iPSC-based hepatic organoids reveal a heterozygous MYO5B variant as driver of intrahepatic cholestasis

Abstract

Background:

Hereditary intrahepatic cholestasis is caused by variants of various genes involved in enterohepatic bile circulation, metabolization, and conjugation. Originally classified into 3 groups, the number of contributing genes is still increasing, underlining the need for a deeper understanding of the molecular interaction during intrahepatic cholestasis.

Methods:

In the present study, we investigate the interplay of heterozygous variants in 3 cholestasis-associated genes (ABCB11, ABCB4, and MYO5B) by exploiting iPSC-based hepatic organoids from a patient suffering from recurrent intrahepatic cholestasis.

Results:

Functional characterization of MRP2-mediated cholyl-lysyl-fluorescein (CLF) and BSEP-mediated Tauro-nor-THCA-24-DBD transport demonstrated a marked reduction of transport in MYO5B-deficient organoids, in comparison to unaffected control organoids. Moreover, iPSC-based organoids derived from the patient carrying 3 heterozygous variants in ABCB11, ABCB4, and MYO5B also exhibited absence of BSEP-mediated Tauro-nor-THCA-24-DBD transport, but functional MRP2-mediated CLF-transport. Interestingly, CRISPR/Cas9-mediated correction of the mutated ABCB11 allele could not restore the impaired BSEP function, suggesting the heterozygous MYO5B variant as the main driver of the transport deficiency. In fact, CRISPR/Cas-mediated correction of the MYO5B variant finally resulted in a restoration of the BSEP-mediated Tauro-nor-THCA-24-DBD transport.

Conclusions:

iPSC-based organoids serve as an authentic model for functional assessment of the hepatobiliary transport with fluorescent substrates. This allows the characterization of variants of uncertain significance and other variants in cholestasis-associated genes and revealed that a heterozygous MYO5B variant increases the susceptibility to defective hepatobiliary BSEP-mediated transport.

INTRODUCTION

Progressive familial intrahepatic cholestasis (PFIC) comprises a set of genetic liver disorders leading to chronic liver damage and failure to detoxify compounds via the biliary system. The severity of a pathophysiological manifestation depends on several criteria, such as the number of affected alleles and the kind of variation, for example, a truncated variant or a missense variant. ¹ Despite the overall prevalence being rather low, varying from 1:50,000 to 1:100,000, ² genetic PFIC subtypes account for ~10%–15% of neonatal cholestasis. ³ PFIC was historically classified into 3 subgroups related to pathogenic variants in genes encoding hepatobiliary transporters and ATPases, whereas PFIC1 (ATP8B1/FIC1) and PFIC2 (ABCB11/BSEP) usually appear in the first months after birth, and PFIC3 (ABCB4/MDR3) manifests later in childhood or even in adulthood. ² More recent data suggest at least 8 more genes contributing to PFIC phenotypes, including TJP2 (PFIC 4), ⁴ NR1H4 (PFIC 5), ⁵ KIF12 (PFIC 8), ⁶ MYO5B (MVID/PFIC10), ⁷ and others. ⁸ Furthermore, with advances in sequencing technologies, more uncharacterized variants of uncertain significance (VUS) are described, and reliable tools to investigate subcellular trafficking and functional characteristics of the PFIC-associated variants need to be established. To explore the impact of hepatobiliary transport-associated VUS in vitro, a cell culture system with polarized hepatic cells exhibiting functional basolateral and apical membrane domains is essential. Primary human hepatocytes are difficult to culture in a polarized shape for longer periods, and it is difficult to model genetic disorders with primary tissue samples from affected patients. Induced pluripotent stem cells (iPSC), which can be generated from a small blood sample of the patient, can be used to model various liver disorders. ⁹

