Evaluation of a human 3D multicellular hepatic spheroid model as a platform for studying hepatic transporters

Abstract

There is growing demand for improved in vitro liver models to better predict in vivo pharmacology, specifically drug disposition mediated by hepatic transporters and assessment of transporter-mediated drug interaction risk. While 2D sandwich-cultured human hepatocytes (SCHH) remain valuable, they are limited to short-term use due to hepatocyte de-differentiation and absence of non-parenchymal cells. Multicellular hepatic spheroids (MHS) offer a promising alternative, but transporter concentrations, functionality, and suitability for hepatobiliary transport studies remain unclear. We evaluated an all-human MHS model, comprised of transporter-certified™ cryopreserved primary human hepatocytes (PHH), Kupffer, stellate, and endothelial cells, for long-term hepatic transporter assessment. Over a 21-day culture period, we monitored transporter concentrations (targeted proteomics), regulation (RNA-seq), localization (immunofluorescence), bile acid profiles (LC–MS/MS), and functional transport (B-CLEAR^®). This is the first report of protein concentrations of 13 transporters in MHS over 21 days directly compared to freshly thawed PHH and SCHH from the same donor. Most transporters declined in MHS compared to PHH, while SCHH maintained or increased transporter concentrations by day 5. However, multidrug resistance-associated protein (MRP) 4 and organic solute transporter (OST)-α/β were upregulated in MHS, likely reflecting adaptation to bile acid accumulation. Bile acid profiling confirmed functional synthesis, metabolism and excretion. Functional MRP2 efflux into sealed canalicular compartments was demonstrated with the MRP2 substrate, 5(6)-carboxy-2′,7′-dichlorofluorescein (CDF). Tight junction disruption of canaliculi with Ca²⁺-free buffer resulted in CDF release from canalicular compartments, with partial entrapment within MHS, likely due to the 3D architecture. These findings highlight key strengths and limitations of MHS as a model for assessing hepatobiliary transport.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00018-025-05996-z.

Introduction

It is now well-recognized that drug transporters, in addition to drug-metabolizing enzymes, play a key role in clinical pharmacology [1]. Hence, there is a need for reliable in vitro models to predict in vivo drug disposition mediated by transporters and the potential risk for transporter-mediated drug-drug interactions (DDIs). Primary human hepatocytes (PHH) are considered the gold standard for in vitro evaluation of pharmacokinetics and hepatotoxicity in pharmaceutical research [2]. In drug development, cultured PHH (e.g., suspension, monolayer, sandwich-cultured human hepatocytes (SCHH), spheroids, liver-on-a-chip) provide a platform to assess transport pathways, drug metabolism, DDIs, and hepatotoxicity [2, 3]. Cultures of PHH are also utilized for disease modeling to provide insight into disease mechanisms and possible therapeutic interventions [4]. Among the various hepatocyte culture models, the utility of SCHH for hepatic transporter assessment has been well-documented [5]. With functional canalicular networks and correctly localized transporters, SCHH are successfully used for transporter evaluation and hepatobiliary drug transport studies, such as quantifying the hepatic uptake or biliary excretion of a drug [6]. However, de-differentiation of PHH in 2D culture formats limits the use of SCHH to short-term studies, and other liver cell types are not routinely incorporated into this well-established model system. Thus, there is a growing demand in the pharmacological field to establish novel in vitro models that can be implemented for the assessment of drug metabolism and transport interplay over extended culture periods.

To address these concerns, emerging 3D models such as PHH cultured as spheroids, and ‘multicellular hepatic spheroids’ (MHS) containing PHH co-cultured with nonparenchymal cells in a 3D format, are becoming increasingly recognized for their ability to maintain cell viability and preserve an in vivo-like phenotype over time. PHH cultured as spheroids are functionally stable over extended culture periods, maintaining albumin and urea production for up to five weeks in culture [7]. The prolonged hepatic functionality of PHH cultured as spheroids, along with their enhanced capabilities to predict hepatotoxicity [8], clearance of low turnover compounds [9, 10], and DDIs involving drug metabolizing enzymes and transporters [11–13] compared to 2D models, are attributed to increased cell–cell interactions and their ability to more closely mimic the in vivo hepatic microenvironment. Incorporating nonparenchymal cells such as Kupffer, hepatic stellate, and liver sinusoidal endothelial cells into in vitro models may improve in vivo relevance, as nonparenchymal cells play an essential role in liver physiology [4]. To this end, important progress has been made to establish MHS as a novel in vitro liver model for preclinical assessment. MHS have demonstrated significant utility in toxicity screening applications [14] and identification of underlying drug toxicity [15] and disease [16, 17] mechanisms. The presence of nonparenchymal cells in 3D cultures influences the expression of key drug-metabolizing enzymes under both normal [18] and inflammatory [19] conditions, further confirming the importance of considering nonparenchymal cells in predictive in vitro models.

Although previous studies have characterized transporter expression in PHH spheroids and MHS [7, 8, 13, 20–23], these investigations were typically restricted to a few transporters or short-term culture periods, without a comprehensive integration of transcriptomic, targeted proteomic, imaging, and functional endpoints. Consequently, a holistic understanding of transporter abundance, regulation, localization, and function in long-term 3D hepatic spheroid models remains limited. Furthermore, a direct comparison of targeted proteomics results from the MHS model to data generated in SCHH has not been reported previously. In this study, MHS consisting of PHH, human Kupffer cells, human stellate cells, and human endothelial cells were characterized. This fully human, four cell-type culture configuration has been proposed to more closely mimic the structural and functional complexity of the human liver [17, 24, 25], positioning it as a human-relevant preclinical model and promising new approach methodology (NAM). To evaluate the suitability of the system for long-term transporter assessment, we monitored multiple endpoints over an extended 21-day culture period, including transporter concentrations by targeted proteomics, regulation by RNA-sequencing, localization with immunofluorescent analysis, and functional activity using bile acid profiling and transport studies (B-CLEAR^® technology).

