The liver is a highly organized organ that consists of hepatic parenchymal cells, hepatocytes, and non-parenchymal cells such as the liver sinusoidal endothelial cells (LSECs), hepatic stellate cells (HSCs), cholangiocytes, and Kupffer cells. Although previous studies have primarily focused on the hepatocyte dynamics in the injured liver, recent studies have shown that non-parenchymal cells play an essential role in both liver regeneration and liver fibrosis progression. Among the non-parenchymal cells, HSCs directly contribute to the progression of liver fibrosis because the activation of HSCs in response to liver injury or inflammation results in the excess production of extra cellular matrix. LSECs also contribute to modulate the function of hepatocytes, HSCs, and immune cells during liver fibrosis. Therefore, to investigate the mechanisms for liver fibrosis in vitro, it is necessary to develop an appropriate liver model that accurately recapitulates the pathology of human liver fibrosis including HSC activation. However, the supply of human cells is limited and freshly isolated liver cells easily lose their specific characteristics in culture. To overcome this shortage of human liver cells, human induced pluripotent stem cell (hiPSC)-derived liver cells were generated by mimicking the liver developmental process. In this review article, we outline the differentiation system of liver non-parenchymal cells from hiPSCs and development of in vitro liver disease models using hiPSC-derived liver cells. We describe the utility of these liver models as experimental systems to investigate the mechanism of liver fibrosis and development of drugs for the treatment thereof.
The liver is responsible for the maintenance of homeostasis and consists of liver parenchymal cells, hepatocytes, and liver non-parenchymal cells. The non-parenchymal cells are composed of liver sinusoidal endothelial cells (LSECs), hepatic stellate cells (HSCs), cholangiocytes, and Kupffer cells. Since hepatocytes express various metabolic enzymes such as cytochrome P450 oxidases, they have functions in metabolism and detoxification. They also synthesize plasma proteins such as albumin and bile acids which facilitate digestion and the absorption of lipids, respectively. The liver non-parenchymal cells closely interact with hepatocytes and are essential to support the functions of hepatocytes in the liver. LSECs form the hepatic sinusoid which connect to two large blood vessels, the portal and central vein, and provide blood supply to the hepatocytes. They have a high endocytic capacity and contribute to the transfer of molecules from the liver sinusoid to the space of Disse and liver parenchyma . HSCs are liver-specific pericytes that localize in the perisinusoidal space known as the space of Disse. They are quiescent in a healthy liver and store vitamin A in their cytoplasm .
Liver fibrosis is caused by chronic hepatitis virus B or C (HBV or HCV) infection, alcohol abuse, non-alcoholic steatohepatitis (NASH), or cholestasis, etc. HSCs are key players at the onset of liver fibrosis as quiescent HSCs (qHSCs) are converted into activated HSCs (aHSCs) which then produce excessive amounts of extra cellular matrix (ECM) [2,3]. Although the production of ECM by HSCs contribute to wound healing, the excessive accumulation of ECM leads to cirrhosis or hepatocellular carcinoma. LSECs also contribute to liver regeneration after liver injury. They modulate the function of hepatocytes, HSCs, and immune cells during liver fibrosis . For example, LSECs in fibrotic livers produce hepatic cytokines such as hepatocyte growth factor (HGF) and Wnt family member 2 (Wnt2) which lead to liver regeneration . Therefore, to investigate the mechanisms of liver fibrosis in vitro, it is necessary to develop multi-cellular human liver models composed of hepatocytes, HSCs, and LSECs.
Various in vivo mouse fibrosis models are available to investigate the dynamics of parenchymal and non-parenchymal cells in fibrotic livers [5,6]. However, these cellular properties in response to liver injury in mice are different from those in humans. To develop novel therapeutic strategies for human liver fibrosis, accurate human liver models which recapitulate liver structure and functions in vitro are valuable tools. However, no human liver fibrosis models have been established due to the shortage of human liver cells needed to properly recapitulate human livers in vitro. To overcome this challenge, human induced pluripotent stem cells (hiPSCs) have been used to generate human liver cells including non-parenchymal cells [7–13]. These hiPSC-derived liver non-parenchymal cells such as LSECs and HSCs have been used to investigate the cellular mechanisms of liver fibrosis and develop novel therapeutic targets.
