NADPH:Cytochrome P450 reductase (POR) is a ubiquitous microsomal electron transport protein essential for cytochrome P450 (CYP) mediated oxidative metabolism of xenobiotics and physiological processes. This includes biosynthesis of numerous endogenous compounds especially steroids such as cholesterol and bile acids (Riddick et al., 2013). POR plays also a pivotal role for other important redox partners like heme oxygenase HMOX1, squalene monooxygenase or cytochrome b5 CYB5A (Porter, 2012; Pandey and Flück, 2013). Genomic deletion of Por in mice leads to embryonal death around day 13 due to severe disturbances in retinoid and cholesterol homeostasis (Shen et al., 2002; Otto et al., 2003; Schmidt et al., 2009). So far effects of Por knockout in liver on drug as well as endogenous metabolism were extensively studied in mouse models (Gu et al., 2003; Henderson et al., 2003; Wang et al., 2005; Weng et al., 2005; Finn et al., 2009; Cheng et al., 2014b). In addition to diminished CYP activity levels (Gu et al., 2003; Henderson et al., 2003; Finn et al., 2007), these mice developed severe hepatic lipidosis as well as reduced circulating cholesterol, triglyceride levels and gallbladder bile volume (Gu et al., 2003; Henderson et al., 2003; Wu et al., 2003). Moreover, in-depth studies in liver specific Por knockout mice revealed diverse effects on bile acid synthesis. While total amounts of bile acids in mouse liver were only slightly decreased, concentrations of chenodeoxycholic acid (CDCA) and its gut microbial product lithocholic acid (LCA) were increased, probably due to induction of the alternative bile acid synthesis pathway (Cheng et al., 2014b; Gethings et al., 2021). In humans the multistep bile acid biosynthesis is mainly mediated by CYPs 7A1, 7B1, 8B1 and the mitochondrial, POR-independent CYP27A1, resulting in a substantially different composition of the bile acid pool. In contrast mice are displaying a more hydrophilic complement of bile acid species (Li and Dawson, 2019).

Bile acids act as endogenous ligands for FXR, PXR and G proteins-coupled BA receptor (TGR5 (Gpbar1)), which regulate homeostasis of xenobiotics and lipid, glucose and energy metabolism (Pols et al., 2011; Chiang and Ferrell, 2018). Gene expression analyses of liver specific Por knockout mice generally indicate an induction of drug metabolizing as well as bile acid synthesizing Cyp expression (Wang et al., 2005; Weng et al., 2005; Mutch et al., 2007; Henderson et al., 2013). This effect is ascribed to changed bile acid signaling arising from altered bile acid composition (Wang et al., 2005; Weng et al., 2005; Mutch et al., 2007; Henderson et al., 2013). Increased hepatic lipid accumulation as a result of impaired lipid secretion coupled with increased lipid uptake has also been attributed to changed bile acid signaling (Weng et al., 2005; Mutch et al., 2007).

In humans certain POR gene mutations lead to deficient POR (PORD) characterized by disordered steroidogenesis with broad phenotypic spectrum including cortisol deficiency, altered sex steroid synthesis, disorders of sex development and skeletal malformations resembling the Antley-Bixler syndrome phenotype (Flück and Pandey, 2011; Fukami and Ogata, 2014; Miller, 2018). Medical treatment of PORD patients consisting of glucocorticoid and sex steroid replacement therapy requires individually tailored dosing due to diminished CYP mediated hepatic drug metabolism (Tomalik-Scharte et al., 2010; Miller et al., 2011). Yet, no phenotypic changes of livers in PORD patients comparable to the lipidosis observed in mouse liver were reported so far (Miller, 2018). Furthermore, translation of findings from mouse models to the human system is limited due to differences in the composition of CYP enzymes (Takahashi et al., 2016) or the bile acid pool (Li and Dawson, 2019; Straniero et al., 2020). For a better understanding of the role of POR in human liver metabolism, meaningful human cell models are needed.

