Water-soluble oleic acid (OA; catalog #O1257-10 mg), water-soluble free cholesterol (#C4951-30 MG), staurosporine (#S6942), fetal bovine serum (FBS; #F2442), taurocholic acid (TCA; #T4009), and estradiol-17β-glucuronide (E₂17G; #E1127) were purchased from Sigma-Aldrich (St. Louis, MO). Sterile water was used to dissolve the water-soluble compounds. 1-Oleoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (18:1 LysoPE; #846725), 1-oleoyl-2-hydroxy-sn-glycero-3-phosphocholine (18:1 LysoPC; #845875), 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (16:0 LysoPE; #856705), and 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphocholine (16:0 LysoPC; #855675) were purchased from Avanti Polar Lipids (Alabaster, AL). Tumor necrosis factor alpha (TNF-α; #300-01A) and interleukin-6 (IL-6; #200–06) were purchased from PeproTech (Cranbury, NJ). Palmitic acid (PA; #10006627), dexamethasone (#11015), and imatinib (#13139) were purchased from Cayman Chemicals (Ann Arbor, MI). BODIPY™ 493/503 (#D3922), BODIPY™ 558/568 C₁₂ (#D3835), BODIPY™ 665/676 (#B3932), MitoTracker™ Orange CM-H2TMRos (#M7511), CellEvent™ Caspase-3/7 Detection Reagent (#C10423), Hoechst Blue 33342 Solution (#62249), Dulbecco’s modified Eagle’s medium (DMEM: #11995-065), Williams’ Medium E without Phenol Red (#A1217601), penicillin–streptomycin (#15140122), Ca²⁺- and Mg²⁺-free Dulbecco’s phosphate buffered saline (PBS; #14190144), Hank’s Balanced Salt Solution (standard HBSS; #14025–092), and Ca²⁺- and Mg²⁺-free HBSS (Ca²⁺-free HBSS; #14175–095) were purchased from Thermo Fisher Scientific (Waltham, MA). Anhydrous ≥99.9% dimethyl sulfoxide (DMSO; #D12345) used in the dissolution of hygroscopic compounds (e.g., lysophospholipids, BODIPY™ probes) was purchased from Thermo Fisher Scientific, and ≥99.5% DMSO (#41639) used for cell culture was purchased from Sigma-Aldrich. Cryopreserved Transporter Certified™ human hepatocytes from three different donors (JEL, WID, and IWM; Supplementary Table 1) and QualGro™ seeding, thawing, overlay, and maintenance media were purchased from BioIVT (Baltimore, MD). Collagen I-coated BioCoat™ plates (24-well; #356408) were purchased from Corning Life Sciences (Tewksbury, MA). [³H]-TCA (#NET322250UC, 6.5 Ci/mmol, radiochemical purity >97%), and [³H]-E₂17G (#NET1106250UC, 52.9 Ci/mmol, radiochemical purity >97%) were purchased from PerkinElmer Life Sciences (Boston, MA).

Table 1 summarizes all the selected treatment components and the corresponding concentrations evaluated within each in vitro system. Lysophospholipid notation “18:1” indicates a fatty acid chain of 18 carbon atoms with one double bond, whereas “16:0” indicates a fatty acid chain of 16 carbon atoms with zero double bonds. A final media concentration of 0.5% DMSO was selected to limit any potential impact on human hepatocyte gene expression or viability (Sumida et al., 2011). OA, PA, and lysophospholipid concentrations were kept constant and are collectively referred to as the “lipid mix.” All treatments were included in the cell culture media for 72 h before endpoint assays, which is a previously optimized duration between collagen overlay and functional assay assessment in SCHH (Yang et al., 2016). Positive toxicity controls (staurosporine [500 nM] for differentiated HuH-7 cell culture and imatinib [40 μM] for SCHH) were incubated in the cell culture media for 24 h before endpoint assays.