Recent differentiation protocols describe the generation of organoids with polarized hepatic cells,^10,11 carrying the respective genetic variants of the patient for direct investigations of the disturbed hepatobiliary transport and bile formation (Figure 1). After genetic correction in the clonally expandable iPSC stage, the impact of a given variant can be directly compared with its unaffected wild-type variant in the same genetic background. As the organoids exhibit inner luminal structures, they are able to model the hepatobiliary transport to a certain extent. Presence of basolaterally localized members of the solute carrier organic anion transporter family (OATP) for the uptake of bile salt analogs and other compounds, and expression of apically localized transporters such as the bile salt export pump (BSEP) or the multidrug resistance-associated protein 2 (MRP2) has been described in various reports.^10,12–14 To explore the transporter function within such a system, the use of specific fluorophore-labeled substrates and transport inhibitors is fundamental. ¹⁵ In the rat model, several studies described Abcb11-mediated transport of cholyl-L-lysin fluorescein (CLF), which was abrogated in Abcb11-mutant rats.^16–18 However, in the human system, CLF, containing 2 COOH⁻ groups, is considered to be transported by MRP2, which transports conjugated organic anions with 1 or, preferably, 2 negative charges. ¹⁹ Thus, we chose the bile salt analogs Tauro-nor-THCA-24-DBD (Tauro-DBD) as a specific substrate for BSEP-mediated transport ²⁰ and exploited CLF as a substrate for MRP2-mediated transport analyses. To mimic the 3-dimensional shapes of liver tissue, a polarized layer of hepatic cells is required. Adjacent hepatocytes contribute an apical plasma membrane forming the bile canaliculus (BC), sealed by tight junctions. ²¹ Due to the polarization of the cell, proteins necessary for a specific task are transported to their destination by motor proteins. Among this large protein family, unconventional myosin-Vb is associated with liver-related diseases, including cholestasis.^7,22 As unconventional myosin-Vb and Rab11a are crucial for the formation of BC at the apical membrane, ²³ defects in one of these proteins can alter the polarization and lead to major pathological consequences. Pathogenic variants in MYO5B can lead to microvillus inclusion disease (MVID) ²⁴ or cholestatic liver disease mimicking BSEP-associated PFIC.^25–27 Moreover, in vitro data show that polarization of cells is required for intracellular trafficking of BSEP to the BC and that this intracellular transport is provided by the interaction of unconventional myosin-Vb and its regulator Rab11a.^23,28 Recent whole-exome and whole-genome analyses revealed that various VUS need to be considered as contributors to cholestatic PFIC phenotypes. In the present study, we describe a patient diagnosed with MASLD, suffering from recurrent episodes of intrahepatic cholestasis and cholangiocarcinoma. Panel sequencing revealed 3 heterozygous variants in ABCB4, ABCB11, and MYO5B. The identified ABCB4 variant (NM_018849: c.1769G>A, p.[Arg590Gln]) is likely pathogenic (score 0.76 according to VASOR ²⁹ ). The identified ABCB11 variant (NM_003742: c.2628C>G, p.[Ile876Met]) is a VUS, which is not yet described in the literature, but it is scored with 0.099 by the AlphaMissense Pathogenicity prediction tool and, therefore, is rather classified as benign. Our hypothesis was that the underlying reason for the cholestasis is a decreased BSEP transport activity together with a disturbed polarization of the hepatocytes caused by the heterozygous MYO5B VUS (NM_001080467: c.2122C>T, p.[Arg708Trp]). To assess the hepatobiliary transport characteristics of this patient, we aimed to generate iPSC-based organoids for a comparison to organoids from a healthy donor and a donor suffering from loss of polarization in hepatocytes. ⁶ Subsequent comparative analyses using patient-specific iPSC-derived hepatic organoids and corresponding gene-corrected organoids would allow us to identify the influence of the given MYO5B VUS as a susceptibility factor for impaired bile salt transport.

FIGURE 1.

Flow chart for the generation of patient-specific iPSC-derived hepatic organoids and in vitro organoid-based disease modeling. Abbreviation: iPSC, induced pluripotent stem cell.

METHODS

Cell line generation

Human material and data used in this study was obtained in accordance with the principles of the Declaration of Helsinki, has been approved by the ethics committee of Hannover Medical School (No 2967-2015) and all patients and/or their parents provided written informed consent. Characterization of the established iPSC lines and quality assessment have been performed according to guidelines of the International Society for Stem Cell Research (ISSCR ³⁰ ). All iPSC lines acquired typical iPSC-like morphology, expressed pluripotency-associated nuclear and surface markers, and exhibited a normal karyotype (Supplemental Figure S1, http://links.lww.com/HC9/C129). Furthermore, all cell lines showed clearance of Sendai Virus RNA latest after passage 8, confirmed by qRT-PCR for each reprogramming factor (Supplemental Figure S1, http://links.lww.com/HC9/C129 and Supplemental Table S1, http://links.lww.com/HC9/C129). Confirmation of triple heterozygous variants (MYO5B^+/−, ABCB11^+/−, and ABCB4^+/−) was performed by Sanger sequencing (Supplemental Figure S1, http://links.lww.com/HC9/C129). Previously established iPSCs from a healthy control and a patient suffering from double heterozygous MYO5B mutations (c.1323-2A>G; c.2014A>T) were used during control experiments (Supplemental Figures S1–S6, http://links.lww.com/HC9/C129 and Suupplemental Tables S1-S3 http://links.lww.com/HC9/C129).

Hepatic differentiation

Differentiation was performed as previously described ¹¹ with minor changes. In short, cells were seeded as single cells on GelTrex-coated dishes (TPP) in E8 media containing 40 ng/mL FGF2 and 10 µM RHO-Kinase Inhibitor (Tocris) with a density of 120,000 cells/cm2. Obtained foregut endoderm cells were seeded with a density of 100,000 cells/50 µL in a GelTrex dome containing 10 µM Rho-Kinase inhibitor. After polymerization, hepatic liver organoid media was added for 96 hours with daily media change. Subsequently, liver specification media 1 and 2 were added for 24 and 72 hours, respectively. For final differentiation, the domes were cultivated with liver maturation media for another 72 hours.