This integrated experimental design, which combines transcriptomic, proteomic, and imaging-based assessments within a single, long-term study, exceeds the scope of prior investigations that have traditionally focused on one or two transporter-related endpoints in isolation. By integrating multiple endpoints focused on hepatobiliary transport, this study provides a more complete and mechanistic understanding of transporter biology in a more physiologically relevant 3D microenvironment. These data are necessary for an improved understanding of this model’s capabilities and limitations and is critical information for scientists already utilizing MHS for disease modeling, metabolism, and toxicity studies. Furthermore, targeted proteomic assessment of transporters and drug-metabolizing enzymes in this advanced in vitro model compared to SCHH and freshly thawed PHH in suspension is required to provide data for informative in vitro-to-in vivo extrapolation (IVIVE) by informing scaling factors to facilitate extrapolation across systems [26]. Together, this study provides the groundwork for establishing MHS as an additional in vitro pharmacological tool to assess hepatobiliary drug transport. Furthermore, by advancing the characterization of hepatic transporters in this model, the present study helps clarify the translational utility of MHS data for hepatotoxicity testing, DDI risk assessment, disease modeling, and regulatory decision-making.

Materials and methods

Chemicals and reagents

Dimethyl sulfoxide (DMSO; Cat. No. 41639), bovine serum albumin (BSA; Cat. No. A3059), and chenodeoxycholic acid (CDCA; Cat. No. C9377) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Paraformaldehyde (Cat. No. 15710) was purchased from Electron Microscopy Sciences (Hatfield, PA, USA). Acetonitrile (Cat. No. A995), methanol (Cat. No. A456), water (Cat. No. W6), formic acid (Cat. No. A117), phosphate-buffered saline (PBS; Cat. No. 14190), standard Hank’s balanced salt solution (HBSS; Cat. No. 14025), Ca²⁺-free HBSS (Cat. No. 14175), 5(6)-carboxy-2′,7′-dichlorofluorescein diacetate (CDFDA; Cat. No. C369), and qPCR TaqMan probes were purchased from Thermo Fisher Scientific (Waltham, MA, USA).

MHS formation and treatment

MHS were formed as previously described [17, 24] with slight modifications. Cryopreserved Transporter Certified™ human hepatocytes were purchased from BioIVT (Baltimore, MD) [donor WID, JEL, BXW]. After cell thawing, MHS were seeded using 1,200 total cells at the following composition: 75% Transporter Certified™ human hepatocytes, 10% Kupffer, 10% hepatic stellate, and 5% liver sinusoidal endothelial cells. For human hepatocyte and nonparenchymal cell donor information, see Supplemental Table S1. MHS were cultured in 384-well ultra-low attachment (ULA) plates with U-shaped wells (Corning, USA) or Akura plates (InSphero, ME) and kept in Lonza maintenance medium (cat. #CC-3198) for up to 21 days. Media was exchanged every 2–3 days starting on culture day 6 by gently aspirating 30 µL (60% volume) and replacing it with fresh media. CDCA stock solution was prepared in DMSO and diluted in maintenance medium. CDCA (20 µM final concentration) treatment was initiated on a subset of MHS with the first media change on day 6 and dosed with each subsequent media change until day 14 of culture (1 week exposure). The 20 µM dose was chosen based on previous reports indicating toxicity at higher doses in spheroid models [21]. The final concentration of DMSO was 0.2% in all MHS experiments. The donor(s) used in each experiment are specified in brackets within the Methods section.

Sandwich-cultured human hepatocytes (SCHH)

MHS endpoints, including liquid chromatography tandem mass spectrometry (LC–MS/MS)-based bile acid profiling and quantitative targeted absolute proteomics (QTAP), were compared to data generated using hepatocytes from the same donors (WID, JEL, and BXW) cultured in a sandwich configuration using QualGro™ culture media from BioIVT in a 24-well format as previously described [27]. SCHH were cultured for 5 days and exposed to 0.5% DMSO control for 24 h prior to harvest on culture day 5.

Morphology and cell viability

MHS morphology was evaluated using brightfield images captured at 10X magnification (Nikon Eclipse TS-100 Phase Contrast Microscope) [donors WID, JEL, BXW]. Cellular adenosine triphosphate (ATP) was measured in n = 3–6 individual MHS per time point using the CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, WI, USA) and assessed relative to baseline with a threshold of 75% used as a general threshold to indicate that adequate cell viability was maintained [28] [donors WID, JEL, BXW]. Media (25 µL) was removed from each MHS-containing well followed by the addition of 25 µL of CellTiter-Glo reagent. Samples were mixed by pipetting up and down three times and then incubated at 37 °C for 20 min. Following incubation, all contents were transferred to a white opaque plate, and the luminescence of each well was recorded using a SpectraMax M3 microplate reader (Molecular Devices, San Jose, CA, USA). ATP concentration per well (nM) was calculated using an ATP standard curve prepared in maintenance medium.

