Department of Developmental and Regenerative Biology, Black Family Stem Cell Institute, Mount Sinai School of Medicine, New York, New York, USA
Received Date: July 06, 2012; Accepted Date: August 13, 2012; Published Date: August 15, 2012
Citation: Han S, Bourdon A, Hamou W, Dziedzic N, Goldman O, et al. (2012) Generation of Functional Hepatic Cells from Pluripotent Stem Cells. J Stem Cell Res Ther S10:008. doi:10.4172/2157-7633.S10-008
Copyright: © 2012 Han S, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at Journal of Stem Cell Research & Therapy
Liver diseases affect millions of people worldwide, especially in developing countries. According to the American Liver Foundation, nearly 1 in every 10 Americans suffers from some form of liver disease. Even though, the liver has great ability to self-repair, in end-stage liver diseases including fibrosis, cirrhosis, and liver cancer induced by viral hepatitis and drugs, the liver regenerative capacity is exhausted. The only successful treatment for chronic liver failure is the whole liver transplantation. More recently, some clinical trials using hepatocyte transplantation have shown some clinical improvement for metabolic liver diseases and acute liver failure. However, the shortage of donor livers remains a life-threatening challenge in liver disease patients. To overcome the scarcity of donor livers, hepatocytes generated from embryonic stem cell or induced pluripotent stem cell differentiation cultures could provide an unlimited supply of such cells for transplantation. This review provides an updated summary of hepatic differentiation protocols published so far, with a characterization of the hepatic cells generated in vitro and their ability to regenerate damaged livers in vivo following transplantation in pre-clinical liver deficient mouse models.
Hepatic cells; Pluripotent stem cells; Embryonic stem cells
The liver is a remarkable regenerative organ that can restore normal mass and function after an injury or partial hepatectomy. After mild injury, this ability is supported by the proliferation of fully mature hepatocytes, biliary epithelial cells  as well as the proliferation and help of surrounding cells including stellate cells, endothelial sinusoidal cells  and macrophages . After severe injury or chronic disorder, hepatic progenitors or “oval cells-ductular hepatocytes” are recruited to ensure regeneration of the hepatocytic and biliary cell mass . However, in end-stage liver diseases, such as fibrosis, cirrhosis, or liver cancer induced by viral hepatitis or drugs, the regenerative capacity of the liver is exhausted. Consequently, liver diseases are becoming one of the most common causes of mortality all over the world, especially in developing countries. Until now, liver transplantation is the most successful treatment for patient with chronic liver failure [4,5]. Alternatively to the whole organ transplantation, hepatocyte transplantation has been recently used in clinical trials for the treatment of acute failure and life-threatening metabolic liver diseases [6-9]. Unfortunately, these two options face the major concern of shortage of human donor livers .
To overcome the scarcity of donor livers, hepatocytes generated from pluripotent stem cell (PSC) differentiation cultures could provide an unlimited supply of such cells for transplantation in liver diseases. Embryonic Stem Cells (ESC) established from the inner cell mass of the blastocyst from early embryos or induced pluripotent stem cells (iPSC) from somatic adult cells are pluripotent, and proliferate indefinitely in an undifferentiated state in vitro [11,12]. Since the first establishment of mouse iPSC from embryonic fibroblasts by Takahashi and Yamanaka using viral vectors of Oct3/4, Sox2, Klf4 and c-Myc transcription factors, this technology has been widely and successfully applied to different species including human, established from different terminally differentiated adult cell types using various gene delivery systems (Recently discussed by Yamanaka, ).
Numerous studies summarized in this manuscript have shown that functional hepatocyte-like cells could be derived from ESCs (ESCHep) or induced pluripotent stem cells (iPSC-Hep), which express transcription factors and markers for mature hepatocytes as well as possess the metabolism and secretion functions both in vitro and in vivo (Table 1 and Table 2). Therefore, with their pluripotency and self-renewal abilities, ESCs and iPSCs have been proposed as a very valuable and unlimited transplantable hepatic cell source for patients with end-stage liver diseases . The patient-specific iPSC-Hep have the advantage over the ESC-Hep to be immunologically compatible with the host for cell therapy approaches, and also to provide in vitro liver disease models . As a proof of principle that iPSC can be a valuable source of functional hepatic cells, Espejel et al.  have injected iPSC into the blastocysts of the liver deficient Fah-/- mice, and found that by 70 days after birth, 100% of the hepatocytes in adult livers derived from the iPSC. These liver-chimeric mice display very healthy liver functions, demonstrating the definite ability of iPSC to generate functional hepatocytes in vivo avoiding the limitations of current in vitro differentiation protocols. Similar experiments were performed by Duncan’s group by producing embryos by tetraploid complementation from mouse iPSC . All embryos including their livers were derived from the iPSC except the extra embryonic tissues that were derived from the donor tetraploid embryos. The authors demonstrated that fetal mouse livers derived from iPSCs were indistinguishable from wildtype livers based on histological and gene expression assays, thereby supporting the ability of iPSCs to generate functional hepatocytes as ESCs do.