Gene correction

Cells were cultivated as single cells under E8 conditions until they reached confluence. Cells were detached with Accutase, and 800,000 cells were resuspended in 1 mL E8 Media containing 10 µM RHO-kinase inhibitor. Ten micrograms recombinant Alt-R S.p. Cas9 nuclease (Integrated DNA Technologies) and 4 µg sgRNA (Supplemental Table S5, http://links.lww.com/HC9/C129) were incubated at RT for 10 minutes. After incubation, 200 µM ssODN (Supplemental Table S5, http://links.lww.com/HC9/C129) was added and placed on ice until electroporation. For electroporation, the Amaxa P3 primary cell 4D-Nucleofector X kit was used according to the manufacturer’s protocol. Immediately after pulse, cells were transferred to 4 mL E8 containing 40 ng/mL FGF2 and 3 µM RHO-kinase inhibitor and 0.5 µM Alt HDR enhancer (Integrated DNA Technologies). Cells were seeded with 200,000 cells/well in 12-well cell culture plates coated with GelTrex (1:200 diluted). Cells were incubated at 32 °C for 72 hours with daily media change (E8 with 20 ng/mL FGF2). After 72 hours, part of the bulk population was analyzed for gene correction efficiency and transferred to 37 °C. From a positive bulk population, clonal cell lines have been generated by seeding (limited dilution) on CF1 feeder cells in a 96-well plate with an initial final 10 µM RHO-kinase inhibitor. Single colonies were expanded and analyzed for proper gene correction and pluripotency.

CLF and Tauro-DBD assay

If not mentioned otherwise, the following protocol was always performed with the indicated inhibitor and RHO-kinase inhibitor (final 10 µM) within the given ice-cold buffer or media. All centrifugation steps were performed at 4 °C. All tubes and pipette tips were blocked with 1% BSA before use. After differentiation before assay, RHO-kinase inhibitor (final 10 µM) and inhibitor/no inhibitor were added to the media and incubated for an additional 1 hour at 37 °C. The cell culture plate containing gel domes with organoids was placed on ice and the supernatant was removed. Domes were washed twice with PBS, and Cell Recovery Solution (Corning) was added in the middle of the domes (domes should swell and detach) with a 1 mL pipette and slight pressure. The plate was placed on ice with orbital shaking (40 rpm). Within the next 45 minutes, GelTreX dissolved. After GelTrex dissociation, the mixture was transferred to 14 mL ice-cold PBS and centrifuged for 10 minutes at 70g w/o brake and medium acceleration. The obtained pellet was resuspended in ice-cold PBS and transferred into 1.5 mL reaction tubes for the hepatobiliary transport assays (all time points ±inhibition) with the respective substrate. Organoids were centrifuged for 2 minutes at 300g, supernatant was removed, and 100 µL media with or without inhibitor, but containing the respective fluorescent substrates cholyl-L-lysine fluorescein disodium salt (Merck, #810389P) or Tauro-nor-THCA-24-DBD (GenoMembrane, #GM7001), was added to each sample and incubated for the given time points at 37 °C. After incubation, substrate transport was stopped with 1 mL ice-cold PBS and centrifuged at 300 rpm for 2 minutes. Supernatant was removed, samples were washed with 1 mL PBS, and the procedure was repeated 3 times. Obtained organoids were seeded in 300 µL 37 °C warm HCM (containing inhibitor) in a 24-well plate and analyzed using an inverted fluorescence microscope (Olympus IX70) and CellSens software. For semi-quantitative analyses, at least 7 pictures with at least 7 organoids were taken in the bright-field channel. Corresponding fluorescent pictures were taken afterward with the same fluorescent settings for each channel. Using ImageJ, organoids from bright-field pictures were marked and their respective area was calculated. Next, an intensity threshold was defined for the fluorescence images (red or green for CLF or Tauro-DBD) and not altered during analysis. Subsequently, the fluorescence-positive area was determined by ImageJ, and the ratio of fluorescence-positive over the total organoid area was calculated.

Statistical analysis

Statistical analyses were performed with GraphPad Prism using a 1-way ANOVA. Tukey’s method was employed as a post hoc test. The results were obtained by at least 3 measurements (mean and SD) from at least 3 independent experiments. Differences were considered statistically significant at p values below 0.05 and are marked with asterisks (*p<0.05, **p<0.01, ***p<0.001, ****p<0.0001).

RESULTS

Generation of hepatic organoids

For hepatic differentiation, we adapted a protocol from Shinozawa et al ¹¹ and differentiated iPSCs under 2D conditions toward the hepatic lineage until an anterior-foregut endodermal-like stage was reached. These endodermal progenitor cells exhibited robust expression of the transcription factor SOX17, hepatocyte nuclear factor 3-beta (FOXA2), and hematopoietically-expressed homeobox protein (HHEX) (Supplemental Figure S2A, http://links.lww.com/HC9/C129 and Supplemental Table S3, http://links.lww.com/HC9/C129). Flow cytometry analyses verified the expression of the endodermal marker C-X-C chemokine receptor type 4 (CXCR4; CD184) and the stem cell growth factor receptor Kit (C-KIT; CD117) with a population of 98.4% double-positive cells (Supplemental Figure S2B, http://links.lww.com/HC9/C129 and Supplemental Table S7, http://links.lww.com/HC9/C129). After passaging or thawing of cryopreserved aliquots, cells were seeded in GelTrex domes, and a 3D differentiation protocol was applied. Mature hepatic organoids showed round shape morphology with an inner lumen (Supplemental Figure S2C, http://links.lww.com/HC9/C129) and exhibited strong mRNA expression of the affected canalicular transporters BSEP and phosphatidylcholine translocator/multidrug resistance protein 3 (MDR3) as well as the motor protein unconventional myosin-Vb. Furthermore, hepatic organoids showed expression of albumin (ALB), hepatocyte nucleus factor 4α (HNF4α), and transthyretin (TTR), further supporting a phenotype of functional hepatic cells (Supplemental Figure S2D, http://links.lww.com/HC9/C129 and Supplemental Table S4, http://links.lww.com/HC9/C129).

