A synthetic nanofibrillar matrix promotes in vitro hepatic differentiation of embryonic stem cells and induced pluripotent stem cells

The liver is an important organ that performs many complex functions, including the metabolism of carbohydrates, proteins and lipids, as well as storage of essential nutrients and biotransformation of drugs. Drug biotransformation involves detoxification and bioactivation, where the metabolite becomes more toxic. Therefore, drug biotransformation plays an important role in the early stages of drug discovery processes. Primary hepatocyte cultures are often used for pharmacological assays, but they are short-lived and cannot be maintained in long-term culture. In addition, there are considerable donor-dependent variations. By contrast, embryonic stem (ES) cells or induced pluripotent stem (iPS) cells can proliferate infinitely and maintain their pluripotent ability to differentiate into various cell types. There is evidence that ES or iPS cells recapitulate normal developmental processes, and can serve as an alternative resource for hepatological researches, drug development and clinical uses. Through our present knowledge of developmental biology, efficient induction of hepatic lineage cells has been established. For example, based on the evidence that TGFβ–activin–Smad2 signaling is involved in definitive endoderm formation in the mouse (Tremblay et al., 2000), the activation of Activin–Nodal signaling was used for endoderm induction (D'Amour et al., 2005; Kubo et al., 2004). Fibroblast growth factor (FGF) and bone morphogenetic protein (BMP) were added for the specification of liver lineages (Jung, 1999; Mfopou et al., 2010; Shiraki et al., 2008a); this helped to mimic the mesodermal signals from the septum transversum mesenchyme in normal development (Katsumoto et al., 2010; Shin et al., 2007; Rossi et al., 2001). Because hepatocyte growth factors are known to be important effectors in the specification of cell fate and organogenesis of the liver (Schmidt et al., 1995; Sonnenberg et al., 1993), hepatocyte growth factor (HGF), dexamethasone and oncostatin M have been used for induction of hepatocyte maturation (Basma et al., 2009; Kamiya et al., 1999; Si-Tayeb et al., 2010). Compared with the factors described above that direct hepatic differentiation, the role of extracellular matrices (ECMs) and scaffolds remains unclear.

We have previously reported that culturing ES/iPS cells on the mesonephric M15 cell line, in the presence of specific growth factors, resulted in an efficient induction of endoderm-derived tissues, such as the liver or pancreas (Shiraki et al., 2008a; Shiraki et al., 2008b; Umeda et al., 2013). We further suggested that the basement membrane components, including lama5, play an important role in guiding the differentiation of ES cells into regional-specific lineages of the definitive endoderm (Higuchi et al., 2010). We also successfully established an alternative hepatic differentiation procedure without using feeder cells, but with a synthesized basement membrane (sBM) substratum (Higuchi et al., 2010; Shiraki et al., 2011). Together, these results revealed the importance of the ECM for differentiation of ES cells.

The basement membrane, a highly integrated three-dimensional structure composed of ECM molecules, is known to regulate various cellular processes. It is known that electrospun nanofibers provide not only three-dimensional microenvironments mimicking the ECM, but also appropriate guidance cues to modulate cell behavior. Here, we tested the effects of synthetic nanofiber (sNF) matrices on ES/iPS cell differentiation. We found that ES/iPS cells grown on the sNF were induced into endoderm and then hepatic fates. Overall, we conclude that the sNF is more potent in promoting hepatic differentiation, compared with the traditional two-dimensional culture surfaces, and is able to substitute for the sBM or M15 cells.

We first tested the sNF matrix for its potency to mimic the basement membrane substratum of the cells. Murine SK7 ES cells (Shiraki et al., 2008a) were seeded onto the sNF matrix, and allowed to differentiate into the hepatic lineage by sequential changes of medium containing specific growth factors (Fig. 1A). We found that the expression of the pluripotent marker Oct3/4 was downregulated, whereas the mesendoderm marker Gsc and definitive endoderm markers Sox17 and Foxa2 were expressed at day 4 (d4) of differentiation (Fig. 1B). Whereas Gsc was downregulated rapidly, Sox17 and Foxa2 showed peak expressions around d7 and were downregulated afterwards. Notably, the hepatic progenitor marker gene, alpha-fetoprotein (Afp), and the mature hepatocyte marker, albumin (Alb1) were detectable from d12 and d16, respectively, and their expression levels were increased with time. Although the Afp transcript level was decreased after d22, the Alb1 expression continued increasing beyond d22. The immunocytochemical analysis further confirms that ALB and AFP were present in the cytoplasm of differentiated ES cells (Fig. 1C). In addition, periodic-acid–Schiff (PAS) staining and the Indocyanin Green (ICG) test were also conducted to investigate the hepatocyte functions of differentiated ES cells. The former reflects glycogen storage by showing positive populations as magenta in the cytoplasm, and the latter is used to examine cellular uptake activities, which are regarded as a hepatocyte detoxification function. As shown in Fig. 1D, glycogen storage was observed as the accumulation of magenta staining in the cytoplasm of the differentiated cells (top panel) and the ICG test also shows a similar result (bottom panel).

