Nutrient environment improves drug metabolic activity in human iPSC-derived hepatocytes and HepG2

Abstract

Induced pluripotent stem cells (iPSCs) have emerged as a transformative tool in regenerative medicine, in liver research. The perspective of a stable and functional source of hepatocytes has led to developing protocols for human iPSC-derived hepatocytes-like cells (HLCs). Yet, hepatic models remain one of most challenging systems to functionally reproduce with iPSCs, due to its resulting limited metabolic function. Using an adapted nutrient regimen, two human hepatocyte models were characterized: HLCs (derived from iPSCs) and metabolically active HepG2 (mHepG2, derived from the cell line HepG2), for their drug metabolism activity. In these cell systems, the transcriptome, proteome, and metabolome of 11 drug-relevant cytochrome P450 (CYP) isoenzymes were studied. A liquid chromatography-mass spectrometry (LC-MS)-based metabolomics approach, using model drugs as isoenzyme reporters, was applied, achieving a comprehensive overview of mHepG2 and HLCs drug metabolism. Drugs used in this study to characterize xenobiotic machinery were: bupropion (25 µM), phenacetin (30 µM), rosiglitazone (10 µM), diclofenac (75 µM), dextromethorphan (15 µM), chlorzoxazone (60 µM), midazolam (15 µM), benzydamine (15 µM), coumarin (250 µM) and 7-ethoxycoumarin (60 µM). Being HepG2 notorious for its limited metabolic capacity, our study raises mHepG2 as a highly performant cell model, with activity on 8 drug-metabolizing CYPs. Modulation by nutrient environment in improving metabolic function of in vitro models is here proven as a key determinant. Likewise, HLCs hold the widest CYP coverage at the transcript level and were able to cope with a wide variety of chemical insults, making them a promising model for personalized metabolic studies.

Keywords: Cytochrome P450 (CYP); Drug metabolism; Induced pluripotent stem cells (iPSCs); LC–MS; Liver; Metabolomics.

Fig. 1

Expression of drug-metabolizing CYPs differs across hepatic cell systems. Gene expression obtained by qPCR, relative to expression of housekeeping gene GAPDH for CYP isoenzymes family 1: CYP1A1 and CYP1A2, family 2: CYP2A6, CYP2B6, CYP2D6, CYP2C8, CYP2C9, CYP2C19 and CYP2E1, and family 3: CYP3A4, CYP3A5. Cell models tested were primary human hepatocytes (A, PHHs); B HepaRG; iPSC-derived HLCs, following a differentiation protocol for 22 days (C, day 22 HLCs) and for 44 days (D, day 44 HLCs); E HepG2; and metabolically active HepG2 (F, mHepG2). Results are represented as average ± SD; n.d. = not detected. N = 3 replicates from the same cell line or batch of differentiation (3 independent RNA isolated wells). Only for PHHs: N = 6 replicates from the same RNA isolation of a mixed donor pool cryopreserved vial

Fig. 2

Protein levels of drug-metabolizing CYPs differ across hepatic cell systems. Protein quantification (pmol of protein/mg total protein) obtained by targeted proteomics for CYP isoenzymes family 1: CYP1A1 and CYP1A2, family 2: CYP2A6, CYP2B6, CYP2D6, CYP2C8, CYP2C9, CYP2C19 and CYP2E1, and family 3: CYP3A4, CYP3A5, on primary human hepatocytes (A, PHHs); B HepaRG; iPSC-derived HLCs, following a differentiation protocol for 22 days (C, day 22 HLCs) and for 44 days (D, day 44 HLCs); E HepG2; and metabolically active HepG2 (F, mHepG2) (average log10 ± SD); n.d. = not detected; n.m. = not measured. N = 3–5 replicates from the same cell model or batch of differentiation (samples prepared as independent lysates). In case some, but not all biological replicates were below the quantification limit, (x), ½ limit of quantification was used for the analysis (Bailey and Michelson)

Fig. 3

Kinetics of phase I metabolism of model drugs in human liver microsomes (HLMs). Model drugs were used as substrates in enzyme (isoform)-specific metabolic reactions, forming a marker metabolite: A phenacetin to acetaminophen (CYP1A2); B coumarin to 7-hydroxycoumarin (CYP2A6); C bupropion to hydroxybupropion (CYP2B6); D rosiglitazone to N-desmethylrosiglitazone (CYP2C8); E diclofenac to hydroxydiclofenac (CYP2C9); F dextromethorphan to dextrorphan (CYP2D6); G chlorzoxazone to hydroxychlorzoxazone (CYP2E1); H midazolam to hydroxymidazolam (CYP3A4); I benzydamine to benzydamine-N-oxide (FMOs); J 7-ethoxycoumarin to 7-hydroxycoumarin (phase I and phase II metabolism). Drugs (in black), marker metabolites (in red) and other identified metabolites (multiple colors) were analyzed over five-time points: 15, 30, 60, 120 and 180 min using LC–MS metabolomics, in either positive or negative ion mode. Drug and assigned putative metabolite LC–MS intensities were normalized using min–max scaling (average ± SD, indicated by the width of the line). N = 3 replicates, independent incubations per drug. The drugs coumarin (B) and 7-ethoxycoumarin (J) showed poor LC–MS ionization and thus were not displayed