Apical-basolateral polarization of iPSC-derived hepatic organoids

Immunofluorescence analyses from liver tissue and iPSC-derived organoids reflected important alterations in the subcellular distribution of canalicular transporters. In the healthy liver, OATP1B1 ³¹ is expressed at the basolateral membrane, while BSEP^32,33, MRP2 ³⁴ , and ZO-1-positive tight junctions are located around the canalicular/apical membrane, demonstrating a normal polarization of healthy hepatocytes (Figure 2A). In the patient suffering from an intrahepatic cholestasis caused by 2 heterozygous MYO5B variants, depolarized hepatocytes showed a disturbed phenotype. Canalicular structures were rarely detectable, and the clear canalicular expression of MRP2 surrounded by ZO-1 is scarce, while blurry coexpression of both proteins indicated a hampered polarization (Figure 2B). Moreover, most of the MRP2 signal is within the cytoplasm of the cells (Figure 2B). Similar observations were made for OATP1B1, which showed some areas of basolateral localization, whereas in other areas, no membrane-located OATP1B1 was detectable (Figure 2B). However, all hepatocytes expressed considerable amounts of OATP1B1 in the cytoplasm. In line with these findings, the localization of BSEP was only partially at canalicular structures, while the larger proportion of cells demonstrated an intracellular BSEP signal, indicating that the protein is not properly processed to the canalicular membrane (Figure 2B). In the triple heterozygous patient’s liver, the immunofluorescence analyses reveal a mixed phenotype. There are areas with clear basolateral OATP1B1 expression and apical BSEP expression (Figure 2C). However, in close proximity, both proteins show a cytoplasmic localization without a clear border (Figure 2C). Both membrane and cytoplasmic expression show areas in which the signal suggests a granulated localization. Similarly, ZO-1 and MRP2 expression shows apical localization, but some areas also show a more cytoplasmic localization (Figure 2C).

FIGURE 2.

(A–C) Immunofluorescence analyses of liver tissue. FFPE sections were analyzed for expression of the BSEP and ZO-1 (red) as well as OATP1B1 and MRP2 (green) in a healthy liver sample (A), a MYO5B control liver (B), and the triple heterozygous patient’s liver (HC01) (C). All sections were counterstained with DAPI (blue). (D–I) Immunofluorescence analyses of hepatic organoids. FFPE sections were analyzed for expression of BSEP, OATP1B1, and MRP2 (green). ZO-1 (red) served as a control for canalicular structures. All sections were counterstained with DAPI (blue). Healthy control hepatic organoids (D) and MYO5B^−/− organoids (E) served as controls. HC01 organoids are shown without gene correction (F), ABCB11 gene corrected (G), MYO5B gene corrected (H), and ABCB11 and MYO5B double gene corrected (I). Abbreviations: BSEP, bile salt export pump; FFPE, formalin-fixed paraffin-embedded; MYO5B, Myosin-Vb.

Regarding the histology of the iPSC-derived organoids, GelTrex domes consisting of multiple organoids were embedded in paraffin before sectioning. Individual sections (n≥5) were analyzed, and representative organoids, where the luminal structures have been cut in the section plane, were depicted (Figures 2D–I). Those inner luminal membranes reflect the apical/canalicular membrane, whereas the cell borders toward the outside and neighboring cells represent the basolateral membrane. Organoids differentiated from healthy control iPSCs reflect the proper canalicular localization of MRP2, ZO-1, and BSEP as well as the basolateral-specific localization of OATP1B1 (Figure 2D), indicating a normal polarization of the hepatic cells within the organoid. In contrast, organoids derived from the MYO5B patient (c.1323-2A>G; c.2014A>T [MYO5B ^−/− control]) reflected a disturbed localization pattern. Most of the canalicular transporters MRP2 and BSEP were located in the cytoplasm, and only a few membrane-associated fluorescent signals were detected at the organoids’ lumen (Figure 2E). OATP1B1 was not restricted to the basolateral parts but also to the membrane facing the lumen (Figure 2E). Although ZO-1 was located at the canalicular space in most organoids, some organoids showed expression at the basolateral membrane (Figure 2E). Interestingly, most of the triple heterozygous patient iPSC-derived organoids (HC_01) exhibit an inner lumen surrounded by ZO-1, indicating a polarized phenotype (Figure 2F). Moreover, MRP2 was predominantly located at the membrane of the inner lumen and flanked by ZO-1 (Figure 2F). The basolateral localization of OATP1B1 differed within the organoid population. Some organoids showed basolateral localization, whereas others exhibited a more cytoplasmic localization. However, BSEP demonstrated no distinct localization pattern within the organoids (Figure 2F). Taken together, these patient iPSC-derived organoids reflect, in terms of transporter localization, the 2 patients’ phenotypes.