To study systematically the effect of diminished POR levels on drug metabolising CYP activity and expression in a human cell background, we recently established two genetic POR HepaRG knockout cell lines using CRISPR/Cas9 technology (Heintze et al., 2021). HepaRG cells have the unique ability to differentiate into hepatocyte- and biliary-like cells (Gripon et al., 2002). Compared to other hepatoma cell lines, HepaRG cells demonstrate stable and functional expression of a broad range of liver specific genes comparable to primary human hepatocytes (Rogue et al., 2012). These include several CYP enzymes as well as phase II enzymes, drug transporters, and liver-specific transcription factors comprising dedicated ligand-activated nuclear receptors. HepaRG cells have therefore become widely accepted as a highly useful model to study various aspects of drug metabolism, transport and its regulation (Kanebratt and Andersson, 2008; Andersson et al., 2012; Klein et al., 2015; Rubin et al., 2015; Tolosa et al., 2016; Tanner et al., 2018; Kugler et al., 2020). Best to mimic the human situation, HepaRG cells are ideally suited to investigate the effects of diminished POR on drug as well as endogenous metabolism. We recently showed differential effects on CYP activity levels and isoform dependent effects on CYP expression in the newly established genetic POR knockout HepaRG cell model (Heintze et al., 2021). Most strikingly, contrasting to increased CYP expression seen in mice, CYP expression levels were generally decreased except for CYPs 1A1 and 1A2.

The aim of our present study was to analyze comprehensively the impact of diminished POR levels in the HepaRG cell background with focus on genes involved in the absorption, distribution, metabolism, and excretion (ADME) and related regulators. We analysed mRNA expression, measured intracellular cholesterol precursers and bile acid secretion by mass spectrometry and protein expression using an untargeted proteomics approach. Our genetic POR knockout HepaRG cell model provides first human data on changes in gene and protein expression and the metabolite composition, which reveal marked differences to rodents. The results indicate widespread effects on gene transcription and protein abundance in certain pathways that may reflect coordinated responses to changed endogenous metabolites like bile acids.

The influence of Por on hepatic drug and endogenous metabolism has been extensively studied in mouse knockout models and recent publications addressed effects of Por knockout even on metabolome and proteome (Gray et al., 2020; Gethings et al., 2021). However, studies in human liver models are still missing. In our previous study we established two different CRISPR/Cas9 induced genetic POR knockout cell lines in the human liver cell model HepaRG, which revealed diminished CYP activities as well as a changed drug metabolizing CYP expression pattern (Heintze et al., 2021). The aim of the present study was an in-depth analysis of effects of deminished POR levels in the HepaRG cells with focus on genes involved in the absorption, distribution, metabolism, and excretion (ADME) and related regulators. We found consistent decrease of mRNA expression of several regulators of hepatic CYP expression, bile acid and lipid homeostasis in both HepaRG^−POR cell lines, indicating that the observed downregulation of many CYP enzymes may thus be an indirect effect of diminished POR (Figure 1).

The transcriptional receptors downregulated in HepaRG^−POR cells (Figure 1) belong to the NR1H (LXR, FXR) and NR1I (PXR, CAR) subfamilies, both of which are involved in regulation of bile acid metabolism by sensing the metabolic environment (Endo-Umeda and Makishima, 2019). Expression of other regulators involved in sterol and lipid homeostasis, namely PPARA, NR5A2 (LRH-1) and SREBF1, were also decreased. The main axis of bile acid regulation consists of NR1H2 (FXR), NR0B2 (SHP) and NR5A2 (LRH-1), which repress the expression of the rate limiting enzyme for bile acid synthesis, CYP7A1 and, in cooperation with HNF4α, CYP8B1 (Shin and Wang, 2019). As both NR1H2 (FXR) and NR5A2 (LRH-1) are downregulated in HepaRG^−POR cells, a severe deregulation of bile acid homeostasis was conceivable. While NR1H2 (FXR) downregulation was also seen in liver specific Por knockout mice (Wang et al., 2005), FXR target gene expression (CYP8B1, 7A1) was decreased in HepaRG^−POR cells suggesting FXR activation rather than inactivation. CYP7A1 and 8B1 as FXR targets were among the four strongest decreased phase I enzymes measured in HepaRG^−POR, while expression of CYP27A1, involved in the alternative bile acid synthesis pathway, was not changed. In addition, deregulation of lipid and cholesterol biosynthesis is indicated due to the decreased mRNA expression of other enzymes involved in these metabolic processes (CYP4A11, CYP51, SCD, CYB5A, FABP1), which are target genes for PPARα, LXRα and SREBF1 (Debeljak et al., 2000; Antoun et al., 2006; Paton and Ntambi, 2009; Guzmán et al., 2013). By contrast, AHR/ARNT as transcriptional components of the AhR (aryl hydrocarbon receptor) pathway, which regulate CYP1A1 and 1A2 in particular, were not changed significantly. Therefore, induction of the AhR pathway by endogenous ligands such as arachidonic acid derivatives (Larigot et al., 2018) or chenodeoxycholic acid (CDCA) (Ibrahim, 2015) may lead to a compensatory increase of CYP1A expression.