Quantitative targeted absolute proteomics

Freshly thawed PHH (150,000 cell aliquots) were harvested on day 0, SCHH (450,000 cells/well) were harvested on culture day 5, and approximately 128 MHS (153,600 cells) were harvested and pooled on days 7, 14, and 21 of culture. All samples were prepared in triplicate (technical replicates) for QTAP by nano- and micro-liquid chromatography with LC–MS/MS as previously described [29, 30] [donors WID, JEL, BXW]. Briefly, samples were fractionated into cytosolic and membrane proteins by sequential use of digitonin and Triton-X-100. Extracted membrane protein (10–20 μg) was spiked with stable isotope labeled peptide standards followed by trypsin digestion and solid phase extraction cleanup. Analysis was performed on either a nanoACQUITY (Waters, Milford, MA, USA) interfaced with a SCIEX (Framingham, MA, USA) QTRAP 5500 hybrid mass spectrometer operated in the positive MRM mode and equipped with a NanoSpray III source [all MHS samples], or on an M-Class Acquity (Waters) interfaced with a SCIEX Triple Quadrupole 7500 hybrid mass spectrometer also operated in the positive MRM mode and equipped with an OptiFlo Pro ion source [all freshly thawed PHH and SCHH samples] (See Supplemental Fig. S2 for cross-instrument validation demonstrating equivalent concentration measurements from the same samples). System control for the nanoLC-QTRAP 5500 was with SCIEX Analyst 1.5 software and nanoACQUITY UPLC Console. MultiQuant 2.0.2 (SCIEX) was used for peak integration. For the microLC-Triple Quadrupole 7500, system control was with SCIEX OS software, and SCIEX Analytics software embedded in OS was used for peak integration. The lower limit of quantification (LLOQ) was set at 0.1 pmol/mg membrane protein for all peptides measured on both systems. A human liver microsome sample was used as an assay quality control (data not shown). For signature peptide multiple reaction monitoring (MRM) transitions, see Supplemental Table S2.

Immunohistochemistry and image acquisition

Approximately 24 MHS were harvested and pooled after 14 days of culture, washed three times with PBS, fixed with 4% paraformaldehyde for 1 h at room temperature, and subsequently washed with PBS. For whole spheroid staining, samples were permeabilized with 0.1% Triton X-100 for 1 h at room temperature, then blocked with PBS containing 1% BSA and 0.5% Triton X-100 (blocking buffer) for 1.5 h at room temperature and incubated overnight at 4 °C with primary antibodies diluted in blocking buffer. Following overnight incubation, samples were washed four times with PBS and incubated with secondary antibodies diluted in blocking buffer (1:100) at room temperature for 1 h. Samples were washed three times with PBS, stained with 4′,6-diamidino-2- phenylindole (DAPI) (0.1% in PBS) and transferred to a glass bottom dish for imaging [donors WID, JEL, BXW]. For immunohistochemical staining of sectioned MHS, following fixation, samples were embedded in paraffin and sectioned (5 µm). Staining of sectioned MHS was performed as previously described [31] [donor JEL]. For antibody information, see Supplemental Table S3. Confocal images were taken using a Zeiss LSM 880 Confocal Laser Scanning Microscope (Plan Apo 20X/0.8NA or EC Plan-Neofluar 40X/1.30 Oil DIC M27).

Differential gene expression analysis

MHS [donor JEL] were treated for 7 days with the farnesoid X receptor (FXR) agonist, CDCA (20 µM). Differential gene expression analysis of CDCA-treated vs. DMSO control MHS was performed to assess transporter and drug-metabolizing enzyme regulatory mechanisms. After one week of ± CDCA treatment, approximately 64 MHS were harvested in triplicate per treatment group on culture day 14. RNA was extracted using TRI reagent (Sigma-Aldrich) according to the manufacturer’s instructions. RNA samples were submitted to GENEWIZ (South Plainfield, NJ) for RNA sequencing. Complementary DNA (cDNA) libraries were prepared and sequenced using the Illumina NovaSeq platform. Following sequencing, raw read counts for each gene were obtained and used for differential expression analysis. The DESeq2 package was utilized to compare gene expression between sample groups. The Wald test was applied to calculate p-values and log2 fold changes. Genes were considered differentially expressed if they met the criteria of an adjusted p-value < 0.05 and an absolute log2 fold change greater than 1. Gene ontology (GO) enrichment analysis was conducted using GeneSCF, with the human GO list applied to categorize genes based on biological processes and assess statistical significance. A volcano plot was generated to illustrate overall transcriptional differences between the groups. Statistical significance of selected genes following CDCA treatment was confirmed by qPCR in a separate donor [WID] as previously described [32]. For qPCR TaqMan probe information, see Supplemental Table S4.

LC–MS/MS-based bile acid profiling

SCHH lysate and media (100 µL aliquots) were harvested on culture day 5, and approximately 128–384 pooled MHS lysate and respective culture media (100 µL aliquots) were harvested on days 7, 14, and 21 of culture and prepared for individual bile acid quantification as previously described [27] using culture media as a matrix [donors WID, JEL, BXW]. Cell lysate samples were designated as “cells + bile,” based on the assumption that the measured bile acids originated from within the cells and/or were retained in the bile. Briefly, samples were lysed in a 70:30 acetonitrile:water solution containing internal standards and filtered by centrifugation. Samples were concentrated under nitrogen gas and reconstituted in 50:50 acetonitrile:water containing 0.1% formic acid prior to LC–MS/MS analysis. Data were acquired with MRM, and calibration curves were generated for each bile acid as a series of concentrations diluted in culture media. For MRM transitions, see Supplemental Table S5.