We next investigated whether human ES or iPS cells could differentiate in the sNF system. We used khES3 human ES cells (supplementary material Fig. S1A–D), as well as human iPS cell lines, such as Toe (supplementary material Fig. S1E,G) and 201B7 (supplementary material Fig. S1F), and found that these cells were able to differentiate into hepatocyte-like cells, thereby producing ALB and taking up ICG (supplementary material Fig. S1D–G). Together, these results indicate that sNF is a suitable matrix for potentiating hepatic differentiation, not only in murine cells, but also in human ES cells and iPS cells.

To compare the supportive effects of NFs and other substrata, we seeded murine ES cells onto either the sNF matrix or normal plates precoated with other stubstrata, including collagen I, Matrigel, gelatin and fibronectin and then performed the differentiation experiment as described in Fig. 1A. On differentiation day 22 (d22), quantitative PCR analyses were carried out to quantify the expression profiles of hepatic function marker genes in differentiated cells. Our results indicate that ES cells grown on sNF showed higher expression levels of proteins secreted by hepatocytes, such as serine peptidase inhibitor a1 (Serpina1), Ttr and Alb1, compared with those grown on other substrata (Fig. 2). Similar results were observed for the expression of several other genes, including glucose 6-phosphatase (G6p) and fatty-acid binding protein (Fabp1), or transporters, such as organic anion-transporting polypeptide 1 (Oatp1), Na⁺-taurocholate cotransporting polypeptide (Ntcp) and UDP-glucuronosyltransferase (Ugt1a1), as well as multidrug resistance-associated protein family proteins 2 and 3 (Mrp2 and Mrp3) (Fig. 2). By contrast, little change was found in the expression of glucose transporter 2 (Glut2).

To determine the hepatic functions of differentiated ES cells grown on Matrigel, collagen I or sNF, we next measured their ALB secretions, ICG uptake and cytochrome p450 (CYP) activities. Our results show that ES cells grown on sNF secreted ALB at a higher level compared with those on Matrigel or collagen I (Fig. 3A). By day 26, the percentage of ICG-positive ES cells on sNF was also higher than in ES cells grown on the other two substrates (Fig. 3B,C).

To measure CYP activities, the differentiated ES cells were treated with a CYP1A inducer, 3-methylcholantrene (3MC), for 48 hours during d22–d24 or d64–d66, as shown in supplementary material Fig. S2A. We found that the differentiated cells cultured on sNF had higher CYP1A1 activities and responses to the inducer than those cultured on fibronectin, Matrigel or collagen I (supplementary material Fig. S2B). We also assayed the effects of sNF on the maintenance of the mature hepatic cells. ES cells cultured on sNF were able to maintain their CYP1A1 activities and responses to 3MC even on d66, whereas those cultured on other matrices did not survive in long-term cultures (supplementary material Fig. S2C). It is also worth noting that the differentiated cells on sNF could be maintained in culture for more than 100 days. Specifically, we show that the ES cells cultured on sNF for 129 days were able to uptake and secrete ICG (supplementary material Fig. S2D). Based on these findings, we conclude that sNF is an excellent matrix, not only for the differentiation of ES cells into the hepatic lineage but also for maintaining the mature state of ES-cell-derived hepatocytes.

Next, we investigated the effects of sNF on hepatic differentiation of ES cells. It was previously reported that the undifferentiated murine ES cells cultured on sNF exhibited spheroid morphologies and formed dome-like structures, and proliferated well (Nur-E-Kamal et al., 2006). Therefore, we checked the morphological changes between the undifferentiated and differentiated states of the murine ES cells. To exclude the effect of the extracellular matrix, we compared gelatin-coated normal two-dimensional (2D) plates with either gelatin-coated or uncoated sNF. We found that the undifferentiated ES cells grown on gelatin-coated 2D plates, with cytoplasmic spreading morphologies and attached to the plate surface in large areas (Fig. 4A, left). By contrast, the ES cells grown on sNF showed aggregated morphologies (Fig. 4A, middle and right).