In this review, we introduce the history of the development of hiPSC-derived liver non-parenchymal cells and the application of human iPSC-derived liver models in human liver fibrosis research. We also describe new insights of cellular mechanisms of non-parenchymal cells in liver fibrosis using in vitro human iPSC-derived liver models.
hiPSC-derived liver sinusoidal endothelial cells
LSECs are liver-specific endothelial cells which form the wall of sinusoidal and capillary vessels in the liver parenchyma. LSECs are unique in their functional and structural characteristics and are distinct from endothelial cells of other organs. Since LSECs possess fenestrae and form discontinuous vessel structures, liver sinusoidal vessels allow for the transport of substances between the blood and liver parenchyma . Functionally, LSECs produce a blood clotting protein, factor VIII (F8), and scavenge molecules from the blood .
Some protocols for induction of LSECs from hiPSCs have recently been established [7,9]. Our group induced endothelial progenitor cells from hiPSCs through mesodermal induction and isolated FLK1+CD31+CD34+ LSEC progenitors using a cell sorter. We found that LSEC progenitors matured into LSECs through the inhibition of transforming growth factor beta (TGFβ) signaling under hypoxic culture conditions. The mature hiPSC-derived LSECs highly expressed LSEC-specific marker genes such as Fc gamma receptor IIb (FCGR2B), Stabilin 2 (STAB2), Lymphatic vessel endothelial hyaluronan receptor 1 (LYVE1), and F8 compared to the primary human LSECs and human umbilical vein endothelial cells (HUVECs) and produced F8 proteins in vitro. We co-cultured hiPSC-derived LSECs with the hiPSC-derived hepatic progenitor cells and found that the hiPSC-derived LSECs promote the maturation of hepatocytes in vitro. Although cell characteristics such as fenestration formation and scavenger functions have not been fully examined, we demonstrated that hiPSC-derived LSECs exhibited LSEC-like properties in vitro (Figure 1) . Gage et al.  identified the differentiation capacity of venous and arterial endothelial cells to transform into LSECs and generated functional LSECs from hiPSCs. They generated hiPSC-derived endothelial cells and then isolated the arterial and venous endothelial cells using a cell sorter. They showed hiPSC-derived venous endothelial cells have robust potential to differentiate into LSECs in response to cyclic adenosine monophosphate (cAMP) and TGFβ inhibition under hypoxic culture conditions. The hiPSC-derived LSECs expressed some LSEC-specific marker genes such as STAB2, FCGR2B, LYVE1, and F8. They transplanted these hiPSC-derived venous endothelial cells into a mouse liver and showed that the endothelial cells gave rise to mature LSECs. They also demonstrated that hiPSC-derived LSECs produced the F8 protein in mice and formed fenestration structures (Figure 1) .
Differentiation of LSECs from hiPSCs. Graphical summary for differentiation of LSECs from hiPSCs.
HBV infection is a trigger of liver fibrosis as it promotes inflammation and leads to the damage of hepatocytes and fibrogenesis. An in vitro HBV infection model has been established using sodium-taurocholate co-transporting polypeptide (NTCP)-expressing HepG2 cells and hiPSC-LSECs. Using this model, our group demonstrated that HBV infection of the HepG2 cells is promoted by the co-culturing with hiPSC-derived LSECs. We demonstrated that the epidermal growth factor (EGF) produced by hiPSC-derived LSECs modulates HBV infection [14,15]. Therefore, hiPSC-derived LSECs and their co-culture models are effective tools to study the mechanisms of liver diseases including liver fibrosis to develop novel therapeutic strategies for liver fibrosis.
hiPSC-derived hepatic stellate cells
HSCs are quiescent in the normal healthy liver and store vitamin A in their cytoplasm. They express specific markers such as low-affinity nerve growth factor receptor (NGFR) and lecithin retinol acyltransferase (LRAT) which catalyze the esterification of retinol for storage . In addition, qHSCs express a limited amount of collagen. However, qHSCs are converted into aHSCs in response to liver injury and begin to increase the expression of activation markers such as alpha smooth muscle actin (αSMA) and collagens . The deposition of collagens produced by the aHSCs results in liver fibrosis. Although ECM supports the reconstruction of liver structure and the improvement of liver function after liver injury, the continuous activation of HSCs results in severe liver disease like cirrhosis or hepatocellular carcinoma. Therefore, the prevention of the conversion into aHSCs is an effective strategy for the treatment of liver fibrosis [17,18]. However, therapeutic drugs inhibiting the activation of HSCs have not been developed due to the lack of available human qHSCs that have similar characteristics to in vivo human qHSCs. This lack of an adequate in vitro drug screening model that recapitulates HSC activation has hampered the development of such therapeutic drugs.