|References||Species||Protocol specificity||EB/monolayer||Endoderm induction||Hepatic specification and maturation||In vitro functional assays||In vivo assay|
|Inducers||Efficiency & marker||Inducers||Efficiency & marker|
|Gouon-Evans et al. ||mESC||EB||Act||55% Foxa2/ckit/ Cxcr4||BMP4, bFGF, HGF, Dex,
TGFa, EGF, VEGF
|70% Afp, 60% Alb||Alb, Glycogen||yes|
|Cai et al. ||hESC||monolayer||Act, ITS||80% Foxa2/Sox17||FGF4, BMP2, HGF, OSM, Dex||70% Alb||Alb, Glycogen, ICG, LDL, p450||yes|
|Hay et al. ||hESC||monolayer||Act, Wnt3a||Serum, Insulin, HGF, OSM,||up to 90% Alb||Urea, Gluconeogenesis, Afp||yes|
|Agarwal et al. ||hESC||monolayer||Act, low serum||72% Cxcr4||FGF4, HGF||70%||yes|
|Basma et al. ||hESC||EB/monolayer||Act, bFGF||FGF, DMSO, Dex||26% Asgr-1||Alb, Urea, AAT, P450||yes|
|Si-Tayeb et al. ||hiPSC||monolayer||Act||80%||BMP4 FGF2, HGF||80%||Glycogen, LDL, oil red O storage, ICG,
|Touboul et al. ||hESC||monolayer||Act, FGF2, BMP4, LY294002||80% Cxcr4||FGF10, RA, SB431542, FGF4,
|50-60% Hnf4a, Afp or
Ck19, 35% Asgr1
|Glycogen, p450, ICG, LDL||yes|
|Rashid et al. ||hiPSC||EB||Act||DMSO, HGF, Dex||p450, Alb, Afp, urea, LDL, ICG,
Glycogen, oil red O storage
|Mfopou et al. ||hESC||monolayer||Act, Wnt3a, low serum||60-80% Cxcr4/Foxa2||FGF10, BMP, Cyclo, DAPT,
Exendin-4, IGF1, HGF
|40-60% Afp||Urea, Glycogen|
|Sullivan et al. ||hiPSC||monolayer||Act, Wnt3a||β-ME, DMSO, Insulin, HGF,
|70-90% Alb||p450, Fibrinogen, Fibronectin,
Transthyretin and Afp secretion
|Rashid et al. ||hiPSC||monolayer||Act, FGF2, BMP4, LY294002||Act, HGF, OSM||83% Alb||Alb, p450, Glycogen, LDL,|
|Liu et al. ||hiPSC||monolayer||Act, low serum||90% Cxcr4||FGF4, HGF, OSM, Dex||90% Afp||Alb, p450||yes|
|Li et al. ||miPSC||monolayer||Act, ITS||FGF4, HGF, OSM, Dex||~90% Afp, ~90% Alb||LDL, Glycogen, P450||yes|
|Woo et al. ||hESC / hiPSC||EB||LiCL, GSK-3 inhibitor (BIO),
|35% Foxa2/Sox17||HGF, OSM, Dex||~69% Alb/CK18||Urea, Alb, Glycogen||yes|
|Pauwelyn et al. ||mESC||monolayer||Act, Wnt3a, low serum||FGF2, BMP4, FGF8b, FGF1,
|Alb, Urea, Glycogen, p450|
|Sancho-Bru et al. ||miPSC||monolayer||Act, Wnt3a, low serum||70% Foxa2||BMP4, FGF2, FGF1, FGF4,
FGF8, HGF, Follistatin
|30% Hnf4a||Alb, Glycogen, Urea, p450|
|He et al. ||mESC||EB/monolayer||Act||75% Cxcr4/cKit
|BMP4, bFGF, EGF, TGFα,
VEGF, HGF, Dex
|31% Alb (monolayer)||Alb, Ammonia, Glycogen, LDL, ICG||yes|
|Chen et al. ||hiPSC||monolayer||Act, Wnt3a, HGF||61-64%||OSM, Dex, ITS||p450, Urea, LDL, Glycogen||yes|
|Kubo et al. ||mESC||Hex expression||EB||Act||Hex expression, BMP4||Alb, Transferrin|
|Inamura et al., 2011 ||hESC / hiPSC||Hex expression||monolayer||Act, bFGF, Hex expression||46%||BMP4, FGF4, HGF, OSM, Dex||p450|
|Takayama et al. ||hESC / hiPSC||Sox17 and Hex expression||monolayer||Act, Sox17 and Hex
|67% Cxcr4/ckit and 57%
|BMP4, FGF4, HGF, OSM, Dex||50% p450||LDL, p450|
|Takayama et al. ||hESC / hiPSC||Sox17 and Hex expression||monolayer||Act, Sox17 and Hex
|BMP4, FGF4, HGF, OSM, Dex||80% p450, 80% Asgr1, cmet