MRP2 function and tight junction modulation

To assess multidrug resistance-associated protein 2 (MRP2) transporter function and bile canalicular network formation within MHS, the fluorescent MRP2 substrate, 5(6)-carboxy-2′,7′-dichlorofluorescein (CDF), and modulation of tight junctions using standard Ca²⁺-containing and Ca²⁺-free buffer (B-CLEAR^® technology) [6] were utilized. CDF-diacetate (CDFDA) passively diffuses into cells and is hydrolyzed to fluorescent CDF by intracellular esterases [33]. CDF is an ideal compound to evaluate biliary excretion because it is not subject to hepatic metabolism, and MRP2-mediated excretion into the bile networks can be tracked via fluorescent signal [34]. On culture days 7, 14, and 21, media was slowly aspirated from MHS [donor WID] followed by a single wash with 50 µL of Ca²⁺-containing standard HBSS within the Akura plate. MHS were incubated in standard HBSS containing 5 µM CDFDA for 30 min at 37 °C. Following CDFDA incubation, MHS were washed three times and maintained in 50 µL of standard HBSS (“cells + bile”) or Ca²⁺-free HBSS (“cells only”) for imaging using a CellInsight CX7 High Content Scanner (10 µm step size, 12 steps, 120 µm extended depth of field, 10X objective). Disruption of tight junctions throughout the MHS was evaluated qualitatively by visual observation.

Statistical analysis

Statistical analyses for ATP and QTAP endpoints were performed using GraphPad Prism statistical software version 9.4.1 (GraphPad Software Inc.). Mean values of ATP concentration from day 7, 14, and 21 of culture were compared using a one-way analysis of variance (ANOVA) by donor. Mean values of protein concentrations from SCHH (day 5) and MHS (day 7, 14, 21) were compared to protein concentrations in freshly thawed PHH (day 0) using a one-way ANOVA by donor. When needed, a Dunnett’s post hoc test was employed for multiple comparisons within each donor group.

Results

Morphology and cell viability

MHS were cultured over a 3-week culture period as shown in Fig. 1a. Spheroids demonstrated stable morphology over 21 days in culture (Fig. 1b; Supplemental Fig. S1). Cell viability (measured by ATP levels) remained above the pre-specified 75% threshold for donor WID and JEL MHS over the 21-day culture period; however, a significant decline in culture viability below this threshold was observed over time in donor BXW MHS (Fig. 1c).

Fig. 1.

Human multicellular hepatic spheroids (MHS) experimental workflow, morphology, and cell viability. a) MHS formation process from seeding to endpoint analysis. MHS were generated by seeding 75% Transporter-Certified™ primary human hepatocytes (PHH) from 3 donors [WID (pink), JEL (blue), and BXW (black)], and human nonparenchymal cells (10% Kupffer, 10% stellate, and 5% endothelial cells) into ultra-low attachment plates. Created with BioRender.com. b) Representative brightfield images of MHS from donor WID taken with a 10X objective on days 7, 14, and 21 of culture. Scale bar = 200 µm. c) Cellular ATP content was measured as a marker for cell viability and normalized to culture day 7. Each data point is the mean SD of n = 3–6 spheroids. Statistically significant differences were measured using a one-way analysis of variance (ANOVA) by donor with Dunnett’s multiple comparison [*, P < 0.05]

Transporter concentrations

Transporter (Fig. 2) and select drug metabolizing enzyme (Supplemental Fig. S3) concentrations were measured in freshly thawed PHH (day 0), SCHH (day 5), and MHS (days 7, 14, and 21) using targeted proteomics. Overall, transporter concentrations were significantly lower in MHS over time compared to day 0 freshly thawed PHH and day 5 SCHH and exhibited substantial donor variability (Fig. 2). Organic anion transporting polypeptide (OATP)1B1 and OATP1B3 concentrations were significantly reduced in MHS across all culture weeks compared to day 0 PHH (Fig. 2a). OATP2B1 concentrations remained stable in day 5 SCHH but significantly declined in MHS over time in all three donors. Organic solute transporter (OST)-α concentrations were consistently low (near or below the LLOQ [0.1 pmol/mg membrane protein]) across all culture systems, while OST- concentrations were low in day 5 SCHH but elevated in MHS over time compared to day 0 PHH. P-glycoprotein (P-gp) concentrations were stable or elevated in day 5 SCHH across all three donors compared to day 0 PHH (Fig. 2b). In MHS, P-gp concentrations remained stable in donor JEL and decreased over time in donor WID and BXW. Bile salt export pump (BSEP), MRP2, and MRP3 concentrations were reduced across all MHS culture days compared to day 0 PHH. MRP4 concentrations were generally below the LLOQ in day 0 PHH and day 5 SCHH across all donors but increased in two of three donors (WID and BXW) in MHS. Overall, nine of 13 measured transporters were either stable or increased in day 5 SCHH compared to day 0 PHH, whereas only OST-α, OST-β, and MRP4 were increased (1.5-fold, 3.0-fold and 3.6-fold, respectively) in MHS by culture day 21 (mean of 3 donors); all other transporters declined over time in MHS compared to day 0 freshly thawed PHH (Fig. 2c).

Fig. 2.