The initial optimization of the in vitro MASH culture model utilized a previously described differentiated HuH-7 cell system that displays hepatocyte-like morphology and bile canalicular-like formation, and is amenable to economical high-throughput testing (Saran et al., 2022a). The HuH-7 cell line (JCRB0403) was acquired from Sekisui Xenotech. Cell line identity was confirmed by the UNC Vironomics Core. HuH-7 cell cultures were maintained and passaged in T-25 (Sarstedt [Newton, NC] #83.3910) or T-75 (Sarstedt #83.3911) cell culture flasks using maintenance media (DMEM supplemented with 10% FBS and 1% penicillin–streptomycin) until fully confluent. On culture day 0, HuH-7 cells were seeded at 0.075 million cells/well in a 96-well collagen type-I-treated opaque-walled plate (Greiner Bio-One #655956). On culture day 2, maintenance medium was supplemented with 1 μM dexamethasone and 0.5% DMSO (differentiation media). On culture day 7, the cells were overlaid with 0.25 mg/mL Matrigel Basement Membrane Matrix Phenol Red-Free (Corning #356237, lot #3068002) in ice-cold differentiation media. The culture was then maintained for one additional week, with differentiation media replaced every 2–3 days. To ensure optimal hepatocyte-like characteristics, MASH culture treatments were initiated on day 14 of culture. This occurred 1 week after Matrigel overlay (Saran et al., 2022a). Additional amounts of DMSO were added to each treatment to ensure a final working concentration of 0.5% for comparison to the non-MASH-inducing treatment control group that contained 0.5% DMSO in differentiation media. Treatment and control differentiation media were replaced every 24 h for 72 h on culture days 14–16. All terminal cellular toxicity and phenotypic endpoints were assessed on day 17 of culture. All HuH-7 cell cultures were between passage numbers 16–24.

Following the identification of optimal treatment concentrations using the differentiated HuH-7 cell culture model, MASH culture treatments were tested in SCHH. On culture day 0, cryopreserved Transporter Certified^® human hepatocytes were thawed using QualGro™ thawing media, diluted to 0.9–1.0 million cells/mL in QualGro seeding media, and seeded at a density of 0.45–0.5 million cells/well in 24-well BioCoat Collagen-I-coated plates. Approximately 16 h–20 h following seeding (culture day 1), cells were overlaid with 0.25 mg/mL of Matrigel Basement Membrane Matrix Phenol Red-Free (Corning #356237, lots #3068002, 3068003) in ice-cold QualGro maintenance media. The following day (culture day 2), treatments were dissolved in QualGro maintenance media supplemented with additional DMSO to reach a final working concentration of 0.5%. Treatments and control QualGro maintenance media supplemented with 0.5% DMSO were administered to SCHH on the same day. Treatment and control maintenance media were exchanged every 24 h on culture days 2–4. All terminal cell toxicity and phenotypic endpoints were assessed on culture day 5.

Following the culmination of the 72-h treatment on culture day 17 (differentiated HuH-7 cells) or day 5 (SCHH), phenotypic endpoints were assessed using fluorescent probes. Given their documented roles in the progression of MASL to MASH (Ramanathan et al., 2022; Pierantonelli and Svegliati-Baroni, 2019), the following four endpoints (fluorescent probes) were evaluated:

Intracellular neutral lipid accumulation/lipid droplet formation (BODIPY™ 493/503).

A Hoechst Blue 33342 stain was used to identify cell nuclei. Fluorescent probes were evaluated using fluorescence-based high-content imaging (HCI) with the CellInsight™ CX7 LZR High Content Analysis platform (Thermo Fisher Scientific #CX7A1110LZR). HCI was performed using a 10X imaging objective, with the HCS Studio™ Cell Analysis Software (Thermo Fisher Scientific; version 6.6.1) used for cell identification, subsequent fluorescence object detection/surface rendering, and automated data analysis. Specifically, the fluorescence signal from each probe was rendered into a two-dimensional object. The Hoechst Blue nuclei stain was used to detect cell nuclei, followed by the application of an optimized software-based algorithm to estimate the surrounding cell area. All fluorescence-based HCI data are reported as the mean rendered surface area (μm²) per cell or the mean rendered object count per cell (see caspase-3/7 detection assay below).

To perform fluorescent probe and Hoechst Blue staining, treatment and control media were aspirated on the terminal culture day and cells were washed twice with PBS. Fluorescent probe solutions (Supplementary Table 2) were freshly prepared in FBS-free DMEM (for differentiated HuH-7 cells) or Williams’ Medium E (for SCHH) and added to the cells at 37 °C/5% CO₂ for 30 min–45 min with the plate covered in tin foil to minimize light exposure. For the BODIPY™ 558/568 C₁₂ probe, cells were incubated in the absence of culture treatments for 24 h to allow for adequate FA cellular uptake per manufacturer instructions. Then, the FBS-free dosing solutions were aspirated and replaced with FBS-free media and immediately imaged using appropriate laser emission and excitation wavelengths (Supplementary Table 2). In differentiated HuH-7 cells, six replicate measures were obtained for each treatment and control group, whereas in SCHH, three replicate measures per donor were obtained.