|LDL, p450, ICG, Glycogen, metabolismmediated
|Hay et al. ||hESC||NaBu||monolayer||Act, NaBu||70% Cxcr4||DMSO, HGF, OSM, 8.3%
|71% Alb, 65% Hepar-1||Glycogen, generation and secretion of
plasma proteins, p450
|Li et al. ||mESC||NaBu||monolayer||Act, NaBu||100% Cxcr4, 65% cKit,
|BMP4, FGF2, HGF, EGF,
|51% Afp, 62% Alb||ICG, Glycogen||yes|
|Duan et al. ||hESC||NaBu||monolayer||Act, NaBu||85% Cxcr4, Sox17, and
|HGF, BMP2, FGF4, BMP4,
DMSO, serum, OSM, Dex
|90% Alb||ICG, drug metabolism|
|Ren et al. ||mESC||NaBu||EB||NaBu||HGF, Dex||50%||Glycogen|
|Fang et al. ||mESC||scaffold-alginate microbeads||EB||FGFa||HGF, OSM, Dex, Insulin,
|49% Alb, 50% Ck18||Alb, Urea|
|Li et al. ||mESC||scaffold-polyacrylamide
|monolayer||DMSO||NaBu, cells placed on
|70% functional hepatocytelike
|Mizumoto et al. ||mESC,
|scaffold-hollow fibers||fibers||NaBu||Ammonia, Alb|
|Matsumoto et al. ||mESC||scaffold-Polyurethane Foam||EB||aFGF, HGF, OSM, Dex, ITS||Glycogen, Ammonia, Alb|
|Lee et al. ||mESC||Fibronectin / collagen1,
collagen 4 / laminin
|Farzaneh et al. ||hESC||scaffold-ultraweb nanofiber||monolayer||Act||86% Foxa2, 94% Sox17,
|FGF4, HGF, OSM, Dex||66% Afp||Afp, Alb, Urea, LDL, Glycogen, ICG,
|Liu et al. ||mESC||scaffold-50% poly-l-lactic
and 50% polyglycolic acid
|EB||Dex, DMSO||FGF4, rmHGF, OSM, Dex, ITS||Alb, LDL, Glycogen|
|Shafritz et al. ||miPSC/hESC||scaffold||monolayer||Act, bFGF||40% Cxcr4/E-cad
|RA, Act, bFGF, ITS-DMEM,
Dex, HGF, DMSO, NA, AsP,
Akt inhibitor treated cells
|Miki et al. ||hESC||scaffold-hydrophilic hollow
|monolayer||Act||FGF4, BMP2, HGF, OSM, Dex||30% Asgpr||Alb, Ammonia, Glycogen, p450|
|Haque et al. ||mESC||E-cadherin substratum||monolayer||Act, bFGF||~ 55%||HGF, OSM, Dex||92% Alb||Glycogen|
|Amimoto et al. ||mESC /
|cellulose triacetate hollow
|fibers||NaBu, Dex, OSM, ITS||Ammonium, Alb|
|Co-culture or conditional medium|
|Cho et al. ||mESC||Co-culture with rat
|Zhao et al. ||hESC, primate
|Co-culture with STO feeder
|monolayer||Act||FGF4, BMP2, HGF, OSM, Dex||90% Afp||Alb, Glycogen, ICG, LDL, PROD assay|
|Fukumitsu et al. ||mESC||Co-culture with murine fetal
liver stromal stromal cell line
|monolayer||trans retinoic acid||HGF, bFGF, Nicotinamide, lascorbic
acid phosphate, ITS,
|Glycogen, Ammonia, Alb|
|Huang et al. ||primate ESC||Co-culture with human ESCderived
fibroblast-like cells coculture
|monolayer||Haptoglobin, Urea, Alb, EROD,
Glycogen, HBV infection
|Tuleuova et al. ||mESC, hESC||Culture with protein
|Ishii et al. ||mESC, hESC||Co-culture with murine fetal
liver stromal cells and
|monolayer||Act||HGF||Glycogen, Ammonia, p450|
|Yu et al. ||hESC||Co-culture with mitomycin
treated 3T3-J2 feeder cells
|Nishiofuku et al. ||mESC, ratESC||Co-culture with rat hepatic
|Han et al. ||mESC||Co-culture with D4T
|monolayer||Act||~ 50% Foxa2/Foxa3||EGF, TGFa, Dex, VEGF, bFGF,
|Pal et al. ||hESC||Conditioned medium from
|EB||bFGF||Glycogen, Afp, SGOT, SGPT, GGT|
Table 1: Summary of PSC-Hep differentiation protocols in serum free media.
