A Critical Perspective on 3D Liver Models for Drug Metabolism and Toxicology Studies

fcell-09-626805 February 25, 2021 Time: 14:28 # 1 REVIEW published: 22 February 2021 doi: 10.3389/fcell.2021.626805 Edited by: Emmanuel S. Tzanakakis, Tufts University, United States Reviewed by: Natesh Parashurama, University at Buffalo, United States Salman Khetani, University of Illinois at Chicago, United States *Correspondence: Joana P. Miranda [email protected]; [email protected] †These authors have contributed equally to this work Specialty section: This article was submitted to Stem Cell Research, a section of the journal Frontiers in Cell and Developmental Biology Received: 06 November 2020 Accepted: 21 January 2021 Published: 22 February 2021 Citation: Serras AS, Rodrigues JS, Cipriano M, Rodrigues AV, Oliveira NG and Miranda JP (2021) A Critical Perspective on 3D Liver Models for Drug Metabolism and Toxicology Studies. Front. Cell Dev. Biol. 9:626805. doi: 10.3389/fcell.2021.626805 A Critical Perspective on 3D Liver Models for Drug Metabolism and Toxicology Studies Ana S. Serras1†, Joana S. Rodrigues1†, Madalena Cipriano2†, Armanda V. Rodrigues1, Nuno G. Oliveira1and Joana P. Miranda1* 1Research Institute for Medicines (iMed.ULisboa), Faculty of Pharmacy, Universidade de Lisboa, Lisbon, Portugal, 2Fraunhofer Institute for Interfacial Engineering and Biotechnology IGB, Stuttgart, Germany The poor predictability of human liver toxicity is still causing high attrition rates of drug candidates in the pharmaceutical industry at the non-clinical, clinical, and postmarketing authorization stages. This is in part caused by animal models that fail to predict various human adverse drug reactions (ADRs), resulting in undetected hepatotoxicity at the non-clinical phase of drug development. In an effort to increase the prediction of human hepatotoxicity, different approaches to enhance the physiological relevance of hepatic in vitro systems are being pursued. Three-dimensional (3D) or microfluidic technologies allow to better recapitulate hepatocyte organization and cellmatrix contacts, to include additional cell types, to incorporate fluid flow and to create gradients of oxygen and nutrients, which have led to improved differentiated cell phenotype and functionality. This comprehensive review addresses the drug-induced hepatotoxicity mechanisms and the currently available 3D liver in vitro models, their characteristics, as well as their advantages and limitations for human hepatotoxicity assessment. In addition, since toxic responses are greatly dependent on the culture model, a comparative analysis of the toxicity studies performed using two-dimensional (2D) and 3D in vitro strategies with recognized hepatotoxic compounds, such as paracetamol, diclofenac, and troglitazone is performed, further highlighting the need for harmonization of the respective characterization methods. Finally, taking a step forward, we propose a roadmap for the assessment of drugs hepatotoxicity based on fully characterized fit-for-purpose in vitro models, taking advantage of the best of each model, which will ultimately contribute to more informed decision-making in the drug development and risk assessment fields. Keywords: in vitro liver model, fit-for-purpose models, hepatotoxicity, paracetamol, diclofenac, troglitazone, three-dimensional culture INTRODUCTION The process of development of new drugs is a costly investment with the pharmaceutical industry facing considerable challenges regarding the balance between the political pressure to increase drugs safety while reducing the cost of medicines. According to a recent study by Wouters et al. (2020), the median investment of bringing a new drug into the market, also accounting for failed Frontiers in Cell and Developmental Biology | www.frontiersin.org 1February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 2 Serras et al. 3D Liver Models in Toxicology Studies trials, was estimated at $985.3 million over the period of 2009– 2018. It is a process that usually takes 10–15 years, with a success rate from phase I to launch of less than 10% (Dowden and Munro, 2019). This is mostly due to lack of drug efficacy or safety issues that occur essentially in the clinical phases IIb and III of drug development (Kola and Landis, 2004;Paul et al., 2010). Even after reaching the market (phase IV), there is still a relevant number of drug withdrawals for toxicological reasons. Approximately 18–30% of such withdrawals are caused by hepatotoxic effects, showing that the liver is the most frequent organ for adverse drug reactions (ADRs) (Onakpoya et al., 2016;Siramshetty et al., 2016;Zhang X. et al., 2020). Importantly, about 40–50% of the drug candidates associated with hepatotoxicity in humans did not present the same toxicological concern in animal models (van Tonder et al., 2013). Indeed, besides raising ethical issues, animal models often fail to correlate with human toxicity, since several toxic features disclosed in human trials were not predicted by animal studies (Olson et al., 2000;Shanks et al., 2009). One of the reasons for this discrepancy is the differential expression and activity of drug metabolizing enzymes between animals and humans that might confound the extrapolation of data derived from non-clinical species (Martignoni et al., 2006;Ruoß et al., 2020). Moreover, drug-induced liver injury (DILI) is a rare, but potentially fatal event, resultant from the poor translation between clinical trials and clinical practice and highlights the importance of targeting population variability at non-clinical stages (Jones et al., 2018). Within DILI, the idiosyncratic category is particularly difficult to identify by the pharmaceutical industry as it is almost undetectable in animal models (Kuna et al., 2018; Walker et al., 2020). Altogether, this has led to the proposal that the better the quality of non-clinical safety profiles, the higher the success rates for moving phase II upward (Cook et al., 2014; Walker et al., 2020). Consequently, in vitro liver models are growing strong while new drugs advance into clinical trials. The search for more accurate non-clinical models along with the concern about animal welfare, reducing time and cost associated to drug development and the ever-increasing number of chemicals that need testing, made the establishment of relevant in vitro culture systems a priority in the toxicology assessment of drugs by the pharmaceutical industry, as these allow a higherthroughput capacity. Novel cell culture and tissue engineering technologies along with integrated endpoints have been adopted for improving liver cell metabolic performance in vitro and are expected to generate more robust data on the potential risks of pharmaceuticals (Davila et al., 2004;Andersen and Krewski, 2009, 2010;Krewski et al., 2009;Giri et al., 2010;Shukla et al., 2010;Balls, 2011;Mandenius et al., 2011). Existing strategies include three-dimensional (3D) structures, flow-based cultures, co-cultures and stem-cell differentiation. In this review, we discuss the dissimilarities of the 3D in vitro hepatic systems currently used in research and drug development and their actual contribution for unraveling the mechanisms of drug-induced hepatotoxicity. Special emphasis is given to the features of 3D culture systems, cell organization and architecture, the effects of stirring and perfusion and how these characteristics modulate the phenotype and functionality of liver cells. In addition, we take a step forward by presenting a comparative analysis of the IC50 values for cytotoxicity and mechanistic endpoints, obtained either with two-dimensional (2D) and 3D in vitro systems for the classical hepatotoxic drugs paracetamol (acetaminophen), diclofenac and troglitazone. In this context, it seems clear the need for harmonized and fully characterized models. Moreover, it is also important to highlight that the hepatotoxicity assessment and the choice of the in vitro liver models depend on the questions that need to be addressed. These strategies stand out as crucial when evaluating the model’s relevance value for mechanism-based hepatotoxicity assessment. DRUG-INDUCED HEPATOTOXICITY: OVERVIEW, LIVER METABOLISM AND MECHANISMS OF TOXICITY The liver is responsible for most of the metabolism of orally administered drugs since its anatomical proximity to the gastrointestinal tract and histological structure, including the sinusoidal space and the blood supply from the portal vein, allows the efficient transport of drugs and other xenobiotics (Vernetti et al., 2017). It is a complex organ composed by ∼60% of hepatocytes, parenchymal cells responsible for multiple functions, including metabolism. Non-parenchymal cells (NPCs) include cholangiocytes lining the bile ducts; sinusoidal endothelial cells, which constitute a permeable barrier between the blood and the space of Disse; Kupffer cells, the liverresident macrophages; and stellate cells, which synthesize fat and produce vitamin A and collagen (Kuntz and Kuntz, 2008). Drug-induced hepatotoxicity is defined as the hepatic damage caused by the exposure to prescription-only or over-thecounter medicines, herbs or other xenobiotics. DILI represents a major challenge for clinicians, the pharmaceutical industry, and regulatory agencies worldwide. As above mentioned, it corresponds to the leading cause of attrition of compounds in drug development, being also frequently associated to drug withdrawals from market or to use restrictions (Stevens and Baker, 2009;Devarbhavi, 2012;Jones et al., 2018). Classically, DILI can be classified as intrinsic (e.g., caused by paracetamol) or idiosyncratic (e.g., caused by troglitazone) hepatotoxicity. Intrinsic hepatotoxicity is direct, dose-dependent and predictable, whereas idiosyncratic hepatotoxicity occurs without obvious dose-dependency, in an unpredictable fashion and with a short latency time, particularly after re-exposure (Russmann et al., 2009;Roth and Ganey, 2010). Idiosyncratic DILI can be an allergic immune-mediated hypersensitivity or the result of a non-allergic metabolic injury (Larson, 2010). DILI may also be categorized according to the duration (i.e., acute or chronic) and location/typology of the injury. The latter can be classified as hepatitis (mostly due to hepatocyte necrosis), cholestatic (i.e., bile duct damage or cholangiolitis) or mixed injury (Stefan and Hamilton, 2010). Despite the variety of its clinical presentations, DILI still does not display specific biomarkers, leading to abnormal liver tests and often the dysfunction is only identified by exclusion of other etiologies, which can lead to life-threatening clinical situations (Devarbhavi, 2012;Fu et al., 2020). Indeed, the identification of new molecular Frontiers in Cell and Developmental Biology | www.frontiersin.org 2February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 3 Serras et al. 3D Liver Models in Toxicology Studies biomarkers has been investigated in order to improve diagnosis and treatment of DILI. However, its applicability is still limited (Fu et al., 2020). Thus, DILI is largely unrecognized and underreported, such that the true incidence is unknown. There are several examples of clinically relevant drugs that have received prescription restrictions or the inclusion of a black box warning for potential hepatotoxicity. Among hepatotoxic drugs, paracetamol is the most frequently studied. Nevertheless, the most commonly hepatotoxicity-associated pharmacological groups of orally administrated drugs are antibiotics (e.g., amoxicillin-clavulanate and rifampicin), antiretrovirals (e.g., nevirapine), non-steroidal anti-inflammatory drugs (NSAIDs, e.g., diclofenac and ibuprofen), antidepressants (e.g., paroxetine), and anticonvulsants (e.g., phenytoin, carbamazepine, and valproic acid) (EMEA, 2000;Paniagua and Amariles, 2018). Among intravenous administration, antibiotics, and antineoplastic drugs are the pharmacological groups mostly associated with hepatic toxicity (Ghabril et al., 2013). It should be mentioned that during the past decades, particularly in the last 20 years, several medicines such as troglitazone, bromfenac, trovafloxacin, ebrotidine, nimesulide, nefazodone, ximelagatran, lumiracoxib, pemoline, tolcapone, and sitaxentan have also been removed from the market in some countries in Europe and in the United States due to severe DILI (Fung et al., 2001;Qureshi et al., 2011;Babai et al., 2018). Liver Metabolism Drug metabolism is a major determinant of hepatotoxicity, as both detoxification and bioactivation processes can occur, and are most frequently responsible for inter-individual differences in drug-induced toxicity. Liver metabolism encompasses phase I biotransformation reactions, also known as functionalization reactions, leading to the hydrolysis, oxidation, and reduction of a given drug or xenobiotic. Key enzymes in this phase belong to the CYP450 family, but can also be epoxide hydrolase and monoamine oxidase, among others. The metabolites generated can be detoxified or bioactivated by further phase I biotransformation or by conjugation through phase II metabolism (e.g., glucuronidation, sulfation, and acetylation). The role of liver transporters (e.g., organic anion-transporting polypeptides, OATP, multidrug resistance-associated proteins, and MRP) is of great importance for the excretion, being this step also known as phase III (Gomez-Lechon et al., 2010;Yuan and Kaplowitz, 2013). A significant feature of liver drug metabolism is that it may transform the parental compounds into chemically reactive intermediates or electrophilic metabolites (i.e., bioactivation) that attack tissue constituents, potentially leading to mutations, cancer or tissue necrosis (Pessayre, 1993). Drug-induced hepatotoxicity can thus be consequence of the toxicity of the parental drug per se or the result of one or more of its metabolites that arise from liver metabolism (Figure 1). Therefore, the toxicity of a given xenobiotic greatly depends on the equilibrium between detoxification and bioactivation. Hence, in a new drug development scheme, the biotransformation processes should be widely studied in order to predict the physiological effect of the new compound. There are several prodrugs that take advantage of liver metabolism, e.g., cyclophosphamide (Preissner et al., 2015) and L-Dopa (Di Stefano et al., 2011), as the initial molecule is only active after biotransformation near the target site, decreasing its potential toxicity and also increasing its bioavailability. On the other hand, paracetamol is an interesting example in which hepatotoxicity is dose-dependent and occurs since its metabolic pathway switches at a high dose exposure from the detoxifying phase II metabolism to phase I metabolism, generating the hepatotoxic metabolite N-acetyl-p-benzoquinone imine (NAPQI). This metabolite can covalently react with proteins, leading to necrosis, apoptosis and, ultimately, to liver failure (Hinson et al., 2010). Additionally, phase II metabolism may also lead to hepatotoxic derivatives, such as for example carboxylic acids, e.g., bromfenac (and other NSAIDs) or valproic acid (Sidenius et al., 2004;Skonberg et al., 2008). These can be bioactivated to acyl-coenzyme A thioesters, which are intermediates in phase II conjugation reaction, and may lead to reactivity toward reduced glutathione (GSH) and covalent binding to endogenous proteins (Sidenius et al., 2004; Skonberg et al., 2008). Hence, factors including the inhibition or induction of any of the biotransformation enzymes, drug-drug interactions, or genetic polymorphisms, may lead to increased activity and toxicity or, on the other hand, to an absence of effect. Several widely prescribed drugs are themselves potent CYP450 enzyme inducers, e.g., phenobarbital, carbamazepine and rifampicin, or inhibitors, e.g., fluoxetine, ritonavir, fluconazole, and ciprofloxacin (Baxter et al., 2010;Wooten, 2015;Wolverton and Wu, 2020). Subsequently, in the context of multiple drug prescription, the biotransformation of drugs that are substrates of CYP450 enzymes or other phase II enzymes and hepatic transporters can be severely altered when administered simultaneously. Some antiretroviral drugs, such as efavirenz (Grilo et al., 2017) or nevirapine (Pinheiro et al., 2017), may be simultaneously the substrate and the inducer of an enzyme, such as CYP3A4 and CYP2B6, and can regulate its own biotransformation (auto-inducer) (Kappelhoff et al., 2005). Indeed, enzyme induction is included within the pharmacokinetic (PK) tolerance concept, as it can lead to overdose reactions (higher parent drug/metabolite activation) or to sub-therapeutic exposures (lower parent drug/metabolite inactivation) to drugs when normal doses are administered (Dumas and Pollack, 2008;Omiecinski et al., 2011;Jaeschke, 2013). However, the effect that xenobiotics can exert on the induction