1. Introduction
Humans are continuously and increasingly exposed to a variety of xenobiotics such as drugs, chemicals, pesticides, and environmental pollutants. Before commercialisation, drugs undergo a thorough testing process to evaluate their effects and toxicity (Khetani et al., 2015). Since 2007, enforcement of the REACH (Registration, Evaluation, Authorisation and Restriction of CHemical substances) legislation imposes the evaluation of risks of all chemical substances (Zeller et al 2016). The need for toxicological evaluation is thus increasing. Animal models are widely used as reference tools for predictive studies in drug development and risk assessment (Messelmani et al., 2022). However, due to differences between animal and human metabolism and physiology, animal models fail to accurately reproduce the human condition, and this issue challenges the extrapolation of data to humans (Son et al., 2020). For example, the predictivity of animal models for chemical-induced hepatotoxicity is only 50% (Ruoss et al., 2020). Moreover, animal experiments are costly, time-consuming, and, most importantly, raise ethical and regulatory issues (Ruoss et al., 2020; Soldatow et al 2013). To decrease the use of laboratory animals, the REACH legislation and the 3R rules, recommended to reduce as much as possible the use of animal models, have pressed industrial companies and scientists to develop alternative approaches to animal testing (Son et al., 2020). Consequently, developing reliable methods not based on using animals and in vivo experimentation has become necessary.
The liver is the main site involved in the metabolism of xenobiotics and is therefore the most commonly used organ in toxicological and pharmacological tests. It is a multifunctional and complex organ performing a variety of vital functions focused on biotransformation, storage, and synthesis (Polidaro et al., 2021; Bale, S. S., & Borenstein 2018). The liver is composed of several cell types, the main ones being hepatocytes (parenchymal cells) and non-parenchymal cells (NPCs): sinusoidal endothelial cells (LSECs), Kupffer cells (KCs), hepatic stellate cells (HSCs), and biliary epithelial cells (LeCluyse et al., 2012, Moradi et al., 2020). Hepatocytes represent approximately 60% of the total liver cells, and are the main cell type, ensuring most metabolic activities (Beckwitt et al., 2018). The NPCs are involved in several key functions, such as the production of growth factors and mediators of cellular functions, maintenance of tissue architecture, and regulation of liver response to xenobiotics (LeCluyse et al., 2012, Moradi et al., 2020).
Given that hepatocytes ensure the major functions of the liver, especially xenobiotic metabolism, most of the current in vitro liver models are focused on hepatocytes and do not include NPCs (Bale et al., 2016). Moreover, the models used for drug screening and risk assessment are mainly based on cell culture in static two-dimension (2D) monolayers using conventional Petri dishes (Messelmani et al., 2022). These 2D cultures present some advantages, such as allowing high throughput analyses, ease of manipulation, and a lower cost (Milner et al., 2020; Moradi et al., 2020). However, 2D monocultures of hepatocytes or of hepatic cell lines suffer from several disadvantages associated with the loss of tissue-specific architecture, mechanical and biomechanical cues, and cell-cell and cell-matrix interactions. Consequently, these models fail to recapitulate the complexity of the in vivo physiological environment, show limited prediction capacity for xenobiotics, and cells are prone to dedifferentiation within 48-72 h (Messelmani et al., 2022. Milner et al., 2020; Panwar et al., 2021).
Recently, several approaches have been proposed to overcome the drawbacks associated with 2D monolayer cultures of hepatocytes. Microfluidic devices, or organ-on-chip (OoC) technology, are a promising tool for building more relevant in vitro liver models aimed at mimicking the in vivo environment (Merlier et al., 2017). The microfluidic perfusion improves the exchanges and transport of nutrients, oxygen, metabolites, and other chemicals, and creates a controlled micro-environment and physiological-like features, including the liver zonation, cell-cell interactions, shear stress, and chemical concentration gradients (Moradi et al., 2020; Messelmani et al., 2022; Lee et al., 2021; Lee et al., 2019). Several studies have reported that perfused microfluidic cultures enhance the long-term viability and functionality of hepatocytes (Jellali et al., 2016; Schepers et al., 2016; Yu et al., 2017). The three-dimensional (3D) cell culture (spheroids/organoids), with and without polymer matrix, also makes it possible to maintain tissue architecture similar to the in vivo situation and maintains liver-specific functions. This organisation enhances cell-cell and cell-matrix interactions and the creation of chemical gradients (Polidaro et al., 2021; Fang et al., 2017; Yun et al., 2023). Among other approaches used to maintain hepatocyte functions, cocultures with NPCs are commonly used strategies (Ruoss et al., 2020). Among NPCs, LSECs participate in liver metabolic functions and maintain hepatocyte phenotype and functions through paracrine communication (Ortega-Ribera et al., 2018). The benefits of coculturing LSECs and hepatocytes have been reported in several works (Ortega-Ribera et al., 2018; Bale et al., 2015; Xiao et al., 2015).
Previously, we developed a liver-on-chip model integrating a hydroscaffold containing key liver extracellular matrix (ECM) components (Messelmani et al., 2022). This device made possible the dynamic culture of HepG2/C3a organised into 3D spheroids for the long-term, while maintaining their functionalities. Here, to better reproduce the physiology of the liver, our liver-on-chip model was cocultured with liver sinusoidal endothelial cells. The coculture was performed using a fluidic platform making it possible to connect the liver biochip previously developed by our laboratory (Bricks et al., 2014) to a new LSEC barrier insert. The behaviour and functionalities of the LSECs barrier (SK-HEP-1 cell line) and hepatocyte biochip (HepG2/C3a cells) in monoculture and coculture were studied and compared. Then, the coculture model was exposed to paracetamol (APAP), and the crosstalk between both compartments was studied and compared to monocultures exposed to APAP.
2. Materials and methods
2.1. Manufacturing of biochip
The biochip fabrication and design details were described in our previous work (Jellali et al., 2016). The biochip consists of two polydimethylsiloxane layers (PDMS, Sylgard 184 kit, Dow Corning) manufactured by soft lithography and sealed via plasma treatment. The biochip consists of two parts: the microstructured bottom layer contains chambers and microchannels (height of 100 µm), and the top layer, with a 100 µm-deep reservoir, includes an inlet and outlet for culture medium perfusion (Figure S1A).
To promote 3D cell organisation, a hyaluronic acid (HA)-based hydroscaffold (RGDS-grafted HA, galactosamine-grafted HA, collagen type I and collagen type IV) was integrated into the biochip (BIOMIMESYS® Liver, HCS Pharma, Loos, France). The hydroscaffold preparation was performed in accordance with a previously patented process (Souguir et al., 2016). Briefly, the pseudo-hydrogel solution (containing HA, collagen and crosslinker: adipic acid dihydrazide) was injected into the biochip and the hydroscaffold crosslinking was performed in situ. The biochips were then washed, freeze-dried, and sterilised using ultraviolet (UV) exposure. Pictures and microscope images of the biochips with and without hydroscaffolds are presented in Figure S1B.