|Reference||Animal model||Type and number of transplanted cells||Percentage of repopulation||Improved liver functions||Mouse survival after transplantation|
|Teratani et al. ||129X1/SvJ and BALB/c nude mice (DMN, cirrhosis)||mESC-Hep 5 e6/mouse (IV)||ND||Suppression of onset of fibrosis/cirrhosis plasma fibrinogen Alb||~40% after 12 weeks|
|Heo et al. ||MUP-uPA/SCID + CCL4||mESC-Hep 1 e6/mouse (IS) isolated Alb+ cells||1.94% +/- 5.81 after 82 days||ND||ND|
|Gouon-Evans et al. ||Dpp4-/-, Rag2-/- (CCl4 and retrorsine) and Fah-/-||mESC-Hep 0.25 - 1.5 e6/mouse (IS)||Scattered endothelial and hepatocytic clusters||ND||ND|
|Agarwal et al. ||NOD/SCID (CCl4 and retrorsine, acute liver failure)||hESC-Hep 1 e6/mouse (PV)||sporadic hepatocytic clusters||ND||ND|
|Hay et al. ||NOD/SCID||hESC-Hep 1 e6/mouse (IS)||Clusters of CK18/CK19 and Alb cells after 3 days in the spleen||Alb|
|Cai et al. ||SCID (CCl4, acute liver failure)||hESC-Hep 1 e6/mouse (IS)||scattered AAT human cells in livers||ND||ND|
|Duan et al. ||NOD/SCID||hESC-Hep 0.5 e6/site (IH)||ND||Alb||ND|
|Haridass et al. ||Alb-uPA, Rag2-/-, IL2Rγ-/-||ES-Hep 0.5-1 e6/mouse (IS)||Few cells scattered and teratoma formation||ND||ND|
|Basma et al. ||NOD/SCID (retrorsine and 50% hepatectomy) Alb-uPA SCID||hES-Hep 0.1-0.2 e6/mouse (IS) Isolated ASGPR+ cells||Few clusters of albumin+ cells after 28 days||Alb, AAT||ND|
|Li et al. ||Fah-/-||mES-Hep 1 e6/mouse (IS) Isolated ckit-Epcam+||24%+/-15%||ND||at least 10 weeks|
|Touboul et al. ||Alb-uPA, Rag2-/-, IL2Rγ-/-||hES-Hep 0.5 e6/mouse (IH)||Small and large clusters throughout the liver after 8 weeks||AAT Alb||ND|
|Si-Tayeb et al. ||hiPSC-Hep 0.3 e6/mouse (IH)||ND||ND||ND|
|Huang et al. ||Fah-/-, Rag2-/-||m iHep from fibroblasts 0.833 e6/mouse (IS)||5%-80%||ALT, AST, Tyrosine, Phenylalanine, Bilirubin||5/12 mice survived after 8 weeks|
|Sekiya and Suzuki ||Fah-/-||m iHep from fibroblasts (IS)||Large clusters of||Bilirubin, ALT, ALP, Alb||40% at 10 weeks|
|Li et al. ||BALB/c nude (TAA, liver fibrosis)||m iPSC-Hep ~0.25 e6/mouse (IV)||Scattered cells in the liver after 24h00||ALT, AST, Bilirubin, Ammonia, decreased ROS levels and necrosis||~80% after 14 days|
|Chang et al. ||BALBc nude (CCl4, acute liver failure)||m iPSC-Hep 0.2-5 e6/mouse (IP)||Necrosis : 70% control, 20% transplant||ALT, AST, Bilirubin||90% at 14 days|
|Woo et al. ||BALBc nude (CCl4, acute liver failure)||hES-Hep / hiPSC-Hep 2 e6/mouse (IS) Isolated ICG+ cells||~20% at day 3 versus ~10% after 35 days||Alb||ND|
|Chen et al. ||NOD/SCID (CCl4, acute liver failure)||hiPSC -Hep (IS)||ND||rescue of hepatic necrosis glutamyl oxaloacetic aminotransferase glutamyl pyruvic aminotransferase Bilirubin lactate dehydrogenase HepPar1 Alb||5/7 (71%) after 21 days|
|Liu et al. ||NSG (DMN, cirrhosis)||hiPSC-Hep 0.1-2 e6/mouse (IV)||2 to 17%||Alb, CYP2E1||90% after 8 weeks|
|He et al. ||Fah-/-||mESC-Hep (IS)||0.001%-12.5% after 8-10 weeks||ALT, AST, Bilirubin, Alb, succinylacetone, tyrosine and phenylalanine secretion 30% of wt FAH activity in recipients||8/20 survived after 2nd transplantation|
|Bandi et al. ||CD17.NOD/SCID (Rif + Phen + MCT, acute liver failure)||hESC-MEC 4-6 e6/mouse (IP with microcarriers)||Detection of transplanted cells in the peritoneal cavity||Endogeneous hepatocytes proliferation Phenobarbital metabolism Ammonium chloride||11/11 after 14 days (MCT 125, mild injury) 38% after 2 weeks (MCT 160, strong injury)|
Table 2: Summary of in vivo functions of PSC-Hep in mice.