or inhibition of biotransformation enzymes is especially difficult to predict with the currently existing in vitro and in vivo drug testing models, mainly due to interspecies and inter-individual differences, or decreased cells functionality (Reder-Hilz et al., 2004;Zanger et al., 2007;Godoy et al., 2013). Genetic polymorphisms are particularly relevant risk factors regarding drug-metabolizing enzymes and may represent susceptibility biomarkers, important for predicting potential hepatotoxicity risks. Genetic polymorphisms are common gene variations that might encode for impaired/altered metabolic enzymes and generate different population subgroups in terms of metabolism assessment (Meyer and Zanger, 1997; Frontiers in Cell and Developmental Biology | www.frontiersin.org 3February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 4 Serras et al. 3D Liver Models in Toxicology Studies FIGURE 1 | Schematic representation of the mechanisms of hepatotoxicity including examples of associated drugs. Drug biotransformation (phase I and II metabolism) is based on the chemical modification of a parent drug into a metabolite which may become inactive (detoxification), leading to its rapid and innocuous excretion, or reactive (bioactivation), leading to potential toxicity. Specifically, hepatotoxicity may result from direct damage, from failure of repairing mechanisms or from immune-mediated responses, leading to alterations in lipids metabolism, mitochondrial dysfunction, oxidative stress and accumulation of bile, amongst others. Moreover, the saturation of cells stress defense mechanisms may lead to carcinogenic events and promote tissue necrosis or fibrosis, resulting in liver’s functions impairment. For a given hepatotoxic compound different mechanisms of toxicity can be involved. GSH, reduced glutathione; ROS, reactive oxygen species. Andrade et al., 2009;Ahmad and Odin, 2017). As this event is not rare, these subgroups need to be accounted in a drug development scheme and, thus, properly mimicked at the non-clinical stage. Interindividual variability concerning phase I, II, and III enzyme expression can also justify some cases of hepatotoxicity. Genetic polymorphisms are reported to affect the biotransformation of drugs dependent on CYP2C9, CYP2C19, CYP2B6, CYP2D6, CYP3A4, and CYP3A5 subfamilies, phase II enzymes uridine 50-diphosphate glucuronosyltransferase (UGT) 1A1, UGT2B7 and N-acetyltransferase (NAT) 2, and hepatic transporters multidrug resistance protein (MDR) 1, breast cancer resistance protein (BCRP), MRPs, and OATP1B1, amongst others (Wienkers and Heath, 2005;Zanger et al., 2007;Brockmöller and Tzvetkov, 2008;Shah et al., 2015;Krasniqi et al., 2016; Saiz-Rodríguez et al., 2020). Some classical examples include CYP2D6, due to its clinical impact in the bioactivation of drugs such as codeine, tramadol, or tamoxifen within low or extensive metabolizers (Cavallari et al., 2019). Another classical example are NAT2 polymorphisms, reflected in slow, intermediate, and rapid acetylators of drugs, particularly isoniazid (antituberculosis drug), in which the former presents potentially more ADRs than the latter (Brockmöller and Tzvetkov, 2008). Moreover, an inherited mutation in the adenosine triphosphate (ATP)-binding cassette subfamily B (ABCB) 11 gene, which encodes for bile salt export pump (BSEP), may lead to the diminishing of the bile acids transport and clearance, potentially leading to cholestasis (Kenna and Uetrecht, 2018). Mechanisms of Hepatotoxicity The liver is a prime target for drug-induced damage due to its central role for concentrating and metabolizing the majority of drugs. Therefore, earlier and better understanding of drug modes of action and toxicity are essential (Kola and Landis, 2004;Paul et al., 2010;Padda et al., 2011). As above mentioned, following exposure, the toxic effect of a given drug may be attributed directly to the interaction of the parent drug or the product of its biotransformation, with an endogenous target through covalent or non-covalent binding, hydrogen abstraction, electron transfer, or enzymatic reactions, resulting in dysfunction or destruction of the target molecules (Chan and Benet, 2017). Moreover, besides arising from direct damage by the molecule, hepatotoxicity may also be resultant from a failure of repair mechanisms or due to immune-mediated responses. The mechanisms of hepatotoxicity more frequently described are depicted in Figure 1 and involve: i) Mitochondrial dysfunction, an effect that may occur upon the exposure to different drugs, particularly amiodarone (Bethesda, 2012), nimesulide (Singh et al., 2010), troglitazone (Smith, 2003), or valproic acid (Xu et al., 2019); Frontiers in Cell and Developmental Biology | www.frontiersin.org 4February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 5 Serras et al. 3D Liver Models in Toxicology Studies ii) Oxidative stress, as observed for instance upon paracetamol or nitrofurantoin administration (Bethesda, 2012;Bruderer et al., 2015;Ramachandran and Jaeschke, 2018); iii) Covalent binding with proteins that may impair their transporter function leading to accumulation of toxic elimination products and intrahepatic cholestasis (Boelsterli, 2003;Padda et al., 2011), as reported for ethinylestradiol and cyclosporine (Bethesda, 2012). It may also alter their conformation or structure as observed on the inhibition of hepatic synthesis of coagulation factors by exposure to coumarins (Grattagliano et al., 2009;Gregus, 2013); iv) DNA damage, as suggested in the context of nevirapine toxicity (Kranendonk et al., 2014;Pinheiro et al., 2017; Marinho et al., 2019); v) Depletion of enzymes or co-factors as observed upon paracetamol overdose (Mazaleuskaya et al., 2015; Ramachandran and Jaeschke, 2018); vi) Dysfunction of cell repairing mechanisms that can result in: tissue necrosis, as for example by sulfasalazine, ketoconazole, or valproic acid (Kleiner, 2017); in fibrosis, by e.g., chronic exposure to methotrexate, high doses of retinol (vitamin A), and iron intoxication (Zhang et al., 2016); or in carcinogenesis, as a consequence of aflatoxin B1 exposure (Gregus, 2013;Jaeschke, 2013;Cai et al., 2020); vii) Immunological-mediated tissue damage, that has been linked to NSAIDs such as diclofenac (Aithal et al., 2004), antibiotics such as amoxicillin-clavulanate (Bethesda, 2012) or flucloxacillin (Woolbright and Jaeschke, 2018) and anticonvulsants such as carbamazepine or lamotrigine (Bethesda, 2012). These molecular mechanisms may intersect with each other leading to a cascade of key events. Indeed, an initial drugrelated reactive oxygen species (ROS) formation may lead to lipid peroxidation on fatty acids chains in the cell membrane. In parallel, β-oxidation of lipids and oxidative stress may cause mitochondrial membrane permeabilization and dysfunction, ultimately leading to hepatocyte apoptosis. The rupture of the mitochondrial membrane can result in ATP depletion that accompanied by an increase in intracellular calcium concentration may generate liver necrosis. Conversely, inhibition of peroxisomal fatty acid β-oxidation may result in abnormal triglycerides accumulation in the hepatocyte and result in liver steatosis (Gregus, 2013). Adverse outcome pathways (AOPs) are promising tools in that regard, as they describe existing knowledge concerning the linkage between a direct molecular initiating event (MIE) and an adverse outcome through a number of key events (KEs) at a biological level of organization relevant to risk assessment (Gijbels and Vinken, 2017). At the cellular level, the paracrine communication between hepatocytes and NPCs is also crucial for the response to a toxic insult. It has been reported that NPCs, after a primary injury of the hepatocyte, exhibit a secondary response that may aggravate or ameliorate the initial lesion, e.g., metabolic alterations and activation of immune cells, such as Kupffer cells and lymphocytes (Figure 1;Godoy et al., 2013;Kostadinova et al., 2013;Messner et al., 2013;Leite et al., 2016;Proctor et al., 2017;Bell et al., 2020; Li et al., 2020). LIVER IN VITRO MODELS FOR TOXICOLOGICAL STUDIES Both liver metabolism and the mechanisms of initial liver injury are important to comprehend the potential toxicity of a drug. Therefore, the development of efficient and fit-for-purpose in vitro models should mimic the complexity of the in vivo hepatic milieu. As such, when building a relevant liver in vitro model, the hepatic cell sources and tissue architecture, flow dynamics and the formation of molecular gradients need to be carefully considered. No universally accepted hepatocyte source that provides robust, predictive and significant toxicological and pharmacological results is currently available. Cell source selection depends on cell availability and study requirements while understanding the limitations associated to each cell origin, namely metabolic competence, stability, and population representativeness (Soldatow et al., 2013). Regarding culture architecture, efforts have been focused in better mimic the in vivo microenvironment, giving special attention to culture three-dimensionality either by taking advantage of cell selfassembling capacity or by using natural polymers. More complex systems, such as bioreactors, micropatterning techniques, or microfluidic devices can also be employed (Miranda et al., 2010;Bell et al., 2016;Knospel et al., 2016;Adiels et al., 2017). Those platforms should also allow acute toxicity studies and long-term assessment so that the exposure to a xenobiotic generates relevant responses (Jiang et al., 2019). Overall, the value of an in vitro model depends on how well it reproduces the key physiological characteristics of an in vivo system. However, the criteria for defining liver function maintenance in vitro are not consensual, ranging from focusing on the preservation of hepatocyte phase I and II enzyme functions to the inclusion of a broader spectrum of tissue characteristics involved in human liver toxicity, such as the incorporation of NPCs for mimicking cells’ crosstalk (Bale et al., 2014;Zeilinger et al., 2016;Langhans, 2018;Bell et al., 2020). Some common evaluated features to compare hepatic cell-based in vitro culture systems’ value for toxicological applications include cell morphology, viability, and functional stability; metabolic capacity; preservation of hepatic-specific gene expression under long-term cultures; and response to a panel of well-accepted reference drugs (e.g., paracetamol and valproic acid) capable of replicating human in vivo intrinsic DILI (Miranda et al., 2009, 2010;Leite et al., 2011;Mueller et al., 2011;Tostoes et al., 2011;Cipriano et al., 2017b; Pinheiro et al., 2017;Vinken and Hengstler, 2018;Bell et al., 2020). Moreover, the generated data should be able to be correlated to clinical observations, reproducible, comparable among laboratories, and analyzed properly to support decisionmaking with a clear definition of the models’ applicability and limitations (Dash et al., 2009;Vinken and Hengstler, 2018; Albrecht et al., 2019). Frontiers in Cell and Developmental Biology | www.frontiersin.org 5February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 6 Serras et al. 3D Liver Models in Toxicology Studies Liver Cell-Based Versus Stem Cell-Based Models Over the past decades, large efforts have been made to establish predictive in vitro liver test models. However, despite the number of reports available, a comprehensive and systematic comparison between cell culture systems adequate to objectively rank or select them for pharmacological and toxicological applications is still scarce. Several in vitro human-based models for the prediction of hepatotoxicity have been developed using a range of cell sources and endpoints. These include the use of liver slices, genetically engineered cells, human hepatoma cell lines (e.g., HepG2, THLE, and HepaRG cells), primary hepatocytes or stem cell (SC)-derived models (Gomez-Lechon et al., 2008;Asha and Vidyavathi, 2010;Sirenko et al., 2016;Gao and Liu, 2017;Pinheiro et al., 2017;Nudischer et al., 2020). Figure 2 summarizes the advantages and limitations of each cell source for in vitro testing, as well as their in vivo physiological relevance. Liver slices and isolated perfused livers, containing both parenchymal and NPCs, retain liver’s structure and thus maintain zone-specific enzymatic activity. However, within hours, the cell functionality decreases and necrosis takes place (LercheLangrand and Toutain, 2000;Boess et al., 2003;Haschek et al., 2009). It is associated with limited throughput and requires continuous animal experimentation and personnel expertise (Vernetti et al., 2017). Alternatively, cell-based models are less complex and associated to higher throughput screening for the identification of hepatotoxic compounds. Primary hepatocytes, either obtained from human liver autopsies or biopsies or from animal livers, have been used for cytotoxicity, biotransformation, and PK studies (Vernetti et al., 2017). Human primary hepatocytes (hpHep), in particular, are considered the gold standard in human-relevant liver in vitro models for cytotoxicity and drug metabolism testing, retaining most of the native tissue’s functionality, namely phase I and phase II enzymes (Godoy et al., 2013;Zeilinger et al., 2016). However, both the limited availability of primary human cells and its suitability only for shortterm studies under monolayer cultures are major disadvantages. Indeed, in 2D conditions, it is observed a progressive loss of the hepatic phenotype in a process called de-differentiation, which is a consequence of the disruption of cell–cell and cell-matrix connections (Zeilinger et al., 2016). Additionally, hpHep display inter-donor variability and thus the use of different cell batches to validate results is advised, covering several metabolic genetic polymorphism and phenotypes (Godoy et al., 2013;Zeilinger et al., 2016). On the other hand, rat primary hepatocytes (rpHep), despite being more easily available, present relevant interspecies differences (Sandker et al., 1994;Li et al., 2008;Ménochet et al., 2012;Shen et al., 2012). Human hepatoma cell lines, such as HepG2 and HepaRG, have no limitations in terms of cell numbers and are easy to culture, but display poor phenotype and functional match to in vivo hepatocytes (Gerets et al., 2012). The use of these cell lines do not consider populational differences and may reflect characteristics that primary cells do not have, e.g., being more sensitive to compounds with anti-proliferative properties (Sirenko et al., 2016). HepG2 present low levels of CYPs and normal levels of phase II enzymes except for UGTs (Westerink and Schoonen, 2007a,b), which make them appropriate for testing the toxicity of the parent compound but less suited for metabolite toxicity testing. Instead, HepaRG cell line composed of a mixture of both hepatocyte-like and biliary-like cells, have been reported to maintain hepatic functions and expression of liver-specific genes comparable to hpHep without the inter-donor variability and functional instability issues (Guillouzo et al., 2007). Nevertheless, it should be noted that a cell characterization study at the mRNA/gene expression and CYP activity levels, by Gerets et al. (2012), revealed that although it is a suitable model for induction studies, these cells were not as indicative as hpHep for the prediction of human hepatotoxic drugs, being comparable to HepG2 cells. On the other hand, Lübberstedt et al. (2011) showed that HepaRG presented similar or even higher CYP2C9, CYP2D6, and CYP3A4 enzyme activity than that of hpHep, whereas Aninat et al. (2006) confirmed the presence of relevant UGT1A1 and GST activity levels. Still, high metabolic capacity in cell lines does not necessarily correlate with high sensitivity for the hepatotoxicity detection (Gerets et al., 2012). Thus, unfortunately, even the most promising and differentiated hepatoma cells do not constitute an ideal surrogate system for human hepatocytes for hepatotoxicity studies, as they do not reproduce the drug-metabolizing enzyme pattern of human hepatocytes. An alternative approach to overcome the limitations of hepatic cell lines is to genetically modify cells with vectors encoding for human CYP enzymes and other genes involved in xenobiotic metabolism (Coecke et al., 2001;Kanamori et al., 2003;Gomez-Lechon et al., 2008;Prakash et al., 2008;Godoy et al., 2013). However, the number of enzymes that can be satisfactorily transfected into cells is low and the metabolic profiles differ from those of primary hepatocytes (Frederick et al., 2011;Godoy et al., 2013). To overcome the limitations of the above mentioned cell sources, SC-derived human hepatocyte-like cells (HLCs) have been suggested as a reliable alternative (Szkolnicka et al., 2014; Takayama et al., 2014;Freyer et al., 