2.2. Coculture platform: IIDMP fluidic device
We used the previously described Integrated Insert in a Dynamic Microfluidic Platform (IIDMP, Bricks et al., 2014) coculture system which consists of a polycarbonate fluidic platform with three subunits (Figure S2). Each subunit is composed of the association of an insert and a biochip linking two wells. The insert was placed in the first well and defined an apical pole (the LSECs barrier) and a basal pole making possible the exchange of culture medium between the LSECs barrier and the hepatocyte compartment (biochip, Figure S2). The biochip connected the first and second well (acting as a reservoir). The overall IIDMP device consisted of three subunits and made possible the coculture of three inserts and three biochips simultaneously. The volume of culture medium was 10 mL: 1 mL placed in the apical insert, 5 mL below the insert, and 4 ml in the second well. Culture medium flowed through the biochip from the basal compartment in the first well towards the second well. The perfusion fluid was provided by a cover connected to a peristaltic pump (Ismatec™) via PTFE (polytetrafluoroethylene) tubing. This cover made it possible to close the IIDMP device hermetically. The other components of the IIDMP platform were silicone gaskets sealing the device, and a bottom layer composed of the well subunits, thanks to which the biochips were connected (at the bottom, Figure S2).
2.3. Cells and culture media
HepG2/C3A, a clone of the HepG2 line derived from human hepatocellular carcinoma (ATCC CRL-10741), were used as the hepatocyte model. They were cultured in Minimal Essential Medium (MEM) with phenol red (Pan Biotech), 10% fetal bovine serum (FBS, Gibco), 1 mM hydroxy-ethylpiperazine-N-2-ethanesulfonic acid (HEPES, Gibco), 2 mM L-glutamine (Gibco), 0.1 mM non-essential amino acids (Gibco), 1 mM sodium pyruvate (Gibco), and 100 U/mL penicillin, 100 μg/mL streptomycin (Pan Biotech).
The SK-HEP-1 cell line (ATCC, HTB-52) derived from an adenocarcinoma of the liver was used as the LSECs model. For maintenance, SK-HEP-1 cells were cultured in a mixture of 75% EGM-2 medium (Lonza) and 25% MEM (complemented as mentioned above).
All cells were cultured in 75 cm² flasks at 37°C in a humidified atmosphere with 5% of CO2. The culture medium was renewed every 2 days and the cells were passaged weekly at a confluence of 80-90%. To decrease the potential variability in the results, the cells were used between passages 10 and 20.
2.4. Experimental procedure of the cell cultures
2.4.1. Optimisation of common culture medium for HepG2/C3A and SK-HEP-1 cells
Culture medium optimisation was performed in static conditions, and different MEM/EGM-2 ratios were tested. The SK-HEP-1 cells were seeded in cell culture inserts (6-well format, polyethylene terephthalate membrane, 0.4 µm pore, THINCERT Greiner) at a density of 0.35 105 cell/cm2. The culture medium was renewed every 2 days in the apical (1 mL) and basal (2 mL) compartments, and the culture was maintained until confluence was attained (6-8 days). The HepG2/C3A were seeded in the wells of a 6-well plate at a density of of 105 cell/cm2. The culture was maintained for 4 days, and the medium (2 mL) was changed every 2 days. The cultures were continuously maintained at 37°C in a 5% CO2 supplied incubator and the assays were performed at the end of the experiments.
2.4.2. Dynamic monoculture and coculture in the IIDMP device
Each experiment lasted two days (Figure 1). The SK-HEP-1 inserts were maintained for 8 days for the formation of a confluent barrier, before performing the dynamic experiments, as mentioned in section 2.4.1. In parallel, 24 h before the dynamic experiments, HepG2/C3a cells were seeded in the biochips containing the hydroscaffold (4 105 cell/biochip), and the biochips were incubated overnight at 37°C in a humidified atmosphere with 5% of CO2.
On Day 0 of the experiment, the SK-HEP-1 previously grown for 8 days on inserts were transferred into the first well of the IIDMP device and the HepG2/C3a biochips were connected to the bottom of the device. As shown in Figure 1, three conditions were established: SK-HEP-1 monoculture (IIDMP with insert alone), HepG2/C3a monoculture (IIDMP with biochip alone) and coculture (IIDMP containing insert and biochip). Culture medium was added (1 mL in the apical insert side, 5 mL in the basal side and 4 mL in the reservoir well), the IIDMP was closed and connected to the pump. The entire setup was placed in the incubator and perfusion started at 10 µL/min for 48 h in a closed loop. For exposure to drugs, acetaminophen (APAP, Sigma-Aldrich) was loaded into the apical compartment of the insert at 1 mM before perfusion started (an insert without cells was used for HepG2/C3a monoculture experiments). After dilution in the total medium in the circuit (10 mL), the systemic concentration of APAP was 100 µM.
For each experiment, three IIDMP devices were used simultaneously to achieve 9 replicates/condition per experiment. At the end of the experiment, the pump was stopped, and the cover removed. The culture media were sampled, and the SK-HEP-1 inserts and HepG2/C3a biochips were detached from the device to perform the assays.
2.5. Lucifer Yellow permeability assay
Lucifer Yellow (LY) is a fluorescent molecule used to test paracellular transport across the endothelial (SK-HEP-1) barrier. LY (LY CH dipotassium salt, Sigma-Aldrich) was diluted in Hanks' balanced salt solution (HBSS, with CaCl2 and MgCl2, Gibco) at 50 µM and loaded into the apical compartment of an empty insert and inserts with cells were cultured for 4-15 days. The basal compartment was filled with HBSS. The inserts were then incubated at 37°C and 5% of CO2. After 90 min, medium from the apical and basal compartments was collected. The fluorescence intensity was measured using a microplate reader (TECAN Spectafluor Plus) at excitation/emission wavelengths of 485/530 nm. The flow of LY was expressed by the calculation of the apparent permeability (Papp, m/s) as follows:
Papp = (dQ/dt) x (1/AxCa)
Where dQ/dt is the amount of LY transported during a given time (mol/s), Ca is the initial concentration of LY solution (mol/m3) and A is the surface of the insert (m2).
2.6. Permeability to dextrans
The SK-HEP-1 barrier’s permeability to molecules of different molecular weights was assessed using fluorescein isothiocyanate-dextrans (FITC-dextran 4, 70 and 150 kDa, Sigma-Aldrich). The assays were performed using confluent SK-HEP-1 barriers (8 days of culture) in static and dynamic (IIDMP device) conditions. The dextrans were diluted in the culture medium at a concentration of 100 µg/mL and deposited in the apical compartment of the culture inserts. Then, culture medium was sampled in the apical and basal compartments at different times. The FITC-dextran fluorescence intensity was measured using a microplate reader (TECAN Spectafluor Plus) at excitation/emission wavelengths of 490/525 nm.
2.7. Immunostaining assays
At the end of the experiments, the SK-HEP-1 inserts were washed with phosphate buffer saline solution (PBS, Gibco), fixed in PBS, 4% para-formaldehyde (PFA, MP biomedicals) for 30 min at room temperature, washed and stored at 4°C in PBS. Cells were permeabilised with 0.5% Triton X-100 in PBS solution for 30 min and blocked in PBS, 1% bovine serum albumin (BSA, Sigma Aldrich) for 30 min. The samples were incubated with primary antibodies overnight at 4°C in the dark. After washing with PBS, the secondary antibodies were added, and the samples incubated for a further 12 h at 4°C in the dark. Actin filaments were stained with Alexa Fluo 488 Phalloidin for 3h (Thermo Fisher, 1/50). Nuclei were stained with 10 µg/mL 4′,6-diamidino-2-phenylindole (DAPI, D1306, Invitrogen) for 30 min at room temperature in the dark. Imaging was obtained with a laser scanning confocal microscope (LSM 710, Zeiss).
