Hepatocyte Selection Medium

does not contain hazardous reagents or introduce genetic material. Our results show that hiPS cells die after three days of culture in HSM. Prior to performing the experiments, we compared the hiPS cell viability in media containing crude or dialyzed KSR or combination of insulin (10 μM), dexamethasone (10 μM) and aprotitin (5000 U/ml) (IDA) Unexpectedly, the KSR dialysis and IDA had no effect on hiPS cell survival. As expected, primary human hepatocytes survived in HSM as well as in HCM, which is the recommended medium for their culture. HSM can be used in clinical practices in situations when hepatocytes differentiated from human iPS cells are transplanted to patients suffering from liver failure.


Introduction
Embryonic stem (ES) cells have the potential to differentiate to hepatocytes [1]. However, the use of ES cells may pose ethical problems because they are derived from human embryos. The use of human induced pluripotent stem (hiPS) cells that have been generated from adult somatic cells [2], on the other hand, does not create ethical controversies. HiPS cells are useful tools in drug discovery and regenerative medicine because they can differentiate into functional somatic cells [3]. If hiPS cells could be differentiated into hepatocytes, they would be useful for transplantation into patients suffering from hepatic failure [4]. Complications such as graft-versus-host disease as well as ethical issues could be avoided because patient-specific somatic cells could be generated from hiPS cells isolated from the patient.
The ES and hiPS cells that survive among the differentiated hepatocytes and are transplanted to patients may be tumorigenic [5]. Therefore, methods need to be developed to eliminate ES and iPS cells from the population of differentiated cells used for transplantation. To overcome these problems, a new medium, called "hepatocyte selection medium" (HSM), has been developed and will be discussed in this chapter [6].
First, pluripotency and tumorigenicity of ES and iPS cells will be discussed [7]. Next, current methods of eliminating pluripotent cells will be outlined [8,9]. All the cells, including human iPS cells, require glucose and arginine to live [10,11]. They will die without glucose or arginine. Hepatocytes have enzymes to produce glucose from galactose and arginine from ornithine. The unique features of hepatocytes compared with other cells will next be discussed. It was expected that hepatocytes would survive in a medium without glucose or arginine, and supplemented with galactose and ornithine [12] [13]. After this introduction, the formulation of HSM will be described [14]. Finally, the application of HSM for the selection of cells differentiated from mouse ES and human iPS cells will be presented.

Pluripotency and tumorigenicity
The link between pluripotency and tumorigenicity was reported in 1960 based on a study of teratocarcinoma [15]. ES and iPS cells are pluripotent and are capable of self-renewal as well as differentiation into a variety of cell types. Pluripotent cells can, however, be tumorigenic because they proliferate rapidly and exhibit telomerase activity [7]. Therefore, one of the problems faced while using ES and iPS cell-derived cells for transplantation into patients is the risk of tumorigenicity [5]. For example, transplantation of mouse hepatocytes differentiated from ES cells into liver resulted in the formation of teratoma [16]. Tumorigenicity was initially attributed to genomic integration of the viral vectors used for the induction of pluripotency [17]. The Sendai virus was also used to generate iPS cells because it posed no risk of altering the host genome [18]. To reduce this risk, plasmid vectors have been used to introduce reprogramming factors such as Oct3/4, Sox2, Klf4, and c-Myc [19]. In addition, the ES cell-specific microRNA, miR-302, has been used to reduce the iPS cells tumorigenicity by suppressing cyclin E-CDK2 and cyclin D-CDK4/6 [20]. Furthermore, Yakubov et al. introduced RNA synthesized from the cDNA of the four reprogramming transcription factors [21]. Several combinations of reprogramming factors have also been investigated. Nakagawa et al. omitted c-Myc to generate iPS cells, thereby reducing the tumorigenicity because c-Myc is a wellknown oncogene [22]. Despite these efforts, the risk of tumorigenicity has not yet been eliminated. The process of pluripotency and tumorigenicity involve self-renewal, proliferation, and active telomerase mechanisms [7]. It is, therefore, necessary to develop methods for the efficient eradication of iPS cells that survive among differentiated somatic cells.