Substrate-specific transporter activity

Next, we performed transport assays in iPSC-derived organoids exploiting fluorescent substrates that were transported into luminal canalicular structures (Supplemental Figure S2E, http://links.lww.com/HC9/C129). ¹¹ As MRP2 was not affected in any of the cell lines, we used the MRP2-mediated export of the fluorescein-conjugated bile acid CLF as the control for proper hepatobiliary transport within our cell system. Organoids generated from the healthy control cell line started CLF transport already after 5 minutes of incubation (9.1%±5.6%). Over time, the number of organoids positive for CLF increased constantly up to 87.2%±8.4% after 45 minutes (Figure 3A). As the percentage of positive organoids did not significantly increase from 45 to 60 minutes (87.2%±8.4% vs. 89.2%±4.8%), we considered 45 minutes as sufficient to reach saturation in this assay. Therefore, time points of 0, 5, 10, 20, and 45 minutes have been used to analyze transporter activity in further experiments. Applying 100 µM of the inhibitor Cyclosporin A (CsA) decreased the transport of CLF significantly (p≤0.001) for the analyzed time points of 15 minutes and later (Figure 3A). As CsA inhibits MRP2 and BSEP at the used concentration of 100 µM,^34,35 we performed the same experiment with the MRP (1, 2, 4, 5)-specific inhibitor MK571 (L-660711) as well as with the BSEP-specific inhibitor PSC833 (Valspodar) after 45 minutes of incubation for CLF and Tauro-DBD (Figures 3B, C). For the CLF transport, the inhibition of BSEP with 0.4 µM PSC833 did not significantly reduce the CLF transport (86.6%±5.6% vs. 84.9%±6.1%; Figure 3B) but reduced the Tauro-DBD transport significantly (92.9%±5.9% vs. 0.1%±0.02% p≤0.001; Figure 3C). In contrast, the inhibition of MRP2 with MK571 reduced the CLF transport significantly (86.6%±5.6% vs. 1.9%±0.9%, p≤0.001; Figure 3B) but did not significantly affect the Tauro-DBD transport (92.9%±5.1% vs. 87.7%±8.4%; Figure 3C). These data provide strong evidence for a specific transport of CLF by MRP2 and specific transport of Tauro-DBD by BSEP in our human iPSC-derived hepatic organoids.

FIGURE 3.

(A–F) Substrate-specific transport assays and transport-specific inhibition. MRP2-specific CLF transport assay was analyzed in a time frame of 60 minutes in organoids from a healthy control iPSC line without inhibition (filled circles) and with inhibition (empty circles) using 100 µM CsA (A). Transporter inhibition in healthy control (black circles) after 45 minutes with BSEP-specific inhibitor PSC833, broad range inhibitor CsA and MRP2-specific inhibitor MK571 for CLF transport (B) and Tauro-DBD transport (C). The MYO5B control (blue triangle) and the HC_01 organoids (red square) were compared with the healthy control (black circle) for their CLF transport capacity over 45 min. Depolarized MYO5B control shows reduced CLF transport capacity (D), whereas the HC_01 and healthy control organoids are fully functional for MRP2-dependent transport (E). For the Tauro-DBD transport, the depolarized control shows only minor but constantly increasing Tauro-DBD transport over time, comparable to the CLF transport (F); the patient-specific organoids exhibit no detectable BSEP-mediated transport activity at any time point (G). Abbreviations: BSEP, bile salt export pump; CLF, cholyl-lysyl-fluorescein; iPSC, induced pluripotent stem cell; MRP2, multidrug resistance-related protein 2; MYO5B, Myosin-Vb.

Finally, we applied the CLF transport assays to the triple heterozygous (HC_01) organoids in comparison to the MYO5B control organoids and the healthy control organoids. As expected, the MYO5B control exhibited a significantly reduced number of CLF-positive organoids after 20 and 45 minutes of incubation (54.9%±15.2% vs. 5.3%±3.9% at 20 min and 85.9%±2.7% vs. 16.1%±1.9% at 45 min, p≤0.0001; Figure 3D). Interestingly, the HC_01 organoids showed no significant differences in CLF transport over all time points (Figure 3E), demonstrating the proper uptake and MRP2-dependent export of conjugated substrates in these organoids. To investigate the BSEP-mediated export, we performed another set of experiments with the BSEP-specific substrate Tauro-DBD. Again, MYO5B ^−/− control exhibited a marked reduction of BSEP-mediated Tauro-DBD transport (Figure 3F). Interestingly, the triple heterozygous organoids HC_01 showed no BSEP-mediated transport at any time point (Figure 3G), indicating a functional loss of BSEP-mediated substrate transport.

Ursodeoxycholic acid treatment of patient-specific organoids

As we could observe barely any BSEP-dependent transport in our in vitro system of HC_01 organoids, we investigated whether BSEP activity might be induced by ursodeoxycholic acid (UDCA), a commonly used drug in patients with cholestasis, including our patient. This hydrophilic bile acid is nontoxic to hepatocytes, induces BSEP and MDR3 expression, and the (partial) response in patients is ~35%–40% in low-GGT PFIC and ~70% in high-GGT PFIC.^36,37 We treated the organoids with low doses (150 µM) and high doses (450 µM) of UDCA for 72 hours in fully differentiated organoids (Figure 4A). Here, the Tauro-DBD assay reproduced the almost complete absence of transport (0.56%±0.38%) in the untreated control, whereas a slight response in the low-dose group (2.9%±1.4%) and a significant response (9.8%±5.1%, p≤0.01) in the high-dose group could be observed (Figure 4A). However, corresponding qRT-PCR analyses for the expression of BSEP mRNA showed no significant difference in BSEP mRNA expression between the untreated control and the low-dose and high-dose UDCA-treated organoid samples (Figure 4B), indicating UDCA’s mode of action is rather via chaperone effects supporting BSEP functional expression. ³⁸

FIGURE 4.