In the differentiated state, the ES cells were found to form a monolayer on sNF. However, the morphological differences were still observed between ES cells cultured on sNF or normal 2D plates (Fig. 4B). Because these morphological changes are known to be regulated by cytoskeletal molecules, such as small Rho GTPase family member proteins, we next examined Rac1 activities in undifferentiated and differentiated ES cells. Our western blot analyses demonstrate that the GTP-bound active form of Rac1 was expressed at a higher level in ES cells cultured on sNF than those cultured on normal 2D plates, even though total Rac1 expression levels were similar (Fig. 4C,D). Interestingly, both the undifferentiated (Fig. 4C) and the differentiated ES cells on d8 (Fig. 4D) showed higher Rac-GTP activities. These results not only agree with the morphological differences of the ES cells, but also suggest that activated Rac1 plays a crucial role in potentiating the differentiation activity of ES cells cultured on sNF into hepatic lineages.

NSC23766 is a selective inhibitor of Rac1 activation that is mediated by the Rac-specific guanine nucleotide exchange factors (GEFs) TrioN and Tiam1, without affecting other Rho family members, such as RhoA or Cdc42 (Gao et al., 2004). We confirmed that 100 µM NSC23766 inhibited Rac1 hydroxylation (supplementary material Fig. 3A). To test whether sNF potentiates the differentiation of ES cells into the hepatic lineages through Rac1 activation, we treated murine ES cells with 100 µM NSC23766 at various stages and then determined the expression of stage-specific markers (Fig. 5A–C). We first added NSC23766 for 4 days at stage I to examine the effect of Rac1 activation on endoderm induction (Fig. 5A). We found that Foxa2 expression was downregulated by the Rac1 inhibitor in ES cells cultured on sNF (Fig. 5A). We next examined the effects of Rac1 inhibition on hepatic differentiation. Our results show that the Rac1 inhibitor added at stage II (Fig. 5B) or stage III (Fig. 5C) downregulated the expression of hepatic markers, Afp or Alb1, on d10 or d14, respectively.

These results suggested that Rac1 activation is crucial for endoderm and hepatic differentiation. We subsequently examined the stage dependency of hepatic differentiation on Rac1 activities. The Rac1 inhibitor was added at different stages (I, II, III or IV) and Alb1 expression was assayed on day 18 (Fig. 5D). We found that Rac1 inhibition at all four stages blocked the potentiating effects of sNF, and resulted in decreases in Alb1 expression. These results further confirm the important role of Rac1 and demonstrate that continuous activation of Rac1 is crucial for the potentiation of hepatic differentiation.

Then we confirmed whether NSC23766 had any effect on the proliferation of ES cells. NSC23766 decreased the proportion of EdU-positive cells in stages I, II and III, particularly in stage I, without apparent toxicity (supplementary material Fig. S3C). Interestingly, the total numbers of cells in the NSC23766-treated groups was smaller in stages I and II, which became greater than that of control groups used in stages III and IV (supplementary material Fig. S3B). Taken together, these findings suggest that Rac1 differentially contributes to proliferation in the early differentiation stages and promotes differentiation in the late stage.

Our previous study suggested that although addition of soluble growth factors is sufficient to promote the differentiation of ES cells into the definitive endoderm, further differentiation from the definitive endoderm into hepatic and pancreatic fates appears to require a direct contact with M15 cells (Shiraki et al., 2008a). We previously showed the importance of basement membrane substratum by culturing ES cells on sBM (Higuchi et al., 2010; Shiraki et al., 2011). Specifically, ES cells grown on sBM were able to differentiate into hepatic and pancreatic lineages. These results imply that the basement membrane structure plays a major role in the differentiation of ES cells. Although the sBM used previously was constructed by overexpressing recombinant laminin-511 (laminin α5, laminin β1 and laminin γ1) in H293 cells (Doi et al., 2002), ES cells or iPS cells could be induced into the hepatic and pancreatic lineages. The efficacy of such an sBM for differentiation was high, and the differentiated cells could perform liver-specific functions, such as protein secretion, detoxification and glycogen storage (Higuchi et al., 2010; Shiraki et al., 2011).

The nanofiber produced by the electrospinning technique is a chemically and physically stable synthetic three-dimensional surface that mimics the structural geometry and porosity of the basement membrane or ECM (Schindler et al., 2005; Schindler et al., 2006). NF scaffolds have been shown to recapitulate the structural features of stem cell niche (Lim and Mao, 2009) and have been used for the ex vivo expansion of various types of stem cells such as murine ES cells (Hashemi et al., 2011; Nur-E-Kamal et al., 2006) and human tissue stem cells. In addition, the ECM was found to deposit as an extensive scaffold on the basal surface of the cells attached to NFs (Shih et al., 2006; Chua et al., 2007; Ma et al., 2008). Importantly, the sNF system has been reported to enhance not only the differentiation of murine ES cells into neural lineages (Lim et al., 2010; Purcell et al., 2012; Xie et al., 2009), but also differentiation from human MSCs into hepatoblasts (Ghaedi et al., 2012; Kazemnejad et al., 2009).

