Human Induced Pluripotent Stem Cell-derived Hepatocyte-like Cells Provide Insights on Parenteral Nutrition Associated Cholestasis in the Immature Liver

Parenteral nutrition-associated cholestasis (PNAC) significantly limits the safety of intravenous parenteral nutrition (PN). Critically ill infants are highly vulnerable to PNAC-related morbidity and mortality, however the impact of hepatic immaturity on PNAC is poorly understood. We examined developmental differences between fetal/infant and adult livers, and used human induced pluripotent stem cell-derived hepatocyte-like cells (iHLC) to gain insights into the contribution of development to altered sterol metabolism and PNAC. We used RNA-sequencing and computational techniques to compare gene expression patterns in human fetal/infant livers, adult liver, and iHLC. We identified distinct gene expression profiles between the human feta/infant livers compared to adult liver, and close resemblance of iHLC to human developing livers. Compared to adult, both developing livers and iHLC had significant downregulation of xenobiotic, bile acid, and fatty acid metabolism; and lower expression of the sterol metabolizing gene ABCG8. When challenged with stigmasterol, a plant sterol found in intravenous soy lipids, lipid accumulation was significantly higher in iHLC compared to adult-derived HepG2 cells. Our findings provide insights into altered bile acid and lipid metabolizing processes in the immature human liver, and support the use of iHLC as a relevant model system of developing liver to study lipid metabolism and PNAC.


Introduction
addition, genome-wide association studies suggest that ABCG5 and ABCG8 genetic variation in the normal population may play a biological role in altered plasma lipid concentrations 15,16 . Thus, proper expression of ABCG5 and ABCG8 is required for plant sterol clearance.
Normal mechanisms of eliminating sterols may be impaired in the immature liver and contribute to the development of PNAC. Higher plant sterol levels have been described in infants receiving soy lipids compared to older children with soy lipids 6 .
During a short period of soy lipid therapy, we found higher accumulation of plant sterols in very preterm infants compared to more mature infants 17 . These findings suggest that the developmental expression of critical sterol-regulating genes may be of particular significance in the development of PNAC in infants dependent on TPN.
There are little data regarding the expression of sterol-regulating genes, including ABCG5 and ABCG8, during normal human development. A developmental inability to regulate xenosterol levels may predispose the immature liver to sterol accumulation and PNAC when challenged with intravenous lipids containing plant sterols. Furthermore, our understanding of hepatic lipid metabolism during development is limited because of difficulties obtaining liver samples in fetal and infant populations for experimental studies. Stem cell technology has enabled researchers to recapitulate hepatogenesis from human stem cells in vitro to allow studies that are difficult to perform in humans.
Human induced pluripotent stem cells (iPSC) can be differentiated into a variety of tissue-specific cell types, and iPSC-derived hepatocyte-like cells (iHLC) are a promising system for modeling the developing human liver and hepatic metabolism 18,19 . Reports suggest that iHLC more closely resemble immature hepatocytes than adult hepatocytes 20 , supporting their potential as a model for PNAC in infants. The primary objective of this study was to identify transcriptome differences in sterol metabolizing pathways between fetal/infant and adult livers. The secondary objective was to explore the contribution of development to altered sterol metabolism and PNAC using iHLC.