2016;Cipriano et al., 2017a,b, 2020;Figure 2). SCs represent normal primary cells with a mostly stable genotype than hepatoma cell lines. Moreover, compared to hpHep, present unlimited supply, can be maintained for long-term and may also represent a broad patient population (Godoy et al., 2013;Horvath et al., 2016). As such, stem or progenitor cells are an exciting prospect for drug metabolism studies and cell transplantation, providing that high levels of hepatocyte-like functions can be induced and tumorigenicity concerns are overcome. Many protocols have been developed for differentiating SCs into HLCs with different approaches, such as mimicking liver development through the sequential addition of growth factors and cytokines (Cai et al., 2007;Hay et al., 2008b;Brolén et al., 2010), modulation of signaling pathways (Hay et al., 2008a) or by using epigenetic modifiers (Sharma et al., 2006;Dong et al., 2009;Norrman et al., 2013). Currently, most work has been developed using induced pluripotent SCs (iPSCs) isolated from adult tissues in an non-invasive way, with Frontiers in Cell and Developmental Biology | www.frontiersin.org 6February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 7 Serras et al. 3D Liver Models in Toxicology Studies FIGURE 2 | Summary of the advantages and limitations of commonly used cell sources for in vitro liver models. HLCs, hepatocyte-like cells; hpHep, human primary hepatocytes; NPCs, non-parenchymal cells. Frontiers in Cell and Developmental Biology | www.frontiersin.org 7February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 8 Serras et al. 3D Liver Models in Toxicology Studies promising outcomes (Sauer et al., 2014;Sirenko et al., 2016; Yamashita et al., 2018;Pareja et al., 2020). An example is the work from Gao and Liu (2017), that revealed that iPSC-derived HLCs resembled hpHep more closely than most hepatoma cell lines in global gene expression profiles, specifically in the expression of genes involved in hepatotoxicity, drug-metabolizing enzymes, transporters, and nuclear receptors. Interestingly, Freyer et al. (2016) detected CYP1A2, CYP2B6, and CYP3A4 activities in iPSC-derived HLCs, but also at a lower level than in hpHep. Likewise, Takayama et al. (2014) showed that iPSC-derived HLCs retained donor-specific drug metabolism capacity and drug responsiveness, reflecting interindividual differences, but lower CYP1A2, CYP2C9, CYP2D6, and CYP3A4 activities when compared to the correspondent hpHep donors. Besides hepatocytes, efforts have also been made to generate NPCs from iPSCs, including cholangiocytes (Ogawa et al., 2015;Sampaziotis et al., 2015), Kupffer cells (Tasnim et al., 2019), LSECs (Koui et al., 2017), and hepatic stellate cells (Koui et al., 2017;Coll et al., 2018). Nevertheless, iPSC technology has some limitations related to the genomic instability and to residual iPSC-specific methylation patterns that links these cells to their tissue of origin, which ultimately may affect their final differentiation (Robinton and Daley, 2012). Still, iPSC-derived HLCs show powerful value not only for toxicology applications but also for disease modeling and personalized drug therapy. Alternatively, adult liver SCs (LSCs) are a particularly interesting SC source. LSCs can be obtained from liver biopsies, propagated in vitro and differentiated into mature hepatocytes (Huch et al., 2015;Wang et al., 2015;Luo et al., 2018). LSCs are located in the epithelium of the canals of Hering and contribute to liver regeneration in response to an injury (Overi et al., 2018). LSCs are bipotent, being able to differentiate into hepatocytes or cholangiocytes. As such, these cells express SC (e.g., SRY-box transcription factor 9, Sox9), cholangiocyte (CK19), and hepatocyte (CK-18) markers (Overi et al., 2018). The identification of populations of proliferating and self-renewing cells that can replace injured hepatocytes can be performed with lineage tracing approaches using Wnt-responsive genes such as Axin2 or Lgr5 (Huch et al., 2013;Wang et al., 2015). Mesenchymal SCs (MSCs) including liver, bone-marrow, adipose, or umbilical cord tissue-derived MSCs have also been used for deriving human HLCs (Snykers et al., 2006, 2007;Banas et al., 2007;Kazemnejad et al., 2008;Okura et al., 2010;Yin et al., 2015;Fu et al., 2016;Yang et al., 2020). From those, human neonatal MSCs stand as a promising choice due to the noninvasive access and to its more primitive origin (Hass et al., 2011;Lee et al., 2012;Cipriano et al., 2017a;Yu Y. B. et al., 2018. The first report using human neonatal umbilical cord tissue-derived MSCs (hnMSCs) was from Campard et al. (2008). Therein, hnMSCs were differentiated into HLCs with impressive results, i.e., presenting hepatic-specific markers, urea production, glycogen accumulation, and CYP3A4 activity. Afterward, other researchers also differentiated hnMSCs into HLCs exhibiting hepatic markers, urea and albumin (ALB) production. However, their biotransformation activity was not assessed (Zhang et al., 2009;Zhao et al., 2009;Zhou et al., 2014). More recently, Cipriano et al. (2017a) generated hnMSC-derived HLCs with more partial hepatic phenotype, sharing expression of gene groups with hpHep that was not observed between HepG2 and hnMSCs, as shown by genome-wide analysis (Cipriano et al., 2017a). Importantly, when resorting to the 3D culture technology, MSC-derived HLCs demonstrate an improvement in phase I biotransformation activity, urea and ALB production, as well as relevant diclofenac and nevirapine biotransformation capacity, which supports its potential usefulness for toxicological studies (Cipriano et al., 2017b, 2020). Nevertheless, despite the growing efforts made in this research field a complete mature hepatocyte phenotype of HLCs derived from MSCs has not yet been achieved. Perhaps liver MSCs may be the best choice, because they are originally committed to hepatic lineage, but an accurate comparison of hepatocytes derived from human liver MSCs and other sources must still be done (Kholodenko et al., 2019;Shi et al., 2020). All these strategies are not deprived of challenges as they require specialized personnel and expensive culture medium supplementation, whereas a complete mature phenotype has not yet been achieved. The fetal HLC phenotype is still a challenge, revealing the need to further understand hepatic differentiation mechanisms and optimizing differentiation strategies (Raju et al., 2018;Raasch et al., 2019). Moreover, the use of diverse differentiation protocols across different laboratories hinders the robustness assessment of the use of HLCs for toxicology applications. To address this issue, some authors proposed a set of cellular markers and functional assays to control the quality of iPSC-derived cells, since these are the most common type of SCs used in vitro (Daston et al., 2015;Beken et al., 2016). Although the specific metrics to monitor cell characteristics may vary according to the differentiation protocol and cell line used, this guide provides an important reference for quality control of other types of SC-based models. For HLCs, the most important markers to be analyzed are CYP3A4, CYP2B6, CYP1A1/2, CYP2C9, CYP2C19, CYP2D6, alpha-fetoprotein (AFP), ALB, Sox17, C-X-C motif chemokine receptor 4 (CXCR4), hepatocyte growth factor (HGF), hepatocyte nuclear factor 4 alpha (HNF4α), tyrosine aminotransferase (TAT), transthyretin (TTR) while functional assays include urea and ALB synthesis, glycogen uptake, fibrinogen secretion, ATP, and GSH levels, CYP3A activity in particular, phase II activities and drug transporter capacity (Beken et al., 2016). Nonetheless, due to overall unsatisfactory phenotype of the currently available cell sources, at least for some hepatic features, the improvement of the cell culture system has been explored as will be further described in the following sections. Three-Dimensional Liver Systems The major shortcoming of the currently available in vitro liver preparations lays on insufficient hepatocyte-like functions and metabolic competence. In fact, none of the hpHep-, HepG2-, or HepaRG-based 2D models are suitable to indicate the risk of hepatotoxicity for novel chemical entities unless PK data are incorporated in the study, supporting the need to employ more sophisticated technologies to increase prediction sensitivity (Sison-Young et al., 2017). Accordingly, recent reports emphasize a shift, by the industry, from 2D in vitro Frontiers in Cell and Developmental Biology | www.frontiersin.org 8February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 9 Serras et al. 3D Liver Models in Toxicology Studies FIGURE 3 | Summary of the characteristics of complex 3D in vitro cell culture systems for hepatotoxicity studies. (A) Sandwich cultures; (B) static spheroid cultures; (C) dynamic spheroid cultures; (D) bioreactors; (E) bioprinting; (F) microfluidic platforms. PBPK, physiologically based pharmacokinetic modeling; TD, toxicodynamics; TK, toxicokinetics. approaches to more complex 3D assays where multicellular microphysiological devices are being evaluated within a vision to replicate the characteristics and response of human tissues in vivo (Vivares et al., 2015). Traditionally, 2D cultures are employed as in vitro models due to their ease of use to quickly screen large numbers of compounds. However, this culture approach negatively impacts cell expression profiles (Engler et al., 2006) and causes primary hepatocytes to rapidly lose their differentiation markers (Treyer and Müsch, 2013), compromising long-term and repeated dose studies. On the other hand, 3D cell culture systems have been shown to improve the biotransformation capacities in primary hepatocytes (Tibbitt and Anseth, 2009;Miranda et al., 2010; Mandenius et al., 2011;Mueller et al., 2011;Zeilinger et al., 2011;Schyschka et al., 2013), hepatoma cell lines (Fey and Wrzesinski, 2012;Molina-Jimenez et al., 2012;Wrzesinski et al., 2014) and SC-derived HLCs (Gieseck et al., 2014;Freyer et al., 2016;Cipriano et al., 2017b, 2020) over time in culture. In general, as summarized in Figure 3 and Table 1,3D cell culture systems are prone to high-throughput adaptation and scale up but vary in complexity and on remote monitoring of cell culture parameters. Three-dimensional systems can comprise extracellular matrix (ECM) sandwich cultures (Chatterjee et al., 2014;Deharde et al., 2016), spheroid and organoid cultures (Miranda et al., 2009;Leite et al., 2011, 2012;Tostoes et al., 2011;Wrzesinski et al., 2014;Huch et al., 2015;Bell et al., 2016; Peng et al., 2018;Ramli et al., 2020), cells adherent to a scaffold (Kazemnejad et al., 2008;Lin and Chang, 2008;Haycock, 2011), or more complex cellular systems such as hollow-fiber bioreactors (Darnell et al., 2011, 2012;Lübberstedt et al., 2011;Mueller et al., 2011;Zeilinger et al., 2011;Hoffmann et al., 2012;Cipriano et al., 2017b), bioartificial livers (Chan et al., 2004), multi-well perfused bioreactors (Domansky et al., 2010;Vivares et al., 2015; Aeby et al., 2018;Mannaerts et al., 2020), and more recently bioprinted systems (Lauschke et al., 2016;Goulart et al., 2019) and microfluidic platforms (MP) (Rennert et al., 2015;Ma C. et al., 2016;Bauer et al., 2017;Danoy et al., 2019). Three-dimensional cell cultures can also be achieved using either static or dynamic systems. Static systems are less complex and do not include medium flow, while dynamic systems might be stirred and/or perfused, depending on the cell culture system complexity (Miranda et al., 2009, 2010;Tostoes et al., 2011). Static culture systems are compliant with high-throughput and are usually adopted for the optimization of culture medium constitution, to test a diversity of toxic compounds using fewer cells or as a step to produce spheroids to be used in more complex 3D culture systems, e.g., bioreactors (Wrzesinski et al., 2014; Fey et al., 2020). In contrast, stirring conditions facilitate oxygen diffusion as well as medium homogenization, further resembling the physiological blood flow, and create a hydro-dynamic shear stress that must be balanced, by improving cell performance while avoiding cellular stress and death (Conway et al., 2009). Also, a continuously perfused system is particularly interesting in hepatocyte cell culture and in xenobiotic metabolism studies, avoiding fluctuations of basic cell culture parameters such as pH, oxygen, glucose, and lactate concentration and the accumulation Frontiers in Cell and Developmental Biology | www.frontiersin.org 9February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 16 Serras et al. 3D Liver Models in Toxicology Studies biomarkers are essential to identify also the models’ ability to mimic processes related to cholestasis, steatosis, genotoxicity, and viral hepatitis (Shen et al., 2012;Bell et al., 2016;Hendriks et al., 2016;Leite et al., 2016;Prill et al., 2016;Williams et al., 2020), amongst others. Paracetamol Paracetamol is a widely used antipyretic and non-opioid analgesic agent that constitutes an example of a safe drug at therapeutic doses, but overdosage causes predictable and reproducible hepatotoxicity through mitochondrial dysfunction and centrilobular necrosis in the liver (Hinson et al., 2010). Paracetamol is metabolized mainly by conjugation with sulfate and glucuronic acid (Riches et al., 2009) and, in a less extent, by oxidation by CY2E1, CYP1A2, CYP2D6, CYP2A6, and CYP3A4 (Mazaleuskaya et al., 2015). As previously stated, its oxidation generates NAPQI that is detoxified by GSH conjugation, through glutathione S-transferases (GSTs) GSTP1, GSTT1, and GSTM1. When large quantities of NAPQI are formed, liver GSH pool can be critically depleted, meaning that excess NAPQI is not detoxified and cell injury occurs, namely trough the modification of cellular proteins. Protein binding leads to oxidative stress and mitochondrial damage (McGill and Jaeschke, 2013;Caparrotta et al., 2018). Paracetamol toxicity is also related to calcium accumulation and activation of endonucleases, DNA damage (Boelsterli, 2003), ATP depletion, Jnk activation, up-regulation of electron transport chain protein components and activation of p53 signaling (Davis and Stamper, 2016). Supplementary Table 1 summarizes the collected in vitro data for paracetamol. It suggests that mouse primary hepatocytes are more sensitive to paracetamol, with lower IC50 values (Jemnitz et al., 2008;Kuˇ cera et al., 2017), followed by rpHep, hepatic cell lines HepG2 and HepaRG, hpHep and HLCs (Lewerenz et al., 2003;Jemnitz et al., 2008;Riches et al., 2009;Zhang et al., 2011;Tasnim et al., 2015;Bell et al., 2017), highlighting not only the interspecies differences but also the importance of choosing a representative cell type (Carmo et al., 2004; Reder-Hilz et al., 2004). Paracetamol IC50 values compiled in Supplementary Table 1 further suggest that 2D cultures are less sensitive to paracetamol toxicity than 3D cultures (Gunness et al., 2013;Jang et al., 2015;Gaskell et al., 2016;Bell et al., 2017;Li et al., 2020). It was demonstrated increased sensitivity of 3D human liver microtissues, with subsequently lower IC50 values for a panel of known hepatotoxicants, including paracetamol, in comparison with 2D-plated hpHep (Proctor et al., 2017). Moreover, by using a 3D liver-sinusoid-on-a-chip of HepG2, Deng et al. (2019) showed that not only this system was able to improve cell functions, but also its sensitivity to paracetamol when compared to 2D-plated HepG2. Besides, even within 3D systems, different culture strategies may lead to different results for the same drug. Jang et al. (2015) tested HepG2 in a 3D static MatrigelR  culture and in a 3D microfluidic chip and obtained a higher sensitivity for hepatotoxicity in the latter, justified by its improved maintenance of hepatic functions. Foster et al. (2019) found also an increased sensitivity in 3D co-culture systems of hpHep and NPCs compared to hpHep spheroids. Likewise, Li et al. (2020) observed an augmented toxicity to paracetamol in the coculture spheroids group (hpHep and Kupffer cells). Interestingly, when both systems were co-treated with lipopolysaccharides (mimicking inflammatory conditions), a higher protective role was detected in the co-culture system, mainly due to Kupffer cells, when comparing to hpHep spheroids. In fact, exposure and response to paracetamol under healthy or pre-inflammatory states may lead to different cytokine release profiles with distinct activation of immune cells (Kim et al., 2017;Li et al., 2020). Although it is generally acknowledged that paracetamol has only very weak anti-inflammatory properties, it should be highlighted that is commonly administered already under an inflammatory condition. This reinforces the need of considering co-cultures for an early identification of possible drug-induced hepatotoxic immunological responses depending on the patient health state. Although mechanistic