The primary antibodies used were mouse anti-PECAM-1/CD31 (ab24590, abcam, 1 µg/mL), rabbit anti-stabilin-2 (ab121893, abcam, 1 µg/mL), mouse anti-vimentin (ab8978, abcam, 1 µg/mL). Donkey anti-mouse Alexa Fluor 647 (ab150107, Abcam, 2 µg/mL) and goat anti-rabbit Alexa Fluor 488 (A11034, Invitrogen, 2 µg/mL) were used as secondary antibodies.
2.8. Albumin and interleukin-6 measurements
ELISA sandwich assays were used to quantify the albumin and IL-6 concentrations in the culture media collected at the end of the experiments. The assays were performed using a human albumin ELISA Quantitation Set (E80-129, Bethyl Laboratories) and a human IL-6 ELISA Kit (ab718013, abcam) for albumin and IL-6, respectively, following the protocols recommended by the manufacturers. The results were acquired using a Spectafluor Plus microplate reader (TECAN) set to a wavelength of 450 nm.
2.9. RNA extraction and RTqPCR analysis
At the end of the experiments, the culture media were removed from the biochips and inserts, and the cells (HepG2/C3a and SK-HEP-1) were lysed and recovered using 500 µL of TRIzol (Thermofischer Scientific) and stored at -80 °C until use.
Total RNA was purified by phenol/chloroform extraction followed by alcohol precipitation, and RNA concentrations measured using a NanodropOne (Thermofisher Scientific). Reverse transcription reactions were performed using a High-capacity cDNA reverse transcription kit with RNase inhibitor (Applied Biosystems, ThermoFisher Scientific). Quantitative PCRs were performed using a StepOnePlus machine (Applied Biosystems, Thermofisher Scientific) in duplex reactions, mixing the cDNA with the TaqMan FAM-labelled probes of the selected gene (Applied Biosystems, ThermoFisher Scientific) and a β2-microglobulin-VIC-labeled probe (Table S1). The threshold cycle (CT) values were calculated at the upper linear range of the logarithm−2 amplification curve using the StepOne v2.3 software (Thermofisher scientific). The data were then expressed as 2−∆∆CT with ∆CT as the difference between the CT of the transcript of interest and the CT of the reference, and ∆∆CT is the difference between the mean ∆CT of the experimental samples and the mean ∆CT of the control samples (Livak and Schmittgen 2001). The relative quantity (RQ) corresponds to 2−∆∆CT which transforms the logarithmic−2 data into decimal values.
2.10. HPLC-HRMS
Detection and quantitative evaluation of APAP and APAP metabolites was performed with LC-HRMS. The HPLC system (Infinity 1290, Agilent Technologies, France) with DAD, was connected to a Q-TOF micro hybrid quadrupole time of flight mass spectrometer (Agilent 6538, Agilent Technologies, France) with electrospray ionisation (ESI). HPLC was carried out on a Thermo Hypersyl Gold C18 (USP L1) column (150 × 4.6 mm ID, 5 µm, 175 A), connected to an Agilent Infinity 1290 HPLC at 40°C. The solvent system was A: 0.1% formic acid in H2O and B: acetonitrile. The gradient programme began with 5% B, held at 5% for 1 min and ramped up to 20% B at 5 min and to 95% at 7 min, held at 95% for 3 min, then decreased to the initial condition and held at 5 % for 1 min. The flow rate was set at 1 mL/min. All compound responses were measured in ESI+ and ESI- alternatively and were calibrated externally. The ESI Gas Temp was 350 °C, at electrospray voltage +3800 V or – 3500V. Drying gas was set at 10 L/min and the nebuliser was at 30 psi. Fragment voltage was set at 110 V. The HRMS spectrum was registered at 5 Hz in the mass range of 50 to 1200 m/z with internal calibration. Software MassHunter (Version B.07.00, Agilent Technologies, Santa Clara, CA 95051, United States) was used for data processing, quantification, and data acquisition. APAP and APAP metabolites were validated by the conjunction of exact mass and retention time from standards. Software MassHunter (Version B.07.00, Agilent Technologies, Santa Clara, CA 95051, United States) was used for data processing, quantification, and data acquisition. APAP and APAP metabolites were validated by the conjunction of exact mass and retention time from standards.
2.11. Statistical analysis
All experiments were performed at least three times and a minimum of 2 bio-chips/inserts/cocultures were performed in each experiment (N = 3 experiments and n = 6 HepG2/C3a biochips, SK-HEP-1 inserts and HepG2/C3a/SK-HEP-1 cocultures). Data are presented as means ± standard deviations (SD) of the 6 replicates (for RTqPCR assays, only 3 replicates from 3 different experiments were used). To determine statistical differences, a one-way ANOVA test and unpaired t-test were performed using GraphPad software (Prism 8). Data with P values < 0.05 were identified as statistically significant and highlighted in the figures.
3. Results
3.1. Selecting a culture medium for SK-HEP-1 and HepG2/C3a coculture
The culture of cells of different origin in the same system requires an adapted coculture medium capable of maintaining both cell types in good conditions, without impairing their characteristics and functionalities. The routine culture medium used in our conditions for SK-HEP-1 is EGM-2/MEM (75%/25%) and the cells formed a well-structured cell monolayer at confluence, as needed for the barrier function (Figure 2A, Figure S3). On the other hand, when SK-HEP-1 cells were cultured in HepG2/C3a medium, which is based on MEM only, the endothelial cell morphology was greatly altered, and the cells failed to form a confluent monolayer (Figure S3). In an attempt to, first, create the endothelial barrier, and then to switch to a hepatocyte culture medium, SK-HEP-1 cells were cultured in their normal medium for 6 days, followed by culture in MEM for 3 days (as the coculture period). In these conditions again, the endothelial cells failed to maintain a confluent monolayer (Figure S3). Finally, when cells were maintained in EGM-2/MEM (25%/75%) medium for 7 days, the SK-HEP-1 cells formed a confluent monolayer (Figure 2A and Figure S3) and exhibited the characteristic morphology of SK-HEP-1, as when cultured in their original medium. The gene expression levels of several LSECs markers were investigated. No major differences were observed for most of the genes when cells were cultured in EGM-2/MEM (25%/75%) when compared to their original medium. A downregulation of CLEC4M and VCAM1 was observed when cells were maintained in EGM-2/MEM (25%/75%) in comparison with native medium, with fold changes (FC) of 0.25 and 0.48, respectively (Figure 2B).
The EGM-2/MEM (25%/75%) medium was also tested on HepG2/C3a cells and compared to culturing in MEM. After 4 days of static culture, the HepG2/C3a presented a typical morphology and formed a monolayer in both conditions (Figure 2C). Additionally, secretion of albumin was measured to assess whether HepG2/C3a cells retained their hepatic properties. Similar albumin secretion levels were observed in both conditions. The levels were approximately 125 ± 11 and 114 ± 17 ng/h for cells cultured in MEM and EGM-2/MEM 1/3 mixture, respectively (Figure 2D).