Results
Global expression differences: developing liver vs adult liver. We focused on gestational ages eligible for neonatal intensive care unit (NICU) care and examined available, non-diseased liver tissue at clinically-relevant gestational ages from the National Institutes of Health (NIH) Neurobiobank. One liver sample each at 20, 22, 25, and 39 weeks gestation were available for this study. Information regarding fetal/infant liver tissue used in this study are shown in Table 1.
We first determined whether gene transcripts determined by RNA-sequencing (RNA-seq) of human developing livers were different from human adult liver (downloaded from the UCSD Human Reference Epigenome Mapping Project and reanalyzed in parallel to minimize differences because of analytic pipelines). For a global comparison, we first performed principal component analysis (PCA) of RNA-seq data from fetal/infant samples and adult liver (Fig. 1a). All 4 fetal/infant samples clustered together and were distinct from the adult liver, suggesting that the various stages of fetal/infant liver were substantially more similar to each other than to adult liver. Given the clustering of fetal/infant samples, normalized gene expression data were averaged for fetal/infant samples to represent the developing liver stage in subsequent comparisons between developing liver and adult liver stages.
To focus on the most substantial transcriptome differences between developmental stages, differentially expressed genes (DEG) were defined as those with at least a 4-fold change difference in expression (|Log2FC| > 2) and adjusted p-value < 0.05, in the developing liver compared to adult liver. A total of 1,928 genes were differentially expressed between stages (1239 (64%) were upregulated and 689 (36%) downregulated in developing liver versus adult liver). All up-and downregulated DEG in developing liver compared to adult are shown in Supplementary Tables S1 and S2, respectively. Interestingly, several genes involved in bile acid and sterol metabolism were among the most significantly downregulated DEG in developing liver compared to adult. Specifically, developing liver had lower expression of genes involved in canalicular transport of conjugated bilirubin and bile acids (ABCC2, ABCB11) and phospholipids (ABCB4); alternative export of bile acids (ABCC3); elimination of sterols (ABCG8); conversion of cholesterol to bile acids (CYP8B1); and bile acid conjugation and detoxification (SLC27A5, SULT2A1) as shown in Fig. 1b. The expression of select genes were validated using real-time qRT-PCR of the fetal/infant samples and six adult cadaveric liver samples, and confirmed significantly lower expression of ABCB4, ABCC3, and ABCG8; and trends in lower expression of ABCC2, ABCB11, and CYP8B1 in fetal/infant compared to adult cadaveric liver tissue (Fig. 1c). The significantly lower expression of these genes suggests decreased metabolism of bile acids and sterols in the immature compared to adult liver.
Downregulation of xenobiotic, bile acid, and fatty acid metabolism pathways in developing liver. To examine directional enrichment of gene sets differentially expressed between developing liver and adult liver stages, untargeted analysis of the RNA-seq dataset was performed. All gene expression data were analyzed with Gene Set Enrichment Analysis software (GSEA) 21 using the Molecular Signatures Database Hallmark gene set collection 22 . Seven gene sets were enriched in developing liver and 16 were enriched in the adult liver with FDR q-value < 0.05. Developing liver was associated with upregulation of heme metabolism and processes involved in cell division including E2F targets, G2M checkpoint, mitotic spindle, DNA repair, and spermatogenesis. In contrast, several metabolic pathways including xenobiotic, bile acid, and fatty acid metabolism and peroxisome function were among the most downregulated gene sets associated with developing liver as shown in Table 2. The full list of gene set enrichment between developing and adult liver is shown in Supplementary Table S3. Collectively, this indicates that the immature liver has inhibition of key aspects of hepatic function which may play a role in the metabolism of lipids.
Gene expression differences: iHLC vs developing liver vs adult liver. We next determined whether human iHLC were more similar to human developing liver or adult liver. Using a previously described protocol 18 , iPSC were differentiated into iHLC  S2b). These data show that we are able to recreate procedures from published literature to produce highly differentiated hepatocytes ( Supplementary Fig. S2a-d).
RNA-seq was performed on iHLC from six independent differentiations and compared to gene expression data from the fetal/infant and adult liver samples to characterize the similarities and differences among the three groups. We found that iHLC were globally much more similar to fetal/infant liver samples by PCA (Fig. 2a).
Among the DEG with at least a 4-fold difference in expression (|Log2FC| > 2) and  Supplementary Table S4. Within the bile acid metabolism gene set, ABCG8 was among the most downregulated differentially ranked genes in both developing liver and iHLC compared to adult (Fig. 2c). These data show that iHLC are an appropriate model system for human immature liver.

Lipid accumulation in iHLC and HepG2 cells exposed to exogenous stigmasterol.
Given the similarities between developing liver and iHLC in global gene expression, gene set enrichment, and downregulation of ABCG8, the susceptibility to steatosis with immaturity was tested in iHLC exposed to an exogenous plant sterol, stigmasterol Intracellular lipid accumulation was then compared in iHLC and HepG2 cells exposed to stigmasterol for 24 hours. While the intensity of fluorescence of iHLC increased with increasing concentrations of stigmasterol and approached that of the positive control (Fig. 3b), the fluorescence of the stigmasterol-exposed HepG2 cells resembled that of the vehicle (Fig. 4a). Notably, 75 µM stigmasterol induced a greater degree of steatosis in iHLC than in HepG2 cells (Fig. 4b).