endpoints are not often represented in the majority of studies using paracetamol as hepatotoxicant, the reports that present this important information assess mainly mitochondrial dysfunction and oxidative stress (Bruderer et al., 2015;Goda et al., 2016;Zhang C. et al., 2020), followed by reactive metabolites formation and liver cholestasis or steatosis (Lewerenz et al., 2003;Prot et al., 2012;Prill et al., 2016;Kuˇ cera et al., 2017;Williams et al., 2020) and liver fibrosis (Leite et al., 2016). Nevertheless, the drastic differences in sensitivity to paracetamol, may be due not only to the cell type but also to the different culture conditions, highlighting the importance of both the cell architecture and the presence of other liver cell types for the study of distinct pathways that may be involved in drug toxicity. Diclofenac Diclofenac is one of the most worldwide prescribed NSAID. It has been linked with rare, albeit significant, cases of severe hepatotoxicity with a fatality rate of 10% (Aithal, 2004). Diclofenac is mainly metabolized by CYP2C9 into 4-OHdiclofenac and in a lower extent converted into 5-OH-diclofenac by CYP3A4. Diclofenac and its metabolites are conjugated with glucuronic acid by UGT2B7 and excreted across the canalicular plasma membrane into the bile via MRP-2 (Aithal, 2004). There is evidence that individuals that present increased glucuronidation activity as a consequence of a genetic polymorphism in UGT2B7 (C-161T allele) exhibit a 9-fold increased risk of adverse hepatic reactions (Daly et al., 2007). Diclofenac-acylglucuronide may also conjugate with GSH forming a diclofenac glutathione thioester (Syed et al., 2016). Thus, diclofenac metabolism comprises phase I, II, and III activities and its toxic effects can be associated to the reactive metabolites 4-OHdiclofenac, diclofenac-acyl-glucuronide, diclofenac glutathione thioester, and 4-OH-diclofenac-acyl-glucuronide (Aithal, 2004) that cause ATP depletion resulting in mitochondrial toxicity (Syed et al., 2016). As such, diclofenac constitutes an example of a hepatotoxicant dependent on bioactivation. Compared to paracetamol, the available studies of diclofenac addressing the toxicity in 2D versus 3D cell cultures are fewer (Supplementary Table 2). Most importantly, regarding mechanistic biomarkers of toxicity, only a minority of studies assess endpoints for mitochondrial dysfunction (Goda et al., 2016), reactive metabolite formation, calcium homeostasis Frontiers in Cell and Developmental Biology | www.frontiersin.org 16 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 17 Serras et al. 3D Liver Models in Toxicology Studies (Ponsoda et al., 1995), and liver cholestasis or steatosis (Bell et al., 2016;Williams et al., 2020). As shown in Supplementary Table 2,Gaskell et al. (2016) and Cipriano et al. (2017b) obtained IC50 values for the 3D spheroid cultures of HepG2/C3A and HLCs, respectively, lower than for the corresponding 2D cultures. This may be due to an increase in phase II (glucuronidation) activity, indicating that the 3D system may be more representative of the biological response. Ramaiahgari et al. (2014) attained higher sensitivity to diclofenac in the 3D culture of HepG2 compared to 2D but the IC50 values were higher than those reported by Gaskell et al. (2016).Yu K. N. et al. (2018) also demonstrated that the addition of both phase I and II enzymes in a 3D miniaturized Hep3B cell system led to a more predictive assessment of diclofenac’s toxicity when compared to the 3D system groups with only CYP450 enzymes or human liver microsomes added and its 2D counterpart. Also, Atienzar et al. (2014) proved that the co-culture of dog hepatocytes with NPCs present similar sensitivity to diclofenac compared to HepG2 but less sensitivity than hpHep in a 5-day exposure culture. This IC50 variation is transversal to the majority of reports presented in Supplementary Table 2, in which most of the studies only evaluate one type of culture system (Wang et al., 2002;Xu et al., 2003;Lauer et al., 2009;Lin et al., 2012;Goda et al., 2016;Knospel et al., 2016;Sarkar et al., 2017), hindering the comparison between both types of cultures. Besides the relevance of the culture system, this observation reinforces the importance to include mechanistic insights in early toxicity assessments rather than rely solely on cell viability quantification. A competent and complete in vitro model may provide a higher amount of valuable information if more variables are taken into account. Troglitazone Troglitazone (TGZ) is a thiazolidinedione derivative developed for the treatment of type II diabetes. Soon after being approved, TGZ was withdrawn from the marked in Europe and 3 years afterward in the United States due to non-immune idiosyncratic toxicity (Chojkier, 2005). It is a classic example of a drug whose toxicity failed to be predicted during drug development. TGZ is extensively metabolized by CYP3A4 and GST, it is a CYP3A and 2B6 inducer and is able, as well as its sulfate conjugate, to inhibit BSEP (Sahi et al., 2000;Kassahun et al., 2001). This leads to an increased formation of TGZ metabolites along with its intracellular accumulation, resulting in intrahepatic cholestasis, mitochondrial dysfunction, covalent binding to proteins, and macromolecular damage ultimately leading to apoptosis (Smith, 2003;Chojkier, 2005). TGZ is more toxic in humans than in rodent models (Shen, 2007;Kostadinova et al., 2013). In view of this, TGZ was only identified as hepatotoxic after reaching the market. TGZ is, thus, an example of the importance of identifying species-specific toxicity. In fact, Shen et al. (2012), using 3D gel entrapped rat and human hepatocytes, observed that at clinical doses of TGZ, hepatotoxicity was absent in rat hepatocytes, but present in human hepatocytes. Similarly, Kostadinova et al. (2013) reported TGZ-induced cytotoxicity in human 3D liver cells but only minor effects in rat 3D liver. In addition, as summarized in Supplementary Table 3, and in accordance with clinical observations, 3D human liver models show increased sensitivity to TGZ toxicity which may be related to higher formation of metabolites as consequence of the higher induction ability of 3D models. Nonetheless, Gunness et al. (2013) obtained discrepant results, with the 3D model of HepaRG cells being 10-fold less sensitive to TGZ than the monolayer model. On the other hand, this decreased sensitivity might reflect a higher clearance, commonly increased in 3D cultures due to the cellular architecture. Moreover, HepaRG cells present higher CYP3A4 activity (Aninat et al., 2006) that might interfere with drug and metabolite clearances since TGZ metabolites may also be hepatotoxic (Tolosa et al., 2018). Another explanation could be the known TGZ-induced BSEP inhibition (Jackson et al., 2018), not investigated in this study. Hendriks et al. (2016) assessed BSEP inhibition in two long-term 3D spheroid models of HepaRG and hpHep with repeated drug exposure and bile acids co-exposure and was able to detect the cholestatic effect of several compounds, including TGZ, in both cell types. Independently of the studied mechanisms, the co-culture of hpHep (Kostadinova et al., 2013;Proctor et al., 2017;Li et al., 2020) or hepatoma cell lines (Granitzny et al., 2017) with NPCs displayed a higher sensitivity to TGZ toxicity (Kostadinova et al., 2013). Using HLCs as alternative sources to hpHep, Tasnim et al. (2016) found similar sensitivity between 3D hESC/hiPSC-HLCs and hpHep upon exposure to TGZ for 24 h, whereas when comparing to its 2D counterpart, 3D cellulosic scaffold cultured hiPSC-HLCs showed a slightly lower sensitivity. On the other hand, Takayama et al. (2013) showed that 3D-cultured iPSCHLCs presented a decreased sensitivity to TGZ after 24 h of exposure when compared to hpHep, but a better sensitivity compared to 3D-cultured HepG2. Moreover, Holmgren et al. (2014) was able to use hiPSC-HLCs in a long-term culture and successfully detected steatosis in the cells exposed to TGZ for 2 days. This is an important step to show that these HLCs may reveal the mechanistic pattern of TGZ hepatotoxicity and stand as a good alternative for hpHep. CHALLENGES WITHIN THE ASSESSMENT OF DRUGS HEPATOTOXICITY USING IN VITRO LIVER MODELS The pharmaceutical industry is clearly interested on the early identification of toxicity cues in models covering different aspects of human liver (patho)physiology (Bale et al., 2014). On the other hand, the chemical/cosmetic industry has been politically stressed to use advanced alternatives for animal testing for hazard identification and characterization. Although each type of industry presents different needs and goals, the early identification of potential hepatotoxic substances and deep understanding of hepatotoxic mechanisms in relevant in vitro models will profit both industries. In this section, we propose a roadmap with the essential steps to assess drugs hepatotoxicity using in vitro liver models. Frontiers in Cell and Developmental Biology | www.frontiersin.org 17 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 18 Serras et al. 3D Liver Models in Toxicology Studies FIGURE 4 | Roadmap for assessing drugs hepatotoxicity mechanisms using in vitro models that might be used alone or in combination at different points and on different scales. Tier 1 comprises single-cell systems that report on immediate chemical/biological effects such as cytotoxicity and bioactivation while Tier 2 includes more complex systems containing liver cells in a more physiologic state, enabling assessment of the consequences of acute and chronic drug exposure. Moreover, phenotypic characterization and the pharmacological and toxicological functionality of a system and the ability to identify toxicity mechanisms needs to be considered before undertaking toxicological investigations to ensure that the most appropriate methods are used. Depending on the complexity, each model might be able to represent one or more liver functional endpoints and can be used alone or in combination depending on the hepatotoxicity mechanisms that are intended to study. To integrate findings from different test systems and to dissect the multilevel impact of compounds, bioinformatics and machine learning models may also be useful, which will ultimately contribute to more informed decision-making in the drug development and risk assessment fields. SC, stem cells; HLCs, hepatocyte-like cells; MPS, microphysiological system. As no single currently used model can recapitulate all human hepatotoxicity mechanisms, we further support the need for a systematic tiered approach for drug hepatotoxicity assessment combining more than one model. Indeed, the model systems with increased biological complexity might be efficiently and effectively used alone or in combination, at different points, and on different scales of the drug development process. The final goal is thus to cover the different human hepatotoxicity mechanisms needed to provide accurate hazard identification and risk assessment (Figure 4). Finally, the data generated in such non-clinical assays should be integrated to deliver robust information to drug regulators to improve the decision-making process (Figures 4,5). Firstly, the cell source needs to be carefully considered to ensure the appropriate tissue context. Hepatotoxicity involves different mechanisms in a multistep and multicellular process (Grattagliano et al., 2009;Gregus, 2013). As such, there is no single model or test that can evaluate a chemical’s risk of inducing liver injury, but rather a set of well-characterized hepatic models with well-defined purposes to be used in a multistep manner. This approach may vary in terms of cell source and culture complexity, allowing to assess specific toxicity mechanisms as well as to properly mimic the different types and stages of liver toxicity. In particular, human-derived cells, namely hpHep or SCderived HLCs, contribute to more relevant in vitro systems by allowing to capture population heterogeneity and to represent healthy and disease conditions. On the other hand, simpler and single-cell system based on, e.g., hepatic cell lines and monolayer cultures enable high-throughput applications for testing a wide set of conditions and allow to preliminarily understand a drug’s basic biological effects in human cells (Figure 4, tier 1). Conversely, more complex systems, such as 3D-based cultures improve data accuracy not only concerning cytotoxicity, but also biotransformation activity and drug accumulation by mimicking tissue architecture, mechanical forces and gradients of oxygen, nutrients, and drugs that are found in vivo (Figure 4, tier 2). Additionally, complex models that include different liver cell types, e.g., hepatocytes and stellate cells or immune cells, allow to explore additional biological responses such as cholestasis, steatosis, fibrosis, and inflammation (Leite et al., 2016; Jeon et al., 2020). A crucial step within the development and selection of a relevant in vitro model for non-clinical studies is its thorough characterization. The systems’ phenotype and functionality need to be assessed to understand which pharmacology and toxicology mechanisms can be accurately represented Frontiers in Cell and Developmental Biology | www.frontiersin.org 18 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 19 Serras et al. 3D Liver Models in Toxicology Studies FIGURE 5 | Data integration from non-clinical assays for prediction of clinical conditions. A shift in paradigm where fit-for-purpose human-based in vitro models, particularly using in vitro 3D systems and causality-inferring bioinformatic approaches, might provide high-quality data for relevant extrapolation of human toxicokinetics and toxicodynamics, ultimately leading to the prediction of human hepatotoxicity mechanisms and molecules’ risk assessment. As a consequence, animal models will then become progressively less used with the increasing complexity and relevance of these strategies. AOP, adverse outcome pathway; AUC, area under the curve; Cmax, maximum plasma concentration; IC50, half maximal inhibitory concentration; PBPK, physiologically based pharmacokinetic; PK, pharmacokinetic; PD, pharmacodynamic; TD, toxicodynamics; TK, toxicokinetics. and evaluated in each model. Importantly, the capacity of the models for both short and/or long-term exposures should be evaluated as well (Figure 4). Indeed, the models’ comparison presented in section “In vitro Hepatotoxicity Studies: 3D Versus 2D” revealed to be challenging and little informative because most of the hepatotoxicity studies available in the literature focus on cytotoxicity estimation instead of analyzing specific functional endpoints of hepatotoxicity such as altered conversion of primary and secondary metabolites, disruption of repairing mechanisms, immunological response, mitochondrial dysfunction, bile salt transporter modification, lipids accumulation, or calcium homeostasis alteration, which highlight hepatic injury mechanisms such as inflammation, cholestasis, steatosis, fibrosis, or genotoxicity (Xu et al., 2004; O’Brien et al., 2006;Khetani et al., 2013;Trask et al., 2014;Schadt et al., 2015;Bell et al., 2016;Leite et al., 2016;Jeon et al., 2020; Williams et al., 2020;Zhang C. et al., 2020). Retrospective analysis of approved or failed drugs with a known mechanism displaying the expected hepatic response is a common approach to determine the capability of an in vitro system to accurately recapitulate the human liver biological response (Jemnitz et al., 2008;Kostadinova et al., 2013;Messner et al., 2013;Hendriks et al., 2016;Bell et al., 2017;Cipriano et al., 2017b;Proctor et al., 2017;Sarkar et al., 2017;Foster et al., 2019;Williams et al., 2020), e.g., paracetamol for induced necrosis, valproic acid for induced steatosis, or cyclosporine A for induced cholestasis (Vinken and Hengstler, 2018). The data generated with these systems should then be compared with clinical data as well as with other in vitro methods to understand applicability and limitations of each system for assessing the inherent risk of a chemical or a new molecular entity (Beken et al., 2016;Figure 5). Models are commonly assessed by direct comparison of in vitro to the in vivo maximum plasma concentration (Cmax) (Godoy et al., 2013;Shah et al., 2015;Proctor et al., 2017), often multiplied by a factor of 20× to 100×(O’Brien et al., 2006;Xu et al., 2008;Khetani et al., 2013;Proctor et al., 2017;Vorrink et al., 2018;Yu K. N. et al., 2018). Thus, knowing that the Cmax of paracetamol, diclofenac, and TGZ is 165.38 µM, 10.13 µM (Regenthal et al., 1999) and 2.82 µM (Loi et al., 1999), respectively, it is possible to comprehend by Supplementary Tables 1–3 that, in general, 3D models present better capability of predicting cytotoxicity than 2D models. However, this comparison assumes that the