This review summarizes the results of various protocols using pluripotent stem cells published so far, and compares their efficiency in generating functional hepatic cells and their therapeutic relevance in pre-clinical animal models of liver disease. The use of PSC-Hep for drug screening and liver disease models will not be discussed here.
In the last decade, numerous protocols have been established to generate ESC-Hep and iPSC-Hep. The most efficient and reproducible hepatic differentiation approaches are those that recapitulate in ESC and iPSC cultures the appropriate signalling pathways uncovered in embryo studies, thereby subsequently mimicking endoderm development, endoderm hepatic specification followed by hepatic cell maturation.
Cytokine-based hepatic differentiation protocols
The use of a specific cocktail of cytokines in a serum-free media became a pre-requisite to control in a timed manner the successive steps of liver organogenesis in differentiation cultures. Studies in Xenopus have demonstrated that different concentrations of the TGF β (Transforming growth factor beta) family member, activin-A used as a surrogate for nodal, induce mesodermal and endodermal fates in animal cap cells in culture [18-23]. Consequently, since its first use in the mouse ESC cultures , high doses of Activin-A are now widely utilized for endoderm induction in ESC and iPSC cultures from human and mouse lines [25-31]. Associated with Activin-A, activation of other pathways has also been shown to promote endoderm development including FGF (Fibroblast growth factor) and Wnt signalings [26,27,29,32-39]. In some protocols, low doses of serum have been reported to be necessary for Activin-A to induce an efficient endoderm program [30,40,41]. The presence of PI3kinase inhibitor in the serum was suspected to be responsible for promoting Activin-A-induced endoderm development .
Originally, most of the ESC differentiations were performed using embryoid bodies (EBs), the three-dimensional structure that was meant to mimic the blastocyst and epiblast architecture [43,44]. Few years ago, D’Amour et al.  demonstrated efficient endoderm induction from monolayers of hESC that was subsequently reproduced by many groups. The monolayer induction is thought to better synchronize the endodermal cell fate by exposing the cells evenly to the endodermal inducer, Activin-A.
Regardless of the induction system used (EBs or monolayers), heterogeneity of endoderm cultures remains inherent to pluripotent stem cell differentiation cultures. To overcome this issue, Activin-Ainduced endoderm generated from pluripotent stem cell differentiation cultures have been enriched using the cell surface receptors CXCR4 , cKit , ENDM-1 , E-cadherin and PDGF receptor-α or through the selection of reporter molecules targeted to the Brachyury, Foxa2, Foxa3, goosecoid, sox17 and Hex loci [30,32,33,47-50]. Interestingly, endoderm induction appears to be always more effective in human cultures than in murine cultures, with the human endoderm fraction reaching ~90% of the total population versus ~60% at the most in murine cultures. The discrepancy of efficiency between both species may be due to the quicker endoderm induction in the mouse embryo compared to the human embryo, thereby restricting the time flexibility to modulate signaling to generate homogenous endoderm cell population in mPSC differentiation cultures. Overall, pluripotent stem cells have been successfully and reproducibly differentiated in serum-free media to endoderm with the help of Activin-A (Table 1).
To recapitulate the liver specification signaling based on developmental studies in the mouse and the Xenopus models [51-55], many groups including ours have used the combination of BMPs (BMP2, BMP4) and FGFs (FGF1, FGF2, FGF4, FGF8, FGF10) to specify endoderm cells generated in pluripotent stem cell cultures [17,25,27,28,35,36,40,50,56-59]. Shh inhibition found in Xenopus to be required for hepatic and pancreas fate decision, has been shown to improve the generation of mESC-Hep with the use of cyclopamine . In the same study, inhibition of Notch by DAPT also helped mESCHep development supposedly by favoring hepatocyte fate over the cholangiocyte fate . Hepatocyte growth factor (HGF) is known to promote hepatoblast proliferation, migration and survival through its tyrosine kinase receptor c-Met [60,61], and is therefore always used in PSC hepatic differentiation protocols. Following hepatic specification, combination of FGF10, retinoic acid and inhibition of Activin-A in the presence of SB431542 has been also reported to improve hepatic endoderm maturation . Maturation of hepatoblasts in the mouse fetal liver requires Oncostatin M (OSM) secreted by hematopoietic cells . Consequently, OSM has been also widely used in PSC hepatic cultures (Table 1).
Genetic modification-based hepatic differentiation protocols
The transcriptional machinery of the specified hepatic endoderm is very well documented including the homeodomain protein Mixer/ Mix.3 , the Sry-related HMG-box transcription factor Sox17 , the zinc-finger transcription factors Gata5, Gata4 and Gata6 [65,66], Hepatocyte nuclear factors HNF4α  and Hex . Some studies have used this information to improve efficiency of hPSC-hep generation by overexpression in a timely manner either Sox17, Hex and HNF4α [56,69,70]. Sox17 and Hex were transduced into the hESCs and hiPSCs differentiation cultures at day 3 and day 6 of differentiation respectively to induce hepatic commitment. Adenovirus vectormediated overexpression of HNF4α in later cultures could further promote the hepatic maturation of the hPSC-Hep. Similarly, using a Hex inducible mESC line, Kubo et al. , have shown that forced expression of Hex in endoderm cells dramatically improved mESCHep hepatic specification and maturation synergistically with BMP-4.
