase inhibitors reduced the increased release of LDH, the reduction of LDH release did not approach the baseline level, suggesting that there may be other mechanisms contributing to the cell death caused by perhexiline. One potential mechanism could be necrosis. To test this hypothesis, we applied cell permeable necrosis inhibitor NecroX 5. HepG2 cells were pre-incubated with 10 μM NecroX 5 for 2 h before exposure to 20 μM perhexiline for 4 h, and cellular damage was assessed by the LDH release assay. As shown in Fig. 3C, pretreatment with NecroX 5 partially rescued the toxic effect of perhexiline, suggesting necrosis may also contribute to perhexiline-induced liver toxicity.

3.4. Perhexiline causes mitochondrial dysfunction in HepG2 cells

The rapid depletion of cellular ATP content indicates that perhexiline may induce mitochondrial dysfunction. One of the classic methods to investigate mitochondrial dysfunction is by changing glucose in the growth media to galactose and comparing cytotoxicity, an approach termed the “glucose-galactose” assay (Marroquin et al., 2007). When grown in galactose-containing medium, cells are forced to rely on oxidative phosphorylation for ATP production, thus, the cells are more susceptible to mitochondrial toxicants. Mitochondrial dysfunction-mediated liver toxicity can be detected with more sensitivity using this “glucose-galactose” assay. Therefore, HepG2 cells were cultured in glucose-containing or galactose-containing medium for 24 h and then were challenged with 5 to 25 μM perhexiline. Again, ATP content, MTS viability, and LDH release were used as endpoints for the toxic effects of perhexiline. As shown in Fig. 4A-C, HepG2 cells cultured in galactose-containing medium displayed a much higher sensitivity towards perhexiline treatment. At ≥15 μM perhexiline, cells in the galactose-containing medium had a significant reduction in cellular ATP content and cell viability, and a significant increase in LDH release, compared to the cells grown in glucose-containing medium under the same treatment. This effect was rapid and profound, as severe cellular damage was observed at 2 h time point. These observations suggest that perhexiline caused mitochondrial dysfunction.

Fig. 4.

Perhexiline induces mitochondrial dysfunction. (A-C) HepG2 cells cultured in high-glucose medium or galactose medium were exposed to 5, 10, 15, 20, and 25 μM perhexiline for 2 or 4 h. DMSO treated cells served as the vehicle control. Cellular damage was assessed by cellular ATP content (A), MTS assay (B) and LDH release (C). (D) Mitochondrial membrane potential was measured by JC-1 staining upon perhexiline treatment for 4 h. (E) HepG2 cells were pre-treated with 10 μM Bongkrekic acid or 1 μM Cyclosporine for 1 h before exposed to 20 μM perhexiline for 4 h. Cytotoxicity of perhexiline was assessed by LDH release. The results shown are the mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

Mitochondrial damage frequently involves the opening of mitochondrial permeability transition pore (MPTP), and results in loss of mitochondrial membrane potentials. To investigate whether or not this occurs in perhexiline-induced cytotoxicity, we used a classic fluorescent assay, JC-1 staining, to detect changes in mitochondrial membrane potential in cells treated with perhexiline. As displayed in Fig. 4D, treatment of as low as 5 μM perhexiline for 4 h caused a small but significant reduction in the JC-1 fluorescence ratio, suggesting a loss in mitochondrial potential. Higher concentrations of perhexiline decreased further the JC-1 fluorescence ratio in a concentration-dependent manner until reaching a plateau at 15 μM.

MPTPs are composed of a voltage-dependent anion channel, cyclophilin D (CypD) and the adenine nucleotide translocator (ANT), along with other components. Cyclosporine A, an inhibitor targeting CypD, and bongkrekic acid, an inhibitor targeting ANT, are two inhibitors that can reduce the opening of MPTP and thus attenuate the cell damage caused by mitochondrial dysfunction. We applied these two inhibitors to decipher further the mechanisms of perhexiline-induced mitochondrial damage. HepG2 cells were pre-treated for 2 h with 10 μM bongkrekic acid or 1 μM Cyclosporine A before exposed to 20 μM of perhexiline, and LDH release (Fig. 4E) was used to measure cellular damage. Only bongkrekic acid showed a protective effect upon perhexiline treatment. These results suggest that perhexiline-induced mitochondrial dysfunction is possibly mediated in an ANT-dependent but CypD-independent manner.