Tauro-DBD assay with UDCA-treated organoids. HC_01 organoids were incubated for 72 hours before the Tauro-DBD assay without UDCA, with 150 µM UDCA and with 450 µM UDCA. Subsequent Tauro-DBD assay shows UDCA response in UDCA-treated organoids (A). Corresponding qRT-PCR analyses show no differences between the treated organoids (B) (n=3; GAPDH internal control, **p<0.01). Abbreviation: UDCA, ursodeoxycholic acid.

ABCB11 gene correction in patient-specific organoids

Next, we aimed to investigate the impact of the heterozygous ABCB11/BSEP variant p.(Ile876Met) by applying CRISPR/Cas9-mediated gene editing to obtain proper syngeneic control cells with the wild-type ABCB11 genotype. We used an ssODN-based gene correction approach in our patient-specific iPSC and generated ABCB11-corrected (MYO5B ^+/− , ABCB4 ^+/− , ABCB11 ^+/+ ) subclones for further investigations (Supplemental Figure S3, http://links.lww.com/HC9/C129 and Supplemental Table S6, http://links.lww.com/HC9/C129). Before differentiation, the obtained cell line was characterized for full pluripotency and genetic stability. When the Tauro-DBD assay was performed with hepatic organoids derived from these cells, we were still unable to detect Tauro-DBD transport in the ABCB11 gene-corrected patient-specific organoids (Figure 5A), whereas the healthy control organoids transported Tauro-DBD in a robust manner in the same assay. To confirm the general transport capacity of the ABCB11 gene-corrected cell line, we performed the CLF assay in parallel (Figure 5B). Here, robust CLF transport in the ABCB11 gene-corrected cell line could be observed, and the specific transporter activity had been evaluated by MRP2 inhibition with MK-571 (Figure 5B). Moreover, no apical (canalicular) localization of BSEP protein was detected in the ABCB11 gene-corrected organoids by immunofluorescence analysis (Figure 2G).

FIGURE 5.

Tauro-DBD and CLF assay with ABCB11 gene-corrected patient-specific cell line. No Tauro-DBD transport was observed using the ABCB11 gene-corrected cell line (red squares), whereas the healthy control (black circles) shows normal Tauro-DBD transport (A). Functional CLF transport was demonstrated in ABCB11 gene-corrected organoids (filled red squares), indicating the proper polarity and maturation. MRP2-specific inhibition (MK571) resulted in a blocked transport (empty red squares) of the CLF substrate in the ABCB11 gene-corrected-organoids (B). Abbreviations: CLF, cholyl-lysyl-fluorescein; MRP2, multidrug resistance-related protein 2.

MYO5B gene correction and partial restoration of transporter activity

As the correction of the ABCB11/BSEP variant p.(Ile876Met) did not result in a restored BSEP-mediated bile salt transport, the loss of functional BSEP transporter activity seems to be caused by a different mechanism. As previously described, variants in motor proteins can lead to disturbed polarization, which ultimately can result in cholestasis. ⁶ Therefore, we further examined the impact of the MYO5B variant in the patient’s (MYO5B ^+/− , ABCB4 ^+/− , ABCB11 ^+/− ) cell line. To investigate the impact of this variant in a syngeneic cell system, we again targeted the iPSC by a CRISPR/Cas9-mediated gene editing approach and characterized the generated subclones for pluripotency and genetic stability corresponding to the previous experiments (Supplemental Figure S4, http://links.lww.com/HC9/C129 and Supplemental Table S6, http://links.lww.com/HC9/C129). The obtained cell line (MYO5B ^+/+, ABCB4 ^+/−, ABCB11 ^+/−) was differentiated into hepatic organoids to investigate the effect of the MYO5B gene in terms of BSEP-mediated Tauro-DBD transport. In these assays, the MYO5B gene correction resulted in a restored BSEP-specific Tauro-DBD transport (Figure 6A). Compared with the healthy control, we observed a slightly delayed and marginally weaker Tauro-DBD transport into the organoids’ lumen. Specifically, we observed a significantly higher Tauro-DBD transport within the healthy control after 20 and 45 minutes (18.2%±12.1% vs. 58.9%±17.8%; p≤0.001 and 52.7%±14.0% vs. 78.9%±10.1%; p≤0.01), respectively. The BSEP-specific transport could be further characterized as PSC833 inhibited the transport at all time points (Figure 6A). The corresponding CLF assay shows similar results compared with the CLF assay in ABCB11 gene-corrected and uncorrected HC_01 organoids (Figures 6B, 5B, and 3D), indicating a proper MRP2-mediated transport in the hepatic organoids derived from the MYO5B-corrected iPSC subclones. In contrast to the ABCB11 gene correction, the MYO5b gene-corrected organoids show a predominant apical localization of both BSEP and MRP2 protein (Figure 2H). Furthermore, OATP1B1 protein is associated with the basolateral membrane domain (Figure 2H).