Epigenetic modification-based hepatic differentiation protocols
In addition to genetic modulation, epigenetic modifications have been shown to improve hepatic differentiation protocols. For instance, sodium butyrate, a well-known specific HDACs inhibitor was used in ESCs and iPSC differentiation a long time ago . In addition to decreasing cell death and promoting viability of ESCs, sodium butyrate has been reported to induce ESC differentiation toward different cell lineages including the neural cells , cardiac cells , pancreatic cells [73,74], and hepatic cells [74-79]. Ren et al. , showed specifically that the cell fate induced by sodium butyrate depends on the concentration and time of the treatment. For instance, a low concentration of sodium butyrate with shorter exposure time induce generation of pancreatic progenitors over the hepatic progenitors, while higher concentrations with longer exposure favor the hepatic lineage fate
Role of supportive cells and matrices in hepatic differentiation protocols
To recapitulate the inductive signals established in liver embryogenesis in PSC hepatic differentiation cultures, some studies have proposed to combine the use of cytokines or chemicals with cocultures with supportive cells or cultures on specific matrices found in the developing liver. For instance, Pal et al. used conditioned media derived from the HepG2 liver carcinoma cell line during hESC differentiation, and obtained a high yield population of mature hESCHep that served as a functional in vitro hepatic cell model to study the effects of ethanol toxicity . Fibroblast cells from different sources (STO feeder cells, 3T3 cells or ESC-derived fibroblast like cells) were utilized in several studies as supportive cells to improve hepatic differentiation of hESC as well as primate ESC and iPSC [81-83]. Huang et al. identified FGF2 and Activin A, as two factors secreted by the ESC-derived fibroblast important for endoderm induction. Additional source of hepatocytes harvested from rats in co-culture with mESC-derived cell cultures have been reported to improve mESC-Hep maturation . Endothelial cells, mesenchymal cells, kupffer cells and stellate cells constitute the microenvironment of the developing hepatocytes, and were also reported to provide support for PSC-hep specification and maturation [85-90]. Our lab used an immortalized endothelial cell line derived from mESC cultures to improve mESCHep generation . We demonstrated that endothelial cells are not only required for hepatic endoderm outgrowth as demonstrated previously by Zaret’s group , but are also essential earlier to induce hepatic specification of endoderm through dual repression of Wnt and Notch signaling in endoderm cells . This was one example of few studies showing that PSC differentiation cultures are not only a source of hepatic cells for future cell therapy, but also an in vitro culture system to understand liver organogenesis as much as embryology studies in animal models help improving PSC differentiation protocols.
Another alternative to produce better PSC-Hep in vitro is to use specific matrices or scaffold that would mimic the proper architecture of the in vivo microenvironment of the developing liver. Using an extracellular matrix microarray platform for the differentiation of mESC toward an early hepatic fate, Flaim et al.  have established combinations of extracellular matrices that synergistically impact both hepatic ESC differentiation and mESC-Hep hepatic functions. Specific studies have identified a wide variety of artificial materials and natural matrices to improve both human and mouse PSC-Hep generation such as collagen type I, vitrogen, matrigel, polyurethane foam , fibronectin, laminin , polyacrylamide , hollow fibers [59,96,97], poly-l-lactic acid plus polyglycolic acid , ultraweb nanofibers , alginate microbeads  and also recombinant E-cadherin substratum . Compared to the 2D culture, the 3D scaffold system provides the physical support to enable spontaneous spheroid formation and mass cultivation of PSC-derived cells. With the support of the scaffold and cytokines, hepatocyte-like cells could be generated with the purity of up to 98% without cell isolation .
Hepatic cell generation from direct reprogramming of fibroblasts
Since the discovery of iPSC by Yamanaka in 2006, many groups have successfully reprogramed somatic cells directly to specialized cells bypassing the pluripotent stem cell stage. Last year, two studies have shown that fibroblasts can be reprogrammed directly to hepatic cells with the help of several transcription factors essential for liver development. Huang et al.  have demonstrated that the transduction of mouse fibroblasts from p19arf-/- mice with GATA4, FoxA3 and Hnf1 leads to the generation of hepatic cells that express hepatic markers, and restore liver functions following transplantation in the Fah liver deficient deficient mouse model. The second study from Sekiya et al.  used a different set of transcription factors, HNF4α, FoxA1 and FoxA2 or FoxA3 to reprogram mouse fibroblasts into hepatic cells, and demonstrated the in vivo ability of the cells to improve 40% of survival 10 weeks after cell transplantation in Fah deficient mice .