ratio of test compounds in blood or plasma in vivo is the same in the cell culture medium in vitro (Vinken and Hengstler, 2018) and that cells response in vivo is the same in vitro. Also, within the same drug, such a big range of concentration (20×to 100×Cmax) may represent different mechanisms of toxicity (Albrecht et al., 2019). Frontiers in Cell and Developmental Biology | www.frontiersin.org 19 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 20 Serras et al. 3D Liver Models in Toxicology Studies A paradigm change from selecting a list of reference chemicals that cause cytotoxicity, to selecting a group of positive and negative mechanistic controls, coupled to a mechanisticbased selection of functional endpoints, including biomarkers assessment, is thus essential for using advanced in vitro models to its full potential. An example of a mechanistic positive control for a test system where biliary excretion is measured would be TGZ since it is a known BSEP inhibitor associated with bile acid accumulation (Funk et al., 2001;Jackson et al., 2018). Moreover, the combination of advanced 3D hepatic in vitro models with other advanced endpoint methodologies and systems biology employing -omics approaches (e.g., RNA-seq, Epigenetics, and ChIP-seq proteomics) could further support better prediction of hepatotoxicity. Finally, the integration of the biological data obtained with the selected in vitro models may support a deeper understanding of the models’ potential to predict specific mechanisms of detoxification and toxicity. The inclusion of relevant positive and negative controls that validate the obtained data will support robust knowledge on chemical’s risks and human toxicity mechanisms that has not been detected so far using monolayers of a single cell type. CONCLUSION There has been major progress toward the development of more physiologically relevant hepatic in vitro models, namely through the application of 3D culture techniques. Despite the wide variety of available cellular models along with distinct cell sources, 3D-based liver systems and co-culture strategies improve hepatic-specific functions and sustain the culture through longer periods than 2D counterparts, emphasizing the value-added capacity of such systems to mimic the multicellular mechanisms involved both in intrinsic and idiosyncratic liver toxicity. This has been demonstrated in paracetamol, diclofenac and TGZ toxicities studies where IC50 values from 3D cultures confirmed its higher sensitivity. Nevertheless, the appropriate in vitro system’s evaluation and proper extrapolation of the in vitro data requires a paradigm shift from quantitative cytotoxicity assessment to a comprehensive mechanistic evaluation of the response to chemicals. Advanced 3D systems provide the opportunity to investigate mechanistic hepatotoxicology if accompanied by a careful selection of adequate positive and negative controls and systems toxicology data coupled to disease-related functional endpoints. Despite the advances in creating more physiologically relevant culture systems, a major difficulty is still the standardization of protocols across laboratories and the selection of critical positive and negative controls to assess the transferability and reproducibility of models across different groups. Importantly, advanced endpoint methodologies are essential to identify the applicability of each hepatic system, further allowing comparison between studies. Such a detailed and complete in vitro evaluation will support future decision on the adequate model to be used in chemical risk assessment and in a non-clinical drug development scheme. AUTHOR CONTRIBUTIONS JM, MC, JR, and AS: conception and design. JM, MC, JR, AS, AR, and NO: writing and critically review the manuscript. JR, AS, and AR: figure design and elaboration. JM: directing manuscript. All authors contributed to the article and approved the submitted version. FUNDING This research has been supported by FCT (Portugal) through the research grants and scholarship PTDC/MEDTOX/29183/2017, UIDB/04138/2020, UIDP/04138/2020, and SFRH/BD/144130/2019 to JR and by the H2020, European Commission, though the MSCA-IF-EF-ST – Standard EF to MC (GA-845147-LIV-AD-ON-A-CHIP). SUPPLEMENTARY MATERIAL The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcell.2021. 626805/full#supplementary-material REFERENCES Adiels, C. B., Goksör, M., Wölfl, S., Paukštyte, J., Banaeiyan, A. A., and Theobald, J. (2017). Design and fabrication of a scalable liver-lobule-on-a-chip microphysiological platform. Biofabrication 9:015014. doi: 10.1088/1758-5090/ 9/1/015014 Aeby, E. A., Misun, P. M., Hierlemann, A., and Frey, O. (2018). Microfluidic hydrogel hanging-drop network for long-term culturing of 3D microtissues and simultaneous high-resolution imaging. Adv. Biosyst. 2, 1–11. doi: 10.1002/adbi. 201800054 Ahmad, J., and Odin, J. A. (2017). Epidemiology and genetic risk factors of drug hepatotoxicity. Clin. Liver Dis. 21, 55–72. doi: 10.1016/j.cld.2016.08.004 Aimar, A., Palermo, A., and Innocenti, B. (2019). The role of 3D printing in medical applications: a state of the art. J. Healthc. Eng. 2019:5340616. doi: 10.1155/2019/ 5340616 Aithal, G. P. (2004). Diclofenac-induced liver injury: a paradigm of idiosyncratic drug toxicity. Expert Opin. Drug Saf. 3, 519–523. doi: 10.1517/14740338.3.6.519 Aithal, G. P., Ramsay, L., Daly, A. K., Sonchit, N., Leathart, J. B. S., Alexander, G., et al. (2004). Hepatic adducts, circulating antibodies, and cytokine polymorphisms in patients with diclofenac hepatotoxicity. Hepatology 39, 1430–1440. doi: 10.1002/hep.20205 Albrecht, W., Kappenberg, F., Brecklinghaus, T., Stoeber, R., Marchan, R., Zhang, M., et al. (2019). Prediction of human drug-induced liver injury (DILI) in relation to oral doses and blood concentrations. Arch. Toxicol. 93, 1609–1637. doi: 10.1007/s00204-019-02492-9 Allen, J. W., Khetani, S. R., and Bhatia, S. N. (2005). In vitro zonation and toxicity in a hepatocyte bioreactor. Toxicol. Sci. 84, 110–119. doi: 10.1093/toxsci/kfi05 Andersen, M. E., and Krewski, D. (2009). Toxicity testing in the 21st century: bringing the vision to life. Toxicol. Sci. 107, 324–330. doi: 10.1093/ toxsci/kfn255 Frontiers in Cell and Developmental Biology | www.frontiersin.org 20 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 21 Serras et al. 3D Liver Models in Toxicology Studies Andersen, M. E., and Krewski, D. (2010). The vision of toxicity testing in the 21st century: moving from discussion to action. Toxicol. Sci. 117, 17–24. doi: 10.1093/toxsci/kfq188 Andrade, R., Agundez, J., Lucena, M., Martinez, C., Cueto, R., and Garcia-Martin, E. (2009). Pharmacogenomics in Drug Induced Liver Injury. Curr. Drug Metab. 10, 956–970. doi: 10.2174/1389200097907 11805 Aninat, C., Piton, A., Glaise, D., Le Charpentier, T., Langouët, S., Morel, F., et al. (2006). Expression of cytochromes P450, conjugating enzymes and nuclear receptors in human hepatoma HepaRG cells. Drug Metab. Dispos. 34, 75–83. doi: 10.1124/dmd.105.006759 Asai, A., Aihara, E., Watson, C., Mourya, R., Mizuochi, T., Shivakumar, P., et al. (2017). Paracrine signals regulate human liver organoid maturation from induced pluripotent stem cells. Development 144, 1056–1064. doi: 10.1242/dev. 142794 Asha, S., and Vidyavathi, M. (2010). Role of human liver microsomes in in vitro metabolism of drugs-a review. Appl. Biochem. Biotechnol. 160, 1699–1722. doi: 10.1007/s12010-009-8689-6 Atienzar, F. A., Novik, E. I., Gerets, H. H., Parekh, A., Delatour, C., Cardenas, A., et al. (2014). Predictivity of dog co-culture model, primary human hepatocytes and HepG2 cells for the detection of hepatotoxic drugs in humans. Toxicol. Appl. Pharmacol. 275, 44–61. doi: 10.1016/j.taap.2013.11.022 Babai, S., Auclert, L., and Le-Louët, H. (2018). Safety data and withdrawal of hepatotoxic drugs. Therapie [Epub ahead of print]. doi: 10.1016/j.therap.2018. 02.004 Bale, S. S., Vernetti, L., Senutovitch, N., Jindal, R., Hegde, M., Gough, A., et al. (2014). In vitro platforms for evaluating liver toxicity. Exp. Biol. Med. 239, 1180–1191. doi: 10.1177/1535370214531872 Balls, M. (2011). Modern alternative approaches to the problem of druginduced liver injury. ATLA Altern. to Lab. Anim. 39, 103–107. doi: 10.1177/ 026119291103900205 Banas, A., Teratani, T., Yamamoto, Y., Tokuhara, M., Takeshita, F., Quinn, G., et al. (2007). Adipose tissue-derived mesenchymal stem cells as a source of human hepatocytes. Hepatology 46, 219–228. doi: 10.1002/hep.21704 Bauer, S., Wennberg Huldt, C., Kanebratt, K. P., Durieux, I., Gunne, D., Andersson, S., et al. (2017). Functional coupling of human pancreatic islets and liver spheroids on-a-chip: towards a novel human ex vivo type 2 diabetes model. Sci. Rep. 7, 1–11. doi: 10.1038/s41598-017-14815-w Bavli, D., Prill, S., Ezra, E., Levy, G., Cohen, M., Vinken, M., et al. (2016). Realtime monitoring of metabolic function in liver-on-chip microdevices tracks the dynamics of mitochondrial dysfunction. Proc. Natl. Acad. Sci. U.S.A. 113, E2231–E2240. doi: 10.1073/pnas.1522556113 Baxter, M. A., Rowe, C., Alder, J., Harrison, S., Hanley, K. P., Park, B. K., et al. (2010). Generating hepatic cell lineages from pluripotent stem cells for drug toxicity screening. Stem Cell Res. 5, 4–22. doi: 10.1016/j.scr.2010.02.002 Becker, H., Schulz, I., Mosig, A., Jahn, T., and Gärtner, C. (2014). Microfluidic devices for cell culture and handling in organ-on-a-chip applications. Microfluid. BioMEMS Med. Microsyst. 8976:89760N. doi: 10.1117/12.2037237 Beken, S., Kasper, P., and van Der Laan, J. W. (2016). Regulatory acceptance of alternative methods in the development and approval of pharmaceuticals. Adv. Exp. Med. Biol. 856, 33–64. doi: 10.1007/978-3-319-33826-2_3 Bell, C. C., Chouhan, B., Andersson, L. C., Andersson, H., Dear, J. W., Williams, D. P., et al. (2020). Functionality of primary hepatic non-parenchymal cells in a 3D spheroid model and contribution to acetaminophen hepatotoxicity. Arch. Toxicol. 94, 1251–1263. doi: 10.1007/s00204-020-02682-w Bell, C. C., Dankers, A. C. A., Lauschke, V. M., Sison-Young, R., Jenkins, R., Rowe, C., et al. (2018). Comparison of hepatic 2D sandwich cultures and 3d spheroids for long-term toxicity applications: a multicenter study. Toxicol. Sci. 162, 655–666. doi: 10.1093/toxsci/kfx289 Bell, C. C., Hendriks, D. F. G., Moro, S. M. L., Ellis, E., Walsh, J., Renblom, A., et al. (2016). Characterization of primary human hepatocyte spheroids as a model system for drug-induced liver injury, liver function and disease. Sci. Rep. 6:25187. doi: 10.1038/srep25187 Bell, C. C., Lauschke, V. M., Vorrink, S. U., Palmgren, H., Duffin, R., Andersson, T. B., et al. (2017). Transcriptional, functional, and mechanistic comparisons of stem cell-derived hepatocytes, HepaRG cells, and three-dimensional human hepatocyte spheroids as predictive in vitro systems for drug-induced liver injury. Drug Metab. Dispos. 45, 419–429. doi: 10.1124/dmd.116.074369 Berto, A., Van der Poel, W. H. M., Hakze-van der Honing, R., Martelli, F., La Ragione, R. M., Inglese, N., et al. (2013). Replication of hepatitis E virus in threedimensional cell culture. J. Virol. Methods 187, 327–332. doi: 10.1016/j.jviromet. 2012.10.017 Bethesda (2012). LiverTox: Clinical and Research Information on Drug-Induced Liver Injury [Internet]. Bethesda, MD: National Institute of Diabetes and Digestive and Kidney Diseases. Bhatia, S. N., and Ingber, D. E. (2014). Microfluidic organs-on-chips. Nat. Biotechnol. 32, 760–772. doi: 10.1038/nbt.2989 Boelsterli, U. A. (2003). Mechanistic Toxicology: The Molecular Basis of How Chemicals Disrupt Biological Targets. First. Boca Raton, FL: CRC Press. Boess, F., Kamber, M., Romer, S., Gasser, R., Muller, D., Albertini, S., et al. (2003). Gene expression in two hepatic cell lines, cultured primary hepatocytes, and liver slices compared to the in vivo liver gene expression in rats: possible implications for toxicogenomics use of in vitro systems. Toxicol. Sci. 73, 386– 402. doi: 10.1093/toxsci/kfg064 Bonn, B., Svanberg, P., Janefeldt, A., Hultman, I., and Grime, K. (2016). Determination of human hepatocyte intrinsic clearance for slowly metabolized compounds: comparison of a primary hepatocyte/stromal cell co-culture with plated primary hepatocytes and hepaRG. Drug Metab. Dispos. 44, 527–533. doi: 10.1124/dmd.115.067769 Brockmöller, J., and Tzvetkov, M. V. (2008). Pharmacogenetics: data, concepts and tools to improve drug discovery and drug treatment. Eur. J. Clin. Pharmacol. 64, 133–157. doi: 10.1007/s00228-007-0424-z Brolén, G., Sivertsson, L., Björquist, P., Eriksson, G., Ek, M., Semb, H., et al. (2010). Hepatocyte-like cells derived from human embryonic stem cells specifically via definitive endoderm and a progenitor stage. J. Biotechnol. 145, 284–294. doi: 10.1016/j.jbiotec.2009.11.007 Brown, L. A., Arterburn, L. M., Miller, A. P., Cowger, N. L., Hartley, S. M., Andrews, A., et al. (2003). Maintenance of liver functions in rat hepatocytes cultured as spheroids in a rotating wall vessel. Vitr. Cell. Dev. Biol. Anim. 39, 13–20. Bruderer, R., Bernhardt, O. M., Gandhi, T., Miladinoviæ, S. M., Cheng, L. Y., Messner, S., et al. (2015). Extending the limits of quantitative proteome profiling with data-independent acquisition and application to acetaminophen-treated three-dimensional liver microtissues. Mol. Cell. Proteomics 14, 1400–1410. doi: 10.1074/mcp.M114.044305 Busche, M., Tomilova, O., Schütte, J., Werner, S., Beer, M., Groll, N., et al. (2020). HepaChip-MP - a twenty-four chamber microplate for a continuously perfused liver coculture model. Lab Chip 20, 2911–2926. doi: 10.1039/d0lc00357c Cai, J., Zhao, Y., Liu, Y., Ye, F., Song, Z., Qin, H., et al. (2007). Directed differentiation of human embryonic stem cells into functional hepatic cells. Hepatology 45, 1229–1239. doi: 10.1002/hep.21582 Cai, P., Zheng, H., She, J., Feng, N., Zou, H., Gu, J., et al. (2020). Molecular mechanism of aflatoxin-induced hepatocellular carcinoma derived from a bioinformatics analysis. Toxins 12:203. doi: 10.3390/toxins12030203 Camenisch, G., and Umehara, K. (2012). Predicting human hepatic clearance from in vitro drug metabolism and transport data: a scientific and pharmaceutical perspective for assessing drug-drug interactions. Biopharm. Drug Dispos. 33, 179–194. doi: 10.1002/bdd.1784 Campard, D., Lysy, P. A., Najimi, M., and Sokal, E. M. (2008). Native umbilical cord matrix stem cells express hepatic markers and differentiate into hepatocyte-like cells. Gastroenterology 134, 833–848. doi: 10.1053/j.gastro.2007.12.024 Caparrotta, T. M., Antoine, D. J., and Dear, J. W. (2018). Are some people at increased risk of paracetamol-induced liver injury? A critical review of the literature. Eur. J. Clin. Pharmacol. 74, 147–160. doi: 10.1007/s00228-017-23566 Carmo, H., Hengstler, J. G., de Boer, D., Ringel, M., Carvalho, F., Fernandes, E., et al. (2004). Comparative metabolism of the designer drug 4methylthioamphetamine by hepatocytes from man, monkey, dog, rabbit, rat and mouse. Naunyn Schmiedebergs Arch. Pharmacol. 369, 198–205. doi: 10. 1007/s00210-003-0850-0 Cavallari, L. H., Van Driest, S. L., Prows, C. A., Bishop, J. R., Limdi, N. A., Pratt, V. M., et al. (2019). Multi-site investigation of strategies for the clinical implementation of CYP2D6 genotyping to guide drug prescribing. Genet. Med. 21:2255–2263. doi: 10.1038/s41436-019-0484-3 Chan, C., Berthiaume, F., Nath, B. D., Tilles, A. W., Toner, M., and Yarmush, M. L. (2004). Hepatic tissue engineering for adjunct and temporary liver support: critical technologies. Liver Transplant. 10, 1331–1342. doi: 10.1002/lt.20229 Frontiers in Cell and Developmental Biology | www.frontiersin.org 21 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 22 Serras et al. 3D Liver Models in Toxicology Studies Chan, R., and Benet, L. Z. (2017). Evaluation of DILI predictive hypotheses in early drug development. Chem. Res. Toxicol. 30, 1017–1029. doi: 10.1021/acs. chemrestox.7b00025 Chan, T. S., Yu, H., Moore, A., Khetani, S. R., and Tweedie, D. (2013). Meeting the challenge of predicting hepatic clearance of compounds slowly metabolized by cytochrome P450 using a novel hepatocyte model, HepatoPac. Drug Metab. Dispos. 41, 2024–2032. doi: 10.1124/dmd.113.053397 Chang, T. T., and Hughes-Fulford, M. (2009). Monolayer and spheroid culture of human liver hepatocellular carcinoma cell line cells demonstrate distinct global gene expression patterns and functional phenotypes. Tissue Eng. Part A 15, 559–567. doi: 10.1089/ten.tea.2007.0434 Chang, T. T., and Hughes-Fulford, M. (2014). Molecular mechanisms underlying the enhanced functions of three-dimensional hepatocyte aggregates. Biomaterials 35, 2162–2171. doi: 10.1016/j.biomaterials.2013.11.063 Chao, P., Maguire, T., Novik, E., Cheng, K. C., and Yarmush, M. L. (2009). Evaluation of a microfluidic based cell culture platform with primary human hepatocytes for the prediction of hepatic clearance in human. Biochem. Pharmacol. 