3.5. Perhexiline alters the expression of genes related to mitochondrial dysfunction

Changes in the expression of mitochondrial proteins are hallmarks of both mitochondrial dysfunction and activation of the intrinsic pathway of apoptosis. Bcl-2 family proteins play critical roles in these processes. Therefore, to understand further the mechanisms of perhexiline-induced liver toxicity, we examined the protein expression levels of several Bcl-2 family proteins.

HepG2 cells were exposed to 5 to 25 μM perhexiline for 2 h, and then total cell extracts were lysed and analyzed using Western blotting. As shown in Fig. 5A and B, the anti-apoptotic Bcl-2 family members Bcl-2 and Mcl-1 showed a concentration-dependent decline upon perhexiline treatment. On the other hand, the protein expression level of Bad, a pro-apoptotic Bcl-2 family member showed significant concentration-dependent increase. Another pro-apoptotic Bcl-2 family member Bax, however, did not show significant change, suggesting different roles of regulation of pro-apoptotic Bcl-2 family members in perhexiline-induced cytotoxicity. The results indicated that perhexiline reduced the protective effect of anti-apoptotic Bcl-2 proteins and increased the pro-apoptotic factors of the same family. These also confirmed the activation of intrinsic pathways of apoptosis.

Fig. 5.

Perhexiline causes mitochondrial damage through the activation of Bcl-2 family proteins. Total cellular extracts from HepG2 cells were harvested after 2 h perhexiline exposure, and the expression levels of Bcl-2 family proteins were analyzed by Western blotting. (A) representative Western blots and (B) quantification in HepG2 cells are shown here. The results shown are the mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

In a previous experiment, we demonstrated that perhexiline treatment increased TNF-α expression at mRNA level. Using Western blotting, we examined the change in the protein level of TNF-α upon perhexiline treatment. As shown in Fig. 5A and B, perhexiline caused a concentration-dependent increase of TNF-α, which confirmed further the involvement of an extrinsic pathway of apoptosis in perhexiline-induced cytotoxicity.

3.6. Perhexiline directly impairs the mitochondrial oxidative phosphorylation complexes

ATP synthesis in mitochondria is conducted through oxidative phosphorylation Complexes I-V. Direct measurement of the activity of each respiratory complex could provide valuable insights of drug effect on mitochondria. Thus, we tested the effect of up to 800 μM perhexiline exposure on the activity of individual respiratory complexes using immunocapture-based OXPHOS activity assays. The results are presented in Fig. 6. As indicated by the calculated 50% inhibitory concentration (IC₅₀), Complex IV and Complex V were particularly vulnerable to the exposure of perhexiline, with the IC₅₀ value at 0.65 μM and 0.01 μM, respectively. Moderate inhibition was observed for Complex II (IC₅₀ = 95.1 μM) and Complex II + III (IC₅₀ = 21.4 μM). Only Complex I was poorly inhibited by perhexiline, with IC₅₀ > 1000 μM.

Fig. 6.

Perhexiline causes direct inhibition of mitochondrial electron transport complexes. Effects of perhexiline on the activities of individual mitochondrial electron transport complexes in isolated mitochondria. The results shown are the mean ± S.D. from 3 independent experiments. The 50% inhibitory concentration (IC₅₀) of perhexiline on Complex I, II, II + III, IV, and V activity was obtained from the dose-response curves.

3.7. Excessive ROS generation does not contribute to perhexiline-induced cytotoxicity

Deficits in mitochondrial function, sometimes lead to excessive reactive oxygen species (ROS) generation. To explore whether this mechanism is underlying the perhexiline induced cytotoxicity, we measured the expression of hallmark genes involved in the ROS generation. As demonstrated in Fig. 7A, the mRNA levels of ROS hallmark proteins NAD(P)H:quinone oxidoreductase 1 (NQO1), heme oxygenase 1 (HMOX1), glutathione synthetase (GSS) and glutamate cysteine ligase catalytic subunit (GCLC) were unaltered upon perhexiline treatment. Western blotting analysis of the protein expression level also suggested that ROS related proteins including SOD1, SOD2, glutathione reductase, γ-GCS, HO1, and NQO1, did not have much changes upon perhexiline treatment (Fig. 7B). Direct measurement of ROS levels using H2DCFDA only showed slight increase (1.2–1.5 folds) at the highest perhexiline concentrations for 2 h and longer time points (Fig. 7C), when excessive cell death was occurring (Fig. 1). These data together suggested that excessive ROS generation might have a very minor role, if any, in perhexiline-induced cytotoxicity at the acute stage.

Fig. 7.