In vitro characterization of the PSC-Hep
To characterize the PSC-Hep in vitro, most of the above studies analyzed hepatic markers that are expressed in the early (alphafetoprotein, transthyretin) and later stages of liver maturation (albumin, cytokeratin 19 (CK19), CK7, CK18, cytochrome p450 enzymes (CyP), α1-antitrypsin (ATT), tyrosine aminotransferase (TAT), γ-glutamyltranspeptidase (GGT), glutathione S transferase (GST), tryptophan 2,3-dioxygenase (TDO), asialoglycoprotein receptor 1 (ASGPR), phosphoenolpyruvate carboxykinase (PPC), glucose-6-phosphatase (G6P), hepar-1 apolipoprotein F, fibrinogen, fibronectin, constitutive androstane receptor (Table 1). In addition to marker expression analyses supporting the differentiation stages, functional in vitro assays are also performed and usually compared to those from primary hepatocytes. Since liver exerts functions related to metabolism, protein synthesis, urea production and detoxification, the functional assays usually include albumin secretion, glycogen storage, low-density lipoprotein uptake, indocyanine green uptake and release, cytochrome P450 enzyme metabolism, urea production and metabolism-mediated toxicity. Results of these in vitro assays for each study are summarized in Table 1.
The definitive functional assay for PSC-Hep is their ability to regenerate diseased livers from animal models and ultimately to increase the animal survival rate. Chronologic studies of repopulation assays (summarized in Table 2 clearly indicate an improvement over the last years of pre-clinical relevance of the PSC-Hep. The differences between pre-clinical outcomes are the result of the combination of the intrinsic functionality of the PSC-Hep generated in vitro discussed above and the liver-deficiency animals used for each study.
To study the liver regenerative ability of the PSC-Hep in vivo, cells are transplanted into animal models in which the liver is injured to provide space for the transplanted cells and a proliferative stimulus for the regeneration to occur. Some of these models provide specific growth advantage for the transplanted cells. The liver injury models used in these studies include two genetic models, some chemical models and partial hepatectomy reviewed recently by Shafritz and Oertel .
The first genetic mouse model was established in 1991 by Sandgren et al.  by overexpressing the transgene urokinase type plasminogen activator under the albumin promoter (Alb-uPA mice). In this mouse model, uPA is expressed strictly in hepatocytes and appeared to be highly cytotoxic for these cells. Most of the uPA is secreted in serum, however a small amount remains in the liver resulting to intense liver damage. Consequently, most of the newborns die from hemorrhaging between 4 and 6 week-of-age, the rest of the newborns survive from extensive liver toxicity due to clonal growth of revertant hepatocytes that lost the transgene. Rhim et al. have subsequently tested the ability of the Alb-uPA mice to host transplanted normal hepatocytes . They demonstrated that Alb-uPA mice provide a growth advantage to transplanted cells over the dying endogenous uPA expressing hepatocytes, and hence offer a permissive environment for survival, expansion and function of transplanted hepatocytes. Alternatively, Sandgren’s group has developed another transgenic mouse model expressing uPA under the major urinary protein (MUP) promoter in which uPA expression was initiated in hepatocytes only from 2 to 4 week-of-age, thereby preventing the uPA-mediated neonatal lethality .
The second genetic mouse model, the fumarylacetoacetate hydrolase deficient mice (Fah-/-mice), was created by Grompe et al.  in 1995 to recapitulate the hereditary tyrosinemia type 1 (HT1) in human characterized by mutation in the Fah gene. The lack of the Fah enzyme, which is involved in the tyrosine catabolism pathway, induces an accumulation in hepatocytes of toxic metabolites including fumarylacetoacetate (FAA), leading to chronic liver damage. Accumulation of FAA can be prevented by adding in drinking water the chemical agent NTBC (2-(2-nitro-4-fluoromethylbenzoyl)-1,3- cyclohexanedione) that is used in clinical settings to treat HT1 patients . NTBC stops the tyrosine catabolism pathway upstream of the Fah enzyme, blocking the accumulation of FAA, thus preventing liver cell toxicity.
Therefore, the liver injury in Fah deficient mice can be easily controlled by the NTBC intake and removal. Overturf et al.  have shown that transplanted wild type hepatocytes have a growth advantage, and hence are able to repopulate mutant livers and restore liver functions. More recently, both the Fah deficient mice and AlbuPA transgenic mice have been crossed to the immunosuppressed mouse models Rag2-/-, Il2Rγ-/- mice or the SCID mice (for AlbuPA) to allow liver repopulation by human hepatocytes and mouse hepatocytes harboring any genetic background [110-112].