hiPSC-derived HSCs have been successfully generated by some researchers. Our research group isolated activated leukocyte cell adhesion molecule+ (ALCAM+) HSC progenitor cells from mesodermal progenitors and generated hiPSC-derived HSCs through the addition of Y27632, a Rho-associated coiled-coil containing protein kinase (ROCK) inhibitor, which induces the maturation of HSCs. Although the Rho signaling pathway was reported to play a role in inhibition of activation and contraction of HSCs [19,20], its effect on HSC differentiation is unknown. They highly expressed the HSC-specific marker genes such as NGFR, LRAT, and Nestin (NES) and exhibited cell-specific characteristics such as vitamin A storage activity. In addition, because hiPSC-derived HSCs produced hepatic cytokines such as HGF, fibroblast growth factor (FGF), bone morphogenetic protein (BMP), and midkine, which induce hepatocyte growth and differentiation, hiPSC-derived HSCs supported the in vitro maturation of hepatocytes . Coll et al.  also developed a differentiation protocol for hiPSC-derived HSCs with typical HSCs features. They demonstrated some aHSC marker genes including αSMA were increased in response to TGFβ stimulation, which is an inducer of HSC activation . Miyoshi et al.  differentiated hiPSCs into HSCs and demonstrated that hiPSC-derived HSCs supported the hepatic maturation of hiPSC-derived hepatocyte progenitors when co-cultured . Although the hiPSC-derived HSCs mimic the HSC-specific function in vitro, the problem that aHSCs marker genes were still expressed in hiPSC-derived HSCs hampered the development of an in vitro human liver fibrosis model. We considered that it is necessary to generate hiPSC-derived qHSCs to recapitulate the progression of liver fibrosis in vitro.
To overcome this problem, our research group developed hiPSC-derived qHSCs which exhibited qHSC-specific characteristics and did not express aHSC marker genes. We mimicked the developmental process of HSCs in a three-dimensional culture system, through the combination of induction factors including Y27632 and the isolation of qHSCs from the mesodermal cell population with NGFR using a flow cytometer. hiPSC-derived qHSCs express several HSC marker genes such as NGFR, NES, LRAT, and Desmin and exhibit vitamin A storage capability, but did not express αSMA and collagens. Importantly, the hiPSC-derived qHSCs can be activated to aHSCs by culturing on a collagen coated dish under two-dimensional conditions. The expression of aHSC marker genes such as actin alpha 2, smooth muscle (ACTA2), collagen type I alpha 1 chain (COL1A1), collagen type II alpha 2 chain (COL1A2), and collagen type III alpha 1 chain (COL3A1) were substantially increased after they were induced into aHSCs (Figure 2) .
Differentiation of HSCs from hiPSCs.
Among these aHSC markers including ACTA2 and collagens, it has been determined that the expression of collagen genes was still detectable in qHSCs isolated from healthy mouse livers and upregulated in mouse aHSCs isolated from fibrotic mouse livers. On the other hand, ACTA2 is known to be expressed in mouse aHSCs, but is not expressed in mouse qHSCs [8,21]. Therefore, ACTA2 is a reliable aHSC marker gene that allows for the clear distinguishing between qHSCs and aHSCs. Because hiPSC-derived qHSCs have the potential to be activated to aHSCs in vitro under two-dimensional conditions, we established a method to quantitatively assess HSC activation using a ACTA2-RFP reporter hiPSC-derived qHSCs. Using this reporter system, an in vitro drug screening system using ACTA2-RFP reporter iPSC-derived qHSCs was developed and used to screen therapeutic agents that inhibit HSC activation. Among the screened agents, Artemisinin, a long-standing anti-malarial drug, demonstrated an improvement of liver fibrosis in mice . Although the mechanism of anti-liver fibrosis is still unknown, this finding provides an indication that anti-HSC activation drugs are effective for the treatment of liver fibrosis.