78, 625–632. doi: 10.1016/j.bcp.2009.05.013 Chatterjee, S., Richert, L., Augustijns, P., and Annaert, P. (2014). Hepatocytebased in vitro model for assessment of drug-induced cholestasis. Toxicol. Appl. Pharmacol. 274, 124–136. doi: 10.1016/j.taap.2013.10.032 Chen, M., Suzuki, A., Thakkar, S., Yu, K., Hu, C., and Tong, W. (2020). U. S. Food and Drug Administration. Drug Induc. Liver Inj. Rank Dataset. Available online at: https://www.fda.gov/science-research/liver-toxicity-knowledge-base-ltkb/ drug-induced-liver-injury-rank-dilirank-dataset (accessed September 25, 2020). Chen, Y., Yu, C., Lv, G., Cao, H., Yang, S., Zhang, Y., et al. (2014). Rapid largescale culturing of microencapsulated hepatocytes: a promising approach for cell-based hepatic support. Transplant. Proc. 46, 1649–1657. doi: 10.1016/j. transproceed.2014.03.002 Choe, A., Ha, S. K., Choi, I., Choi, N., and Sung, J. H. (2017). Microfluidic Gut-liver chip for reproducing the first pass metabolism. Biomed. Microdevices 19, 1–11. doi: 10.1007/s10544-016-0143-2 Chojkier, M. (2005). Troglitazone and liver injury: in search of answers. Hepatology 41, 237–246. doi: 10.1002/hep.20567 Christoffersson, J., Aronsson, C., Jury, M., and Selegård, R. (2019). Fabrication of modular hyaluronan-PEG hydrogels to support 3D cultures of hepatocytes in a perfused liver-on-a-chip device. Biofabrication 11:015013. doi: 10.1088/17585090/aaf657 Cipriano, M., Correia, J. C., Camões, S. P., Oliveira, N. G., Cruz, P., Cruz, H., et al. (2017a). The role of epigenetic modifiers in extended cultures of functional hepatocyte-like cells derived from human neonatal mesenchymal stem cells. Arch. Toxicol. 91, 2469–2489. doi: 10.1007/s00204-016-1901-x Cipriano, M., Freyer, N., Knöspel, F., Oliveira, N. G., Barcia, R., Cruz, P. E., et al. (2017b). Self-assembled 3D spheroids and hollow-fibre bioreactors improve MSC-derived hepatocyte-like cell maturation in vitro. Arch. Toxicol. 91, 1815– 1832. doi: 10.1007/s00204-016-1838-0 Cipriano, M., Pinheiro, P. F., Sequeira, C. O., Rodrigues, J. S., Oliveira, N. G., Antunes, A. M. M., et al. (2020). Nevirapine biotransformation insights: an integrated in vitro approach unveils the biocompetence and glutathiolomic profile of a human hepatocyte-like cell 3D model. Int. J. Mol. Sci. 21, 1–18. doi: 10.3390/ijms21113998 Coecke, S., Bogni, A., Langezaal, I., Worth, A., Hartung, T., and Monshouwer, M. (2001). The use of genetically engineered cells for assessing CYP2D6-related polymorphic effects. Toxicol. Vitr. 15, 553–556. doi: 10.1016/S0887-2333(01) 00061-3 Coll, M., Perea, L., Boon, R., Leite, S. B., Vallverdú, J., Mannaerts, I., et al. (2018). Generation of hepatic stellate cells from human pluripotent stem cells enables in vitro modeling of liver fibrosis. Cell Stem Cell 23, 101.e7–113.e7. doi: 10.1016/ j.stem.2018.05.027 Conway, D. E., Sakurai, Y., Weiss, D., Vega, J. D., Taylor, W. R., Jo, H., et al. (2009). Expression of CYP1A1 and CYP1B1 in human endothelial cells: regulation by fluid shear stress. Cardiovasc. Res. 81, 669–677. doi: 10.1093/cvr/cvn360 Cook, D., Brown, D., Alexander, R., March, R., Morgan, P., Satterthwaite, G., et al. (2014). Lessons learned from the fate of AstraZeneca’s drug pipeline: a five-dimensional framework. Nat. Rev. Drug Discov. 13, 419–431. doi: 10.1038/ nrd4309 Daly, A. K., Aithal, G. P., Leathart, J. B. S., Swainsbury, R. A., Dang, T. S., and Day, C. P. (2007). Genetic susceptibility to diclofenac-induced hepatotoxicity: contribution of UGT2B7, CYP2C8, and ABCC2 genotypes. Gastroenterology 132, 272–281. doi: 10.1053/j.gastro.2006.11.023 Danoy, M., Bernier, M. L., Kimura, K., Poulain, S., Kato, S., Mori, D., et al. (2019). Optimized protocol for the hepatic differentiation of induced pluripotent stem cells in a fluidic microenvironment. Biotechnol. Bioeng. 116, 1762–1776. doi: 10.1002/bit.26970 Danoy, M., Poulain, S., Lereau-Bernier, M., Kato, S., Scheidecker, B., Kido, T., et al. (2020). Characterization of liver zonation-like transcriptomic patterns in HLCs derived from hiPSCs in a microfluidic biochip environment. Biotechnol. Prog. 36:e3013. doi: 10.1002/btpr.3013 Darnell, M., Schreiter, T., Zeilinger, K., Urbaniak, T., Söderdahl, T., Rossberg, I., et al. (2011). Cytochrome P450-dependent metabolism in HepaRG cells cultured in a dynamic three-dimensional bioreactor. Drug Metab. Dispos. 39, 1131–1138. doi: 10.1124/dmd.110.037721 Darnell, M., Ulvestad, M., Ellis, E., Weidolf, L., and Andersson, T. B. (2012). In vitro evaluation of major in vivo drug metabolic pathways using primary human hepatocytes and HepaRG cells in suspension and a dynamic three-dimensional bioreactor system. J. Pharmacol. Exp. Ther. 343, 134–144. doi: 10.1124/jpet.112. 195834 Dash, A., Inman, W., Hoffmaster, K., Sevidal, S., Kelly, J., Obach, R. S., et al. (2009). Liver tissue engineering in the evaluation of drug safety. Expert Opin. Drug Metab. Toxicol. 5, 1159–1174. doi: 10.1517/17425250903160664 Daston, G., Knight, D. J., Schwarz, M., Gocht, T., Thomas, R. S., Mahony, C., et al. (2015). SEURAT: Safety Evaluation Ultimately Replacing Animal Testing — recommendations for future research in the field of predictive toxicology. Arch. Toxicol. 89, 15–23. doi: 10.1007/s00204-014-1421-5 Davila, J. C., Cezar, G. G., Thiede, M., Strom, S., Miki, T., and Trosko, J. (2004). Use and application of stem cells in toxicology. Toxicol. Sci. 79, 214–223. doi: 10.1093/toxsci/kfh100 Davis, M., and Stamper, B. D. (2016). TAMH: a useful in vitro model for assessing hepatotoxic mechanisms. Biomed Res. Int. 2016:4780872. doi: 10.1155/2016/ 4780872 Deharde, D., Schneider, C., Hiller, T., Fischer, N., Kegel, V., Lübberstedt, M., et al. (2016). Bile canaliculi formation and biliary transport in 3D sandwichcultured hepatocytes in dependence of the extracellular matrix composition. Arch. Toxicol. 90, 2497–2511. doi: 10.1007/s00204-016-1758-z Deng, J., Zhang, X., Chen, Z., Luo, Y., Lu, Y., Liu, T., et al. (2019). A cell lines derived microfluidic liver model for investigation of hepatotoxicity induced by drug-drug interaction. Biomicrofluidics 13:024101. doi: 10.1063/1.5070088 Devarbhavi, H. (2012). An Update on Drug-induced Liver Injury. J. Clin. Exp. Hepatol. 2, 247–259. doi: 10.1016/j.jceh.2012.05.002 Di Stefano, A., Sozio, P., Serafina Cerasa, L., and Iannitelli, A. (2011). L-dopa prodrugs: an overview of trends for improving parkinsons disease treatment. Curr. Pharm. Des. 17, 3482–3493. doi: 10.2174/138161211798194495 Domansky, K., Inman, W., Serdy, J., Dash, A., Lim, M. H. M., and Griffith, L. G. (2010). Perfused multiwell plate for 3D liver tissue engineering. Lab Chip 10, 51–58. doi: 10.1039/B913221J Domansky, K., Inman, W., Serdy, J., and Griffith, L. G. (2005). Perfused microreactors for liver tissue engineering. Annu. Int. Conf. IEEE Eng. Med. Biol. Proc. 7, 7490–7492. doi: 10.1109/iembs.2005.1616244 Dong, X. J., Zhang, G. R., Zhou, Q. J., Pan, R. L., Chen, Y., Xiang, L. X., et al. (2009). Direct hepatic differentiation of mouse embryonic stem cells induced by valproic acid and cytokines. World J. Gastroenterol. 15, 5165–5175. doi: 10.3748/wjg.15.5165 Dowden, H., and Munro, J. (2019). Trends in clinical success rates and therapeutic focus. Nat. Rev. Drug Discov. 18, 495–496. doi: 10.1038/d41573-019-00074-z Dumas, E. O., and Pollack, G. M. (2008). Opioid tolerance development: a pharmacokinetic/pharmacodynamic perspective. AAPS J. 10, 537–551. doi: 10. 1208/s12248-008-9056-1 Edington, C. D., Chen, W. L. K., Geishecker, E., Kassis, T., Soenksen, L. R., Bhushan, B. M., et al. (2018). Interconnected microphysiological systems for quantitative biology and pharmacology studies. Sci. Rep. 8:4530. doi: 10.1038/ s41598-018-22749-0 EMEA (2000). Public Statement on Viramune (Nevirapine) - Severe and Life-Threatening Cutaneous and Hepatic Reactions. Available online at: Frontiers in Cell and Developmental Biology | www.frontiersin.org 22 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 23 Serras et al. 3D Liver Models in Toxicology Studies https://www.ema.europa.eu/en/news/public-statement-viramune-nevirapinesevere-life-threatening-cutaneous-hepatic-reactions (accessed May 30, 2020). Engler, A. J., Sen, S., Sweeney, H. L., and Discher, D. E. (2006). Matrix elasticity directs stem cell lineage specification. Cell 126, 677–689. doi: 10.1016/j.cell. 2006.06.044 Farzaneh, Z., Abbasalizadeh, S., Asghari-Vostikolaee, M. H., Alikhani, M., Cabral, J. M. S., and Baharvand, H. (2020). Dissolved oxygen concentration regulates human hepatic organoid formation from pluripotent stem cells in a fully controlled bioreactor. Biotechnol. Bioeng. 117, 3739–3756. doi: 10.1002/bit. 27521 Fey, S. J., Korzeniowska, B., and Wrzesinski, K. (2020). Response to and recovery from treatment in human liver-mimetic clinostat spheroids: a model for assessing repeated-dose drug toxicity. Toxicol. Res. 9, 379–389. doi: 10.1093/ toxres/tfaa033 Fey, S. J., and Wrzesinski, K. (2012). Determination of drug toxicity using 3D spheroids constructed from an immortal human hepatocyte cell line. Toxicol. Sci. 127, 403–411. doi: 10.1093/toxsci/kfs122 Foster, A. J., Chouhan, B., Regan, S. L., Rollison, H., Amberntsson, S., Andersson, L. C., et al. (2019). Integrated in vitro models for hepatic safety and metabolism: evaluation of a human Liver-Chip and liver spheroid. Arch. Toxicol. 93, 1021– 1037. doi: 10.1007/s00204-019-02427-4 Frederick, D. M., Jacinto, E. Y., Patel, N. N., Rushmore, T. H., Tchao, R., and Harvison, P. J. (2011). Cytotoxicity of 3-(3,5-dichlorophenyl)-2,4thiazolidinedione (DCPT) and analogues in wild type and CYP3A4 stably transfected HepG2 cells. Toxicol. Vitr. 25, 2113–2119. doi: 10.1016/j.tiv.2011. 09.015 Freyer, N., Knöspel, F., Strahl, N., Amini, L., Schrade, P., Bachmann, S., et al. (2016). Hepatic differentiation of human induced pluripotent stem cells in a perfused three-dimensional multicompartment bioreactor. Biores. Open Access 5, 235–248. doi: 10.1089/biores.2016.0027 Fu, S., Wu, D., Jiang, W., Li, J., Long, J., Jia, C., et al. (2020). Molecular biomarkers in drug-induced liver injury: challenges and future perspectives. Front. Pharmacol. 10:1667. doi: 10.3389/fphar.2019.01667 Fu, Y., Deng, J., Jiang, Q., Wang, Y., Zhang, Y., Yao, Y., et al. (2016). Rapid generation of functional hepatocyte-like cells from human adipose-derived stem cells. Stem Cell Res. Ther. 7:105. doi: 10.1186/s13287-016-0364-6 Fung, M., Thornton, A., Mybeck, K., Wu, J. H. H., Hornbuckle, K., and Muniz, E. (2001). Evaluation of the characteristics of safety withdrawal of prescription drugs from worldwide pharmaceutical markets-1960 to 1999. Ther. Innov. Regul. Sci. 35, 293–317. doi: 10.1177/009286150103500134 Funk, C., Ponelle, C., Scheuermann, G., and Pantze, M. (2001). Cholestatic potential of troglitazone as a possible factor contributing to troglitazoneinduced hepatotoxicity: in vivo and in vitro interaction at the canalicular bile salt export pump (Bsep) in the rat. Mol. Pharmacol. 59, 627–635. doi: 10.1124/ MOL.59.3.627 Gao, B., Yang, Q., Zhao, X., Jin, G., Ma, Y., and Xu, F. (2016). 4D bioprinting for biomedical applications. Trends Biotechnol. 34, 746–756. doi: 10.1016/j.tibtech. 2016.03.004 Gao, X., and Liu, Y. (2017). A transcriptomic study suggesting human iPSC-derived hepatocytes potentially offer a better in vitro model of hepatotoxicity than most hepatoma cell lines. Cell Biol. Toxicol. 33, 407–421. doi: 10.1007/s10565-0179383-z Gaskell, H., Sharma, P., Colley, H. E., Murdoch, C., Williams, D. P., and Webb, S. D. (2016). Characterization of a functional C3A liver spheroid model. Toxicol. Res. 5, 1053–1065. doi: 10.1039/c6tx00101g Gerets, H. H. J., Tilmant, K., Gerin, B., Chanteux, H., Depelchin, B. O., Dhalluin, S., et al. (2012). Characterization of primary human hepatocytes, HepG2 cells, and HepaRG cells at the mRNA level and CYP activity in response to inducers and their predictivity for the detection of human hepatotoxins. Cell Biol. Toxicol. 28, 69–87. doi: 10.1007/s10565-011-9208-4 Gerlach, J. C., Encke, J., Hole, O., Müller, C., Ryan, C. J., and Neuhaus, P. (1994). Bioreactor for a larger scale hepatocyte in vitro perfusion. Transplantation 58, 984–988. doi: 10.1097/00007890-199411150-00002 Gerlach, J. C., Mutig, K., Sauer, I. M., Schrade, P., Efimova, E., Mieder, T., et al. (2003). Use of primary human liver cells originating from discarded grafts in a bioreactor for liver support therapy and the prospects of culturing adult liver stem cells in bioreactors: a morphologic study. Transplantation 76, 781–786. doi: 10.1097/01.TP.0000083319.36931.32 Ghabril, M., Fontana, R., Rockey, D., Jiezhun, G., and Chalasani, N. (2013). Drug-induced liver injury caused by intravenously administered medications: the drug-induced liver injury network experience. J. Clin. Gastroenterol. 47, 553–558. doi: 10.1097/MCG.0b013e31827 6bf00 Gieseck, R. L. III, Hannan, N. R. F., Bort, R., Hanley, N. A., Drake, R. A. L., Cameron, G. W. W., et al. (2014). Maturation of induced pluripotent stem cell derived hepatocytes by 3D-culture. PLoS One 9:e86372. doi: 10.1371/journal. pone.0086372 Gijbels, E., and Vinken, M. (2017). An update on adverse outcome pathways leading to liver injury. Appl. Vitr. Toxicol. 3, 283–285. doi: 10.1089/aivt.2017. 0027 Giri, S., Nieber, K., and Bader, A. (2010). Hepatotoxicity and hepatic metabolism of available drugs: current problems and possible solutions in preclinical stages. Expert Opin. Drug Metab. Toxicol. 6, 895–917. doi: 10.1517/ 17425251003792521 Glicklis, R., Merchuk, J. C., and Cohen, S. (2004). Modeling mass transfer in hepatocyte spheroids via cell viability, spheroid size, and hepatocellular functions. Biotechnol. Bioeng. 86, 672–680. doi: 10.1002/bit.20086 Goda, K., Takahashi, T., Kobayashi, A., Shoda, T., Kuno, H., and Sugai, S. (2016). Usefulness of in vitro combination assays of mitochondrial dysfunction and apoptosis for the estimation of potential risk of idiosyncratic drug induced liver injury. J. Toxicol. Sci. 41, 605–615. doi: 10.2131/jts.41.605 Godoy, P., Hewitt, N. J., Albrecht, U., Andersen, M. E., Ansari, N., Bhattacharya, S., et al. (2013). Recent advances in 2D and 3D in vitro systems using primary hepatocytes, alternative hepatocyte sources and non-parenchymal liver cells and their use in investigating mechanisms of hepatotoxicity, cell signaling and ADME. Arch. Toxicol. 87, 1315–1530. doi: 10.1007/s00204-013-1078-5 Gomez-Lechon, M., Donato, M., Lahoz, A., and Castell, J. (2008). Cell lines: a tool for in vitro drug metabolism studies. Curr. Drug Metab. 9, 1–11. doi: 10.2174/138920008783331086 Gomez-Lechon, M. J., Lahoz, A., Gombau, L., Castell, J. V., and Donato, M. T. (2010). In vitro evaluation of potential hepatotoxicity induced by drugs. Curr. Pharm. Des. 16, 1963–1977. doi: 10.2174/138161210791208910 Goulart, E., de Caires-Junior, L. C., Telles-Silva, K. A., Araujo, B. H. S., Rocco, S. A., et al. (2019). 3D bioprinting of liver spheroids derived from human induced pluripotent stem cells sustain liver function and viability in vitro. Biofabrication 12:015010. doi: 10.1088/1758-5090/ab4a30 Granitzny, A., Knebel, J., Müller, M., Braun, A., Steinberg, P., Dasenbrock, C., et al. (2017). Evaluation of a human in vitro hepatocyte-NPC co-culture model for the prediction of idiosyncratic drug-induced liver injury: a pilot study. Toxicol. Rep. 4, 89–103. doi: 10.1016/j.toxrep.2017.02.001 Grattagliano, I., Bonfrate, L., Diogo, C. V., Wang, H. H., Wang, D. Q. H., and Portincasa, P. (2009). Biochemical mechanisms in drug-induced liver injury: certainties and doubts. World J. Gastroenterol. 15, 4865–4876. doi: 10.3748/wjg. 15.4865 Gregus, Z. (2013). “Chapter 3: mechanisms of toxicity,” in Casarett and Doull’s Toxicology: Basic Science of Poisons -, 8th Edn, ed. C. D. Klaassen (San Francisco: McGraw-Hill, Medical Publishing Division), 45–107. Grilo, N. M., João Correia, M., Miranda, J. P., Cipriano, M., Serpa, J., Matilde Marques, M., et al. (2017). Unmasking efavirenz neurotoxicity: time matters to the underlying mechanisms. Eur. J. Pharm. Sci. 105, 47–54. doi: 10.1016/j.ejps. 2017.05.010 Grix, T., Ruppelt, A., Thomas, A., Amler, A., Noichl, B., Lauster, R., et al. (2018). Bioprinting Perfusion-Enabled Liver Equivalents for Advanced Organ-on-aChip Applications. Genes 9:176. doi: 10.3390/genes9040176 Guillouzo, A., Corlu, A., Aninat, C., Glaise, D., Morel, F., and Guguen-Guillouzo, C. (2007). The human hepatoma HepaRG cells: a highly differentiated model for studies of liver metabolism and toxicity of xenobiotics. Chem. Biol. Interact. 168, 66–73. doi: 10.1016/j.cbi.2006.12.003 Gunness, P., Mueller, D., Shevchenko, V., Heinzle, E., Ingelman-Sundberg, M., and Noor, F. (2013). 3D organotypic cultures of human heparg cells: a tool for in vitro toxicity studies. Toxicol. Sci. 133, 67–78. doi: 10.1093/toxsci/ kft021 Hafiz, E. O. A., Bulutoglu, B., Mansy, S. S., Chen, Y., Abu-Taleb, H., Soliman, S. A. M., et al. (2020). Development of liver microtissues with functional biliary ductular network. Biotechnol. Bioeng. 