Early-stage cellular apoptosis was assessed using a CellEvent™ Caspase-3/7 Detection Reagent with the CellInsight™ CX7 High-Content Screening platform (10X objective) using the same fluorescent probe staining and imaging protocol described above. The PBS wash step was omitted to prevent the detachment of apoptotic cells, and the probe solution was incubated for 45 min as per manufacturer instructions. Cell viability/ATP content was determined with the CellTiter-Glo^® Luminescent Cell Viability Assay (Promega #G7570) using a PowerWave XS Microplate spectrophotometer (BioTek Instruments). A predetermined “threshold” luminescence value relative to the control of 70% was used to define the undesirable cellular toxicity for this assay. Although no toxicity threshold values have been established specifically for the ATP assay, a cell viability threshold of at least <75% was used to avoid false positives when assessing cytotoxicity based on previous literature (Azqueta et al., 2022). In differentiated HuH-7 cells, six replicate measures were obtained for each treatment and control group, whereas in SCHH, three replicate measures per donor were obtained.

For toxicity endpoints, a one-way ANOVA adjusted for multiple comparisons using Dunnett’s test was performed to compare each treatment group to the control. For phenotypic endpoints evaluated in HuH-7 cells, a one-way ANOVA with a post hoc Tukey’s test was used to facilitate comparison across all treatment groups. Specifically, after confirming an ANOVA F statistic of p < 0.05, comparisons between treatments with and without cytokines and different concentrations of cholesterol in the presence of the lipid mix were evaluated for significance (p < 0.05). Statistical testing was performed in GraphPad Prism (version 10.1.2). For phenotypic endpoints in SCHH, a linear mixed-effects model was fit for each outcome with a random slope on hepatocyte donor to account for inter-donor variability. Models describing the lipid droplet formation outcome (BODIPY™ 493/503) were fit using a gamma distribution and log link function (lme4 package (Bates et al., 2015)) in RStudio (R version 4.3). Models describing the lipid peroxidation outcome (BODIPY™ 558/568) were fit using a Gaussian distribution. When comparing treatments with and without cytokines for both outcomes, models were fit using a Gaussian distribution.

SCHH membrane protein fractions were extracted following the 72-h treatment period (culture day 5) for QTAP analysis using a previously described differential detergent fractionation (DDF) method (Qasem et al., 2021). Briefly, SCHH cultured at 450,000 cells/well were washed twice with 1 mL of PBS containing 1 mM phenylmethylsulfonyl fluoride. To release non-membrane bound intracellular content, 250 μL of digitonin buffer (Qasem et al., 2021) was added to each well, and the cells were incubated at 4 °C for 10 min under gentle agitation. The digitonin buffer, now enriched with cytosolic proteins (cytosolic protein fraction), was transferred to 1.5 mL tubes and stored at −80 °C (no analysis was performed on the cytosolic protein fraction in this study). To extract the membrane fraction, 150 μL of Triton X-100 buffer (Qasem et al., 2021) was added to each well and incubated at 4 °C for 30 min under gentle agitation. The Triton X-100 buffer, now enriched with cellular membrane proteins, was collected. The Pierce™ bicinchoninic acid Protein Assay Kit (#23225, Thermo Fisher Scientific) was used to quantify the total amount of protein in each sample. Membrane protein recovery per 450,000 cells ranged as follows for each hepatocyte donor across all sample treatments and replicates: 57.9 μg–68.0 μg (JEL), 70.7 μg–77.3 μg (WID), and 39.0 μg–56.6 μg (IWM). Ultimately, 20 μg of membrane protein was used from each sample for QTAP analysis. Triplicate samples were obtained for each treatment (three lipid–cytokine treatments + non-treated control [0.5% DMSO]) and donor (JEL, WID, and IWM) combination (12 samples per donor).

