Customised in vitro model to detect human metabolism-dependent idiosyncratic drug-induced liver injury

Abstract

Drug-induced liver injury (DILI) has a considerable impact on human health and is a major challenge in drug safety assessments. DILI is a frequent cause of liver injury and a leading reason for post-approval drug regulatory actions. Considerable variations in the expression levels of both cytochrome P450 (CYP) and conjugating enzymes have been described in humans, which could be responsible for increased susceptibility to DILI in some individuals. We herein explored the feasibility of the combined use of HepG2 cells co-transduced with multiple adenoviruses that encode drug-metabolising enzymes, and a high-content screening assay to evaluate metabolism-dependent drug toxicity and to identify metabolic phenotypes with increased susceptibility to DILI. To this end, HepG2 cells with different expression levels of specific drug-metabolism enzymes (CYP1A2, CYP2B6, CYP2C9, CYP2C19, CYP2D6, CYP2E1, CYP3A4, GSTM1 and UGT2B7) were exposed to nine drugs with reported hepatotoxicity. A panel of pre-lethal mechanistic parameters (mitochondrial superoxide production, mitochondrial membrane potential, ROS production, intracellular calcium concentration, apoptotic nuclei) was used. Significant differences were observed according to the level of expression and/or the combination of several drug-metabolism enzymes in the cells created ad hoc according to the enzymes implicated in drug toxicity. Additionally, the main mechanisms implicated in the toxicity of the compounds were also determined showing also differences between the different types of cells employed. This screening tool allowed to mimic the variability in drug metabolism in the population and showed a highly efficient system for predicting human DILI, identifying the metabolic phenotypes associated with increased DILI risk, and indicating the mechanisms implicated in their toxicity.

Keywords: CYP; Cell model; Drug-induced liver injury; Hepatotoxicity mechanisms; Idiosyncrasy.

Fig. 1

Multiparametric assessment of troglitazone and tienilic acid toxicity in HepG2 cells that express variable levels of a single CYP. Individually, AdCYP-transduced cells were treated for 24 h with increasing concentrations of troglitazone (a–c) or tienilic acid (d–f). Then the effects on the apoptotic cell number (apoptosis), mitochondrial superoxide production (Mit. superoxide), intracellular calcium level (Ca), ROS generation (ROS) and mitochondrial membrane potential (MMP) were analysed by HCS. a Heatmap of troglitazone (0, 50, 75, 100, 150, 200, 250, 300 and 400 µM) and CYP3A4. Differential effects of troglitazone on apoptosis (b) and mitochondrial superoxide production (c) according to the level of CYP3A4. d Heatmap of tienilic acid (0, 25, 100, 200, 400, 600, 800 and 1200 µM) and CYP2C9. Effects on mitochondrial superoxide (e) and ROS production (f) depending on the CYP2C9 level. CYP: 0 (non-transduced cells); 1, 4 and 10 correspond to the cells transduced with a single AdCYP to reach the same level, four-fold or ten-fold, respectively, of CYP activity in human hepatocytes. *p < 0.01; **p < 0.001 compared to non-transduced cells (Student’s t test)

Fig. 2

Multiparametric assessment of flutamide and perhexiline toxicity to the HeoG2 cells that express variable levels of a single CYP. Individually, AdCYP-transduced cells were treated for 24 h with increasing concentrations of flutamide (a–c) or perhexiline (d–f). Dose-dependent effects on the mitochondrial membrane potential (MMP), ROS generation (ROS), number of apoptotic cells (apoptosis), mitochondrial superoxide production (Mit. superoxide) and intracellular calcium level (Ca) were determined by the HCS analysis. a Heatmap of flutamide (0, 50, 75, 100, 125, 250, 500, 750 and 1000 µM) and CYP3A4, CYP1A2 or CYP2C19. Differential effects of flutamide on ROS production (b) and induction of apoptosis (c) according to the level of CYPs. d Heatmap of perhexiline (0, 10, 12.5, 15, 17.5, 20, 22.5, 25 and 27.5 µM) and CYP2D6; neutral lipids accumulation (lipids) was also analysed. Effects on lipid overaccumulation (e) and mitochondrial superoxide production (f) depending on the CYP2D6 level. CYP: 0 (non-transduced cells); 1, 4 and 10 correspond to the cells transduced with a single AdCYP to reach the same level, four-fold or ten-fold, respectively, of CYP activity in human hepatocytes. **p < 0.001 compared to non-transduced cells (Student’s t test)