Perhexiline causes minimal change in cellular ROS response. HepG2 cells were exposed to 5, 10, 15, 20, and 25 μM perhexiline for 1, 2, or 4 h. DMSO treated cells served as the vehicle control. The mRNA levels (A) and protein levels (B) of ROS-response hallmark proteins were analyzed. (C) Cellular ROS levels were measured using H2DCFDA and expressed as fold change to the DMSO vehicle control. The results shown are the mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

3.8. Perhexiline causes mitochondrial dysfunction at 24 h

To elucidate further the mechanisms of perhexiline induced liver injury, especially at lower doses, we exposed HepG2 cells to 2.5 to 10 μM perhexiline for 24 h. Cell viability was again assessed by cellular ATP levels, MTS viability assay, and LDH release. As shown in Fig. 8, HepG2 cellular ATP levels were significantly reduced when exposed to 6.25 μM and higher concentrations of perhexiline at 24 h (Fig. 8A). Similar results were observed in the MTS assay (Fig. 8B). Significant increase in LDH release also occurred at 7.5 μM and higher concentrations of perhexiline treatment. The levels in caspase 3/7 activity, however, showed a different pattern comparing to that in shorter exposure time. Only a 1.3-fold increase in caspase 3/7 activity was observed at 8.75 μM, whereas for all the other concentrations tested the caspase 3/7 activity remained at the same level as vehicle control. This observation suggested that the caspase signaling pathway was only modestly involved in the toxicity of perhexiline at 24 h time point.

Fig. 8.

Perhexiline induces cytotoxicity at 24 h in HepG2 cells. HepG2 cells were exposed to 2.5, 3.75, 5, 6.25, 7.5, 8.75, and 10 μM perhexiline. DMSO treated cells served as the vehicle control. The cytotoxicity of perhexiline in HepG2 cells was measured using cellular ATP levels (A), MTS assay (B) and LDH assay (C) at 24 h. (D) Caspase 3/7 activity was measured at 24 h. The results shown are mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

When tested in the glucose-galactose assay, for the cells grown in galactose medium, the cellular ATP levels reduced faster and more prominently upon exposure to perhexiline at 24 h, compared with those grown in glucose containing medium (Fig. 9A). This observation suggested that mitochondrial dysfunction still played a role in the toxicity of perhexiline at 24 h. We then used JC-1 staining to assess the changes in mitochondrial membrane potential for the 24 h treatment. As shown in Fig. 9B, perhexiline induced concentration-dependent reduction in JC-1 ratio, suggesting the loss of mitochondrial membrane potential by the perhexiline treatment. Furthermore, the protein levels of Bcl-2 family members were significantly changed upon the treatment of perhexiline (Fig. 9C and D). The anti-apoptotic proteins Bcl-2 and Mcl-1 showed a decreased expression level, whereas the pro-apoptotic protein Bad demonstrated a significant increase. Together, these observations suggested that at 24 h, perhexiline treatment caused mitochondrial dysfunction in HepG2 cells.

Fig. 9.

Perhexiline causes mitochondrial dysfunction at 24 h in HepG2 cells. (A) HepG2 cells cultured in high-glucose medium or galactose medium were exposed to 2.5, 3.75, 5, 6.25, 7.5, 8.75, and 10 μM perhexiline for 24 h. DMSO treated cells served as the vehicle control. Cellular damage was assessed by cellular ATP content. (B) Mitochondrial membrane potential was measured by JC-1 staining upon perhexiline treatment for 24 h. (C and D) Total cellular extracts from HepG2 cells were harvested after 24 h perhexiline exposure, and the expression levels of Bcl-2 family proteins were analyzed by Western blotting. (C) representative Western blots and (D) quantification in HepG2 cells are shown here. The results shown are the mean ± S.D. from 3 independent experiments. *p < .05 compared to DMSO control.

4. Discussion

Drug-induced liver injury and its underlying mechanisms have received increasing research interest in recent years, as drug-induced liver toxicity is a cause of clinical development failure and drug withdrawal from the market (Bissell et al., 2001; Dara et al., 2016; Holt and Ju, 2006; Yuan and Kaplowitz, 2013). Among all the mechanisms, apoptosis and mitochondrial dysfunction are closely related and have been shown to contribute to the toxicity of multiple drugs (Apostolova et al., 2010; Chen et al., 2014; Gomez-Lechon et al., 2003; Isomoto et al., 2004; Jaeschke et al., 2012; Li et al., 2012; Xuan et al., 2018). In the current study, we assessed the cytotoxicity of perhexiline in hepatic cells including HepG2, HepaRG, and primary human hepatocytes. Our results suggested that both apoptosis and mitochondrial dysfunction are involved in the hepatotoxicity of perhexiline.

