Generation of HepG2 Cells with High Expression of Multiple Drug-Metabolizing Enzymes for Drug Discovery Research Using a PITCh System

HepG2 cells are an inexpensive hepatocyte model that can be used for repeated experiments, but HepG2 cells do not express major cytochrome P450s (CYPs) and UDP glucuronosyltransferase family 1 member A1 (UGT1A1). In this study, we established CYP3A4–POR–UGT1A1–CYP1A2–CYP2C19–CYP2C9–CYP2D6 (CYPs–UGT1A1) knock-in (KI)-HepG2 cells using a PITCh system to evaluate whether they could be a new hepatocyte model for pharmaceutical studies. To evaluate whether CYPs–UGT1A1 KI-HepG2 cells express and function with CYPs and UGT1A1, gene expression levels of CYPs and UGT1A1 were analyzed by using real-time PCR, and metabolites of CYPs or UGT1A1 substrates were quantified by HPLC. The expression levels of CYPs and UGT1A1 in the CYPs–UGT1A1 KI-HepG2 cells were comparable to those in primary human hepatocytes (PHHs) cultured for 48 h. The CYPs and UGT1A1 activity levels in the CYPs–UGT1A1 KI-HepG2 cells were much higher than those in the wild-type (WT)-HepG2 cells. These results suggest that the CYPs–UGT1A1 KI-HepG2 cells expressed functional CYPs and UGT1A1. We also confirmed that the CYPs–UGT1A1 KI-HepG2 cells were more sensitive to drug-induced liver toxicity than the WT-HepG2 cells. CYPs–UGT1A1 KI-HepG2 cells could be used to predict drug metabolism and drug-induced liver toxicity, and they promise to be a helpful new hepatocyte model for drug discovery research.


Introduction
An in vitro hepatocyte model for evaluation of drug metabolism and drug-induced liver injury would be useful for drug discovery research [1][2][3]. Most drugs used in clinical practice are metabolized by one of five cytochrome P450s (CYPs): cytochrome P450 family 1 subfamily A member 2 (CYP1A2), cytochrome P450 family 2 subfamily C member 9 (CYP2C9), cytochrome P450 family 2 subfamily C member 19 (CYP2C19), cytochrome P450 family 2 subfamily D member 6 (CYP2D6) and cytochrome P450 family 3 subfamily A member 4 (CYP3A4) [4,5], all associated with phase I, or drug-activating metabolic reactions. The role of P450 oxidoreductase (POR, a coenzyme of CYPs) is also important in Green Master Mix was purchased from Thermo Fisher Scientific (Waltham, WA, USA). The ReveTra Ace ® qPCR RT kit was purchased from Toyobo (Osaka, Japan). Fetal bovine serum (FBS) was purchased from Sigma-Aldrich (Munich, Germany). Tks Gflex DNA polymerase was purchased from Takara Bio (Kusatsu, Japan). Collagen I was purchased from Corning (New York, NY, USA). PEI MAX TM -Transfection Grade Linear Polyethylenimine Hydrochloride (MW, 40,000) was purchased from Polysciences (Warrington, PA, USA). All other reagents were purchased as commercially available.

HepG2 Cells
The human hepatoblastoma cell line, HepG2 cells (RCB1648), was provided by the RIKEN BRC through the National BioResource Project of the MEXT/AMED, Japan. The HepG2 cells were cultured with DMEM containing 10% FBS, 1% Minimum Essential Medium Non-Essential Amino Acids Solution, and 1% Antibiotic-Antimycotic Mixed Stock Solution.

Primary Human Hepatocytes
Cryopreserved human hepatocytes (HUCPI, Lonza) were used in this study. The PHH culture protocol was described previously [25,26]. The human hepatocytes were seeded at 1.0 × 10 5 cells/cm 2 onto collagen I-coated 48-well plates. The PHHs cultured for 48 h after plating were used in the experiments.