Select transport protein concentrations in human multicellular hepatic spheroids (MHS) over a 21-day culture period compared to sandwich-cultured human hepatocytes (SCHH) on culture day 5 and freshly thawed primary human hepatocytes (PHH) from the same Transporter Certified™ hepatocyte donors. Data represent n = 3 technical replicates (mean SD) at each timepoint for each of 3 hepatocyte donors [WID (pink), JEL (blue), and BXW (black)] for a) solute carrier (SLC) transporters and b) ATP-binding cassette (ABC) transporters. Statistically significant differences were measured using a one-way analysis of variance (ANOVA) by donor compared to day (D) 0, with Dunnett’s multiple comparison [*, P < 0.05]. The dotted line represents the LLOQ of 0.1 pmol/mg membrane protein. When all three technical replicates were equal to zero, mean values were imputed as 0.01 pmol/mg membrane protein for visualization on a log scale. c) Log₂ fold change of day 5 SCHH and day 7, 14, and 21 MHS compared to day 0 freshly thawed PHH, presented as a mean of all three donors. BSEP, bile salt export pump; MRP, multidrug resistance–associated protein; OAT, organic anion transporter; OATP, organic anion transporting polypeptide; OCT, organic cation transporter; OST, organic solute transporter; P-gp, P-glycoprotein

Transporter regulation

Differential gene expression analysis of CDCA treated vs. DMSO control MHS was performed to assess transporter and drug-metabolizing enzyme regulatory mechanisms. MHS treated with 20 µM CDCA revealed a total of 43 upregulated genes including SLC51A (the gene encoding OST-α) and SLC51B (the gene encoding OST-β) compared to DMSO control MHS (Fig. 3a). A total of 415 genes were downregulated, including CYP7A1 and CYP2C19. Upregulation of SLC51A and SLC51B was confirmed by qPCR in a separate donor (Fig. 3b). The most differentially expressed gene with CDCA treatment was RTP3 with a log2 fold change of + 2.14. This gene is involved in protein targeting and insertion into the membrane and is thought to be involved in the progression of metabolic steatohepatitis [35]. For the full gene ontology analysis, see Supplemental File 1.

Fig. 3.

RNA-sequencing of human multicellular hepatic spheroids (MHS) on culture day 14. a) Volcano plot depicting differential gene expression analysis of 0.2% dimethyl sulfoxide (DMSO)- vs. 20 µM chenodeoxycholic acid (CDCA)-treated MHS for 7 days [donor JEL]. Each data point in the scatter plot represents a single gene. The log2 fold change of each gene is represented on the x-axis, and the log10 of its adjusted p-value is on the y-axis. Genes with an adjusted p-value less than 0.05 and a log2 fold change greater than 1 are indicated by red dots; these represent up-regulated genes. Genes with an adjusted p-value less than 0.05 and a log2 fold change less than −1 are indicated by green dots; these represent down-regulated genes. b) qPCR analysis of MHS of selected differentially expressed genes to confirm results of RNA-sequencing analysis in a separate hepatocyte donor [WID]. Statistically significant differences were measured using an unpaired t-test [*, P < 0.05, ***, P < 0.001]. CYP, cytochrome P450; OST, organic solute transporter; SLC, solute carrier

Transporter localization

Whole and sectioned MHS were stained to assess transporter localization and plasma membrane integrity (Fig. 4; Supplemental Fig. S4). In whole MHS on culture day 14, the bile acid uptake transporter, sodium-taurocholate cotransporting polypeptide (NTCP), was co-localized with the basolateral membrane marker, sodium–potassium pump (Na⁺/K⁺-ATPase), in select regions (Fig. 4a). In contrast, the basolateral transporter, OATP1B3, predominantly showed intracellular localization with minimal co-localization with the basolateral membrane marker in whole MHS (Fig. 4b). Apical transporters, MRP2 and BSEP, were co-localized with the tight junction protein, zonula occludens-1 (ZO1), in whole MHS, suggesting the presence of bile canalicular structures on the MHS outer surface (Fig. 4c-d). However, BSEP staining appeared to be more diffuse compared to MRP2 staining. In sectioned MHS, NTCP was co-localized with Na⁺/K⁺-ATPase in select regions, consistent with the observations in whole MHS (Fig. 4e). OATP1B3 staining in sectioned MHS demonstrated primarily intracellular localization (Fig. 4f), also consistent with the distribution pattern in whole MHS. Sectioned staining also revealed that MRP2 was expressed and co-localized with tight junction protein, ZO1, throughout the spheroid structure (Fig. 4g). Notably, sectioned MHS demonstrated reduced staining for BSEP and ZO1 compared to whole spheroid staining (Fig. 4h), indicating higher abundance of BSEP and tight junctions near the surface. Staining of sectioned MHS for the basolateral membrane marker, Na⁺/K⁺-ATPase (Fig. 4i), and apical membrane marker, aminopeptidase N (CD-13) (Fig. 4j), confirmed the integrity of both apical and basolateral membranes within the MHS interior. Positive CD-13 staining indicated the presence of canalicular networks throughout the MHS structure.

Fig. 4.

Cellular localization of transporters in whole and sectioned human multicellular hepatic spheroids (MHS) on culture day 14. Staining of membrane markers and select transporters (NTCP, OATP1B3, MRP2, BSEP) in whole (a, b, c, d) and sectioned (e, f, g, h, i, j) MHS from 3 donors [WID, JEL, BXW]. Images were taken with a Zeiss 880 Confocal Laser Scanning Microscope with a 20X objective (a, b, c, d, e, h, j) or a 40X objective (e, h, i). DAPI = Nuclei stain. Scale bar = 100 µm. BSEP, bile salt export pump; CD-13, aminopeptidase N; MRP, multidrug resistance–associated protein; Na⁺/K⁺-ATPase, sodium-potassium pump; NTCP, sodium-taurocholate cotransporting polypeptide; OATP, organic anion transporting polypeptide; ZO1, zonula occludens-1

Bile acid profiling

To evaluate the metabolic and efflux capabilities of the MHS model in comparison to SCHH, the endogenous bile acid profile was evaluated using LC–MS/MS in both “cells + bile” and media samples of MHS over the 21-day culture period and compared to those in SCHH after 5 days in culture. Glycocholic acid (GCA) and glycochenodeoxycholic acid (GCDCA) were the most predominant bile acid species in both SCHH and MHS “cells + bile” across all donors (Fig. 5), with GCDCA-3-O-β-glucuronide (GCDCA-3G), cholic acid (CA), and CDCA (data not shown) contributing the least to the overall cellular bile acid composition. Interestingly, CA was undetectable in donor BXW SCHH “cells + bile” and media (Fig. 5a). Both culture systems favored glycine conjugated bile acids across all donors. However, GCA, TCA, GCDCA-S, and GCDCA-3G were higher in day 5 SCHH media than MHS media over time (Fig. 5). In contrast, TCDCA was higher in MHS media over time compared to day 5 SCHH media (Fig. 5).