118, 17–29. doi: 10. 1002/bit.27546 Frontiers in Cell and Developmental Biology | www.frontiersin.org 23 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 24 Serras et al. 3D Liver Models in Toxicology Studies Hammond, T., Allen, P., and Birdsall, H. (2016). Is there a space-based technology solution to problems with preclinical drug toxicity testing? Pharm. Res. 33, 1545–1551. doi: 10.1007/s11095-016-1942-0 Haque, A., Gheibi, P., Gao, Y., Foster, E., Son, K. J., You, J., et al. (2016). Cell biology is different in small volumes: endogenous signals shape phenotype of primary hepatocytes cultured in microfluidic channels. Sci. Rep. 6, 1–15. doi: 10.1038/srep33980 Haschek, W. M., Rousseaux, C. G., and Wallig, M. A. (2009). Fundamentals of toxicologic pathology: second edition. Fundam. Toxicol. Pathol. 2009, 1–691. doi: 10.1016/C2009-0-02051-0 Hass, R., Kasper, C., Böhm, S., and Jacobs, R. (2011). Different populations and sources of human mesenchymal stem cells (MSC): a comparison of adult and neonatal tissue-derived MSC. Cell Commun. Signal. 9:12. doi: 10.1186/1478811X-9-12 Hay, D. C., Fletcher, J., Payne, C., Terrace, J. D., Gallagher, R. C. J., Snoeys, J., et al. (2008a). Highly efficient differentiation of hESCs to functional hepatic endoderm requires ActivinA and Wnt3a signaling. Proc. Natl. Acad. Sci. U.S.A. 105, 12301–12306. doi: 10.1073/pnas.0806522105 Hay, D. C., Zhao, D., Fletcher, J., Hewitt, Z. A., McLean, D., UrruticoecheaUriguen, A., et al. (2008b). Efficient differentiation of hepatocytes from human embryonic stem cells exhibiting markers recapitulating liver development in vivo. Stem Cells 26, 894–902. doi: 10.1634/stemcells.2007-0718 Haycock, J. W. (2011). 3D Cell Culture: A Review of Current Approaches and Techniques. Totowa, NJ: Humana Press. Hendriks, D. F. G., Puigvert, L. F., Messner, S., Mortiz, W., and IngelmanSundberg, M. (2016). Hepatic 3D spheroid models for the detection and study of compounds with cholestatic liability. Sci. Rep. 6:35434. doi: 10.1038/srep35434 Hinson, J. A., Roberts, D. W., and James, L. P. (2010). Mechanisms of acetaminophen-induced liver necrosis. Handb. Exp. Pharmacol. 196, 396–405. doi: 10.1007/978-3-642-00663-0_12 Hoffmann, S. A., Müller-Vieira, U., Biemel, K., Knobeloch, D., Heydel, S., Lübberstedt, M., et al. (2012). Analysis of drug metabolism activities in a miniaturized liver cell bioreactor for use in pharmacological studies. Biotechnol. Bioeng. 109, 3172–3181. doi: 10.1002/bit.24573 Holmgren, G., Sjögren, A. K., Barragan, I., Sabirsh, A., Sartipy, P., Synnergren, J., et al. (2014). Long-term chronic toxicity testing using human pluripotent stem cell-derived hepatocytes. Drug Metab. Dispos. 42, 1401–1406. doi: 10.1124/dmd. 114.059154 Horvath, P., Aulner, N., Bickle, M., Davies, A. M., Del Nery, E., Ebner, D., et al. (2016). Screening out irrelevant cell-based models of disease. Nat. Rev. Drug Discov. 15, 751–769. doi: 10.1038/nrd.2016.175 Huch, M., Dorrell, C., Boj, S. F., Van Es, J. H., Li, V. S. W., Van De Wetering, M., et al. (2013). In vitro expansion of single Lgr5 + liver stem cells induced by Wnt-driven regeneration. Nature 494, 247–250. doi: 10.1038/nature11826 Huch, M., Gehart, H., Van Boxtel, R., Hamer, K., Blokzijl, F., Verstegen, M. M. A., et al. (2015). Long-term culture of genome-stable bipotent stem cells from adult human liver. Cell 160, 299–312. doi: 10.1016/j.cell.2014.11.050 Hultman, I., Vedin, C., Abrahamsson, A., Winiwarter, S., and Darnell, M. (2016). Use of HµREL human coculture system for prediction of intrinsic clearance and metabolite formation for slowly metabolized compounds. Mol. Pharm. 13, 2796–2807. doi: 10.1021/acs.molpharmaceut.6b00396 Ingber, D. E. (2020). Is it time for reviewer 3 to request human organ chip experiments instead of animal validation studies? Adv. Sci. 2020:2002030. doi: 10.1002/advs.202002030 Jackson, J. P., Freeman, K. M., St. Claire, R. L., Black, C. B., and Brouwer, K. R. (2018). Cholestatic drug induced liver injury: a function of bile salt export pump inhibition and farnesoid X receptor antagonism. Appl. Vitr. Toxicol. 4, 265–279. doi: 10.1089/aivt.2018.0011 Jaeschke, H. (2013). “Chapter 13: toxic responses of the liver,” in Casarett and Doull’s Toxicology: Basic Science of Poisons -, 8th Edn, ed. C. D. Klaassen (San Francisco: McGraw-Hill, Medical Publishing Division), 557–583. Jang, M., Kleber, A., Ruckelshausen, T., Betzholz, R., and Manz, A. (2019). Differentiation of the human liver progenitor cell line (HepaRG) on a microfluidic-based biochip. J. Tissue Eng. Regen. Med. 13, 482–494. doi: 10. 1002/term.2802 Jang, M., Neuzil, P., Volk, T., Manz, A., and Kleber, A. (2015). On-chip threedimensional cell culture in phaseguides improves hepatocyte functions in vitro. Biomicrofluidics 9:034113. doi: 10.1063/1.4922863 Jemnitz, K., Veres, Z., Monostory, K., Kóbori, L., and Vereczkey, L. (2008). Interspecies differences in acetaminophen sensitivity of human, rat, and mouse primary hepatocytes. Toxicol. Vitr. 22, 961–967. doi: 10.1016/j.tiv.2008.02.001 Jeon, J. W., Choi, N., Lee, S. H., and Sung, J. H. (2020). Three-tissue microphysiological system for studying inflammatory responses in gut-liver Axis. Biomed. Microdevices 22:65. doi: 10.1007/s10544-020-00 519-y Jiang, J., Pieterman, C. D., Ertaylan, G., Peeters, R. L. M., and de Kok, T. M. C. M. (2019). The Application of Omics-Based Human Liver Platforms for Investigating the Mechanism of Drug-Induced Hepatotoxicity in Vitro. Berlin: Springer. Jones, S. C., Kortepeter, C., and Brinker, A. D. (2018). “Postmarketing surveillance of drug-induced liver injury,” in Methods in Pharmacology and Toxicology, ed. K. Y. James (New York, NY: Springer), 459–474. Kamihira, M., Yamada, K., Hamamoto, R., and Iijima, S. (1997). Spheroid formation of hepatocytes using synthetic polymer. Ann. N. Y. Acad. Sci. 831, 398–407. doi: 10.1111/j.1749-6632.1997.tb52213.x Kanamori, Y., Fujita, K. I, Nakayama, K., Kamataki, T., and Kawai, H. (2003). Large-scale production of genetically engineered CYP3A4 in E. coli: application of a Jarfermenter. Drug Metab. Pharmacokinet. 18, 42–47. doi: 10.2133/dmpk. 18.42 Kang, Y. B. A., Sodunke, T. R., Lamontagne, J., Cirillo, J., Rajiv, C., Bouchard, M. J., et al. (2015). Liver sinusoid on a chip: long-term layered co-culture of primary rat hepatocytes and endothelial cells in microfluidic platforms. Biotechnol. Bioeng. 112, 2571–2582. doi: 10.1002/bit. 25659 Kappelhoff, B. S., Van Leth, F., MacGregor, T. R., Lange, J. M. A., Beijnen, J. H., and Huitema, A. D. R. (2005). Nevirapine and efavirenz pharmacokinetics and covariate analysis in the 2NN study. Antivir. Ther. 10, 145–155. Kassahun, K., Pearson, P. G., Tang, W., McIntosh, I., Leung, K., Elmore, C., et al. (2001). Studies on the metabolism of troglitazone to reactive intermediates in vitro and in vivo. Evidence for novel biotransformation pathways involving quinone methide formation and thiazolidinedione ring scission. Chem. Res. Toxicol. 14, 62–70. doi: 10.1021/tx000180q Kazemnejad, S., Allameh, A., Seoleimani, M., Gharehbaghian, A., Mohammadi, Y., Amirizadeh, N., et al. (2008). Functional hepatocyte-like cells derived from human bone marrow mesenchymal stem cells on a novel 3-dimensional biocompatible nanofibrous scaffold. Int. J. Artif. Organs 31, 500–507. doi: 10. 1177/039139880803100605 Kenna, J. G., and Uetrecht, J. (2018). Do In Vitro assays predict drug candidate idiosyncratic drug-induced liver injury risk? Drug Metab. Dispos. 46, 1658– 1669. doi: 10.1124/dmd.118.082719 Khetani, S. R., and Bhatia, S. N. (2008). Microscale culture of human liver cells for drug development. Nat. Biotechnol. 26, 120–126. doi: 10.1038/nbt1361 Khetani, S. R., Kanchagar, C., Ukairo, O., Krzyzewski, S., Moore, A., Shi, J., et al. (2013). Use of micropatterned cocultures to detect compounds that cause druginduced liver injury in humans. Toxicol. Sci. 132, 107–117. doi: 10.1093/toxsci/ kfs326 Kholodenko, I. V., Kurbatov, L. K., Kholodenko, R. V., Manukyan, G. V., and Yarygin, K. N. (2019). Mesenchymal stem cells in the adult human liver: hype or hope? Cells. 8:1127. doi: 10.3390/cells8101127 Kim, D. E., Jang, M.-J., Kim, Y. R., Lee, J.-Y., Cho, E. B., Kim, E., et al. (2017). Prediction of drug-induced immune-mediated hepatotoxicity using hepatocyte-like cells derived from human embryonic stem cells. Toxicology 387, 1–9. doi: 10.1016/j.tox.2017.06.005 Kizawa, H., Nagao, E., Shimamura, M., Zhang, G., and Torii, H. (2017). Scaffoldfree 3D bio-printed human liver tissue stably maintains metabolic functions useful for drug discovery. Biochem. Biophys. Rep. 10, 186–191. doi: 10.1016/j. bbrep.2017.04.004 Kleiner, D. E. (2017). The histopathological evaluation of drug-induced liver injury. Histopathology 70, 81–93. doi: 10.1111/his.13082 Knospel, F., Jacobs, F., Freyer, N., Damm, G., De Bondt, A., van den Wyngaert, I., et al. (2016). In vitro model for hepatotoxicity studies based on primary human hepatocyte cultivation in a perfused 3D bioreactor system. Int. J. Mol. Sci. 17:584. doi: 10.3390/ijms17040584 Kola, I., and Landis, J. (2004). Can the pharmaceutical industry reduce attrition rates? Nat. Rev. Drug Discov. 3, 711–716. doi: 10.1038/nrd1470 Kostadinova, R., Boess, F., Applegate, D., Suter, L., Weiser, T., Singer, T., et al. (2013). A long-term three dimensional liver co-culture system for improved Frontiers in Cell and Developmental Biology | www.frontiersin.org 24 February 2021 | Volume 9 | Article 626805 fcell-09-626805 February 25, 2021 Time: 14:28 # 25 Serras et al. 3D Liver Models in Toxicology Studies prediction of clinically relevant drug-induced hepatotoxicity. Toxicol. Appl. Pharmacol. 268, 1–16. doi: 10.1016/j.taap.2013.01.012 Koui, Y., Kido, T., Ito, T., Oyama, H., Chen, S. W., Katou, Y., et al. (2017). An in vitro human liver model by iPSC-derived parenchymal and nonparenchymal cells. Stem Cell Rep. 9, 490–498. doi: 10.1016/j.stemcr.2017.06. 010 Koyama, S., Arakawa, H., Itoh, M., Masuda, N., Yano, K., Kojima, H., et al. (2018). Evaluation of the metabolic capability of primary human hepatocytes in threedimensional cultures on microstructural plates. Biopharm. Drug Dispos. 39, 187–195. doi: 10.1002/bdd.2125 Kranendonk, M., Alves, M., Antunes, P., and Rueff, J. (2014). Human sulfotransferase 1A1-dependent mutagenicity of 12-hydroxy-nevirapine: the missing link? Chem. Res. Toxicol. 27, 1967–1971. doi: 10.1021/tx5003113 Krasniqi, V., Dimovski, A., Domjanovic, I. K., Bilic, I., and Bozina, N. (2016). How polymorphisms of the cytochrome P450 genes affect ibuprofen and diclofenac metabolism and toxicity. Arh. Hig. Rada Toksikol. 67, 1–8. doi: 10.1515/aiht2016-67-2754 Kratochwil, N. A., Triyatni, M., Mueller, M. B., Klammers, F., Leonard, B., Turley, D., et al. (2018). Simultaneous assessment of clearance, metabolism, induction, and drug-drug interaction potential using a long-term in vitro liver model for a novel hepatitis b virus inhibitor. J. Pharmacol. Exp. Ther. 365, 237–248. doi: 10.1124/jpet.117.245712 Krewski, D., Andersen, M. E., Mantus, E., and Zeise, L. (2009). Toxicity testing in the 21st century: implications for human health risk assessment: perspective. Risk Anal. 29, 474–479. doi: 10.1111/j.1539-6924.2008.01150.x Kuˇ cera, O., Endlicher, R., Rychtrmoc, D., Lotková, H., Sobotka, O., and ˇ Cervinková, Z. (2017). Acetaminophen toxicity in rat and mouse hepatocytes in vitro. Drug Chem. Toxicol. 40, 448–456. doi: 10.1080/01480545.2016. 1255953 Kuna, L., Bozic, I., Kizivat, T., Bojanic, K., Mrso, M., Kralj, E., et al. (2018). Models of drug induced liver injury (DILI) – current issues and future perspectives. Curr. Drug Metab. 19, 830–838. doi: 10.2174/1389200219666180523095355 Kuntz, E., and Kuntz, H.-D. (2008). Hepatology Textbook and Atlas, 3rd Edn. Berlin: Springer. Langhans, S. A. (2018). Three-dimensional in vitro cell culture models in drug discovery and drug repositioning. Front. Pharmacol. 9:6. doi: 10.3389/fphar. 2018.00006 Larson, A. M. (2010). Diagnosis and management of acute liver failure. Curr. Opin. Gastroenterol. 26, 214–221. doi: 10.1097/MOG.0b013e32833847c5 Lauer, B., Tuschl, G., Kling, M., and Mueller, S. O. (2009). Species-specific toxicity of diclofenac and troglitazone in primary human and rat hepatocytes. Chem. Biol. Interact. 179, 17–24. doi: 10.1016/j.cbi.2008.10.031 Lauschke, V. M., Hendriks, D. F. G., Bell, C. C., Andersson, T. B., and Ingelman-Sundberg, M. (2016). Novel 3D culture systems for studies of human liver function and assessments of the hepatotoxicity of drugs and drug candidates. Chem. Res. Toxicol. 29, 1936–1955. doi: 10.1021/acs.chemrestox. 6b00150 Lee, H. J., Jung, J., Cho, K. J., Lee, C. K., Hwang, S. G., and Kim, G. J. (2012). Comparison of in vitro hepatogenic differentiation potential between various placenta-derived stem cells and other adult stem cells as an alternative source of functional hepatocytes. Differentiation 84, 223–231. doi: 10.1016/j.diff.2012.05. 007 Lee, P. J., Hung, P. J., and Lee, L. P. (2007). An artificial liver sinusoid with a microfluidic endothelial-like barrier for primary hepatocyte culture. Biotechnol. Bioeng. 97, 1340–1346. doi: 10.1002/bit.21360 Leite, S. B., Roosens, T., El Taghdouini, A., Mannaerts, I., Smout, A. J., Najimi, M., et al. (2016). Novel human hepatic organoid model enables testing of druginduced liver fibrosis in vitro. Biomaterials 78, 1–10. doi: 10.1016/j.biomaterials. 2015.11.026 Leite, S. B., Teixeira, A. P., Miranda, J. P., Tostões, R. M., Clemente, J. J., Sousa, M. F., et al. (2011). Merging bioreactor technology with 3D hepatocytefibroblast culturing approaches: improved in vitro models for toxicological applications. Toxicol. Vitr. 25, 825–832. doi: 10.1016/j.tiv.2011.02.002 Leite, S. B., Wilk-Zasadna, I., Zaldivar, J. M., Airola, E., Reis-Fernandes, M. A., Mennecozzi, M., et al. (2012). Three-dimensional HepaRG model as an attractive tool for toxicity testing. Toxicol. Sci. 130, 106–116. doi: 10.1093/ toxsci/kfs232 Lerche-Langrand, C., and Toutain, H. J. (2000). Precision-cut liver slices: characteristics and use for in vitro pharmaco-toxicology. Toxicology 153, 221– 253. doi: 10.1016/S0300-483X(00)00316-4 Lewerenz, V., Hanelt, S., Nastevska, C., El-Bahay, C., Röhrdanz, E., and Kahl, R. (2003). Antioxidants protect primary rat hepatocyte cultures against acetaminophen-induced DNA strand breaks but not against acetaminopheninduced cytotoxicity. Toxicology 191, 179–187. doi: 10.1016/S0300-483X(03) 00256-7 Li, F., Cao, L., Parikh, S., and Zuo, R. (2020). Three-Dimensional Spheroids With Primary Human Liver Cells and Differential Roles of Kupffer Cells in DrugInduced Liver Injury. J. Pharm. Sci. 109, 1912–1923. doi: 10.1016/j.xphs.2020. 02.021 Li, M., Yuan, H., Li, N., Song, G., Zheng, Y., Baratta, M., et al. (2008). Identification of interspecies difference in efflux transporters of hepatocytes from dog, rat, monkey and human. Eur. J. Pharm. Sci. 35, 114–126. doi: 10.1016/j.ejps.2008. 06.008 Lin, J., Schyschka, L., Mühl-Benninghaus, R., Neumann, J., Hao, L., Nussler, N., et al. (2012). Comparative analysis of phase I and II enzyme activities in 5 hepatic cell lines identifies Huh-7 and HCC-T cells with the highest potential to study drug metabolism. Arch. Toxicol. 86, 87–95. doi: 10.1007/s00204-0110733-y Lin, R.-Z., and Chang, H.-Y. (2008). Recent advances in three-dimensional multicellular spheroid culture for biomedical research. Biotechnol. J. 3, 1172– 1184. doi: 10.1002/biot.200700228 Liu, H., Wang, Y., Cui, K., Guo, Y., Zhang, X., and Qin, J. (2019). Advances in hydrogels in organoids and organs-on-a-chip. Adv. Mater. 31, 1–28. doi: 10.1002/adma.201902042 Loi, C. M., Alvey, C. W., Vassos, A. B., Randinitis, E. J., Sedman, A. J., and Koup, J. R. (1999). Steady-state pharmacokinetics and dose proportionality of troglitazone and its metabolites. J. Clin. Pharmacol. 39, 920–926. doi: 10.1177/ 00912709922008533 Loskill, P., Marcus, S. G., Mathur, A., Reese, W. M., and Healy, K. E. (2015). µOrgano: a legor-like plug & play system for modular multi-organ-chips. PLoS One 10:e0139587. doi: 10.1371/journal.pone.0139587 Lübberstedt, M., Müller-Vieira, U., Biemel, K. M., Darnell, M., Hoffmann, S. A., Knöspel, F., et al. (2015). Serum-free culture of primary human hepatocytes in a miniaturized hollow-fibre membrane bioreactor for pharmacological in vitro studies. J. Tissue Eng. Regen. Med. 9, 1017–1026. doi: 10.1002/term.1652 Lübberstedt, M., Müller-Vieira, U., Mayer, M., Biemel, K. M., Knöspel, F., Knobeloch, D., et al. (2011). HepaRG human hepatic cell line utility as a surrogate for primary human hepatocytes in drug metabolism assessment in vitro. J. Pharmacol. Toxicol. Methods 63, 59–68. doi: 10.1016/j.vascn.2010. 