In liver specific Por knockdown mice gene expression phenotypes are ascribed to activation of CAR, SREBP, PPARγ and Nrf1 and repression of PPARα and FXR (Wang et al., 2005; Weng et al., 2005; Cheng et al., 2014a). This results in an induction of alternative detoxification enzymes in liver and the small intestine, which appears to partially compensate for the loss of microsomal P450 function. In addition, Nr1h4 (FXR) inhibition leads to increased Cyp7a1 and 8b1, two major CYP enzymes involved in bile acid synthesis, but decrease of Nr0b2 (SHP) mRNA expression. Alterations in PPARα target gene expression may be causative to a decrease in fatty acid oxidation but an increase in fatty acid uptake, which subsequently accounts for the severe hepatic lipidosis observed in mice and cell culture experiments (Gu et al., 2003; Henderson et al., 2003; Weng et al., 2005; Porter et al., 2011). However, we did not observe a comparable response of lipid accumulation in our model (Supplementary Figure S2), which may be explained not only by the apparent FXR inactivation instead of activation (Porter et al., 2011) but also by more pronounced effects on lipid droplet formation in the rodent system (Li and Chiang, 2014).

De novo cholesterol biosynthesis is POR dependent due to the enzymatic activity of the lanosterol-14α-demethylase (CYP51). POR knockdown in HepaRG cells lead to the accumulation of lanosterol while the cholesterol precursor lathosterol was unchanged (Figure 2). This finding is in accordance with observations in hepatic Por-null mice or in rat hepatoma cells with suppression of Por expression by siRNA (Porter et al., 2011; Porter, 2012).

In humans, POR influences bile-acid homeostasis via CYP7A1, 7B1, 8B1, and indirectly impacts FXR and PXR signaling via bile acids as endogenous ligands (Pols et al., 2011; Chiang and Ferrell, 2018). CDCA, TCA, TCDCA are known agonists of FXR, while GCDCA and other more hydrophilic derivates are more antagonistic factors (Juřica et al., 2016; Chiang and Ferrell, 2018). In contrast to humans, mice have an additional primary bile acid form, muricholic acid (MCA), a product of the microsomal Cyp2a70 (Takahashi et al., 2016). MCA and taurine conjugates of MCA are no agonists or even antagonists of FXR (Sayin et al., 2013; Li and Dawson, 2019) which underlines that bile acid regulation varies majorly between different species. Taking the mRNA expression data as a basis, changes in the bile acid composition of HepaRG^−POR cells is suggested. While the whole bile acid pool in HepaRG^−POR cells was only slightly decreased, individual bile acid species were highly differentially affected by POR knockdown (Figure 3). Slightly increased CDCA conjugates in HepaRG^−POR cells may indicate increased bile acid synthesis via the alternative acidic pathway with the POR-independent, mitochondrial CYP27A1 (Monte et al., 2009) predominantly leading to CDCA synthesis (Alnouti, 2009; Pikuleva and Waterman, 2013). Enhanced alternative bile acid synthesis was also reported in Por knockout mice, where an increase in CDCA and TCDCA in the liver was observed (Cheng et al., 2014b). The observed downregulation of nuclear receptor expression in our cell model could be in response to the altered bile acid composition with decreased levels of TCA and GCA and unchanged TCDCA levels while levels of GCDCA were increased. Under physiological healthy conditions, FXR is usually not activated. Higher levels of CDCA lead to FXR activation (Juřica et al., 2016; Chiang and Ferrell, 2018), which may explain decreased CYP7A1 and 8B1 mRNA expression levels. Increased CDCA levels could also influence HNF4α target gene expression, as CDCA reduces the transactivation potential of HNF4α via induction of the expression of the co-repressor SHP, thereby downregulating for instance the expression of endogenous SLC22A7 (OAT2) mRNA in Huh7 cells (Li et al., 2019). We also found a decrease of SLC22A7 mRNA expression up to 0.46-fold lower in HepaRG^−POR#2 (Figure 1), however not statistically significant. Furthermore, we observed a higher percentage of taurine conjugates compared to glyco-conjugates as already described by others (Sharanek et al., 2015; Burban et al., 2019). Thus, regarding bile acid composition with increased GCDCA and reduced TCA levels, HepaRG cells seem to resemble hepatocytes originating from NASH (non alcoholic steatohepatitis) or cholestatic patients (Trottier et al., 2012; Lake et al., 2013) rather than “healthy” primary human hepatocytes (Sharanek et al., 2015) due to antagonistic effect of GCDCA on FXR. In disease state bile acid signaling is known to show large species differences. Cholestasis in humans is associated with suppressed bile acid synthesis, while rodents have increased bile acid synthesis (Straniero et al., 2020). It should be noted that bile acid homeostasis also involves systemic regulation and a tight interplay between liver and intestine (Cheng et al., 2014b; Chiang, 2017). . This aspect is of course not represented by our HepaRG cell model.