As previously described (Khatri et al., 2019; Fashe et al., 2022; Fallon et al., 2013), membrane protein was mixed with stable isotope labeled (SIL) proteotypic human peptide standards representing the transport and DME proteins of interest. The mixture was digested with trypsin (1:20 trypsin: protein ratio) and then recovered using solid-phase extraction. For QTAP analysis, 0.08 μg protein (0.2 µL volume) of each 20 µg digest was injected onto the micro LC-MS/MS system. The system consisted of an M-Class Acquity (Waters, Milford, MA) (microflow LC) coupled to a SCIEX (Framingham, MA) Triple Quadrupole 7500 mass spectrometer operated in the positive multiple reaction monitoring (MRM) mode. The M-Class Acquity trapping column was a Waters nanoEase™ M/Z Symmetry C18, 100Å, 5 μm, 180 μm × 20 mm column (part no. 186008821), and the analytical column was a Waters nanoEase™ M/Z Peptide BEH C18, 130Å, 1.7 µm, 300 μm × 100 mm column (part no. 186009258). The injection run time was 35 min. A QTAP laboratory human liver microsomes quality control (Gentest™; pool of 50) (Discovery Life Sciences, Huntsville, AL) was analyzed with the samples to ensure batch validity. MRM peak areas were determined by Analytics software embedded in SCIEX OS software (version 3.1.6.44), which was used for system control of the instrumentation. The SIL peptides used to report the protein concentration in the analyses and the MRMs for each (two MRMs for unlabeled [endogenous] peptides and two for each corresponding SIL peptides) are shown in Supplementary Table 3. These peptides have been used in previous studies (Khatri et al., 2019; Ohtsuki et al., 2012; Fallon et al., 2013; Fashe et al., 2022). Equal response of the unlabeled and SIL peptides was assumed. The ratio of the sum of the unlabeled MRM peak areas to the sum of the SIL MRM peak areas was calculated, and this was used to determine the peptide concentration based on the amount of SIL peptide added (0.5 pmol of each). For the bile salt export pump (BSEP) peptide, only one MRM transition was adequately detected for some samples from donors JEL and WID. For these samples, the peak area ratio based on that one MRM (unlabeled and SIL) was used to calculate the peptide concentration. Data for other representative peptides (most often one extra peptide) for many of the proteins were also acquired (data not shown), and these data were used as confirmatory data. Proteomics data are available in ProteomeXchange via the proteomics identification (PRIDE) database (Perez-Riverol et al., 2022) with accession number PXD071083. The lower limit of quantification (LLOQ) was set at 0.1 pmol/mg protein. For data analysis, protein concentrations between 0.1 pmol/mg protein and 0.02 pmol/mg protein (the lower limit of detection) were reported as determined, whereas concentrations below 0.02 pmol/mg protein were imputed as 0.02 pmol/mg protein. RStudio (version 2024.04.2 + 764; R version 4.4) was used to convert the peak area ratios to peptide (and, therefore, protein) concentrations (pmol/mg protein). To assess treatment-specific effects on transporter and DME protein concentrations, a linear mixed-effects model was fit for each protein with a random intercept on hepatocyte donor to account for inter-donor variability. Where appropriate, models were fit using a Gaussian distribution (lme4 package (Bates et al., 2015)) in RStudio (R version 4.3), whereas p values, model predicted mean values, and 95% confidence intervals for each treatment and control group were extracted using the lmerTest package (Kuznetsova et al., 2017). For some of the analytes, the donor effects were so strong that a linear mixed-effects model created negative concentration predictions. For those models, we shifted to using a log link function for the linear mixed models, as opposed to the typical identity link function. Gaussian or gamma distributions were used for the models depending on the distribution of the raw data. GraphPad Prism was used to plot the raw data along with the model-derived predicted mean values and 95% confidence intervals; p < 0.05 was used to define statistical significance. Using model-predicted mean values, the mean fold-change across three hepatocyte donors compared to the control (0.5% DMSO) was determined and converted to log₂ fold-change for additional comparisons.