FIGURE 6.

Tauro-DBD and CLF assay in MYO5B gene-corrected HC_01 organoids. HC_01 organoids (filled red squares) show a partially restored specific BSEP-mediated Tauro-DBD transport after MYO5B gene-correction (A) compared with healthy control (filled black circles). The specificity had been proven by PSC833-mediated inhibition of BSEP (empty red squares). The CLF transport remained mostly unaffected by the MYO5B gene-corrected organoids (filled red squares) compared with healthy control (filled black circles) organoids (B). Abbreviations: BSEP, bile salt export pump; CLF, cholyl-lysyl-fluorescein; MYO5B, Myosin-Vb.

ABCB11 and MYO5B gene correction restores bile salt transport

As the MYO5B gene correction did not fully restore the BSEP-specific Tauro-DBD transport, we asked whether the correction of the mutant ABCB11 allele affected the MYO5B-corrected cell line. Thus, we applied the same ABCB11 gene-correction approach as described above to generate a double corrected (MYO5B ^+/+ , ABCB4 ^+/− , ABCB11 ^+/+ ) cell line (Supplemental Figure S5, http://links.lww.com/HC9/C129). Further analyses of the Tauro-DBD transport assays performed with hepatic organoids of these double gene-corrected iPSCs revealed no differences from the healthy control (Figure 7A). This indicates an impact of the BSEP variant p.(Ile876Met) and suggests classifying this VUS as functionally relevant. As expected, the corresponding CLF assay did not show any differences either in kinetics or in percent of CLF-positive area (Figure 7B, Supplemental Figure S6, http://links.lww.com/HC9/C129), which confirms the robustness and reproducibility of the fluorescent substrate–based transport assays in the iPSC-derived hepatic organoids. ABCB11/MYO5B double gene correction leads to a restoration of apical expression of BSEP and MRP2 protein, as well as the basolateral expression of OATP1B1 protein, comparable to the healthy control (Figure 2I).

FIGURE 7.

Tauro-DBD and CLF assay in MYO5B and ABCB11 double gene-corrected organoids. After correction of the ABCB11 gene in the MYO5B gene-corrected HC_01 cell line (filled red squares), no differences in the kinetics of total transport capacity to the healthy control (filled black circles) are recognizable (A). Specificity of BSEP transport is demonstrated by PSC833 inhibition (empty red squares) over all time points (A). The BSEP-independent transport capacity of CLF via MRP2 is the same for both cell lines (B). Abbreviations: BSEP, bile salt export pump; CLF, cholyl-lysyl-fluorescein; MYO5B, Myosin-Vb.

DISCUSSION

In the present study, we investigated the contribution of 2 different gene variants to the complex liver disorder in 1 patient. Aside from a metabolic dysfunction–associated steatotic liver disorder (MASLD) and a curative resection of an early-stage CCA, the patient suffered from episodes of jaundice, elevated liver function tests (LFTs), and intermittent pruritus. Panel sequencing for cholestasis-associated genes revealed no biallelic variants in PFIC-associated genes but 3 different heterozygous variants in ABCB11, ABCB4, and MYO5B. To investigate the impact of multiple potentially relevant heterozygous variants in 3 different genes on intrahepatic cholestasis, we generated patient-specific iPSC-derived hepatic organoids based on a published protocol from Shinozawa et al. ¹¹ Intrahepatic cholestasis is associated with disturbances of various cellular functions and may affect different transport proteins to a different extent. Moreover, the specificity of such transporters for a given substrate varies between species.^16,19 As depicted in Figure 3B, CLF transport was not affected by the BSEP-specific inhibitor PSC833, but strongly reduced in the presence of the MRP2-specific inhibitor MK571. However, Shinozawa et al ¹¹ demonstrated abrogated CLF-transport capabilities after knockout of BSEP/ABCB11, which might be due to subtle differences in the hepatic organoid features or other experimental alterations. In our hands, the conjugated bile salt analog CLF serves as a substrate for human MRP2 (conjugate export pump)-dependent transport, which can be inhibited by MK-571. Human BSEP activity, which is the focus of our presented study, has been analyzed by Tauro-DBD, and we could confirm, as suggested earlier, ¹⁹ that Tauro-DBD is an appropriate substrate for human BSEP-mediated transport, which can be specifically inhibited by PSC833.

As a prerequisite for a vectorial transport, hepatic organoids must be comprised of hepatocyte-like cells exhibiting proper polarization as observed in our healthy control organoids (Figure 2D). In contrast, in MYO5B-deficient cells, which we used as a loss-of-polarization-control, no clear distinction between basolateral and apical membrane domains could be detected (Figure 2E). In these cells, the motor function is impaired due to the variant c.1323-2A>G, causing exon 11 skipping (p.69-761^39,40) and variant c.2014A>T in exon 17, causing nonsense-mediated mRNA decay (NMD).^41,42 In consequence, a disturbed localization and cytosolic presence of the apical transporters MRP2 and BSEP, as well as the basolateral transporter OATP1B1, can be observed in the liver tissue and the organoids of this MYO5B-deficient control (Figures 2B, E). For the triple heterozygous patient-specific HC_01 organoids, we observed a mixed phenotype. On the one hand, we see a distinct inner lumen bordered by ZO-1 and MRP2. On the other hand, BSEP is not clearly localized at the canalicular space, which indicates a dysfunctional intracellular BSEP trafficking (Figures 2C, F).