Methods of eliminating iPS cells
Flow cytometry, which is commonly used to isolate target cells, was used by Yamamoto et al. to isolate hepatocytes differentiated from the mouse ES cells [8]. These workers generated ES cells expressing green fluorescent protein driven by an albumin promoter/enhancer. However, since albumin is expressed in endodermal cells as well [23], this approach led to isolation of cells other than hepatocytes, such as endodermal cells. Therefore, a different strategy was required to improve hepatocyte isolation. Flow cytometry has also been used to analyze surface antigens specific for hepatocytes. For example, delta-like 1 homolog (DLK1) has been used for isolation of hepatoblasts [9]. The issue with DLK1, however, is that this surface antigen is not expressed in the human adult liver [24]. Therefore, it may not be possible to isolate mature hepatocytes differentiated from hiPS cells using DLK-1 as a marker. In our research we focused on other methods to eliminate iPS or ES cells from heptocytes. Sub-lethal heat shock was shown to induce apoptosis in human ES cells [25], but it might also damage differentiated cells intended for transplantation. Cheng et al. reported the same strategy using suicide genes [26]. They introduced a thymidine kinase gene driven by the Nanog promoter into hiPS cells, which were subsequently ablated by ganciclovir treatment. This method may be ideal for differentiated hepatocytes, which do not express Nanog, but the toxicity of ganciclovir may be a potential issue. Conesa et al. screened a library of 1120 small chemicals to identify molecules that caused mouse ES cells to undergo apoptosis [27], and found that benzethonium chloride and methylbenzethonium induced apoptosis in hiPS and mouse ES cells but not in human fibroblasts or mouse embryonic fibroblasts. Both reagents are quaternary ammonium salts used as antimicrobial agents; they are also used in cancer therapy and may have damaging effects on hepatocytes. N-oleoyl serinol (S18), which is a ceramide analogue, eliminated residual pluripotent cells in embryoid bodies [28]. Interestingly, S18 also promoted neural differentiation of embryoid body-derived cells. This strategy is promising because the reagent not only eradicates undifferentiated cells but also promotes their differentiation into the target cell types.

Arginine and urea cycle
Among all the amino acids, the deficiency in arginine is the least tolerated by the cells cultured in vitro [29]. Arginine is produced through the urea cycle, which is exclusive to hepatocytes. Indeed, an arginine-deficient medium was the first one used for the hepatocyte selection [10]. Tyrosine also is produced by hepatocytes, and H4 II E, a hepatoma cell line adapted to growth in serum-, arginine-, and tyrosine-free medium, has been established [30]. This cell line expresses ornithine transcarbamylase (OTC) involved in the urea cycle, and phenylalanine hydroxylase (PAH), which catalyzes the synthesis of tyrosine in the liver and kidney [31].
The major role of urea synthesis is the excretion of ammonium ions generated in the process of protein degradation. Urea synthesis is a cyclic process as shown in Figure 1. Ornithine plays the key role in urea synthesis, and OTC mediates the formation of L-citrulline from L-ornithine and carbamoylphosphate. Importantly, this process occurs in liver mitochondria (area bounded with green line in Figure 1). The OTC deficiency, linked to X-chromosome is the cause of hyperammonemia type 2 [32]. The elevated ammonium levels lead to infantile death or mental retardation later in life.
Consequently, it could be expected that hepatocytes can be selected from ES cells in a medium deficient in arginine and tyrosine.

Glucose and gluconeogenesis
Glucose is an important source of energy for a majority of cells. Glucose deprivation aids in the hepatocyte selection process because hepatocytes are capable of synthesizing glucose [10]. Pyruvate is the final product of glycolysis, which then enters the tricarboxylic acid cycle. It was shown that pyruvate and glucose deficiency led to neural cell death [11]. Galactose enters glycolysis as a substrate for galactokinase, which is expressed in the liver and kidney [33,34]. Therefore, it is expected that hepatocytes can survive in a medium deprived of glucose or pyruvate but supplemented with galactose [12] [13].
Galactose is produced from lactose by hydrolysis in the gastrointestinal tract and is converted to glucose in the liver (Figure 2). Galactokinase catalyzes ATP-dependent phosphorylation of galactose to galactose 1-phosphate which then reacts with uridine diphosphate(UDP)-glucose to produce UDP-galactose converted to UDP-glucose by uridine diposphogalactose 4epimerase. UDP-glucose is used by glycogen synthase to synthesize glycogen, which is stored in the liver and used as a source of glucose.
Deficiency in the enzymes such as galactokinase, galactose-1-phosphate uridyltransferase, or uridine diphosphogalactose 4-epimerase causes galactosemia. Galactose is then reduced to galactitol, which accumulates in the eye lenses causing cataracts. Deficiency in galactose-1phosphate uridyltransferase results in accumulation of galactose-1-phosphate and depletion of inorganic phosphate in the liver causing liver failure. This is the reason why children suffering from galactosemia are kept on a galactose-free diet.