As indicated by the mRNA expression analysis, POR knockdown in HepaRG cells influences gene expression on a global level (Figure 1). Consistently, the untargeted proteomics analysis demonstrated reduced expression of several drug metabolism-related proteins (Figure 6A), again contrasting findings in liver specific Por knockout mice (Wang et al., 2005; Weng et al., 2005; Mutch et al., 2007; Henderson et al., 2013; Gethings et al., 2021). Of particular interest are the data regarding the protein PLIN2 (perilipin-2), one of the top lower expressed (up to 10fold) proteins in HepRG^−POR, and FETUA (AHSG, alpha 2-HS glycoprotein) the most upregulated (up to 5fold) protein with diminished POR. PLIN2 has been described to impair hepatic lipid accumulation and Plin2 whole body gene deletion prevented obesity and insulin resistance in Western diet-fed mice by suppressing hepatic SREBP-1/2 activity (Libby et al., 2016). This may counteract lipid droplet formation upon POR depletion, thus explaining the absence of lipid accumulation in our cell system. On the other hand, FETUA (AHSG, alpha 2-HS glycoprotein), a plasma glycoprotein predominantly synthesized in the liver, is increased in type 2 diabetes mellitus, metabolic syndrome, and nonalcoholic fatty liver disorder (NAFLD) (Bourebaba and Marycz, 2019). Pathway analysis revealed severe downregulation of a series of metabolic pathways, especially fatty acid metabolism, metabolism of lipids, biological oxidations as well as phase I and phase II drug metabolism (Figure 5). This illustrates the general role of POR in endogenous and xenobiotic metabolism which is in line with previous analyses of the lipid and bile acid metabolome as well as proteomic analyses in liver specific Por knockout mice (Gray et al., 2020; Gethings et al., 2021). Only few proteins show higher expression levels upon POR knockdown in the HepaRG cell model. These were mapped to six and three pathways in HepaRG^−POR#1 and HepaRG^−POR#2, respectively, compared to HepaRG^VC. These include metabolism of RNA, mRNA splicing, and processing of pre-mRNAs (Figures 5A,B, Supplementary Figures S5A,B yellow). Nevertheless, both knockdown cell lines are well comparable in downregulated genes and thereby affected pathways. Only effects for a small number of affected pathways are more pronounced in the HepaRG^−POR#2model (Figure 5, Supplementary Figure S5). POR exhibits a particular role in the cholesterol biosynthesis illustrated by protein expression changes in this pathway (Figure 6B). Biosynthesis of cholesterol can occur by two separate pathways, namely the Kandutsch-Russell and Bloch pathways, with the latter being more prominent in the liver (Mitsche et al., 2015). Several enzymes in these pathways were significantly downregulated with 7-dehydrocholesterol reductase (DHCR7), the terminal enzyme in cholesterol biosynthesis in the Kandutsch-Russell pathway being the most affected. Interestingly, the linking enzyme between the two pathways, 24-dehydrocholesterol reductase (DHCR24), was increased by POR knockdown. This compensatory response was in agreement with already described in Dhcr7 knockout mice (Koczok et al., 2019).

Using STRING analysis, we could confirm the central role of POR in drug metabolism (Figure 7). As already discussed above for the mRNA expression analysis (Figure 1), an indirect consequence of POR knockdown is the change in various hepatic signaling pathways (Supplementary Figure S6). This is depicted by the interaction of several alcohol dehydrogenases with CYP2E1, indicating an activation of the AhR/ARNT signaling pathway by AhR ligands (Attignon et al., 2017). In addition, several interactions indicating towards changed PPARα signaling were revealed (eg. FABP1 and SULT2A1, PLIN1 and ALDOB).

In conclusion, gene and protein expression changes in human hepatocytes seem to be a coordinated response to altered endogenous metabolite levels, mainly bile acids, which is generally in line with findings in liver specific Por knockout mice (Mutch et al., 2007; Gray et al., 2020; Gethings et al., 2021). However, species differences make it difficult to translate specific findings concerning biological mechanisms from Por knockout mice to the human system. Our genetic POR knockout human HepaRG cell model illustrates marked differences especially in bile acid composition and signaling to rodent animal models and may provide new insights into the importance of POR in endogenous cell metabolism.