Based on the results obtained with SK-HEP-1 and HepG2/C3a cells, the mixture of EGM-2/MEM (25%/75%) was chosen for the dynamic coculture experiments. To facilitate the comparisons between monoculture and coculture, this medium was also used for SK-HEP-1 and HepG2/C3a maintenance in monocultures.
3.2. Characterisation of the SK-HEP-1 endothelial barrier
LSECs act as a physical barrier to molecules and play a significant role in transportation from circulating blood to the hepatocytes. Therefore, before using SK-HEP-1 to form a liver endothelial barrier in our coculture model, it was essential to characterise the formation, integrity, and permeability of the barrier. The SK-HEP-1 cells were seeded in static inserts using the selected coculture medium and followed over time. The cells proliferated continuously to reach full confluence and form homogenous and continuous monolayers from Days 7-8 and thereafter (Figure S4). Then, overgrowth could be observed, resulting in the formation of a second layer of cells on top of the first one (Day 10, Figure S4). Nevertheless, the formation of continuous layers of confluent cells was confirmed by nuclei, vimentin, and actin stainings. As shown in Figure 3A, the tissue was dense with contiguous cells and a well-developed actin network. The LSECs phenotype of the SK-HEP-1 barrier was confirmed by the positive staining for LSECs markers PECAM-1 and stabilin-2 (Figure 3B).
The formation of a confluent barrier was associated with major modifications in paracellular permeability. The flow through the barrier was directly correlated to the integrity and homogeneity of the barrier. To confirm the formation of the barrier, permeability to Lucifer Yellow was checked using SK-HEP-1 inserts at different times of culture. PET inserts without cells exhibited a permeability value of 177 10-15 ± 9 10-15 m/s (Figure 3C). When SK-HEP-1 cells were added, a significant decrease in Lucifer Yellow paracellular flow from the apical to the basal compartment was observed, with apparent permeability values of 98 10-15 ± 10 10-15 and 35 10-15 ± 10-15 m/s at Days 4 and 8, respectively. This latter value remained stable, at approximately 40 10-15 ± 8 10-15 m/s until Day 15. These results suggested that the SK-HEP-1 cells were capable of forming a barrier which reached relative stability at Day 8, and could be used for coculture with HepG2/C3a and permeability experiments.
The permeability of the SK-HEP-1 barrier to molecules with different molecular weights was also assessed, using FITC-dextran of 4, 70 and 150 kDa. The experiments were performed using confluent SK-HEP-1 cultures at Day 8 in static inserts. For comparison, the same experiments were performed using inserts without cells. When using each of the different molecular weight dextrans, we found that the tracer concentrations decreased from the apical compartment and increased in the basal one over time (Figure 4). Thus, the tracer molecules were able to pass through the insert membranes whether the cells were present or not. However, the FITC dextrans diffused at faster rates into the basal compartment when the inserts were not seeded with endothelial cells, whereas the presence of a SK-HEP-1 cell layer slowed the diffusion process for the three molecular weight markers, confirming that the SK-HEP-1 made an efficient diffusion barrier. As expected, the diffusion rates were dependent on the FITC-dextran molecular weight and were slower when using FITC-dextran of 150 kDa when compared to 4 kDa- dextran.
3.3. Dynamic coculture of the SK-HEP-1 barrier and HepG2/C3a biochip
Following the previous characterisations and optimisations, the coculture of SK-HEP-1 barrier (LSECs compartment) with HepG2/C3a cells cultured in 3D in the biochip (the hepatocyte compartment as previously characterised, Messelmani et al., 2022) was assessed. The coculture was performed for 48 h in the IIDMP platform and the communication between both compartments was ensured by culture medium circulation. In parallel, for comparison, SK-HEP-1 and HepG2/C3a monocultures were also used in the IIDMP platform.
3.3.1. Effect of the dynamic coculture on the SK-HEP-1 barrier
After 8 days of barrier maturation in static conditions followed by 48 h of dynamic coculture or monoculture, the SK-HEP-1 inserts were collected and characterised. Although cells were barely distinguishable because of the density at confluence, the morphology of the SK-HEP-1 tissues appeared similar in coculture and monoculture. In both culture modes, the cells formed homogenous and continuous barriers and grew beyond confluence (Figure S5). Confocal microscopy imaging of actin, vimentin and nuclei staining confirmed the formation of a continuous endothelial barrier, with different cell layers and a developed actin/vimentin network (Figure 5A). Furthermore, no obvious differences were observed between the staining of cocultured and monocultured endothelial barriers. SK-HEP-1 barriers in monoculture and coculture expressed typical LSECs markers without any apparent difference between the two modes of culture, as illustrated by the detection of PECAM-1 and stabilin-2 positive SK-HEP-1 cells (Figure 5B).
Gene expression level analyses of several LSECs markers revealed the significant upregulation of CLEC4M (FC: 2.05) whereas KDR was downregulated (FC: 0.49) in SK-HEP-1 cocultures (Figure 6A). The expression levels of PECAM-1, MRC1 and CD32b were similar in SK-HEP-1 monocultures and cocultures. Finally, the diffusion rates of FITC-dextran 4 kDa through the SK-HEP-1 barrier, in dynamic monoculture and coculture with HepG2/C3a, were compared. The changes in FITC-dextran concentrations in the apical (decrease) and basal (increase) compartments confirmed the permeability of the barrier and the communication between the apical side of the barrier and the HepG2/C3a biochip (Figure 6B). The variations in FITC-dextran concentrations in the apical compartment revealed a lower diffusion rate through the barrier in coculture when compared to that in monoculture, notably after 24 h.
3.3.2. Behaviour and functionality of HepG2/C3a in coculture with the SK-HEP-1 barrier
The day before starting the dynamic monocultures or cocultures in the IIDMP devices, HepG2/C3a cells were seeded into the biochips containing the hydroscaffold and incubated for 24 h in static conditions (adhesion phase). Twenty-four hours after seeding, the cells were embedded in/adhered to the hydroscaffold and started to create spheroid-like aggregates (Figure 7A). Then, the biochips were connected to the IIDMP device, with and without an SK-HEP-1 barrier, and perfusion was started. The hepatocytes maintained in coculture with an endothelial barrier had a similar morphology to cells maintained in monoculture. In both conditions, the HepG2/C3a formed a dense tissue, organised in 3D spheroids ranging between 200 and 500 µm in diameter, similar in both monoculture and coculture (Figure 7A).
To evaluate the effects of coculture on the specific functions of HepG2/C3a, albumin secretion rates were quantified and found to be similar to those in monoculture (Figure 7B). After 48 h of culture, the albumin secretion levels were 127 ± 24 and 134 ± 28 ng/h in monoculture and coculture, respectively. The expression levels of several specific marker genes of HepG2/C3a cells (UGT2B7, UGT1A1, SULT1A2, CYP1A2 and CYP1A1) were also evaluated to determine whether the SK-HEP-1 barrier influenced the HepG2/C3a cultures. As shown in Figure 7C, there were no differences in expression levels in the selected genes between HepG2/C3a maintained as a monoculture and HepG2/C3a in coculture with SK-HEP-1.