Real-Time RT-PCR
Total RNA was isolated from the HepG2 cells using Sepasol ® -RNA I Super G. Total RNA in the PHHs was isolated using ISOGEN. According to the manufacturer's protocol, cDNA was synthesized with ReveTra Ace ® qPCR RT kit. The real-time RT-PCR protocol was described previously [24].

Cell Viability Tests
The HepG2 cells were seeded in 96-well plates at 1.0 × 10 4 cells/well. The next day, the HepG2 cells were treated with various concentrations of amiodarone, benzbromarone, acetaminophen, imipramine, or troglitazone. After 24 h, to examine the cell viability, we performed a WST-8 (2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt) assay by using Cell Count Reagent SF according to the manufacturer's instructions. The cell viability was calculated as the percentage of that in the cells treated with a vehicle only.

Statistical Analysis
Statistical analyses were performed as indicated in figure legends using the v. 1.55 Easy R (EZR) software. A value of p < 0.05 was considered statistically significant. Using the SigmaPlot v. 14.5 statistical software (Systat Software, San Jose, CA, USA), 50% inhibitory concentrations (IC50) were determined by use of nonlinear regression analysis.
Phase images showed that there was no morphologic difference between the WT-HepG2 cells and the CYPs-UGT1A1 KI-HepG2 cells ( Figure 2B). These results suggest that the knocking-in of CYPs and UGT1A1 does not negatively affect HepG2 cells.
FOR PEER REVIEW 6 of 16 h) as positive controls. The expression levels of hepatic marker genes in the CYPs-UGT1A1 KI-HepG2 cells were similar to those in the WT-HepG2 cells. The expression levels of drug-metabolizing enzyme genes (CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP3A4, POR, and UGT1A1) in the CYPs-UGT1A1 KI-HepG2 cells were comparable to those in the PHHs 48 h (Figure 2A). In the CYPs-UGT1A1 KI-HepG2 cells, there was no decrease in the expression levels of drug-metabolizing enzymes during passages 38 to 57 ( Figure S1). Phase images showed that there was no morphologic difference between the WT-HepG2 cells and the CYPs-UGT1A1 KI-HepG2 cells ( Figure 2B). These results suggest that the knocking-in of CYPs and UGT1A1 does not negatively affect HepG2 cells. and hepatocyte nuclear factor 4 alpha (HNF4A), which served as hepatic markers, and for the drugmetabolizing enzymes cytochrome P450 family 1 subfamily A member 2 (CYP1A2), cytochrome P450 family 2 subfamily C member 9 (CYP2C9), cytochrome P450 family 2 subfamily C member 19 (CYP2C19), cytochrome P450 family 2 subfamily D member 6 (CYP2D6), cytochrome P450 family 3 subfamily A member 4 (CYP3A4), P450 oxidoreductase (POR), and UDP glucuronosyltransferase family 1 member A1 (UGT1A1) in the WT-HepG2 cells, CYPs-UGT1A1 KI-HepG2 cells, 48 h-cultured primary human hepatocytes (PHHs 48 h), and PHHs collected immediately after thawing (PHHs 0 h). On the y-axis, the gene expression levels in the PHHs 0 h were taken as 1.0. The data represent the means ± SD (n = 3, technical replicates). N.D., not detected. Statistical significance was evaluated by one-way ANOVA followed by Tukey's post-hoc test (p < 0.05). The groups that do not share the same letter had significantly different results.
Many drugs are metabolized by multiple cytochrome P450s or in different ways by the same P450. We investigated the metabolism of propranolol in the CYPs-UGT1A1 KI-HepG2 cells. In the WT-HepG2 cells, the three main metabolites (4′-hydroxy propranolol, 5′-hydroxy propranolol, and desisopropylpropranolol) were undetectable, whereas they were present at detectable levels in the CYPs-UGT1A1 KI-HepG2 cells ( Figure 5).