Fig. 5.

Endogenous bile acids in a) day 5 sandwich-cultured human hepatocytes (SCHH) and human multicellular hepatic spheroids (MHS) on days b) 7, c) 14, and d) 21 of culture from the same Transporter Certified™ hepatocyte donors. Bile acid concentrations in “cells + bile” (closed bars; pooled 128–384 MHS per donor; mean ± SD, n = 1–3) and in the culture media (open bars; mean ± SD, n = 1–3) are shown for each donor [WID (pink), JEL (blue), and BXW (black)]. TCDCA concentrations in “cells + bile” were not measured in donor JEL MHS due to analytical error. NA, not analyzed; #, below the limit of quantification; CA, cholic acid; GCA, glycocholic acid; GCDCA, glycochenodeoxycholic acid; GCDCA-3G, GCDCA-3-O-β-glucuronide; GCDCA-S, GCDCA-3-sulfate; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid

MRP2 function and tight junction modulation

Canalicular formation and MRP2 function were confirmed in several MHS throughout the 21-day culture period (Fig. 6). Inter-MHS variability in CDF accumulation in canalicular compartments (i.e., bile pockets) was observed, most notably on day 7, where distinct cell clusters appeared to trap CDF within the MHS. Although most MHS demonstrated CDF accumulation in bile pockets on all evaluated culture days, there were some individual MHS where fluorescent CDF appeared to accumulate intracellularly throughout the entire spheroid, with no visible evidence of canalicular compartments or networks. In general, bile pocket formation increased throughout the three-week culture period. The ability to disrupt tight junctions using Ca²⁺-free buffer (B-CLEAR^® technology) was assessed in MHS on culture day 14 (Fig. 7). After a 30-min incubation with CDFDA, switching from standard HBSS to Ca²⁺-free HBSS resulted in the rapid disappearance of fluorescent CDF from bile pockets, with fluorescence dispersing either within the MHS interior or into the surrounding buffer over a 20-min imaging period. Bile pockets began to dissipate within approximately one min of Ca²⁺-free HBSS addition. In contrast, MHS maintained in standard HBSS retained fluorescent CDF within bile pockets throughout the 20-min imaging period.

Fig. 6.

Canalicular formation and multidrug resistance-associated protein (MRP) 2 function in eighteen representative individual human multicellular hepatic spheroids (MHS) on culture days 7, 14, and 21. Donor WID MHS were incubated in standard HBSS containing 5 µM CDFDA for 30 min at 37 °C on days (D) 7, 14, and 21 of culture. Cells were washed and imaged using a CellInsight CX7 High Content Scanner (10 µm step size, 12 steps, 120 µm extended depth of field, 10X objective). Transmitted light (grey) was used to visualize MHS structure. White arrows depict CDF accumulation in representative bile pocket structures. Scale bar = 100 µm. CDFDA, 5(6)-Carboxy-2′,7′-dichlorofluorescein diacetate; HBSS, Hank’s balanced salt solution

Fig. 7.

Tight junction modulation in human multicellular hepatic spheroids (MHS) on culture day 14. Donor WID MHS were incubated in standard HBSS containing 5 µM CDFDA for 30 min at 37 °C on culture day 14. MHS were washed and maintained in standard HBSS (“cells + bile”) or Ca²⁺-free HBSS (“cells only”) for up to 20 min and imaged over time using a CellInsight CX7 High Content Scanner (10 µm step size, 12 steps, 120 µm extended depth of field, 10X objective). Transmitted light (grey) was used to visualize MHS structure. White arrows depict CDF accumulation in bile pockets in standard HBSS (top panel) or CDF dispersing from bile pockets in Ca²⁺-free HBSS (bottom panel). Scale bar = 100 µm. CDFDA, 5(6)-Carboxy-2′,7′-dichlorofluorescein diacetate; HBSS, Hank’s balanced salt solution

Discussion

This study is the first to evaluate an all-human MHS model for the long-term (21-day) assessment of hepatic transporters. This comprehensive evaluation included absolute protein quantification of 13 key transporters, along with investigation of transporter regulation, localization, and functionality in MHS over time. Select endpoints were directly compared to freshly thawed PHH and SCHH from the same Transporter-Certified™ donor, providing an important dataset for comparisons between systems and IVIVE. The long-term viability of MHS generated using Transporter-Certified™ PHH was donor specific. Notably, hepatocytes from BXW, the oldest donor (73 yrs) with the highest BMI (32.6) and on several medications (Supplemental Table S1), exhibited the poorest performance in culture, potentially due to donor-specific factors. These findings highlight the importance of donor selection and the use of high-quality hepatocytes for sustaining long-term 3D cultures.