3.4. Exposure of the coculture and monoculture models to acetaminophen (APAP)
To test the coculture model and demonstrate the crosstalk between the HepG2/C3a biochips and SK-HEP-1 barrier in the configuration of a drug study, we exposed the SK-HEP-1 barrier to APAP with and without coculture with the HepG2/C3a biochip. APAP was chosen because it is i) metabolised by HepG2/C3a cells, ii) widely studied with liver in vitro models, and iii) not adsorbed by the PDMS biochip (Bricks et al., 2015). APAP was introduced into the apical side of the SK-HEP-1 barrier at 1 mM, leading to a systemic theorical concentration of 100 µM after diffusion in the total circuit. For comparative purposes, HepG2/C3a monoculture in the IIDMP was also performed and APAP was deposited into the insert without SK-HEP-1.
SK-HEP-1 cells exposed to APAP for 48 h in coculture or in monoculture exhibited a confluent and continuous barrier composed of several cell layers, forming a dense tissue. The cell morphologies between the treated SK-HEP-1 barrier in coculture and in monoculture showed no significant differences (Figure S6). Moreover, the SK-HEP-1 cells exposed to APAP were similar to those without APAP (monoculture and coculture, Figure S5). As shown in Figure 8A, APAP treatment appeared to affect the actin cytoskeleton of the barrier. Compared to barrier monoculture and coculture without APAP (Figure 5A), the actin network was disordered and composed of more elongated filaments. The immunostaining of specific LSECs markers showed weaker expression levels of PECAM-1 and stabilin-2 in SK-HEP-1 exposed to APAP (Figure 8B), when compared to monoculture and coculture without APAP (Figure 5B). This effect was more striking in the coculture. Gene expression level analyses of cultures treated or not with APAP showed an upregulation of KDR (FC: 1.8) after APAP exposure in coculture (Figure 9A). Conversely, both this gene and CLEC4M were downregulated in the monoculture exposed to APAP (FC: 0.54 and 0.49 for CLEC4M and KDR, respectively). APAP treatment of the monoculture also led to the upregulation of MRC1 (FC 1.4).
Regarding the HepG2/C3a compartment, the cells exposed to APAP (coculture and monoculture) maintained their organisation in 3D spheroids up until the end of the culture (Figure S6). The cells formed dense tissues, without any apparent difference compared to non-treated cultures. Analysis of gene expression levels showed no differences between the biochip monocultures treated or not with APAP (Figure 9B). In HepG2/C3a cocultured with the SK-HEP-1 barrier, UGT2B7 expression levels were downregulated (FC: 0.7). In addition, the ratios of albumin secretion (culture with APAP versus without APAP) were 0.92 ± 0.25 and 0.95 ± 0.09 for monoculture and coculture, respectively (Figure 9C).
The metabolism of APAP was then investigated in SK-HEP-1 and HepG2/C3a cocultures, or monocultures using HPLC coupled to MS. We used the culture medium collected in the basal compartment to confirm the passage of APAP through the SK-HEP-1 barrier. The ratios of APAP (compared to the initial systemic concentration of 100 µM) recovered at the end of the experiment are provided in Figure 9D. For the SK-HEP-1 monoculture, the APAP ratio at the end of the experiment was 1.02 ± 0.07, indicating that SK-HEP-1 did not metabolise APAP. The recovered ratio, corresponding to a concentration of 100 µM, confirmed the passage of APAP through the barrier, allowing the equilibrium of APAP concentration between the apical and basal sides. In the HepG2/C3a monoculture and SK-HEP-1/HepG2/C3a coculture, the APAP ratios were 0.83 ± 0.05 and 0.87 ± 0.08, respectively, illustrating metabolism (Figure 9D). However, for both conditions, the paracetamol sulphate and paracetamol glucuronide concentrations were below detection limits.
3.5. Expression of inflammatory cytokines
The expression of inflammatory cytokines was evaluated in all culture conditions (with and without APAP) by analysing mRNA levels for TNFα, IL-1, IL-6 and IL-8 genes, and by quantifying IL-6 secretion in the culture medium. SK-HEP-1 cells expressed the four genes in all culture conditions (monoculture/coculture and APAP+/APAP-). Gene expression levels of IL8, IL6 and IL1 were similar, regardless of the culture conditions. There was a noticeable significant upregulation of TNFα in SK-HEP-1 cocultured with APAP (Figure 9E). Regarding HepG2C3a cells, there were no significant differences in expression levels of IL-8 in the conditions tested. On the other hand, there was a slight but significant overexpression of TNFα in HepG2/C3a cocultured without APAP when compared to monocultures (Figure 9F). IL-6 protein quantification in culture medium showed that it was only expressed by SK-HEP-1 cells and that HepG2/C3a monocultures with and without APAP did not produce detectable amounts of IL-6 (Figure 9G).
4. Discussion
Classic 2D in vitro coculture models consist of cells randomly mixed and heterogeneously distributed at the bottom of well-plates and dishes. However, in vivo, LSECs and hepatocytes are separated by the space of Disse which, in 3D models, is generally mimicked by a gel or collagen matrix which physically separates LSECs and hepatocytes (Bale et al., 2015). Furthermore, controlling the homotypic and heterotypic cell-to-cell interactions appears to be a key feature for maintaining and enhancing the hepatocyte phenotype (Bhatia et al., 1999; Bhatia 1997). In the present work, we have established a coculture model of liver sinusoidal endothelial cells with liver hepatocytes. Thanks to our platform which integrates a liver-on-chip solution and a barrier insert, we were able to propose technology that physically separated both cell types. In this model, cell-to-cell paracrine-like communication was made possible by exchanges through the insert membrane, as this model did not allow direct contact between LSECs and hepatocytes. Although this type of technology has already been presented for organ-to-organ models such as the intestine barrier-liver (Bricks et al., 2014), to our knowledge, only two other such dynamic LSECs barrier-hepatocyte coculture models have previously been described (van Grunsven 2017; Lauschke et al., 2019).
We demonstrated the functionality of the coculture model using two human cell lines, SK-HEP-1 and HepG2/C3a. For this purpose, we optimised the culture medium, confirmed the innocuity of the fluid flow and coculture on the LSECs barriers, and characterised the cytokine crosstalk between cells. Establishing a coculture medium that is healthy for two or more types of cells is a critical step in in vitro physiological models (Vis et al., 2020), including liver cells (Lauschke et al., 2019). Similarly, it was reported that LSECs are sensitive to serum components (Elvevold, et al., 2008). Our data demonstrated that the HepG2/C3a MEM-based medium which contained serum contributed to damaging the LSECs layer, whereas the conventional LSECs medium (also containing serum) did not. Interestingly, a mixture of the HepG2/C3a and SK-HEP-1 media led to both healthy LSECs and HepG2/C3a. Although we did not identify the specific factors leading to this result, we postulate that the presence of pro-angiogenic factors in EGM-2 medium played a part in stabilising the LSEC cultures. Interestingly, the present dynamic conditions did not affect the cell junctions or the expression levels of LSEC markers.
Endothelial cells are normally exposed to flow, and dynamic in vitro models have largely been reported as regulating their functions and physiology (Akbari et al., 2018, van Duinen et al, 2017). However, a decrease in endothelial barrier permeability was only reported in dynamic cultures coupled to high shear stress (0.7-1 Pa, van Duinen et al, 2017). In the present work, we did not observe significant variations in barrier permeability functions between static and dynamic SK-HEP-1 monocultures (Figure 4A and 6B). Conversely, the permeability was reduced in dynamic LSECs cocultures (Figure 6B), illustrating stronger cell junctions in the presence of HepG2/C3a, and suggesting that there is a synergistic effect of coculturing cells in our conditions. Furthermore, we found that the LSECs produced basal levels of the pro inflammatory cytokines IL6 and TNF without any significant morphology damage. The dynamic coculture also did not play a part in significantly increasing cytokine levels in LSECs. As high levels of production of pro-inflammatory cytokines in the liver by LSECs leads to fibrosis (DeLeve et al., 2015), our result illustrated the fact that the basal dynamic cocultures of LSECs were not pro-inflammatory.