Hepatocyte selection medium
The hepatocyte selection medium (HSM) was made from powdered amino acids following the formulation of Leibovits-15 medium (Life Technologies, Grand Island, NY). HSM did not contain arginine, tyrosine, glucose, and sodium pyruvate, but was supplemented with galactose (900 mg/L), ornithine (1 mM), glycerol (5 mM), and proline (260 mM) (all from Wako Pure Chemicals, Osaka, Japan); proline (30 mg/L) was added as a component necessary for DNA synthesis [35]. Aspartic acid as a nonessential amino acid was not included because it can be synthesized from ornithine and arginine. Fetal calf serum (FCS, Life Technologies) at a final concentration of 10% was used to culture mouse ES cells. For human iPS cells, 10% knockout serum replacement (KSR) (Life Technologies) was used instead of FCS to establish xeno-free conditions. Depending on the experiment, FCS and KSR were dialyzed against phosphate buffered saline (PBS) to remove amino acids and glucose.

Embryoid bodies in HSM
EB5, a mouse ES cell line provided by Dr. H. Niwa (Center for Developmental Biology, Riken, Kobe, Japan) was maintained in the undifferentiated state in gelatin-coated dishes without feeder cells, in Glasgow minimum essential medium (GMEM) (Sigma Aldrich Japan K.K., Tokyo, Japan) supplemented with 10% FCS (Roche Diagnostics K.K., Tokyo, Japan), 1× nonessential amino acids (NEAA), sodium pyruvate (1 mM), leukemia inhibitory factor (LIF) (1000 U/ml) (Invitrogen Japan, Tokyo, Japan), and 2-mercaptoethanol (0.1 mM) (Wako) [36]. Dissociated ES cells were cultured in hanging drops at a density of 1 × 10 3 cells per 30 μμl of media without LIF (ESM) to form embryoid bodies. After four days in hanging drop culture, the resulting embryoid bodies were plated onto plastic dishes (Iwaki-Asahi Techno Glass, Tokyo, Japan) precoated with gelatin (Sigma Aldrich). Seven days after their formation, the embryoid bodies transferred to HSM appeared slightly smaller than those in ESM. The cells  28 days after the formation of embryoid bodies, sizes of colonies in HSM reduced, and the surviving cells appeared cuboidal (Figure 3). Some of these cells were binuclear, which is characteristic of hepatocytes; it was also previously shown that HSM was selective for hepatoblast-like cells [14]. These results suggest that HSM eliminated undifferentiated cells and enriched the population of hepatoblast-like cells.

Expression levels of GALK1, GALK2, and OTC
Human fetal and adult hepatocytes express galactokinase and OTC, and would survive in HSM containing galactose and ornithine. If hiPS cells express similar levels of these enzymes, HSM could not be applied for selection of differentiated hepatocytes. Therefore, we compared the expression levels of galactokinase and OTC in hiPS cells with those in human fetal and adult livers. The hiPS cell line 201B7 (RIKEN Cell Bank, Tsukuba, Japan) was cultured feederfree in ReproFF medium (Reprocell, Yokohama, Japan) in dishes coated with a thin layer of Matrigel (Becton Dickinson, Franklin Lakes, NJ). Two galactokinase isoforms, GALK1 (GenBank: NM_000154) and GALK2 (BC107153), have been identified in humans. The expression levels of GALK1, GALK2, and OTC in the 201B7 cells and fetal and adult livers were compared [6]. The expression levels of these enzymes in the 201B7 cells constituted 22.2% ± 5.0%, 14.2% ± 1.1%, and 1.2% ± 0.2% (mean ± standard deviation) of those in the adult liver, respectively, and the OTC expression was also significantly lower in the 201B7 cells than in the fetal liver. We then cultured 201B7 cells in HSM to assess their survival rates.