Formation of 1-OH Midazolam (pmol/mg/min
Many drugs are metabolized by multiple cytochrome P450s or in different ways by the same P450. We investigated the metabolism of propranolol in the CYPs-UGT1A1 KI-HepG2 cells. In the WT-HepG2 cells, the three main metabolites (4 -hydroxy propranolol, 5 -hydroxy propranolol, and desisopropylpropranolol) were undetectable, whereas they were present at detectable levels in the CYPs-UGT1A1 KI-HepG2 cells ( Figure 5). These results suggest that the CYPs-UGT1A1 KI-HepG2 cells expressed functional CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP3A4, and UGT1A1. In addition, the results demonstrated that CYPs-UGT1A1 KI-HepG2 cells can be utilized in drug-drug interaction studies. Moreover, the CYPs-UGT1A1 KI-HepG2 cells were able to predict the metabolites of the drugs that undergo different reactions under the influence of multiple cytochrome P450s.

Prediction of Drug-Induced Liver Injury Using CYPs-UGT1A1 KI-HepG2 Cells
To examine whether CYPs-UGT1A1 KI-HepG2 cells could be used to predict druginduced liver injury, the WT-HepG2 cells and the CYPs-UGT1A1 KI-HepG2 cells were treated with five hepatotoxic drugs (amiodarone, benzbromarone, acetaminophen, imipramine, and troglitazone), and cell viability was measured by WST-8 assay. We found that the viability of the CYPs-UGT1A1 KI-HepG2 cells was lower than the viability of the WT-HepG2 cells (Figure 6). These results suggest that the CYPs-UGT1A1 KI-HepG2 cells were more sensitive to drug-induced liver toxicity than the WT-HepG2 cells. These results suggest that the CYPs-UGT1A1 KI-HepG2 cells expressed functional CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP3A4, and UGT1A1. In addition, the results demonstrated that CYPs-UGT1A1 KI-HepG2 cells can be utilized in drug-drug interaction studies. Moreover, the CYPs-UGT1A1 KI-HepG2 cells were able to predict the metabolites of the drugs that undergo different reactions under the influence of multiple cytochrome P450s.

Prediction of Drug-Induced Liver Injury Using CYPs-UGT1A1 KI-HepG2 Cells
To examine whether CYPs-UGT1A1 KI-HepG2 cells could be used to predict druginduced liver injury, the WT-HepG2 cells and the CYPs-UGT1A1 KI-HepG2 cells were treated with five hepatotoxic drugs (amiodarone, benzbromarone, acetaminophen, imipramine, and troglitazone), and cell viability was measured by WST-8 assay. We found that the viability of the CYPs-UGT1A1 KI-HepG2 cells was lower than the viability of the WT-HepG2 cells ( Figure 6). These results suggest that the CYPs-UGT1A1 KI-HepG2 cells were more sensitive to drug-induced liver toxicity than the WT-HepG2 cells.