Following the 72-h treatment period on culture day 5, the biliary excretion index (BEI) of [³H]-TCA and [³H]-E₂17G in SCHH was measured using B-CLEAR^® technology (Swift et al., 2010; Liu et al., 1999; Ito et al., 2020). Standard HBSS, termed “cells + bile” or plus (+) buffer, contained Ca²⁺ and Mg²⁺, supporting normal cellular functions, whereas Ca²⁺-free HBSS, termed “cells” or minus (−) buffer, disrupts tight junctions. As described previously (Saran et al., 2022b), cells underwent two washes with either standard or Ca²⁺-free HBSS, with the latter containing 1 mM egtazic acid, before pre-incubation for 10 min at 37 °C. Post-incubation, cells were treated with 2 μM of [³H]-TCA (200 nCi/mL) or 0.3 μM of [³H]-E₂17G (2 μCi/mL) for 10 min at 37 °C in standard HBSS, followed by three washes in ice-cold HBSS before freezing at −20 °C. Cell lysis was then performed with 0.5% Triton X-100% and 0.005% Antifoam-A in PBS. Radioactivity of cell lysates was measured using a Bio-Safe II counting cocktail (Research Products International; Mount Prospect, IL) and the Tri-Carb 3100 TR liquid scintillation analyzer (PerkinElmer Life Sciences). The total protein content was determined using the Pierce bicinchoninic acid Protein Assay Kit. [³H]-TCA and [³H]-E₂17G accumulation in SCHH was normalized to the protein content. The BEI (Equation 1), apparent in vitro uptake clearance (CL_uptake,app) (Equation 2), and apparent in vitro biliary clearance (CL_biliary,app) (Equation 3) were calculated as follows: B E I   % = A c c u m u l a t i o n C e l l s + B i l e − A c c u m u l a t i o n C e l l s A c c u m u l a t i o n C e l l s + B i l e   X   100 (1) C L u p t a k e , a p p = A c c u m u l a t i o n C e l l s + B i l e I n c u b a t i o n T i m e   x   C o n c e n t r a t i o n m e d i a (2) C L b i l i a r y , a p p = A c c u m u l a t i o n C e l l s + B i l e − A c c u m u l a t i o n C e l l s I n c u b a t i o n T i m e   x   C o n c e n t r a t i o n m e d i a (3)

Triplicate samples were obtained for each treatment (three lipid–cytokine treatments + non-treated control [0.5% DMSO]) and donor (JEL, WID, and IWM) combination (12 samples per donor). A one-way ANOVA adjusted for multiple comparisons using Tukey’s HSD test was performed across treatments for each donor.

Various concentrations of free cholesterol, TNF-α, and IL-6 were investigated independently and in combination with each other and the lipid mix in differentiated HuH-7 cells (Table 1). Despite TNF-α and IL-6 demonstrating minimal toxicity at respective concentrations of 25 ng/mL and 30 ng/mL in the HuH-7 cells (Supplementary Figure 1), the concentrations of 12.5 ng/mL and 15 ng/mL were used for initial phenotypic endpoint assessment due to a previous report of cytokine toxicity in primary human hepatocytes at concentrations of 20 ng/mL or higher (Gramignoli et al., 2022). Free cholesterol at a concentration of 0.5 mM in combination with cytokines and the lipid mix significantly increased cellular apoptosis in differentiated HuH-7 cells despite ATP levels remaining above the 70% threshold with this treatment (Figure 1). Therefore, cholesterol concentrations of 0.25 mM or lower were used for phenotypic endpoint assessment in this cell system.

HCI data indicated that a free cholesterol concentration of 0.25 mM enhanced intracellular lipid droplet formation and accumulation and FA transport/uptake (Figures 2A,B). Specifically, significant increases in FA uptake were observed in the lipid mix treatments containing 0.25 mM cholesterol (±cytokines) compared with treatments with 0.05 or 0.1 mM cholesterol (±cytokines). Similarly, lipid droplet formation significantly increased across these comparisons. However, no significant difference was observed between 0.25 mM cholesterol + cytokines and 0.05 mM cholesterol + cytokines. Additionally, there was a statistically significant increase in FA uptake with lipid mix + 0.1 mM cholesterol + cytokines compared with lipid mix + 0.05 mM cholesterol + cytokines. The addition of 12.5 ng/mL TNF-α and 15 ng/mL IL-6 to each lipid mix + cholesterol-containing treatment was associated with a decrease in mitochondrial function, indicating mitochondrial stress (Figure 2C). An increase in lipid peroxidation was also observed with the addition of cytokines to all treatments, except for treatment with 0.5 mM OA:PA (1:2) (Figure 2D). The observed increases were statistically significant for the 0.05 and 0.25 mM cholesterol concentrations but not for the 0.1 mM (p = 0.08) cholesterol concentration. Furthermore, significant increases in FA uptake were observed when cytokines were added to the lipid mix + 0.1 or 0.25 mM cholesterol treatments (Figure 2B). However, only marginal increases were noted in intracellular lipid droplet formation when cytokines were added to all treatments containing the lipid mix and cholesterol (Figure 2A). Based on promising MASH phenotype and toxicity profiles in the differentiated HuH-7 cell culture, lipid mix concentrations were kept constant while 0.05 mM–0.25 mM cholesterol, 12.5 ng/mL TNF-α, and 15 ng/mL IL-6 were selected for the initial testing in SCHH.