Fig. 3

Isoniazid-induced toxicity to the HepG2 cells co-transduced with AdCYP2E1 and AdGSTM1. The HepG2 cells transduced with different amounts of AdCYP2E1, either alone or in combination with AdGSTM1, were treated for 24 h with isoniazid (0, 5, 10, 20, 30, 35, 40, 45 and 50 mM). a Heatmap showing isoniazid toxicity. Dose-dependent effects on the mitochondrial membrane potential (MMP), ROS generation (ROS), number of apoptotic cells (apoptosis), mitochondrial superoxide production (Mit. superoxide) and intracellular calcium level (Ca) were determined by the HCS analysis. Differential effects on ROS (B) and mitochondrial superoxide production (C), depending on the level and combination of enzymes, are shown. CYP2E1: 0 (non-transduced cells); 1, 4 and 10 (cells transduced to reach the same level, four-fold or ten-fold, of CYP2E1 activity in human hepatocytes, respectively); GSTM1: 0 (non-transduced cells) and 1 (cells transduced to reach the same GST activity as human hepatocytes). *p < 0.01, **p < 0.001 compared to non-transduced cells; #p < 0.01 compared with and without GSTM1 (Student’s t test)

Fig. 4

Acetaminophen-induced toxicity to the HepG2 cells co-transduced with AdCYP2E1, CYP1A2 and AdGSTM1. The HepG2 cells transduced with a mixture of AdCYP2E1 and AdCYP1A2, in combination or not with AdGSTM1, were treated for 24 h with acetaminophen (0, 0.5, 1, 2, 4, 6, 8, 10 and 15 mM). a Heatmap of acetaminophen toxicity in the cells transduced with different adenoviruses. Dose-dependent effects on the mitochondrial membrane potential (MMP), ROS generation (ROS), number of apoptotic cells (apoptosis), mitochondrial superoxide production (Mit. superoxide) and intracellular calcium level (Ca) were determined by the HCS analysis. Differential effects on mitochondrial superoxide (b) and ROS production (c) according to the level or combination of enzymes are shown. CYP2E1 + AdCYP1A2: 0 (non-transduced cells); 1, 4 and 10 (cells co-transduced to reach the same level, four-fold or ten-fold, of each CYP activity in human hepatocytes, respectively); GSTM1: 0 (non-transduced cells) and 1 (cells transduced to reach the same GST activity as human hepatocytes). *p < 0.01, **p < 0.001 compared to non-transduced cells; #p < 0.001 compared with and without GSTM1 (Student’s t test)

Fig. 5

Valproate-induced toxicity to the HepG2 cells co-transduced with AdCYP2B6 and CYP2C9. The HepG2 cells transduced with different amounts of AdCYP2B61, either alone or in combination with AdCYP2C9, were treated for 24 h with valproate (0, 0.5, 1, 2, 4, 6, 8, 10 and 12 mM). a Heatmap of valproate toxicity. Dose-dependent effects on the neutral lipid content (lipids), number of apoptotic cells (apoptosis), mitochondrial superoxide production (Mit. superoxide), intracellular calcium level (Ca), ROS generation (ROS), and mitochondrial membrane potential (MMP) were determined by the HCS analysis. Differential effects on mitochondrial superoxide production (b) and lipid overaccumulation (c) in the cells that expressed different levels of CYP2B6 and CYP2C9 are exemplified. CYP2B6: 0 (non-transduced cells); 1, 4 and 10 (cells transduced to reach the same level, four-fold or ten-fold, of CYP2B6 activity in human hepatocytes, respectively); CYP2C9: 0 (non-transduced cells) and 1 (cells transduced to reach the same CYP2C9 activity as human hepatocytes). *p < 0.01, **p < 0.001 compared to non-transduced cells; #p < 0.001 comparing cells expressing different levels of CYP2B6 with and without CYP2C9 (Student’s t test)

Fig. 6

Diclofenac toxicity to the HepG2 cells co-transduced with AdCYP3A4, AdCYP2C9 and AdUGT2B7. The HepG2 cells transduced with different amounts of AdCYP3A4, either alone or in combination with AdCYP2C9 and/or UGT2B7, were treated for 24 h with diclofenac (0, 5, 10, 50, 125, 250, 500, 1000 and 1500 µM). a Heatmap of diclofenac-induced hepatotoxicity. Dose-dependent effects on the intracellular calcium level (Ca), ROS generation (ROS), number of apoptotic cells (apoptosis), mitochondrial superoxide production (Mit. superoxide) and mitochondrial membrane potential (MMP) were determined by the HCS analysis. Differential effects on apoptosis (b) and mitochondrial superoxide production (c) in the cells that expressed different levels, or a combination of enzymes, are shown. CYP3A4: 0 (non-transduced cells); 1, 4 and 10 (cells transduced to reach the same level, four-fold or ten-fold, of CYP3A4 activity in human hepatocytes, respectively); CYP2C9: 0 (non-transduced cells) and 1 (cells transduced to reach the same CYP2C9 activity as human hepatocytes); UGT2B7: 0 (non-transduced cells) and 1 (cells transduced to reach the same UGT2B7 activity as human hepatocytes). *p < 0.01, **p < 0.001 compared to non-transduced cells (Student’s t test)

Fig. 7

Differences in the TR depending on the AdCYP dose. The TR was calculated as the 100 × C _max/MEC ratio and was normalised by the TR value of the HepG2 cells (**p = 0.002, paired t test)