While overall transporter concentrations in MHS were lower than freshly thawed PHH and day 5 SCHH, the increasing expression of OST-α/β and MRP4 in MHS over time suggests an adaptive response to prevent bile acid accumulation. Previous mouse studies demonstrated that Kupffer cells contribute to hepatoprotective mechanisms by inducing hepatic transporters, particularly Mrp4 [36]. Therefore, the presence of nonparenchymal cells in the MHS may drive the upregulation of specific transporters over time, as observed in the present study. This highlights a potential advantage of MHS over spheroids only containing PHH for toxicity studies, as the inclusion of nonparenchymal cells seems to enhance the model’s sensitivity to bile acid-mediated stress. Despite lower protein expression levels, transporters and metabolizing enzymes can remain functional in MHS. For instance, MRP2 actively transported CDF into bile canaliculi in our MHS system (Fig. 6), even though MRP2 protein concentrations were low. Additionally, MHS demonstrated functional bile acid metabolism including generation of GCDCA-S and GCDCA-3G metabolites, and bile acid transporters actively effluxed bile acids out of hepatocytes (Fig. 5). These results highlight the importance of functional characterization of transporters in the donor of interest when using MHS to investigate drugs that undergo active transport.

We compared our targeted proteomics findings to prior studies reporting proteomics in similar 3D models. Bell et al. [8] compared global proteomics of spheroids composed only of PHH and SCHH from the same three (non-Transporter-Certified™) donors up to 14 days in culture. It is important to note that the SCHH cultures maintained by Bell et al. [8] through day 14 used repeated Matrigel overlays and alternative media conditions, whereas the SCHH data in the present study were obtained on culture day 5 using QualGro™ media. While Bell et al. [8] only evaluated a few transporters, the findings of this study align with the present study, showing increased P-gp and OATP1B1 protein abundance in SCHH compared to spheroids. Furthermore, Handin et al. [37] performed global proteomics of spheroids containing only PHH (mean of four donors) over a 21-day culture period to evaluate the impact of different media conditions and observed, in general, decreased abundance of P-gp, BSEP, and OATP2B1 over time, consistent with the three-donor mean data from the present study. Messner et al. [20] is the only study to our knowledge that has reported global proteomic profiling of MHS (PHH, Kupffer, and endothelial cells). Relative protein abundance (mean of three donors) was analyzed over a 35-day culture period and compared to freshly thawed PHH + nonparenchymal cells. While only select transporters were assessed, reported decreases in MRP2, OCT1, and OATP1B1 in MHS over time were consistent with the mean data from the three donors in the present study. Collectively, our findings are consistent with previous proteomic studies of PHH-only spheroids and MHS while expanding upon prior research by providing a comprehensive dataset of 13 different transporter concentrations across SCHH and MHS. These data are informative for IVIVE and demonstrate that transporter expression in 3D culture is generally lower than in SCHH and freshly thawed hepatocytes.

Although MHS demonstrate the presence of transport proteins and functional efflux activity over time, there are several limitations that need to be considered. For example, CDCA, a potent FXR agonist known to regulate transporter expression as a protective mechanism against bile acid accumulation and cholestatic hepatotoxicity [38, 39], did not regulate all expected transporters in MHS. While significant increases in SLC51A and SLC51B mRNA levels in MHS following CDCA treatment were observed, ABCB11 (BSEP) was not significantly upregulated following CDCA treatment in MHS, despite previous reports of increased mRNA expression in SCHH [38]. However, this discrepancy may be due to experimental conditions; the SCHH study by Jackson et al. [38] used a higher dose of CDCA (100 µM) compared to the present study (20 µM), which may have led to varying degrees of nuclear receptor-ligand binding and regulatory responses. Additionally, substantial donor variability in BSEP induction was observed in SCHH [38], which could further contribute to the differences observed between the two systems. It is also noted that the RNA-sequencing analysis in the present study was limited to a single hepatocyte donor. Future work examining transcriptomic profiles of multiple donors will be important to fully elucidate FXR agonist–mediated gene regulation in the MHS model.

Another limitation of MHS is the lack of proper localization of some transporters at the spheroid periphery and within the MHS interior. The bile acid uptake transporter, NTCP, was co-localized with the basolateral membrane in select regions, which could facilitate bile acid uptake into MHS. Conversely, the basolateral uptake transporter, OATP1B3, did not appear to be correctly localized to the membrane. We hypothesize that this may be attributed to the absence of extracellular matrix proteins in the MHS culture, as the Matrigel overlay in SCHH is a critical component for promoting hepatocyte polarization and transporter functionality [40]. These findings underscore the need for further investigation into the localization and trafficking patterns of basolateral transporters in the MHS model, and the potential role of matrix composition, oxygen gradients, and nonparenchymal cells in these processes.

Furthermore, while the apical transporter, BSEP, was visible by immunofluorescent analysis on day 14, it demonstrated preferential localization near the spheroid periphery, and staining was less prominent towards the center of MHS, consistent with previous reports [20]. Notably, the BSEP zonation pattern observed in the MHS model more closely resembled that reported in liver tissue of patients with metabolic dysfunction-associated steatotic liver disease (MASLD) rather than the uniform BSEP distribution typically observed in non-diseased liver tissue [31]. Increased BSEP expression near the surface may represent a compensatory mechanism to enhance bile acid excretion and mitigate bile acid-induced hepatocyte injury in the outer region of the MHS, which likely experiences higher exposure to extracellular bile acids. Interestingly, ZO1 staining of tight junctions revealed the presence of an apical membrane lining the outer surface of the MHS, which may also facilitate bile acid efflux into the surrounding media (Fig. 5). The colocalization of MRP2 with ZO1 was consistent with functionality at the canalicular membrane, as demonstrated with fluorescent CDF accumulation in bile pockets on culture days 7, 14, and 21. Despite confirmed MRP2 functionality in MHS and proper formation and maturation of canalicular compartments over time, notable inter-MHS variability in CDF accumulation was observed, particularly on day 7, where clusters of trapped CDF were apparent within select MHS. Further research is needed to elucidate the underlying cause of inter-MHS variability in canalicular network formation, and the potential role of nonparenchymal cells in this process.