During an initial assessment in SCHH (data not shown), undesirable cytotoxicity (relative ATP luminescence value below 70%) was observed with lipid mix + cholesterol treatments containing cytokines (12.5 ng/mL TNF-α and 15 ng/mL IL-6) or 0.25 mM cholesterol across two SCHH donors. Due to observations of cellular toxicity at higher concentrations, lower concentrations of TNF-α (1 ng/mL), IL-6 (1.2 ng/mL), and cholesterol (0.01 mM–0.1 mM) were selected for additional testing in SCHH. Lower cytokine concentrations were selected to avoid potential toxicity in SCHH while remaining within ranges that were previously observed to induce cytokine-mediated effects in primary human hepatocytes (Gramignoli et al., 2022). The toxicity and phenotype profiles of the lower cytokine and cholesterol concentrations were then assessed in differentiated HuH-7 cells before performing additional experiments in SCHH. All treatments in differentiated HuH-7 cells with the lower cytokine concentrations maintained significant increases in lipid droplet formation and lipid peroxidation (Figures 3A,B) that were also observed with higher concentrations (Figures 2A,D) and resulted in similar toxicity, as indicated by the ATP assay (Figure 3C). Ultimately, 0.01 mM cholesterol was selected for the final assessment in SCHH as this concentration appeared to have similar impact on phenotypic endpoints with a slightly lower toxicity risk than treatments containing 0.05 mM and 0.1 mM cholesterol in differentiated HuH-7 cells (Figure 3C).

The following three treatments were selected for the final assessment of cellular toxicity and their ability to induce MASH-like phenotypes in SCHH: 1) 0.5 mM OA:PA (1:2), 2) lipid mix, and 3) lipid mix with 0.01 mM cholesterol.

All treatments were tested for cellular toxicity with and without cytokine concentrations of 1.2 ng/mL IL-6 and 1 ng/mL TNF-α by assessing their effect on ATP and caspase-3/7 levels in three SCHH donors. Although statistically significant decreases in ATP were observed for some treatments in select donors compared with the control, none of the treatments resulted in ATP reduction below the pre-specified threshold of 70% (Figure 4A). No significant changes in caspase-3/7 activity were observed for all treatments, with and without cytokines, in SCHH donors JEL and IWM, whereas significant decreases in caspase-3/7 activity compared to control were observed in donor WID for multiple treatments with and without cytokines (Figure 4B). The mechanism underlying decreased caspase-3/7 levels in donor WID with these treatments is unknown. Compared to IWM (46 years) and JEL (27 years), WID was older (71 years) but did exhibit a lower liver fibrosis stage (F1C vs. F2) (Supplementary Table 1). Overall, minimal cellular toxicity risk was demonstrated in the three SCHH donors for all treatments with and without cytokines.

To assess the ability of each treatment (with and without cytokines) to induce a MASH-like phenotype, intracellular lipid droplet formation and lipid peroxidation were assessed in two SCHH donors. Statistically significant increases in lipid droplet formation were observed for all three treatments without cytokines (Figure 5A). All treatments containing cytokines (lipid–cytokine treatments) showed statistically significant increases in lipid droplet formation compared with all treatments without cytokines. Statistically significant increases in lipid peroxidation were observed for all three treatments containing cytokines but only for one treatment without cytokines (lipid mix +0.01 mM cholesterol) (Figure 5B). Direct comparisons between the same treatments with and without cytokines revealed inconclusive differences in lipid peroxidation.