In our transporter assays, we could show that the transport capacity of the loss-of-polarization-control was reduced by ~70% compared with healthy control (Figures 3A, D), which suggests that the system sufficiently determines functional alterations between different patient-related and control iPSC lines. In the experiments with the HC_01 patient-specific organoids, immunofluorescence staining revealed BSEP expression, rather in the cytoplasm but not at the apical membrane, and we determined a total loss of BSEP-mediated Tauro-DBD transport (Figure 3F). These findings were unexpected, because the patient was not experiencing cholestasis in early childhood as would be expected in cases with a severe loss of BSEP function, but developed symptoms at a later age in the context of chronic liver injury presumably caused by metabolic dysfunction–associated steatotic liver disease (MASLD). In our iPSC-derived hepatic organoid system, the proper formation of tight junctions, depicted by ZO1, and canalicular localization of MRP2 (Figure 2F) as well as the intact MRP2-mediated CLF transport (Figure 3D) suggest that there are no general disturbances in hepatic specification and polarization within this cell line. Since the patient benefited from treatment with UDCA, which can act by enhancing BSEP expression and by supporting protein processing in a chaperone-like manner, ⁴³ we treated the organoids with different UDCA concentrations to test the influence on Tauro-DBD transport. With increasing UDCA concentration, we were able to detect a slightly increased Tauro-DBD transport (Figure 4A) without any change in the ABCB11 transcript level (Figure 4B). With regard to the immunofluorescence analyses, which depicted a BSEP expression that is not strongly restricted to the apical membrane, but localized in the cytoplasm, we speculate that the enhanced BSEP-mediated transport activity is most probably a result of a slightly facilitated intracellular processing toward a proper apical membrane integration of BSEP. Thus, it remains unclear if the studied BSEP variant results in a misfolded protein, which is slightly better processed after UDCA treatment, or if its intracellular trafficking with involvement of unconventional myosin 5b and its cofactor Rab11a^6,23 is better supported by UDCA. However, after correction of the BSEP p.(Ile876Met) variant, we still could not detect any relevant Tauro-DBD transport (Figure 5), which argues in the direction that the BSEP variant has only a minor effect on the overall loss of BSEP-mediated transport activity. This interpretation is further supported by the cytoplasmic localization of the BSEP protein in the immunofluorescence analysis (Figure 2G).

For subsequent analyses, we generated a MYO5B-corrected iPS cell line and differentiated hepatic organoids from these cells, in which we were able to detect improved intracellular trafficking of BSEP (Figure 2H). Furthermore, Tauro-DBD transport was restored to a significant level, suggesting a relevant rescue of BSEP transport upon MYO5B correction (Figure 6). As in these MYO5B-corrected iPSC-derivatives, the heterozygous BSEP variant p.(Ile876Met) is still present, it was not too unexpected that these hepatic organoids exhibited a slightly delayed and ~25% lower BSEP-mediated Tauro-DBD transport, whereas the MRP2-mediated CLF transport was comparable to the healthy controls. These results emphasize the impact of the unconventional myosin 5b variant p.(Arg708Trp) on BSEP trafficking and give a hint that the BSEP p.(Ile876Met) variant only has minor consequences for the hepatobiliary transport capabilities. This assumption is further supported by the findings of our final experiments, in which we used iPSC derivatives with correction of both variants, unconventional myosin 5b p.(Arg708Trp) and BSEP p.(Ile876Met). In the hepatic organoids obtained from these cells, BSEP expression was restricted to the apical membrane domain (Figure 2I), and neither transport of Tauro-DBD nor CLF was different from healthy control organoids (Figure 7, Supplemental Figure S6, http://links.lww.com/HC9/C129).

Our findings clearly demonstrate the impact of the unconventional myosin 5b variant p.(Arg708Trp), but it remains unclear why our in vitro data show an overpronounced loss of BSEP transport function, whereas the patient developed recurrent episodes of intrahepatic cholestasis mainly in the context of further MASLD-associated chronic liver damage. This discrepancy might be due to the fact that unconventional myosin 5b is configured as a homodimer, resulting in the majority of unconventional myosin 5b-dimers harboring at least 1 mutated molecule. Furthermore, MYO5B is known to be spliced in a tissue-dependent manner, ⁴⁴ which can result in different variants executing different tasks in different tissues. ⁴⁵ Finally, variant-dependent splicing ⁴⁶ may occur as described for the MYO5B c.1323-2A>G variation, which contributed to the phenotype of our polarization-deficient control cells (Figures 2B, 3D). In this respect, alternative splice variants putatively modulated by the rather artificial in vitro differentiation and cultivation systems also may explain this data.

In conclusion, our study highlights the importance of unconventional myosin-Vb as a relevant factor in intrahepatic cholestasis and demonstrates that heterozygous MYO5B variants should also be considered disease-associated factors, as they may sensitize the susceptibility to disturbed hepatobiliary transport processes.

Supplementary Material

hc9-9-e0812-s001.pdf ^{(5.6MB, pdf)}