In the few countries where perhexiline is still being prescribed, the plasma concentration of the drug is < 600 μg/L (2.16 μM) (Ashrafian et al., 2007; Killalea and Krum, 2001). Patients have been reported to develop hepatic or neurological side effect when the plasma concentration of perhexiline is 720 to 2680 μg/L (2.59 to 9.65 μM) (Ashrafian et al., 2007). However, previous studies have suggested that due to the amphipathic nature of the drug, perhexiline tends to accumulate in tissues including the liver (Ashrafian et al., 2007). In addition, a fast-loading regimen exists for perhexiline, which may result in a temporarily higher concentration than the desired plasma concentration (Philpott et al., 2004). It is well recognized that numerous variables contribute to individual susceptibility of idiosyncratic drugs, including sex, age, race, metabolic capacity, pre-existing conditions, and drug-drug interaction (Lee, 2003). Thus to identify the idiosyncratic hepatotoxic potential of a drug, evaluation of doses up to 100-fold of the Cmax is recommended (Xu et al., 2008). In the current study, we used concentrations of 5–25 μM for perhexiline. For the most acute treatments, we observed severe cytotoxic effects starting at 10 μM, whereas for 24 h treatment, significant cellular damages were detected starting at 6.25 μM. We believe these data presented here are clinically relevant.

One of the major contributions of current study is an in-depth understanding of the mitochondrial liability of perhexiline. Mitochondrial damage has been recognized as the underling mechanism of many drugs and has received increasing research focus (Krähenbühl, 2001). Our results showed that exposure to perhexiline induced loss of mitochondrial membrane potential (Fig. 4D), via the formation of MPTP in an ANT-dependent manner (Fig. 4E). In addition, perhexiline treatment changed the protein levels of Bcl-2 family, which led to the activation of apoptosis (Fig. 5). Direct measurement of the individual mitochondrial oxidative phosphorylation complex activity suggested that perhexiline caused the most prominent inhibition in Complex IV and V, whereas little inhibition effect was observed in Complex I (Fig. 6). Previous study by Deschamps et al. (Deschamps et al., 1994) reported that perhexiline tends to concentrate in the mitochondria and results in up to 20-fold concentration higher in the mitochondrial pellet compared to the medium. Considering the low IC₅₀ levels we observed for Complex IV and V, it is of great concern to monitor patients with pre-existing mitochondrial conditions during perhexiline treatment. It is also worth noting that Deschamps and colleagues observed stronger inhibitory effects of perhexiline on Complex I and II, but not on Complex III and IV. This is different from our observation and the discrepancy is not clear, it could stem from the different assays used. The immunocapture assays we used have the capacity to identify directly individual mitochondrial complexes whose inhibition is responsible for the observed toxicity, as discussed by Nadanaciva et al. (Nadanaciva et al., 2007)

Perhexiline carries out its pharmaceutical effects through inhibition of CPT1 and 2, and thus inhibits fatty acid metabolism in the myocardium (Kennedy et al., 1996). Multiple drugs and chemicals that are also CPT inhibitors, such as amiodarone, etomoxir, and tamoxifen, have been reported to have mitochondrial liability (Cardoso et al., 2001; Massart et al., 2013; Prill et al., 2016; Spaniol et al., 2001; Tuquet et al., 2000; Vickers et al., 2006), suggesting that mitochondrial dysfunction is a shared mechanism in the toxicity of drugs in this category. In addition, some of these drugs have been reported to cause excessive reactive oxygen species (ROS) generation as well (Bekele et al., 2016; Lee et al., 2000; Serviddio et al., 2011). For perhexiline, however, excessive ROS generation is unlikely involved in its cytotoxicity. As demonstrated in Fig. 7, neither the mRNA levels nor the protein levels of the hallmarks of ROS generation showed significant changes upon perhexiline treatment. Direct measurement of relative ROS levels showed that only modestly changes occurred at the highest levels tested. Therefore, although excessive ROS production is often closely associated with mitochondrial dysfunction, it is unlikely a causal factor in the hepatotoxicity of perhexiline under our experimental conditions.