Discussion
We succeeded in generating CYPs-UGT1A1 KI-HepG2 cells using the PITCh system. Furthermore, the CYPs-UGT1A1 KI-HepG2 cells showed drug-metabolizing enzyme gene expression levels comparable to those of PHHs 48 h and had a very high drug-metabolizing capacity compared with the WT-HepG2 cells (Figures 2 and 3). Several other groups have also reported attempts to increase the expression of drug-metabolizing enzymes in HepG2 cells by various methods. Most of them are reports of overexpression of single drug-metabolizing enzymes using plasmid vectors or viral vectors [17,[32][33][34][35]. Recently, Kazuki et al. used an artificial chromosome technology to generate HepG2 cells stably expressing CYP2C9, CYP2C19, CYP2D6, CYP3A4, and POR (transchromosomic HepG2 cells) [36]. Consistent with the results of their study, we showed that CYPs-UGT1A1 KI-HepG2 cells can accurately predict drug metabolism and hepatotoxicity. However, transchromosomic HepG2 cells express very little CYP1A2 and UGT1A1, so it is difficult to predict the metabolism of certain drugs such as propranolol ( Figure 5). In the metabolism of propranolol, CYP2D6 and CYP1A2 play critical roles and are involved in three main metabolic reactions, yielding 4 -hydroxy propranolol (CYP2D6), 5 -hydroxy propranolol (CYP2D6), and desisopropylpropranolol (CYP1A2), respectively [37][38][39]. Many drugs are excreted from the body after oxidation by cytochrome P450 (phase I reaction) and conjugation via UGT (phase II reaction) [4]. It is necessary to investigate whether CYPs-UGT1A1 KI-HepG2 cells can predict the effects of metabolism by CYPs and subsequent conjugation reactions.
The expression levels of UGT1A1 in the CYPs-UGT1A1 KI-HepG2 cells were comparable to those of PHHs 48 h, but its activity was lower than that of PHHs 48 h (Figures 2  and 3). It is known that 7 -hydroxy coumarin is glucuronidated not only by UGT1A1, but also by various UGT1A isoforms [40][41][42]. Therefore, if multiple UGT isoforms can be overexpressed, it will be possible to generate a model with a UGT metabolic capacity comparable to that of PHHs.
There are multiple SNPs in cytochrome P450 and UGT1A1. SNPs, which can drastically change drug metabolism, may also cause unexpected side effects [43,44]. If the potential effects of individual differences in SNPs could be predicted at early stages of drug discovery research, it would be possible to develop drugs more efficiently. However, most PHH donors used in general drug discovery research are Caucasian, so safety studies reflecting Japanese-specific SNPs such as UGT1A1*6 and CYP2D6*10 cannot be adequately conducted. We have successfully expressed UGT1A1*6 in Caco-2 cells using the PITCh system [24]. Furthermore, we have shown that UGT1A1*6-expressing Caco-2 cells can be applied to drug metabolism and drug-induced toxicity experiments that reflect the effects of patients with UGT1A1*6 [24]. By generating CYPs-UGT1A1 KI-HepG2 cells reflecting the effects of patients with CYP2D6*10 and UGT1A1*6, it will be possible to construct an in vitro hepatocyte model that can predict liver metabolism and toxicity reflecting a more representative range of individual differences.
The CYPs-UGT1A1 KI-HepG2 cells were more sensitive to drug-induced liver toxicity than the WT-HepG2 cells ( Figure 6). Amiodarone is metabolized by CYP3A4 to produce desethyl amiodarone, which is known to be more hepatotoxic than amiodarone [45,46]. Benzbromarone is metabolized by CYP2C9 and CYP3A4, and its metabolites have been reported to be hepatotoxic [47,48]. N-acetyl-p-benzoquinone imine (NAPQI), a major metabolite and toxicant of acetaminophen, is reported to be produced not only by CYP2E1, but also by CYP3A4 and CYP1A2 [49,50]. It has also been reported that acetaminophen-induced hepatotoxicity is attenuated in CYP3A4-knockout human iPS cell-derived hepatocyte-like cells [51]. Thus, increased NAPQI levels due to CYP3A4 or CYP1A2 in CYPs-UGT1A1 KI-HepG2 cells may have reduced cell viability. Since CYPs-UGT1A1 KI-HepG2 cells have a higher drug-metabolizing capacity than WT-HepG2 cells (Figure 3), they are a better model to evaluate the potential toxicity of metabolites.
Preclinical drug screening should assess the induction of drug-metabolizing enzymes [6,50,52], which would be difficult to predict because CYPs-UGT1A1 KI-HepG2 cells overexpress CYPs and UGT1A1 using the CAG promoter. Therefore, preclinical drug screening using CYPs-UGT1A1 KI-HepG2 cells would be limited to drug metabolism, drug-drug interactions, and drug-induced liver toxicity.

Conclusions
Using the PITCh system, we successfully generated CYPs-UGT1A1 KI-HepG2 cells. The CYPs-UGT1A1 KI-HepG2 cells expressed functional CYPs and UGT1A1. Furthermore, the CYPs-UGT1A1 KI-HepG2 cells showed use for predicting drug metabolism, drug-drug interactions, and drug-induced liver toxicity that cannot be assessed using single drugmetabolizing enzyme overexpression models. CYPs-UGT1A1 KI-HepG2 cells will be a helpful new hepatocyte model for drug discovery research.

Data Availability Statement:
The authors declare that all the data related to this study are available within the paper or can be obtained from the authors upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.