04.013 Luo, X., Gupta, K., Ananthanarayanan, A., Wang, Z., Xia, L., Li, A., et al. (2018). Directed differentiation of adult liver derived mesenchymal like stem cells into functional hepatocytes. Sci. Rep. 8:2818. doi: 10.1038/s41598-018-20304-5 Ma, C., Zhao, L., Zhou, E. M., Xu, J., Shen, S., and Wang, J. (2016). On-chip construction of liver lobule-like microtissue and its application for adverse drug reaction assay. Anal. Chem. 88, 1719–1727. doi: 10.1021/acs.analchem.5b03869 Ma, J., Wang, Y., and Liu, J. (2018). Bioprinting of 3D tissues/organs combined with microfluidics. RSC Adv. 8, 21712–21727. doi: 10.1039/C8RA03022G Ma, X., Qu, X., Zhu, W., Li, Y. S., Yuan, S., Zhang, H., et al. (2016). Deterministically patterned biomimetic human iPSC-derived hepatic model via rapid 3D bioprinting. Proc. Natl. Acad. Sci. U.S.A. 113, 2206–2211. doi: 10.1073/pnas.1524510113 Mandenius, C. F., Andersson, T. B., Alves, P. M., Batzl-Hartmann, C., Björquist, P., Carrondo, M. J. T., et al. (2011). Toward preclinical predictive drug testing for metabolism and hepatotoxicity by using in vitro models derived from human embryonic stem cells and human cell lines - A report on the vitrocellomics EU-project. ATLA Altern. Lab. Anim. 39, 147–171. doi: 10.1177/ 026119291103900210 Mannaerts, I., Eysackers, N., Anne van Os, E., Verhulst, S., Roosens, T., Smout, A., et al. (2020). The fibrotic response of primary liver spheroids recapitulates in vivo hepatic stellate cell activation. Biomaterials 261:120335. doi: 10.1016/j. biomaterials.2020.120335 Marinho, A. T., Miranda, J. P., Caixas, U., Charneira, C., Gonçalves-Dias, C., Marques, M. M., et al. (2019). Singularities of nevirapine metabolism: from Frontiers in Cell and Developmental Biology | www.frontiersin.org 25 February 2021 | Volume 9 | Article 626805 Supplementary Material 2 3D Miniaturized cell-culture array (DataChip) + CYP450 enzymes (MetaChip) 0.068 Calcein-AM (live) / ethidium homodimer- (dead) fluorescence 3D Miniaturized cell-culture array (DataChip) + CYP450 + phase II enzymes (MetaChip) > 1.200 3D Miniaturized cell-culture array (DataChip) + human liver microsomes (MetaChip) >1.200 HepaRG 2D 24h Toxicity starting on 4.0 mM* Alamar Blue Metabolic competence and metabolite quantification | Mitochondrial ROS levels and MMP measurement (Zhang et al., 2020) 24h after 5 days in culture 24h after 22 days in culture 26.3 34.6 ATP Quantification CYP2E1 and MRP-2 activity | NA (Gunness et al., 2013) 48h 7 days 14 days 5.916 1.587 1.311 CYP, GSTT1, UGT1A1, ABCB11, ABCC1 and SLCO1B1 gene expression | NA (Bell et al., 2017) 3D Spheroids 24h after 5 days in culture 24h after 22 days in culture 2.7 10.1 ATP Quantification CYP2E1 and MRP-2 activity | NA (Gunness et al., 2013) 24h after 6/7 days in culture 24h after 21/22 days in culture 11.6 CYP induction | mitochondrial function (OCR) and glycolytic activity (ECAR), fibrosis (HSC activation, collagen secretion and deposition) (Leite et al., 2016) 3D Spheroids with pulverized liver biomatrix scaffolds 24h 20.0 (toxicity starting on 0.8 mM*) Alamar Blue Metabolic competence and metabolite quantification | mitochondrial ROS levels and MMP measurement (Zhang et al., 2020) HLCs (hESC) 2D 24h 46.0 MTT Assay CYP activity | NA (Tasnim et al., 2015) HLCs (hiPSC) 2D 48h 7 days 14 days >10.0 >10.0 9.439 ATP Quantification CYP, GSTT1, UGT1A1, ABCB11, ABCC1 and SLCO1B1 gene expression | NA (Bell et al., 2017) rpHep 2D 12h 30.0 MTT assay NA | Oxidative stress, lipid peroxidation (MDA release), DNA strand breaks (Lewerenz et al., 2003) 24h 7.6 CYP2E1 activity | NA (Jemnitz et al., 2008) 3 14.0 WST-1 assay NA (Wang et al., 2002) ~3.75 LDH Leakage NA | Intracellular GSH, ROS and MDA production, MMP visualization (Kučera et al., 2017) 30.0 Calcein AM (live) / PI (dead) fluorescent stain CYP activity | NA (Zhang et al., 2011) 3D Scaffold perfused bioreactor (RoboTox) 24h 7.0 Calcein AM (live) / PI (dead) fluorescent stain mpHep 2D 24h ~1.25 LDH Leakage NA | Intracellular GSH, ROS and MDA production, MMP visualization (Kučera et al., 2017) 3D Collagen sandwich 24h 3.8 MTT assay CYP2E1 activity | NA (Jemnitz et al., 2008) hpHep 3D Collagen sandwich 24h 28.2 MTT assay CYP2E1 activity | NA (Jemnitz et al., 2008) Cryo hpHep 2D 24h 5-10 Live cell protease /caspase-3/7 NA | Mitochondrial dysfunction (OCR) (Goda et al., 2016) 45.2 MTT Assay CYP activity | NA (Tasnim et al., 2015) ~20.0 ATP Quantification Sulfation and glucuronidation assessment | NA (Riches et al., 2009) >20.0 NA | inflammatory response (Li et al., 2020) 48h 4.596 NA | miR-122, HMGB1 and α-GST (Proctor et al., 2017) 5 days (compound addition at D0 and 2) 2.987 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) 3D Spheroids 48h 7 days 14 days >10.0 2.703 0.644 ATP Quantification CYP, GSTT1, UGT1A1, ABCB11, ABCC1 and OATP-C gene expression | NA (Bell et al., 2017) 24h after repeated dosing at D8, 12 and 15 3.280 NA | inflammatory response (Li et al., 2020) Co-culture of HepaRG and HSC 3D Spheroids 24h after 6/7 days in culture 24h after 21/22 days in culture 10.0 8.0 ATP Quantification CYP induction | mitochondrial function (OCR) and glycolytic activity (ECAR), fibrosis (HSC activation, collagen secretion and deposition) (Leite et al., 2016) Co-culture of dog hepatocytes and NPC 2D 5 days (compound addition at D0 and 2) 8.489 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) Supplementary Material 4 Co-culture of cryo hpHep and NPC 3D Spheroids 24h 72h 7 days 10 days >10 2.9 1.9 1.7 ATP Quantification Metabolite quantification | miR-122, α-GST (Foster et al., 2019) 3D Spheroid human liver microtissues (3D hLiMT) 72h 1.348 Biotransformation phase I, II and III enzymes induction | Mitochondrial oxidative stress, NAPQI-protein adducts quantification (Bruderer et al., 2015) 5-6 days 14 days 0.9110 0.5728 NA | miR-122, HMGB1 and α-GST (Proctor et al., 2017) 14 days 0.7542 NA (Messner et al., 2013) Co-culture of cryo hpHep and KC 3D Spheroids 5 days 2.247 ATP Quantification NA | inflammatory response (Li et al., 2020) Co-culture of cryo hpHep and LSEC 3D Human Liver-Chip model 10 days 2.4 ATP Quantification Metabolite quantification | miR-122, α-GST (Foster et al., 2019) ABC, ATP Binding Cassette; Ca2+, calcium; cryo, cryopreserved; CYP, cytochrome P450; ECAR, extracellular acidification rates; GSH, glutathione; GST, glutathione S-transferase; HepG2, HepG2/C3A, FaO, Huh7, HCCT-T, hepatic cell lines; hESC, human embryonic stem cells; HLC, hepatocyte-like cells (stem cell derived); HMGB, high mobility group box; hnMSC, human neonatal mesenchymal stem cells; hpHep, human primary hepatocytes; KC, Kupffer cells; MDA, malondialdehyde; miR, microRNA; MMP, mitochondrial membrane potential; MRP, multidrug resistanceassociated protein; NA, not applicable; NAPQI, N-acetyl-p-benzoquinone imine; NPC, primary human non-parenchymal cells; OATP, organic-anion-transporting polypeptides; OCR, oxygen consumption rate; ROS, reactive oxygen species; rpHep, rat primary hepatocytes; SCSIT, cell spreading inhibition test; UGT, UDP-glucuronosyltransferase. * in these reports, no IC50 was calculated and the values presented correspond to the concentration levels (mM) in which toxicity was observed. Table S1. Diclofenac cytotoxicity evaluation, metabolism assessment and mechanistic endpoints in different cell types and cell culture systems. Cell Type Cell Culture System Exposure Time IC50 / EC50 / LC50 / TC50 (µM) Cytotoxicity Endpoints Biotransformation | Mechanistic Endpoints References FaO 2D 24h 700 MTT Response to metabolism inhibitors | Intracellular GSH, Ca2+ homoeostasis alteration, lipid peroxidation (Ponsoda et al., 1995) Huh7 2D 24h 686 Alamar Blue NA (Lin et al., 2012) HCCT-T 2D 24h 104 Alamar Blue NA HepG2 2D 24h 750 MTT Response to metabolism inhibitors | Intracellular GSH, Ca2+ homoeostasis alteration, lipid peroxidation (Ponsoda et al., 1995) 399 WST-1 NA (Wang et al., 2002) ~1700 ATP quantification CYP activity, glucuronidation and sulfation activity, hepatobiliary transport | NA (Ramaiahgari et al., 2014) 5 days (compound addition at D0 and 2) 171.72 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) 3D Spheroids in dynamic culture 24h *500 LDH, Glucose secretion,  GT leakage, spheroid SCSIT NA (Xu et al., 2003) 3D Spheroids with Matrigel 24h 7 days ~2000 ~400 ATP quantification CYP activity, glucuronidation and sulfation activity, hepatobiliary transport | NA (Ramaiahgari et al., 2014) HepG2/C3A 2D 24h >400 ATP quantification CYP2E1 expression | NA (Gaskell et al., 2016) 3D Spheroids on Agarose Overlay 24h 295 ATP quantification CYP2E1 expression | NA 3D Spheroids 4 days 177.64 NA | BSEP inhibition, mitochondrial toxicity and bioactivation (Williams et al., 2020) Hep3B 2D 72h 120 MTT NA (Yu et al., 2018) 3D Miniaturized cell-culture array (DataChip + Metachip) 24h 780 Calcein-AM (live) / ethidium homodimer (dead) fluorescence 3D Miniaturized cell-culture array (DataChip) + CYP450 enzymes (MetaChip) 520 Supplementary Material 2 3D Miniaturized cell-culture array (DataChip) + CYP450 + phase II enzymes (MetaChip) 790 3D Miniaturized cell-culture array (DataChip) + human liver microsomes (MetaChip) 860 HLCs (hESC) 2D 24h 1860 MTT Assay CYP activity | NA (Tasnim et al., 2015) 1, 4 and 7 days >200 (Szkolnicka et al., 2014) HLC (hnMSC) 2D 24h 1510 MTS Assay CYP and UGT activity, bupropion and diclofenac conversion | NA (Cipriano et al., 2017b) 3D Spheroids 24h 980 rpHep 2D 24h 393 MTT Response to metabolism inhibitors | Intracellular GSH, Ca2+ homoeostasis alteration, lipid peroxidation, gluconeogenesis (Ponsoda et al., 1995) 138 ATP quantification NA (Lauer et al., 2009) 263 WST-1 NA (Wang et al., 2002) 3D Collagen sandwich 24h 400 Calcein AM (live) / PI (dead) fluorescent stain CYP activity | NA (Zhang et al., 2011) 3D Scaffold perfused bioreactor (RoboTox) 180 3D Spheroids in dynamic culture *500 LDH, Glucose secretion,  GT leakage, spheroid SCSIT NA (Xu et al., 2003) hpHep 2D 24h 1413 Alamar Blue NA (Lin et al., 2012) 3D Hollow-fiber Bioreactor 7 days *300 Lactate secretion was decreased NA (Knöspel et al., 2016) *1000 Lactate secretion, cell density and cell organization were decreased; ammonia release was increased Cryo hpHep 2D 24h 222 ATP quantification NA (Lauer et al., 2009) 529.5 NA | Inflammatory response (Li et al., 2020) 3586 MTT Assay CYP activity | NA (Tasnim et al., 2015) 3 250-500 Live cell protease/caspase-3/7 NA | Mitochondrial dysfunction (OCR) (Goda et al., 2016) 48h >4500 ATP quantification NA | miR-122, HMGB1 and α-GST (Proctor et al., 2017) 5 days (compound addition at D0 and 2) 100.01 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) 1, 4 and 7 days >200 MTT Assay CYP activity | NA (Szkolnicka et al., 2014) 3D Spheroids 48h 7 days 28 days 191 57 46 ATP quantification CYP activity | Bile acid and neutral lipids accumulation, viral infection (Bell et al., 2016) 24h after repeated dosing at D8, 12 and 15 64.2 NA | Inflammatory response (Li et al., 2020) Co-culture of dog hepatocytes and NPC 2D 5 days (compound addition at D0 and 2) 205.7 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) Co-culture of cryo hpHep and NPC 3D Spheroid human liver microtissues (3D hLiMT) 14 days 178.6 ATP quantification NA (Messner et al., 2013) 5-6 days 14 days 100.0 61.4 NA | miR-122, HMGB1 and α-GST (Proctor et al., 2017) Co-culture of hpHep and KC LiverChip bioreactors 48h 227 WST-1 CYP and UGT activity, metabolite identification, diclofenac protein binding and clearance | Bile acid identification and quantification, inflammatory response (Sarkar et al., 2017) 3D Spheroids 5 days NA ATP quantification NA | Inflammatory response (Li et al., 2020) BSEP, bile salt export pump; Ca2+, calcium; cryo, cryopreserved; CYP, cytochrome P450; GSH, glutathione; GST, glutathione S-transferase; HepG2, HepG2/C3A, FaO, Huh7, HCCT-T, hepatic cell lines; hESC, human embryonic stem cells; HLC, hepatocyte-like cells (stem cell derived); HMGB, high mobility group box; hnMSC, human neonatal mesenchymal stem cells; hpHep, human primary hepatocytes; KC, Kupffer cells; miR, microRNA; NA, not applicable; NPC, primary human non-parenchymal cells; OCR, oxygen consumption rate; rpHep, rat primary hepatocytes; SCSIT, cell spreading inhibition test; UGT, UDP-glucuronosyltransferase. * in these reports, no IC50 was calculated and the values presented correspond to the concentration levels (µM) in which toxicity was observed. Table S1. Troglitazone cytotoxicity evaluation in different cell types and cell culture systems. Cell Type Cell Culture System Exposure Time IC50 / EC50 / LC50 / TC50 (µM) Cytotoxicity Endpoints Biotransformation | Mechanistic Endpoints References HepG2 2D 5 days (compound addition at D0 and 2) 95.85 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) 24h ~200 ATP quantification CYP activity, glucuronidation and sulfation activity, hepatobiliary transport | NA (Ramaiahgari et al., 2014) 3D Spheroids with Matrigel 24h after 21 days in culture 7 days after 21 days in culture ~200 ~80 HepG2/C3A 3D Spheroids 4 days 42.71 ATP quantification NA | BSEP inhibition, mitochondrial toxicity and bioactivation (Williams et al., 2020) HepaRG 2D 24h after 5 days in culture 24h after 22 days in culture 41 301.3 ATP quantification CYP2E1 and MRP-2 activity | NA (Gunness et al., 2013) 48h 7 days 14 days >100 36.5 34.6 CYP, GSTT1, UGT1A1, ABCB11, ABCC1 and SLCO1B1 gene expression | NA (Bell et al., 2017) 3D Spheroids 24h after 5 days in culture 24h after 22 days in culture 398 >500 ATP quantification CYP2E1 and MRP-2 activity | NA (Gunness et al., 2013) 8 days 14 days ~15 ~10 NA | BSEP inhibition, bile acid accumulation, F-actin cytoskeleton disruption (Hendriks et al., 2016) HLCs (hESC) 2D 24h 33.7 MTT Assay CYP activity | NA (Tasnim et al., 2015) 24h 4 days 7 days 198.53 101.06 85.74 ATP quantification CYP activity | NA (Szkolnicka et al., 2014) HLCs (hiPSC) 2D 48h 7 days 14 days 46 33.9 18.7 ATP quantification CYP, GSTT1, UGT1A1, ABCB11, ABCC1 and SLCO1B1 gene expression | NA (Bell et al., 2017) Supplementary Material 2 rpHep 2D 24h 140 ATP quantification NA (Lauer et al., 2009) 48h *6 MTT assay LDH leakage CYP3A4 activity | GSH content, MDA and MMP potential measurement, lipid accumulation (Shen et al., 2012) 3D Gel entrapment 48h 21 days No toxic response *30 hpHep 3D Gel entrapment 4 days *6 MTT assay LDH leakage CYP3A4 activity | GSH content, MDA and MMP potential measurement, lipid accumulation (Shen et al., 2012) Cryo hpHep 2D 24h 62.5-125 Live cell protease/ caspase-3/7 NA | Mitochondrial dysfunction (OCR) (Goda et al., 2016) 29.7 MTT Assay CYP activity | NA (Tasnim et al., 2015) 30.9 ATP quantification NA | inflammatory response (Li et al., 2020) 88 NA (Lauer et al., 2009) 24h 4 days 7 days >200 >200 37.93 CYP activity | NA (Szkolnicka et al., 2014) 48h >4500 NA | miR-122, HMGB1 and α-GST (Proctor et al., 2017) 5 days (compound addition at D0 and 2) 55.59 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) 3D Spheroids 24h after repeated dosing at D8, 12 and 15 1.0 ATP quantification NA | inflammatory response (Li et al., 2020) 8 days 14 days ~ 8.0 ~7.8 NA | BSEP inhibition, bile acid accumulation, F-actin cytoskeleton disruption (Hendriks et al., 2016) 48h 7 days 14 days 37.4 4.2 1.5 CYP, GSTT1, UGT1A1, ABCB11, ABCC1 and OATP-C gene expression | NA (Bell et al., 2017) Co-culture of rpHep and NPC 2D 48h *from 10-100 µM ATP quantification LDH leakage CYP activity | NA (Kostadinova et al., 2013) 3D Nylon scaffold 1 to 8 days No marked toxicity Co-culture of dog hepatocytes and NPC 2D 5 days (compound addition at D0 and 2) 57.51 Live cell protease activity CYP activity, glucuronidation and sulfation activity | NA (Atienzar et al., 2014) 3 Co-culture of hpHep and NPC 2D 48h No marked toxicity ATP quantification LDH leakage CYP activity | NA (Kostadinova et al., 2013) 3D Nylon scaffold 1 to 8 days *from 10-100 µM Co-culture of cryo hpHep and NPC 3D Spheroid human liver microtissues (3D hLiMT) 5-6 days 14 days 25.6 14.6 ATP quantification NA | miR-122, HMGB1 and α-GST (Proctor et al., 2017) Co-culture of cryo hpHep and KC 3D Spheroids 5 days *1.0 ATP quantification NA | inflammatory response (Li et al., 2020) ABC, ATP Binding Cassette; BSEP, bile salt export pump; cryo, cryopreserved; CYP, cytochrome P450; ECAR, extracellular acidification rates; GSH, glutathione; GST, glutathione S-transferase; HepG2, HepaRG, HepG2/C3A, hepatic cell lines; hESC, human embryonic stem cells; HMGB, high mobility group box; hpHep, human primary hepatocytes; KC, Kupffer cells; MDA, malondialdehyde; MMP, mitochondrial membrane potential; mpHep, mouse primary hepatocytes; MRP, multidrug resistance-associated protein; NA, not applicable; NPC, non-parenchymal cells; OATP, organic-anion-transporting polypeptides; OCR, oxygen consumption rate; rpHep, rat primary hepatocytes; UGT, UDP-glucuronosyltransferase. * in these reports, no IC50 were calculated and the values presented correspond to the concentration levels (µM) in which toxicity was observed.