Mitochondria play a critical role in the regulation of apoptosis, particularly through the Bcl-2 family proteins (Adams and Cory, 2007; Kroemer et al., 1998; Nieminen, 2003; Ola et al., 2011). In our study, we found that exposure to perhexiline caused a significant increase in caspase 3/7 activity in HepG2 cells at 20 μM for 2 h or 4 h, and even a higher elevation at 25 μM for 2 h, indicating apoptosis contributes to the cytotoxicity of perhexiline at 2 h and 4 h (Fig. 2A). Using Western blotting, we demonstrated significant decrease in the protein levels of pro-survival Bcl-2 family proteins Bcl-2 and Mcl-1, and increase in the expression level of Bad, a pro-apoptotic member of Bcl-2 family (Fig. 5). The orchestrated effects of these proteins led to the activation of caspase 9 and the intrinsic pathway of apoptosis (Fig. 2B and D). It is worth noting that fast and severe mitochondrial damage may deplete cellular ATP and block the activation of caspases (Nieminen, 2003). In current study, we observed that, when the energy source in the medium changed from glucose to galactose and the mitochondrial dysfunction was more prominent, the increase of caspase 3/7 activity was completely abolished despite severe cell death (Supplemental Fig. 1, Fig. 4A-C), suggesting a shift from apoptosis to necrosis as the cell death mechanism. Moreover, necrosis also contributed to the hepatotoxicity of perhexiline, as evidenced by the partial rescue effects of caspase inhibitors and the attenuating effect of necrosis inhibitor Necrox 5 (Fig. 3). In addition to the intrinsic pathway, apoptosis can also be activated through the extrinsic pathway by death receptor ligands such as TNF-α (Nair et al., 2014). We observed significant increased level of TNF-α upon perhexiline exposure (Fig. 2C); inhibition of caspase 8 activity also attenuated perhexiline-induced apoptosis (Fig. 3A and B), suggesting both intrinsic and extrinsic pathways of apoptosis contribute to the cytotoxicity of perhexiline.

It is interesting that at 24 h the changes in caspase 3/7 level were very modest (Fig. 8D). One explanation is that the prominent activation of caspase signaling pathway may only occur at the most acute phase of the perhexiline toxicity, whereas for prolonged exposure other mechanisms take more dominant roles. Previous studies have shown that caspase activation follows a rapid, “all-or-none” manner (Rehm et al., 2002; Tyas et al., 2000). It is also possible that higher concentrations of perhexiline activates different mechanisms in comparison to lower concentrations for a longer exposure time. On the other hand, the results from glucose-galactose assay, JC-1 staining, and the protein levels of Bcl-2 family members all suggested that mitochondrial damage still contributed to the cytotoxicity of perhexiline at 24 h (Fig. 9), further confirmed the importance of mitochondrial dysfunction in the cytotoxicity of perhexiline.

It is worth noting that increase of caspase 3/7 was also observed in HepaRG and primary human hepatocytes at 2 h for 25 μM perhexiline (Supplemental Fig. 2); however, the extent of the increase was smaller compared to that in HepG2 cells. This might result from the different culture systems and also from the difference in metabolizing activities in these two types of hepatic cells, in comparison to HepG2 cells (Ren et al., 2018). Clinically, perhexiline is metabolized by CYP2D6, 3A4, and 2B6, and severe hepatotoxicity has been reported in patients with poor CYP2D6 metabolism, which result in excessive accumulation of the parent drug in plasma (Cooper et al., 1987; Cooper et al., 1984; Gardiner and Begg, 2006; Sorensen et al., 2003). Since the toxicity of perhexiline is possibly due to the parent drug, we do not consider the limited metabolizing capability of HepG2 cells a major concern for the current study. Moreover, all the trends we observed in HepG2 cells upon perhexiline exposure were also observed in primary human hepatocytes, including the rapid decrease in cellular ATP content (Fig. 1D), an increase of caspase 3/7 activity (Supplemental Fig. 2A), a decrease in the JC-1 ratio (Supplemental Fig. 2B), an increase in the TNF-α mRNA level (Supplemental Fig. 2C), and the changes of mitochondrial proteins (Supplemental Fig. 2D), suggesting that the cytotoxic effects we observed were not limited to HepG2 cells. Studies on the metabolism of perhexiline and its effect on the toxicity of the drug are currently undergoing in our laboratory.

Overall, our study demonstrates that both mitochondrial dysfunction and apoptosis contribute to the cytotoxicity of perhexiline. Both the intrinsic and extrinsic pathways of apoptosis were activated upon perhexiline exposure. Perhexiline also directly inhibited the complexes in the oxidative phosphorylation chain, especially Complex IV and V. Our study facilitates the understanding of mitochondrial liability and apoptosis in drug-induced liver injury.