Previous works reported an improved hepatocyte phenotype when cocultured with endothelial cells (Gugen Guillouzo et al., 2010; Xiao et al., 2015; Bale et al., 2015; Ortega‐Ribera et al., 2018). In the present model, we did not detect any striking benefit of the presence of LSECs on the HepG2/C3a phenotype (no albumin increase, no mRNA gene metabolism upregulation, no clear cytokine over secretion). In fact, the enhanced maturation of hepatocytes was mainly reported on primary hepatocytes that tend to rapidly dedifferentiate (Gugen Guillouzo et al., 2010). It is clear that the hepatocarcinoma HepG2/C3a cell line is probably not an ideal model for liver-on-chip approaches. Although it has been widely used in works related to cancer and liver disease (Donato et al., 2015), and shown that interactions between the liver endothelium (include SK-HEP-1) and this liver carcinoma were reported in studies investigating liver disorders (Thomann et al., 2020; Lee et al., 2022), it has a weak maturation profile. It is certainly a robust model for proof-of-principle studies, but the present on-chip approach would clearly benefit from being extended and refined using normal human primary hepatocytes.
Regarding liver toxicology, the liver’s in vivo features suggest that xenobiotics must first pass the endothelial barrier before accessing the hepatocytes. Analysing the kinetics and toxicity of APAP via the LSEC barrier and its subsequent metabolism inside the liver compartment was presented as a proof of concept of our technology. APAP toxicity directly on LSECs has already been reported in the literature (Badmann et al., 2012; Holt et al., 2010). The presence of APAP contributed to modifying the expression of LSECs markers in this work. This was illustrated by degradation of the actin and vimentin network, and the reduction of the PECAM-1 and STAB2 expression levels as shown in the immunostaining images. We also confirmed APAP metabolism in the presence of the HepG2/C3a cells. The barrier led to modulation of the concentration of APAP reaching the liver cells and we did not detect any particular sign of HepG2/C3a toxicity in our experiments. Consistently with the literature, the 100 µM concentration of APAP on HepG2 is not a toxic concentration, as most studies reported effects between 1 to 2 mM (Gonzales et al., 2017; Prot et al., 2012; Odeyemi et al., 2019).
The development of in vitro liver models that mimic the key elements of the in vivo liver environment is very challenging due to the complexity of liver architecture and physiology. With the progress made in tissue engineering, bioprinting and microfluidics, several microfluidic-based in vitro liver models that reproduce a physiologically relevant microenvironment have been developed in recent years (Lee et al., 2021; Moradi et al., 2021). Although most of these models are based on hepatocyte monoculture, a growing number of groups are interested in developing microfluidic cocultures of different liver cells, especially hepatocytes and LSECs, to reproduce liver sinusoid and cell-cell interactions (Kang et al., 2015; Prodanov et al., 2016; Ortega‐Ribera et al., 2018; Du et al., 2017; Lee et al., 2021; Moradi et al., 2021). In these models, the different cell types are randomly mixed or organised in layers separated by a porous membrane, collagen layer or microstructures (Bale et al., 2015; Lee et al., 2021). The present model combines the advantages of the LSECs barrier and hepatocytes cultured in 3D spheroids thanks to the hyaluronic acid hydroscaffold functionalised with ECM components. A critical issue in microfluidic culture development is the balance between model relevance, complexity and practicality. For liver cell cocultures, all cell types are usually seeded in the same irreversibly sealed microfluidic device/membrane, which makes it extremely difficult to analyse the different cell types separately (Ma et al., 2018). Our model consists of two separate compartments easily assembled in IIDMP devices: hepatocytes in PDMS biochips and the LSECs barrier in commercially standard inserts. Each cell type can be cultured and characterised separately before being connected to an IIDMP device for coculture. At the end of the experiments, the inserts and biochips can also be easily removed for separate external analyses.
Conclusion
In conclusion, we present here a new in vitro coculture model for LSECs and hepatocytes using two human cell lines. The coculture was performed inside a platform integrating a cell culture insert to build the endothelial barrier, and liver organ-on-chip technology. The platform made dynamic perfusion possible and reproduced cell-to-cell communications. The biological characterisation confirmed that the integrity and functionality of the LSECs were not altered by either the perfusion or the coculture conditions. Furthermore, the model showed lower LSECs permeability that increased the barrier functions. However, we did not detect any particular effects of the LSECs on the HepG2/C3a phenotypes. Finally, we successfully demonstrated APAP modulation on the LSEC phenotype, its transit to the liver compartment and its metabolism, as an example of a liver first pass metabolism. We believe that our model could be used as a relevant model for investigating drug kinetics and subsequent physio-pathological hepatotoxicity.
Author Contributions
Conceptualization, T.M., A.L.G., C.L., E.L. and R.J.; methodology, T.M., A.L.G., F.S., and R.J.; validation, A.L.G., F.S., C.L., E.L. and R.J.; formal analysis, T.M.; investigation, T.M.; resources, Z.S., F.M. and N.M.; data curation, A.L.G., F.S., E.L. and R.J.; writing original draft preparation, T.M.; writing review and editing, T.M., A.L.G., F.S., E.L., and R.J.; visualization, T.M.; supervision, A.L.G., E.L. and R.J.; project administration, C.L.; funding acquisition, C.L. All authors have read and agreed to the published version of the manuscript.
Acknowledgements
This work and T. Messelmani PhD were funded by the ANR (Agence Nationale de la Recherche, France) through the MIMLIVEROnChip project, grant number ANR-19-CE19-0020-01, and by a grant from the Contrat de Plan Etat-Région (CPER) Cancer 2015–2020. The authors would like to thank the JSPS Core-to-Core Program (JPJSCCA20190006), the Research Department at the Université de Technologie de Compiègne (Research Training Innovation Chair, DOT project - Disruptive Organoids Technologies) and CNRS (CNRS international research team, TEAMS project – Therapeutics and Engineering Against Metabolic Syndrome, between CNRS/UTC BMBI and CNRS/IIS LIMMS) for their support.
Conflict of Interest
HCS pharma is the BIOMIMESYS® Liver owner and is a partner of the ANR MimLiverOnChip (ANR-19-CE19-0020-01). Co-authors Z. Souguir and N. Maubon are employees of HCS Pharma.
T. Messelmani, A. Le Goff, F. Soncin, F. Merlier, C. Legallais, E. Leclerc and R. Jellali declare no conflict of interest.
Data Availability Statement
All data generated or analysed during this study are included in this published article and its additional files.
References
Akbari, E., Spychalski, G. B., Rangharajan, K. K., Prakash, S., & Song, J. W. (2018). Flow dynamics control endothelial permeability in a microfluidic vessel bifurcation model.
Lab on a Chip, 18(7), 1084–1093.
https://doi.org/10.1039/c8lc00130h.