B-CLEAR^® technology is an in vitro approach to determine the biliary excretion index of compounds [5, 6, 41]. However, this assay was originally optimized for 2D culture, and its applicability to 3D culture has not been investigated prior to this study. In 2D culture, exposure to Ca²⁺-free HBSS disrupts tight junctions and opens bile canaliculi, allowing canalicular contents to be released into the media and subsequently removed [5, 6]. This method enables differentiation between the total amount of compound in “cell + bile” and “cells only” by comparing accumulation in standard and Ca²⁺-free HBSS, respectively [5, 6]. Findings in the MHS model revealed that CDF released from bile pockets after addition of Ca²⁺-free HBSS is partially trapped within the spheroid structure, likely due to the 3D architecture, preventing the complete removal of canalicular contents from the system. Furthermore, because CDF is also an OATP substrate [34], it is plausible that CDF excreted from bile pockets is taken back up by hepatocytes in close proximity within the MHS model. Collectively, these data indicate that because canalicular contents are not fully cleared from MHS after tight junction disruption, this approach is not suitable to accurately quantify the biliary excretion index or biliary clearance of compounds using this model. This is likely due to the spheroid architecture and lack of fully externalized canalicular networks that would allow release of content from the MHS interior bile pockets directly into the buffer. From a toxicological perspective, the entrapment of compounds within MHS may make it a more sensitive model for detecting hepatotoxicity compared to 2D cultures. However, a key limitation that remains is the inability to distinguish between intracellular and canalicular contents within the MHS model without the use of imaging techniques.

The observed entrapment of canalicular contents in MHS was further supported by bile acid metabolite profiling, as GCDCA-S and GCDCA-3G were detected at lower levels in MHS media compared to SCHH media, suggesting these metabolites may accumulate within MHS rather than undergoing active efflux out into the media. GCDCA-S and GCDCA-3G are substrates of MRPs, and MRP3 protein concentrations were higher in SCHH compared to MHS. Thus, higher levels of these metabolites in SCHH media may be explained by increased basolateral efflux via MRP3. Conversely, TCDCA, a substrate of MRP4 [42] and BSEP, was higher in MHS media over time compared to SCHH. Considering BSEP protein concentrations were lower and MRP4 concentrations were higher in MHS compared to SCHH, TCDCA likely undergoes preferential basolateral efflux by MRP4 in MHS. These findings highlight distinct differences in transporter-mediated bile acid disposition between the SCHH and MHS model.

Conclusion

In conclusion, this study highlights strengths and limitations of MHS as an advanced in vitro model for studying hepatic transporter abundance, function, regulation, and bile acid metabolism. By integrating targeted proteomics, RNA sequencing, immunofluorescent analysis, bile acid profiling, and functional assessment within a single experimental framework, the current work provides new insight into transporter biology within this complex system, clarifying the translational relevance of MHS data for preclinical testing and regulatory decision-making. Based on our proteomics findings, the presence of nonparenchymal cells may enhance the sensitivity of MHS by modulating transporter expression in response to bile acid accumulation. However, limitations remain, including mislocalization of certain transporters, inter-MHS variability, and challenges in differentiating intracellular from canalicular contents in MHS. Our findings also highlight that the B-CLEAR^® assay, while useful in 2D models, is not readily transferable to MHS due to the 3D architecture and partial entrapment of canalicular contents after disruption of tight junctions. Collectively, these data advance our understanding of MHS as an evolving human relevant, long-term 3D in vitro platform for studying hepatic transporters and provide important insights into the application of MHS as a promising NAM. This work serves as a benchmark to support continued optimization of next-generation liver models for improved utility in pharmacological and toxicological applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors’ contributions

Mattie Hartauer: Conceptualization, Methodology, Formal analysis, Investigation, Writing—Original Draft, Visualization; Henry Ho: Conceptualization, Methodology, Investigation, Writing—Review & Editing; Meimei Wan: Methodology, Investigation; William A. Murphy: Investigation, Writing—Review & Editing; Jacqueline B. Tiley: Conceptualization, Investigation, Writing—Review & Editing; John K. Fallon: Investigation, Writing—Review & Editing; Colin E. Bishop: Supervision, Writing—Review & Editing, Funding acquisition; Kim L.R. Brouwer: Conceptualization, Supervision, Writing—Review & Editing, Funding acquisition.

Funding

Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number R35 GM122576. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Funding was also provided by the Department of Defense, (DTRA) under Award Number HDTRA1-19–1-0013. The information does not necessarily reflect the position or the policy of the federal government, and no official endorsement should be inferred. M. Hartauer is supported, in part, by a Pre-Doctoral Fellowship from the American Foundation for Pharmaceutical Education and a Dissertation Completion Fellowship from the University of North Carolina at Chapel Hill Graduate School.

Data availability

The author confirms that all data generated or analyzed during this study are included in this published article.

Declarations

Conflict of interest

The project or effort depicted was or is partially sponsored by the Department of Defense, (DTRA).

Prof. Kim L.R. Brouwer is a coinventor of the sandwich-cultured hepatocyte technology for quantification of biliary excretion (B-CLEAR^®) and related technologies, which have been licensed exclusively to BioIVT.

All other authors declared no competing interests for this work.