Fig. 4

Phase I metabolic activity in the human hepatic cell systems HepaRG, HLCs, HepG2 and mHepG2. Model drugs (structures in grey box) were used as substrates in enzyme (isoform)-specific metabolic reactions, forming a marker metabolite (with structure): A phenacetin to acetaminophen (CYP1A2); B coumarin to 7-hydroxycoumarin (CYP2A6); C bupropion to hydroxybupropion (CYP2B6); D rosiglitazone to N-desmethylrosiglitazone (CYP2C8); E diclofenac to hydroxydiclofenac (CYP2C9); F dextromethorphan to dextrorphan (CYP2D6); G chlorzoxazone to hydroxychlorzoxazone (CYP2E1); H midazolam to hydroxymidazolam (CYP3A4); I benzydamine to benzydamine-N-oxide (FMOs); J 7-ethoxycoumarin to 7-hydroxycoumarin (phase I and phase II metabolism). Drugs (in a box) and marker metabolites were analyzed in extracellular contents after 24 h using LC–MS metabolomics. Metabolite intensities in HepaRG (pink), HLCs (purple), HepG2 (light blue), and mHepG2 (dark blue) were normalized to drug intensity in the blank and to total DNA (average ± SD, *p < 0.05). N = 3 replicates per cell model (from the same cell line or batch of differentiation) and drug (3 independent extracted wells). In case metabolites were only found intracellularly in a certain cell system, the name of the cell system was added below the drug structure. Bupropion (C) showed poor LC–MS ionization, and thus was not displayed. n.d. = not detected

Fig. 5

Qualitative multi-omics representation of drug-metabolizing CYP transcriptome (T), proteome (P), and metabolome (M) in human hepatic cell systems HepaRG, HLCs, HepG2 and mHepG2. The phase I CYP isoforms CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2D6, CYP2E1, and CYP3A4 were detected (o) or not detected (x) across the three omics, leading to the presence (in green) or absence (in red) of metabolic activity in the respective cell systems. CYP2A6 was not measured at the protein level (−)

Fig. 6

Phase II metabolism in the human hepatic cell systems HepaRG, HLCs, HepG2 and mHepG2. Protein quantification of 4 UGT isoenzymes (A) and 1 SULT isoenzyme: SULT1A1 (B). Protein levels in PHHs, HepaRG, day 22 HLCs and 44, mHepG2, and HepG2 were determined by targeted proteomics (average log10 (pmol of protein/mg protein extracted) ± SD, *p < 0.05, ns = not significant); n.d. = not detected. N = 2–5 replicates from the same cell model or batch of differentiation (samples prepared as independent lysates). In case some, but not all replicates were below the quantification limit (x), ½ limit of quantification was used for representation (Bailey and Michelson); UGT2B7 in PHHs was above quantification limit; therefore, this value was illustrated alternatively. UGT (C) and SULT (D) metabolic activity. Phase II metabolites of model drugs chlorzoxazone, benzydamine, coumarin, rosiglitazone and 7-ethoxycoumarin analyzed in extracellular contents of each cell system, after 24 h using LC–MS metabolomics. Metabolite intensities in HepaRG (pink), HLCs (purple), HepG2 (light blue), and mHepG2 (dark blue) were normalized to drug intensity in the blank and to total DNA (average ± SD, *p < 0.05). N = 3 replicates per cell model (from the same cell line or batch of differentiation) and drug (3 independent extracted wells)

Fig. 7

Drug metabolism map of benzydamine. Metabolic fate of benzydamine (in grey box) with assigned metabolic enzymes and (putative) metabolites found in the literature and experimentally (in this study). LC–MS metabolite intensities of extracellular contents in HepaRG (pink), HLCs (purple), HepG2 (light blue), and mHepG2 (dark blue) were normalized to drug intensity in the blank and to total DNA (average ± SD, *p < 0.05). N = 3 replicates per cell model (from the same cell line or batch of differentiation) and incubation condition (3 independent extracted wells). In case metabolites were only found intracellularly in a certain cell system, the name of the cell system was added below the drug structure; n.d. = not detected; “theorized-proposed structure” indicates metabolites not reported in the literature, but of likely occurrence through phase I and phase II metabolism