Badmann, A., Langsch, S., Keogh, A., Brunner, T., Kaufmann, T., & Corazza, N. (2012). TRAIL enhances paracetamol-induced liver sinusoidal endothelial cell death in a Bim- and Bid-dependent manner.
Cell Death & Disease, 3(12), e447.
https://doi.org/10.1038/cddis.2012.185.
Bale, S. S., & Borenstein, J. T. (2018). Microfluidic Cell Culture Platforms to Capture Hepatic Physiology and Complex Cellular Interactions.
Drug Metabolism and Disposition, 46(11), 1638–1646.
https://doi.org/10.1124/dmd.118.083055.
Bale, S. S., Geerts, S., Jindal, R., & Yarmush, M. L. (2016). Isolation and co-culture of rat parenchymal and non-parenchymal liver cells to evaluate cellular interactions and response.
Scientific Reports, 6, 25329.
https://doi.org/10.1038/srep25329.
Bale, S. S., Golberg, I., Jindal, R., McCarty, W. J., Luitje, M., Hegde, M., Bhushan, A., Usta, O. B., & Yarmush, M. L. (2015). Long-term coculture strategies for primary hepatocytes and liver sinusoidal endothelial cells.
Tissue engineering. Part C, Methods, 21(4), 413–422.
https://doi.org/10.1089/ten.TEC.2014.0152.
Beckwitt, C. H., Clark, A. M., Wheeler, S., Taylor, D. L., Stolz, D. B., Griffith, L., & Wells, A. (2018). Liver 'organ on a chip'.
Experimental Cell Research, 363(1), 15–25.
https://doi.org/10.1016/j.yexcr.2017.12.023.
Bhatia, S. N., Balis, U. J., Yarmush, M. L., & Toner, M. (1999). Effect of cell-cell interactions in preservation of cellular phenotype: cocultivation of hepatocytes and nonparenchymal cells.
FASEB journal, 13(14), 1883–1900.
https://doi.org/10.1096/fasebj.13.14.1883.
Bricks, T., Hamon, J., Fleury, M. J., Jellali, R., Merlier, F., Herpe, Y. E., Seyer, A., Regimbeau, J. M., Bois, F., & Leclerc, E. (2015). Investigation of omeprazole and phenacetin first-pass metabolism in humans using a microscale bioreactor and pharmacokinetic models.
Biopharmaceutics & drug disposition, 36(5), 275–293.
https://doi.org/10.1002/bdd.1940.
Bricks, T., Paullier, P., Legendre, A., Fleury, M. J., Zeller, P., Merlier, F., Anton, P. M., & Leclerc, E. (2014). Development of a new microfluidic platform integrating co-cultures of intestinal and liver cell lines.
Toxicology in vitro, 28(5), 885–895.
https://doi.org/10.1016/j.tiv.2014.02.005.
Donato, M. T., Tolosa, L., & Gómez-Lechón, M. J. (2015). Culture and Functional Characterization of Human Hepatoma HepG2 Cells.
Methods in Molecular Biology, 1250, 77–93.
https://doi.org/10.1007/978-1-4939-2074-7_5.
Du, Y., Li, N., Yang, H., Luo, C., Gong, Y., Tong, C., Gao, Y., Lü, S., & Long, M. (2017). Mimicking liver sinusoidal structures and functions using a 3D-configured microfluidic chip.
Lab on a Chip, 17(5), 782–794.
https://doi.org/10.1039/c6lc01374k.
Elvevold, K., Smedsrød, B., & Martinez, I. (2008). The liver sinusoidal endothelial cell: a cell type of controversial and confusing identity.
American Journal of Physiology. Gastrointestinal and Liver Physiology, 294(2), G391–G400.
https://doi.org/10.1152/ajpgi.00167.2007.
González, L. T., Minsky, N. W., Espinosa, L. E., Aranda, R. S., Meseguer, J. P., & Pérez, P. C. (2017). In vitro assessment of hepatoprotective agents against damage induced by acetaminophen and CCl4.
BMC Complementary and Alternative Medicine, 17(1), 39.
https://doi.org/10.1186/s12906-016-1506-1.
Holt, M. P., Yin, H., & Ju, C. (2010). Exacerbation of acetaminophen-induced disturbances of liver sinusoidal endothelial cells in the absence of Kupffer cells in mice. Toxicology letters, 194(1-2), 34–41.
https://doi.org/10.1016/j.toxlet.2010.01.020.
Jellali, R., Bricks, T., Jacques, S., Fleury, M. J., Paullier, P., Merlier, F., & Leclerc, E. (2016). Long-term human primary hepatocyte cultures in a microfluidic liver biochip show maintenance of mRNA levels and higher drug metabolism compared with Petri cultures.
Biopharmaceutics & Drug Disposition, 37(5), 264–275.
https://doi.org/10.1002/bdd.2010.
Jellali, R., Paullier, P, Fleury, M. J., & Leclerc, E., (2016) Liver and kidney cells cultures in a new perfluoropolyether biochip.
Sensors and Actuators B: Chemical, 229, 396-407.
https://doi.org/10.1016/j.snb.2016.01.141.
Kang, Y. B., Sodunke, T. R., Lamontagne, J., Cirillo, J., Rajiv, C., Bouchard, M. J., & Noh, M. (2015). Liver sinusoid on a chip: Long-term layered co-culture of primary rat hepatocytes and endothelial cells in microfluidic platforms.
Biotechnology and Bioengineering, 112(12), 2571–2582.
https://doi.org/10.1002/bit.25659.
Khetani, S. R., Berger, D. R., Ballinger, K. R., Davidson, M. D., Lin, C., & Ware, B. R. (2015). Microengineered liver tissues for drug testing.
Journal of Laboratory Automation, 20(3), 216–250.
https://doi.org/10.1177/2211068214566939.
Lauschke, V. M., Shafagh, R. Z., Hendriks, D. F. G., & Ingelman-Sundberg, M. (2019). 3D primary hepatocyte culture systems for analyses of liver diseases, drug metabolism, and toxicity: emerging culture paradigms and applications.
Biotechnology Journal, 14(7), e1800347.
https://doi.org/10.1002/biot.201800347.
LeCluyse, E. L., Witek, R. P., Andersen, M. E., & Powers, M. J. (2012). Organotypic liver culture models: meeting current challenges in toxicity testing.
Critical Reviews in Toxicology, 42(6), 501–548.
https://doi.org/10.3109/10408444.2012.682115.
Lee, S. Y., Kim, D., Lee, S. H., & Sung, J. H. (2021). Microtechnology-based in vitro models: Mimicking liver function and pathophysiology.
APL Bioengineering, 5(4), 041505.
https://doi.org/10.1063/5.0061896.
Lee, D., Park, J. S., Kim, D., & Hong, H. S. (2022). Substance P hinders bile acid-induced hepatocellular injury by modulating oxidative stress and inflammation.
Antioxidants, 11(5), 920.
https://doi.org/10.3390/antiox11050920.
Livak, K. J., & Schmittgen, T. D. (2001). Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method.
Methods, 25(4), 402–408.
https://doi.org/10.1006/meth.2001.1262.
Ma, L. D., Wang, Y. T., Wang, J. R., Wu, J. L., Meng, X. S., Hu, P., Mu, X., Liang, Q. L., & Luo, G. A. (2018). Design and fabrication of a liver-on-a-chip platform for convenient, highly efficient, and safe in situ perfusion culture of 3D hepatic spheroids.