Discussion
In this study, we used RNA-seq to characterize developmental differences in the transcriptome of human fetal/infant livers compared to adult liver. We identified developmentally distinct gene expression profiles with significant downregulation of xenobiotic, bile acid, fatty acid metabolism and peroxisomal function in the immature liver compared to the adult. The immature liver also had significantly lower expression of the sterol metabolizing gene ABCG8, involved in the direct excretion of sterols from hepatocytes. In assessing iHLC as a model for the developing liver, we found that gene expression patterns in iHLC closely resemble developing liver, and exhibited a similar inhibition profile for sterol metabolizing genes compared to adult liver. When challenged with exogenous stigmasterol, we found that iHLC accumulate intracellular lipid in a concentration-dependent manner, and that the degree of stigmasterol-induced steatosis was greater in iHLC than in HepG2 cells.
Intravenous nutrition is crucial for survival and neurodevelopment of infants unable to tolerate enteral feeds [25][26][27] . However, PNAC significantly limits the safety of prolonged TPN in critically ill neonates, with preterm infants having the highest diseaserelated morbidity and mortality 1-3 . This complication is strongly associated with serum and liver accumulation of potentially hepato-toxic plant sterols in parenteral soy lipid emulsions 4,8 . Importantly, infants receiving soy lipids have two-to five-times higher plant sterol levels compared to older children with soy lipids 6 . During a short period of TPN with soy lipid, we found greater cumulative exposure to plant sterols in very preterm infants compared to more mature infants 17 . Furthermore, cholesterol response to soy lipid is inversely related to gestational age and birth weight 28 . Therefore, Despite recapitulating many human hepatocyte functions, iHLC have been reported to have a more immature phenotype resembling the fetal or neonatal liver 19,33 and suggest that iHLC may be relevant for modeling the immature liver.
We mimicked hepatocyte development using iPSCs and compared the gene expression profiles among iHLC, developing liver, and adult liver. We found a high degree of overlap between global transcriptome profiles of fetal/infant liver samples and iHLC, indicating a large amount of similarity. By functional annotation, several downregulated pathways that distinguished developing liver from adult liver were also apparent in the comparison between iHLC and the adult liver. Notably, the expression of ABCG8 was among the most downregulated genes in both developing liver and iHLC compared to adult comparisons.
Due to the similarities between iHLC and developing liver, we used iHLC to study the susceptibility to steatosis that may occur in the immature liver by exposing iHLC to an exogenous plant sterol in setting of decreased ABCG8. We found that iHLC accumulate intracellular lipid in a concentration-dependent manner when exposed to stigmasterol in ranges seen in the plasma of infants receiving TPN with soy lipid 17  Obtaining suitable tissue for studies examining organ changes that occur during early human development is complicated by ethical and logistical issues. Although all available, non-diseased liver tissue at clinically-relevant gestational ages from the NIH Neurobiobank were requested for this study, few samples were available and limited our sample size and our ability to assess unique differences between individual weeks of gestation. Due to the clinical relevance that infants are at highest risk for PNAC, we In conclusion, we found significant differences in gene transcripts from human fetal/infant livers compared to adult, and utilized iHLC differentiated from human iPSC to study plant sterol-induced lipid accumulation in the immature liver. Similar gene expression patterns are seen in the human developing liver and iHLC, and are associated with downregulation of functions important in lipid homeostasis and lower expression of key sterol-regulating genes compared to adult liver. Exogenous stigmasterol exposure induces significantly more lipid accumulation in iHLC than HepG2 cells. We speculate that poorly developed mechanisms for sterol metabolism may predispose infants to PNAC due to an inability to adequately eliminate xenosterols in TPN containing soy lipids.

Methods
Use of de-identified human samples without any linkage to subject information in this study was considered exempt from review and informed consent was waived by the For RNA-sequencing and iHLC validation experiments, iHLC were harvested in Trizol (Life Technologies; Carlsbad, CA) and total RNA isolated and purified using the same RNeasy Mini Kit with on-column DNase treatment as noted previously. iHLC from 6 independent differentiations were used for RNA-sequencing.
Human hepatoma HepG2 cells were obtained from ATCC Cell Lines (Manassas, VA) and subcultured on a routine basis according to protocol. Cells for exposures were grown as monolayers to 75% confluence in normoxic culture conditions at 37°C, 95% O2/5% CO2 in modified Eagle's medium (MEM) supplemented with 10% fetal bovine serum (Life Technologies; Carlsbad, CA).  Table   S7 lists antibodies used. Images were acquired using a Nikon Eclipse TE2000-U inverted microscope (Nikon; Tokyo, Japan), and composed using Adobe Photoshop software. Ethics approval. Use of de-identified human samples without any linkage to subject information in this study was considered exempt from review and informed consent was waived by the Children's Wisconsin Institutional Review Board (E377: 441197-1). All methods were carried out in accordance with relevant guidelines and regulations.

Data Availability
RNA-sequencing data from this study has been deposited in the NCBI Gene Expression Omibus (GEO), accession #GSE148790.   Representative flow cytometry profile showing the fluorescence of BODIPY in iHLC exposed to increasing concentrations of 25, 50, and 75 µM of StigAC (green), compared to iHLC exposed to vehicle alone (red) and the positive control iHLC exposed to oleic acid (OA) (orange). (c) Bar graph showing the BODIPY median fluorescence intensity (MFI) of StigAC-exposed cells relative to the MFI of cells exposed to vehicle alone (dotted line). N = 3 independent experiments in all exposures with the exception of n = 2 for the StigAC 25 µM exposure experiment. Error bar indicates standard deviation (SD).
(d) Immunocytochemistry of showing intracellular lipid droplets in iHLC exposed to 50 µM and 75 µM StigAC compared to iHLC exposed to vehicle alone and the positive control (OA). StigAC, stigmasterol acetate; OA, oleic acid.

Figure 4. Lipid accumulation in stigmasterol-exposed iHLC and HepG2 cells. (a)
Representative flow cytometry profile showing the fluorescence of BODIPY in HepG2 cells exposed to 50 µM and 75 µM of StigAC (green), compared to HepG2 cells exposed to vehicle alone (red) and the positive control HepG2 cells exposed to oleic acid (OA) (orange). (b) Bar graph showing the BODIPY MFI of iHLC (white) and HepG2 cells (black) exposed to 75 µM of StigAC and oleic acid, relative to the MFI of cells exposed to vehicle alone. N = 3 independent experiments in all exposures. Error bar indicates standard deviation (SD). *P-value < 0.05. StigAC, stigmasterol acetate; OA, oleic acid.