Human iPS cells in HSM
The 201B7 cells were cultured in 6-well plates coated with Matrigel in the ReproFF medium, which was then changed to HSM (Figure 4). The 201B7 cells started to die and were completely eliminated in three days. Nuclear condensation and fragmentation was observed after staining with hematoxylin and eosin [6]. These nuclei also tested positive by terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labeling (TUNEL). Some of the 201B7 cells that survived in HSM one day after medium change to HSM were immunostained with antibodies against Nanog, SSEA4, and TRA-1-60. The results suggested that the death of undifferentiated 201B7 cells in HSM was caused by apoptosis.

Primary human hepatocytes
Several protocols for the differentiation of iPS cells to hepatocytes have been reported [3,37], which describe the differentiation of iPS cells into hepatocyte-like cells which are different from primary human hepatocytes. Recently, a method to generate three-dimensional vascularized liver from iPS cells has been reported [38]. The authors induced hepatic differentiation of human iPS cells by following the protocol described by Si-Tayeb et al [37]. They mixed the iPS cells with vascular endothelial and mesenchymal stem cells, and transplanted them into a mouse brain. This method is sophisticated and promising, but xenograft rejection may be a problem when the generated liver is transplanted to patients with liver failure. Practical methods for the differentiation of human iPS cell to functional hepatocytes are not available. It is therefore necessary to use primary human hepatoctyes as a model of hepatocytes fully differentiated from iPS cells. Hepatocytes were isolated from a fragment of resected donor liver by using 2-step collagenase perfusion [39].

Co-culture of human iPS cells and primary human hepatocytes
Methods have not been established regarding hepatocye differentiation from human iPS cells. It is impossible to select hepatocytes differentiated from human iPS cells from the mixture of human iPS cells. Primary human hepatocytes were used as a model of hepatocytes differentiated from human iPS cells. It was expected that human iPS cells and hepatocytes differentiated from them were mixed. Therefore, co-culure of primary human hepatocytes and human iPS cells was used as a model of the mixtures. Primary human hepatocytes were purchased from Lonza (Walkersville, MD) and cultured as per the manufacturer's instructions. Briefly, hepatocytes were thawed and spread at a density of 1.5 × 10 5 cells/cm 2 onto CellBIND 24-well plates coated with type I collagen from the bovine dermis (Koken Co., Ltd., Tokyo, Japan) and cultured in the hepatocyte culture medium (HCM, Lonza).
The 201B7 cells and human primary hepatocytes were co-cultured as follows: human primary hepatoctyes were cultured in HCM for 24 h as described above. The 201B7 cells were added to the wells at a density of 3 × 10 4 cells/well. After 24 h of culture in the ReproFF medium, it was changed to HSM. Human primary hepatocytes survived in HSM, while the human 201B7 cells did not ( Figure 5).

Potential application of HSM
The HSM that we developed can be safely used for the elimination of hiPS cells because it does not contain hazardous reagents or introduce genetic material. Our results show that hiPS cells die after three days of culture in HSM. Prior to performing the experiments, we compared the hiPS cell viability in media containing crude or dialyzed KSR or combination of insulin (10 μM), dexamethasone (10 μM) and aprotitin (5000 U/ml) (IDA) Unexpectedly, the KSR dialysis and IDA had no effect on hiPS cell survival. As expected, primary human hepatocytes survived in HSM as well as in HCM, which is the recommended medium for their culture.
HSM can be used in clinical practices in situations when hepatocytes differentiated from human iPS cells are transplanted to patients suffering from liver failure.

Conclusion
HSM can be successfully used for the selection of hepatoblast-like cells derived from mouse ES cells. HSM is an ideal medium for the elimination of hiPS cells and the isolation of differentiated hepatocytes without causing any damage. In the future, methods will be established to produce hepatocytes from human iPS cells. Residual human iPS cells are a potential hazard when the hepatocytes will be transplanted for patients with liver insufficiency because the undifferentiated cell harbor tumorigenicity. At that stage, HSM will be an indispensable medium to select hepatocytes differentiated from residual human iPS cells.