Lab on a Chip, 18(17), 2547–2562.
https://doi.org/10.1039/c8lc00333e.
Merlier, F., Jellali, R., & Leclerc, E. (2017). Online monitoring of hepatic rat metabolism by coupling a liver biochip and a mass spectrometer.
The Analyst, 142(19), 3747–3757.
https://doi.org/10.1039/c7an00973a.
Messelmani, T., Le Goff, A., Souguir, Z., Maes, V., Roudaut, M., Vandenhaute, E., Maubon, N., Legallais, C., Leclerc, E., & Jellali, R. (2022). Development of liver-on-chip integrating a hydroscaffold mimicking the liver's extracellular matrix.
Bioengineering, 9(9), 443.
https://doi.org/10.3390/bioengineering9090443.
Messelmani, T., Morisseau, L., Sakai, Y., Legallais, C., Le Goff, A., Leclerc, E., & Jellali, R. (2022). Liver organ-on-chip models for toxicity studies and risk assessment.
Lab on a Chip, 22(13), 2423–2450.
https://doi.org/10.1039/d2lc00307d.
Milner, E., Ainsworth, M., McDonough, M., Stevens, B., Buehrer, J., Delzell, R., Wilson, C., & Barnhill, J. (2020) Emerging three-dimensional hepatic models in relation to traditional two-dimensional in vitro assays for evaluating drug metabolism and hepatoxicity.
Medicine in Drug Discovery, 8, 100060.
https://doi.org/10.1016/j.medidd.2020.100060.
Odeyemi, S., & Dewar, J. (2019). Repression of acetaminophen-induced hepatotoxicity in HepG2 cells by polyphenolic compounds from Lauridia tetragona (L.f.) R.H. Archer.
Molecules, 24(11), 2118.
https://doi.org/10.3390/molecules24112118.
Ortega-Ribera, M., Fernández-Iglesias, A., Illa, X., Moya, A., Molina, V., Maeso-Díaz, R., Fondevila, C., Peralta, C., Bosch, J., Villa, R., & Gracia-Sancho, J. (2018). Resemblance of the human liver sinusoid in a fluidic device with biomedical and pharmaceutical applications.
Biotechnology and bioengineering, 115(10), 2585–2594.
https://doi.org/10.1002/bit.26776.
Polidoro, M. A., Ferrari, E., Marzorati, S., Lleo, A., & Rasponi, M. (2021). Experimental liver models: From cell culture techniques to microfluidic organs-on-chip.
Liver International, 41(8), 1744–1761.
https://doi.org/10.1111/liv.14942.
Prodanov, L., Jindal, R., Bale, S. S., Hegde, M., McCarty, W. J., Golberg, I., Bhushan, A., Yarmush, M. L., & Usta, O. B. (2016). Long-term maintenance of a microfluidic 3D human liver sinusoid.
Biotechnology and Bioengineering, 113(1), 241–246.
https://doi.org/10.1002/bit.25700.
Prot, J. M., Bunescu, A., Elena-Herrmann, B., Aninat, C., Snouber, L. C., Griscom, L., Razan, F., Bois, F. Y., Legallais, C., Brochot, C., Corlu, A., Dumas, M. E., & Leclerc, E. (2012). Predictive toxicology using systemic biology and liver microfluidic "on chip" approaches: application to acetaminophen injury.
Toxicology and applied pharmacology, 259(3), 270–280.
https://doi.org/10.1016/j.taap.2011.12.017.
Ruoß, M., Vosough, M., Königsrainer, A., Nadalin, S., Wagner, S., Sajadian, S., Huber, D., Heydari, Z., Ehnert, S., Hengstler, J. G., & Nussler, A. K. (2020). Towards improved hepatocyte cultures: Progress and limitations.
Food and Chemical Toxicology, 138, 111188.
https://doi.org/10.1016/j.fct.2020.111188.
Schepers, A., Li, C., Chhabra, A., Seney, B. T., & Bhatia, S. (2016). Engineering a perfusable 3D human liver platform from iPS cells.
Lab on a Chip, 16(14), 2644–2653.
https://doi.org/10.1039/c6lc00598e.
Soldatow, V. Y., Lecluyse, E. L., Griffith, L. G., & Rusyn, I. (2013). In vitro models for liver toxicity testing.
Toxicology Research, 2(1), 23–39.
https://doi.org/10.1039/C2TX20051A.
Son, Y. W., Choi, H. N., Che, J. H., Kang, B. C., & Yun, J. W. (2020). Advances in selecting appropriate non-rodent species for regulatory toxicology research: Policy, ethical, and experimental considerations.
Regulatory Toxicology and Pharmacology, 116, 104757.
https://doi.org/10.1016/j.yrtph.2020.104757.
Souguir, Z., Vidal, G., Demange, E., & Louis, F., (2016) WO Pat., 2016166479A1.
Thomann, S., Weiler, S. M. E., Marquard, S., Rose, F., Ball, C. R., Tóth, M., Wei, T., Sticht, C., Fritzsche, S., Roessler, S., De La Torre, C., Ryschich, E., Ermakova, O., Mogler, C., Kazdal, D., Gretz, N., Glimm, H., Rempel, E., Schirmacher, P., & Breuhahn, K. (2020). YAP orchestrates heterotypic endothelial cell communication via HGF/c-MET signaling in liver tumorigenesis.
Cancer Research, 80(24), 5502–5514.
https://doi.org/10.1158/0008-5472.CAN-20-0242.
van Duinen, V., van den Heuvel, A., Trietsch, S. J., Lanz, H. L., van Gils, J. M., van Zonneveld, A. J., Vulto, P., & Hankemeier, T. (2017). 96 perfusable blood vessels to study vascular permeability in vitro.
Scientific Reports, 7(1), 18071.
https://doi.org/10.1038/s41598-017-14716-y.
Vis, M. A. M., Ito, K., & Hofmann, S. (2020). Impact of Culture Medium on Cellular Interactions in in vitro Co-culture Systems.
Frontiers in Bioengineering and Biotechnology, 8, 911.
https://doi.org/10.3389/fbioe.2020.00911.
Xiao, W., Perry, G., Komori, K., & Sakai, Y. (2015). New physiologically-relevant liver tissue model based on hierarchically cocultured primary rat hepatocytes with liver endothelial cells.
Integrative Biology, 7(11), 1412–1422.
https://doi.org/10.1039/c5ib00170f.
Yu, F., Deng, R., Hao Tong, W., Huan, L., Chan Way, N., IslamBadhan, A., Iliescu, C., & Yu, H. (2017). A perfusion incubator liver chip for 3D cell culture with application on chronic hepatotoxicity testing.
Scientific Reports, 7(1), 14528.
https://doi.org/10.1038/s41598-017-13848-5.
Zeller, P., Legendre, A., Jacques, S., Fleury, M. J., Gilard, F., Tcherkez, G., & Leclerc, E. (2017). Hepatocytes cocultured with Sertoli cells in bioreactor favors Sertoli barrier tightness in rat.
Journal of Applied Toxicology, 37(3), 287–295.
https://doi.org/10.1002/jat.3360.
Figure 1. Experimental procedures for SK-HEP-1 and HepG2/C3A monoculture and coculture.