The impact of the three selected lipid–cytokine treatments on drug transporter concentrations in SCHH from three donors (JEL, WID, and IWM) was assessed. Cytokines were included in all selected treatments due to the observed cytokine-mediated increases in SCHH lipid accumulation in donors WID and IWM (Figure 5A) and their physiologic relevance in MASH (Duan et al., 2022; Vachliotis and Polyzos, 2023). The proteomic concentrations of all basolateral uptake drug transporters of interest, except for organic anion transporter 7 (OAT7), were statistically significantly decreased across all treatments compared with untreated SCHH (control group; 0.5% DMSO) (Figure 6A). Regarding the efflux transporters (Figure 6B), BSEP and multidrug resistance-associated protein (MRP) 2 were significantly decreased while MRP3 was increased; multidrug resistance protein 1 P-glycoprotein (MDR1 P-gp) was unchanged across all treatments. Differences in transporter concentrations associated with each treatment in SCHH were converted to fold changes from control for comparison to previously published data in patients with MASH (Vildhede et al., 2020) and are depicted in Figure 6C and Supplementary Table 4. In this study, apart from OAT7, the observed changes in drug transporter concentrations were consistent in direction (i.e., increase or decrease) with previously reported alterations in liver tissue from patients with MASH (Vildhede et al., 2020).

Across all transporters, MRP2 was the only protein showing significant changes across all treatments and donors (Supplementary Figure 2). However, significant changes were observed for each transporter across all three treatments in at least two donors except for organic anion transporting polypeptide 1B3 (OATP1B3), OAT2, OAT7, and MRP3 (Supplementary Figure 2).

The impact of the three selected lipid–cytokine treatments on the proteomic concentrations of various DMEs was also assessed in SCHH from three donors (Figures 7, 8). The mean concentrations of cytochrome P450 (CYP) and uridine 5′-diphospho-glucuronosyltransferase (UGT) enzymes were significantly decreased by all treatments compared with the control, except for CYP2E1, which was increased by the lipid mix treatment with and without 0.01 mM cholesterol. Changes in UGT1A1 were inconclusive when treated with the lipid mix with 0.01 mM cholesterol. The mean concentrations of other DMEs evaluated (i.e., aldehyde oxidase 1 [AOX1], carboxylesterases [CESs], and sulfotransferases [SULTs]) were all decreased in the presence of lipid–cytokine treatments. While SULT2A1 was decreased across all treatments, no statistically significant differences were observed for this protein. Differences in DME concentrations associated with each treatment in SCHH were converted to fold changes from the control for comparison to previously published data in patients with MASH (Jamwal, 2018) and are depicted in Figures 7B, 8B. CYP3A5 concentrations were noticeably higher in JEL than in WID and IWM. This observation aligns with previous reports that CYP3A5 is expressed predominantly in hepatocyte donors of African descent (Tirona and Kim, 2017). CYP2D6 was also undetectable in all IWM samples, except for one sample treated with lipid mix + cytokines. The reason for this observation is unclear; however, it is possible that donor IWM carries a genetic variant in CYP2D6 that leads to negligible expression, given the highly polymorphic nature of this enzyme (Ingelman-Sundberg, 2005). Overall, the data indicate that changes in DME concentrations in in SCHH exposed to lipid–cytokine treatments are similar to those observed in the liver tissue from patients with MASH compared with normal liver tissue (Jamwal, 2018).

B-CLEAR^® technology was used to evaluate the hepatobiliary disposition of select transporter probe substrates in SCHH from three donors (JEL, WID, and IWM) following 72-h exposure to the three selected lipid–cytokine treatments. [³H]-TCA was used as a probe for sodium taurocholate co-transporting polypeptide (NTCP) and BSEP function. A significant decrease in [³H]-TCA “cells + bile” accumulation was observed across all donors compared with the control following exposure to each lipid–cytokine treatment (Figure 9). Accordingly, [³H]-TCA CL_uptake,app was also significantly lower than that in the control across all treatment–donor combinations (Table 2). The range of [³H]-TCA BEI values was 52.7%–82.2% in the control (Figure 9). Changes in BEI values varied depending on the treatment–donor combination, whereas the CL_biliary,app of TCA was significantly decreased by all treatments in SCHH from all donors (Table 2).

The cellular disposition of [³H]-E₂17G as a probe for OATP1B1, OATP1B3, and MRP2 function also was assessed in SCHH from the same three donors following exposure to the three lipid–cytokine treatments. Both the “cells + bile” accumulation and CL_uptake,app of [³H]-E₂17G were significantly decreased across all lipid–cytokine treatments and donors (Figure 10; Table 3). BEI values for the control group ranged from 13.2% to 30%. However, BEI values were negligible across all donors for each lipid–cytokine treatment (Figure 10); thus, CL_biliary,app of [³H]-E₂17G was not calculated.