These 2 genetic deficient liver mouse models provide not only liver injury but also selective growth advantage for transplanted human and mouse hepatocytes generated from pluripotent stem cells (27,50,78,113-116). Among those studies, Li et al.  demonstrated extensive liver repopulation capacity (24% +/- 15%) of purified mESCHep based on EpCAM expression (and negative for cKit) 10 weeks following transplantation in Fah-/- mice. Percentage of repopulation from PSC-Hep varied from 1.94%+/- 5.81 in MUP-uPA treated with CCl4 at 82 days , or 0.001%-12.5% in Fah-/- mice at 8 weeks , to detection of few clusters of transplanted cells in either Fah deficient mouse model  or uPA transgenic mouse model [27,114,115]. Two interesting studies describing the direct conversion of mouse fibroblasts by transducing a specific set of hepatic transcription factors demonstrated a remarkable ability of these cells to repair the damaged livers from Fah-/- mice [102,103].
The chemical liver injuries used to test the in vivo functionality of PSC-Hep include the carbon tetrachloride (CCl4)-induced hepatic centrilobular necrosis (38,39,117), the dimethylnitrosamine (DMN)- induced chronic cirrhotic injury [41,118] and the thioacetamide (TAA)-induced hepatic fibrosis . The acute liver injury induced by CCl4 has recently been shown to be a successful mouse model to allow high liver repopulation ability of mouse  and human [38,39] PSC-Hep. Transplantation of miPSC-Hep or hPSC-Hep in the CCl4- treated BALBc-nude mice reduced hepatic necrotic area, oxidative stress, and improved hepatic functions and the survival rate , and in a different study, regenerate about 10% of the liver mass 35 days after transplantation . Similarly, Chen et al.  have successfully rescued the acute liver injury induced by CCl4 in NOD/SCID mice following transplantation of hiPSC-Hep. Another acute liver injury was developed by Bandi et al.  using a combination of the hepatotoxic drugs rifampicin and phenytoin that inhibit host hepatocyte proliferation followed by injection of monocrotaline, a pyrrolizidine alkaloid agent which induces sinusoidal endothelial cell injury to facilitate transplanted cell engraftment. This regiment produced 50- 70% liver necrosis in NOD-SCID mice with 90-100% mortality over 2 weeks. Transplantation of hESC-derived meso-endodermal cells (hESC-MEC) intraperitoneally in this acute liver failure mouse model rescued the survival rate 2 weeks following surgery although further hepatic maturation of hESC-MEC was not successful .
Another variable parameter to consider when investigating the regenerative ability of PSC-Hep is the mode of cell delivery. Most of the published studies use the intra-splenic (IS) route to deliver PSCHep to the liver through the portal vein. These studies showed variable repopulation abilities of the transplanted cells, and in some cases partial rescue of the liver damage (38,39,50,78,113-116). Injections of cells through the portal vein deliver cells directly to the liver bypassing the spleen. Agarwal et al.  transplanted hESC-Hep through the portal vein of the CCl4 treated NOD/SCID, and showed some human cells integrated into the host liver. Cells can also be directly and locally targeted to the liver parenchyma or under the liver capsule. This mode of delivery is appropriate to newborn injection as the liver is visible through the skin. Few studies showed liver function improvement using this route [17,27,121]. The least invasive routes of delivery are intra-peritoneal (IP) and intra-venous (IV) injections.
Mostly, IP transplanted cells remain in the intra-peritoneal cavity  where they can still exert hepatic functions, even though one study has shown integration of the IP transplanted cells into the liver . In the first study, cells were delivered IP with microcarriers that eventually revascularized in the peritoneal cavity and secreted proteins in blood circulation improving hepatic functions, and rescuing 100% of animal survival 2 weeks after a mild injury and 38% after a stronger injury . IV injections of PSC-Hep allowed integration of cells into the host livers and have shown some liver function improvement [41,118,119].
In summary, the liver field of pluripotent stem cells has tremendously advanced the last decade with the generation in vitro of cells harboring hepatocytic functions and capable of improving some liver functions in vivo of mouse models with liver deficiency. Even though the hepatic differentiation protocols are presently very efficient and reproducible among different groups, the variability of hepatic differentiation efficiency between PSC lines remains unclear.
A recent publication from Yamanaka’s group has shown that the variation in hepatic differentiation efficiency of hiPSCs was mostly due to donor differences  rather than the tissue from which the iPSCs were derived as indicated by other studies using mouse [123-125] or human  tissues, or the iPSC derivation methods. This finding may impact our current vision of personalized medicine using hiPSC-Hep for liver functions rescue. In addition to differences in the propensity for hepatic differentiation among PSC lines, the consistency of the regenerative ability of PSC-Hep is still a challenge. Before the use of the PSC-Hep in clinical settings, many issues need to be resolved including namely the purity of the PSC-Hep cultures, the risk of teratoma development and the choice of delivery mode based on the liver damage. In addition to these technical limitations, there is a need to understand the interactions between the transplanted cells and the damaged microenvironment to improve PSC-Hep integration into the liver and their proliferation. One can suspect that some of these crosstalks are universal to all liver injuries, while some are specific to particular damages. Consequently, it is conceivable that understanding the response of the damaged microenvironment will help the generation of customized PSC-Hep to specific liver injuries.