Lai et al.  recently developed a protocol to differentiate hiPSCs and human embryonic stem cells (hESCs) into HSCs. They demonstrated that hiPSC or hESC-derived HSCs expressed several HSC markers such as glial fibrillary acidic protein (GFAP), neural cell adhesion molecule (NCAM), and Desmin at the protein level. The expression of aHSCs markers such as αSMA and Collagen I in hiPSC or hESC-derived HSCs was low compared with those of primary human HSCs but was upregulated in response to TGFβ stimulation. Additionally, in a co-cultured system with HepaRG cells, they showed that acetaminophen stimulation induced the expression of fibrogenesis markers including ACTA2 and collagens. They also developed a co-culture system with HepG2-NTCP or Huh7.5 cells infected with HBV or HCV, respectively, and showed that HBV and HCV infection could stimulate the expression of aHSC markers such as αSMA and collagens in hiPSC or hESC-derived HSCs . They showed that HSCs could be differentiated from both endoderm and mesoderm cells. However, the endodermal origin of human HSCs is still under investigation because previous lineage tracing study in mice has clearly shown mesodermal origin of HSCs . These reports indicate that hiPSC-derived qHSCs and their co-culture models could be useful as a model for liver disease.
Co-culture and organoid models using hiPSC-derived liver cells
As described above, HSCs are the main players in the progression of liver fibrosis because the ECM produced by HSCs can be a source of fibers. However, other liver cells also support the progression of liver fibrosis by interacting with each other in fibrotic livers. To recapitulate the pathology of human liver fibrosis in vitro, co-culture and organoid models that consist of hepatocytes and non-parenchymal cells are valuable tools. Our group developed a co-culture system with hiPSC-derived hepatic progenitors, LSECs, and HSCs. We demonstrated that hiPSC-derived non-parenchymal cells promoted the maturation of hiPSC-derived hepatic progenitors. However, the application of a co-culture model for liver fibrosis has not been performed . Ouchi et al.  developed liver organoids composed of hepatocyte-, Kupffer-, and hepatic stellate-like cells. They showed that liver organoids exhibited steatohepatitis-like features and fibrosis phenotypes when exposed to free fatty acids. They have not shown detailed mechanisms of the development of liver fibrosis in vitro by the addition of free fatty acid, but a liver organoid culture system is an excellent tool to examine the cell–cell interaction during liver fibrosis . Investigation of the mechanisms of the development and progression of liver fibrosis using in vitro co-culture or organoid models instead of mouse liver fibrosis models would be the next challenge in the field of liver biology.
Although liver non-parenchymal cells significantly contribute to liver fibrosis, researchers cannot access human liver non-parenchymal cells due to the lack of available human non-parenchymal cells which exhibit cell-specific characteristics. hiPSC-derived liver non-parenchymal cells are valuable biological tools for the study of the mechanisms of liver fibrosis in humans, and development of therapeutic strategies for the treatment of liver fibrosis.
In vitro disease models using hiPSC-derived liver parenchymal cells and non-parenchymal cells such as LSECs and HSCs have been developed as alternative experimental ‘human liver models’ as a replacement for mouse liver injury models. hiPSC-derived human liver models partially mimic the pathology of human liver fibrosis and have been used to investigate the mechanisms of liver diseases and drug screening for liver fibrosis.
Investigation of the development and progression of liver fibrosis at the tissue level including signal transductions between liver cells, cell–cell contact in fibrotic livers, and structural changes of liver tissues in response to liver injury using hiPSC-derived liver models is the next challenge in this field.
The authors declare that there are no competing interests associated with the manuscript.
This research was supported by JSPS KAKENHI under Grant Number 21H02710 and AMED under Grant Number JP22bk0104136.
Open access for this article was enabled by the participation of the University of Tokyo in an all-inclusive Read & Publish agreement with Portland Press and the Biochemical Society under a transformative agreement with Individual.
Y.K. wrote the review and created the figure. T.K. edited the review and supervised the work.
extra cellular matrix
- HBV or HCV
hepatitis virus B or C
human embryonic stem cells
hepatocyte growth factor
human induced pluripotent stem cell
hepatic stellate cells
lecithin retinol acyltransferase
liver sinusoidal endothelial cells
nerve growth factor receptor
sodium-taurocholate co-transporting polypeptide
transforming growth factor beta
alpha smooth muscle actin