Taken together, these previous observations revealed that the sNF matrix is useful as a substratum to replace feeder cells and that it has the ability to potentiate hepatic differentiation. In this study, we show that both murine and human ES cells, as well as human iPS cells, could differentiate on sNF and exhibit liver-specific functions. Furthermore, we demonstrate that Rac1 activation was involved in hepatic differentiation. Rac1, a member of the Rho family GTP-binding proteins, including Rho and Cdc42 (Heasman and Ridley, 2008), functions by activating actin-rich lamellipodial protrusion and membrane ruffling, which are thought to be a major part of the driving force for cell movement (Nobes and Hall, 1995; Ridley et al., 1992). Although Rho family proteins were reported to be expressed by ES cells cultured on sNF (Nur-E-Kamal et al., 2005; Nur-E-Kamal et al., 2006; Schindler et al., 2006), their roles have never been investigated.

In vivo developmental processes occurring in the endoderm and its derivatives cause dynamic migration during gastrulation and later stages of organogenesis (Woo et al., 2012), suggesting that motility and differentiation are closely inter-related. In this study, we observed that ES cells cultured on sNF showed greater Rac1 activation than did cells cultured on the normal 2D surface. Indeed, Rac1 is known to be involved in not only endoderm induction but also hepatic specification and maturation. In particular, Rac1 mutant mice died by mouse embryonic day 9.5 (E9.5) because of severe developmental abnormalities, and Rac1-deficient embryos showed numerous cell deaths in the space between the ectoderm and endoderm at the primitive streak stage (Sugihara et al., 1998). Rac1 is also important for cellular differentiation, for example, epithelial differentiation in the small intestine (Stappenbeck and Gordon, 2000), pancreatic islet morphogenesis (Greiner et al., 2009), myogenic differentiation (Heller et al., 2001), maintenance of the thymic epithelial cells (Hunziker et al., 2011), formation of the lens (Maddala et al., 2011) and neuronal development (Corbetta et al., 2009; Leone et al., 2010). In addition, Rac1 has been shown to crosstalk with many downstream signaling pathways such as Wnt (Clarke, 2006; Malliri et al., 2006), TGF-β1 (Varon et al., 2008), Nodal (Woo et al., 2012), retinoic acid (Lee et al., 2008) and Myc (Hunziker et al., 2011; Nikolova et al., 2008). Interestingly, Rac1 is also known to mediate stem cell-shape-dependent regulation of differentiation to a chondrogenic versus myogenic fate (Gao et al., 2010). On the basis of these studies, we postulate that the sNF system might potentiate ES cells to differentiate into hepatic lineages by interacting downstream of certain growth factors during differentiation processes.

In conclusion, we show that Rac1 was activated in both undifferentiated and differentiated ES cells cultured on sNF plates and that Rac1 inhibition blocked the potentiating effects of sNF on endoderm and hepatic differentiation. These results suggest that continuous activation of Rac1 throughout the differentiation stage is crucial for potentiating differentiation. Our results also highlight the morphological changes during differentiation along the Rac1 pathway, which controls cellular morphology, motility and differentiation into the hepatic lineage. Here, we established a completely chemically defined method that requires no serum or no xenogenic substrata, thereby eliminating the risk of contamination with unknown factors. We believe that this novel method could be an attractive culture model for pharmacological research and research on stem cell biology and therapeutic strategies.

The murine ES cell line, SK7 (Shiraki et al., 2008a) was maintained on mouse embryonic fibroblast (MEF) feeders in Glasgow minimum essential medium (Invitrogen, Glasgow, UK) supplemented with 1000 units/ml leukemia inhibitory factor (LIF; Chemicon, Temecula, CA), 15% knocked-out serum replacement (KSR; Invitrogen), 1% fetal bovine serum (FBS; Hyclone, Logan, UT), 100 µM nonessential amino acids (NEAA; Invitrogen), 2 mM L-glutamine (L-Gln; Invitrogen), 1 mM sodium pyruvate (Invitrogen), 50 units/ml penicillin and 50 µg/ml streptomycin (PS; Invitrogen) and 100 µM β-mercaptoethanol (β-ME; Sigma-Aldrich, St Louis, MO).

Human ES cells (KhES-3) (Suemori et al., 2006) were from Dr Norio Nakatsuji and Dr Hirofumi Suemori (Kyoto University, Kyoto, Japan). They were used in accordance with the human ES cell guidelines of the Japanese government. This human ES work was approved by Kumamoto University institutional review board. Human iPS 201B7 cells were a gift from Dr Yamanaka (Kyoto University, Kyoto, Japan). The human iPS Toe cell line was established by M. Toyoda and colleagues (National Institute for Child Health and Development, Tokyo, Japan). Undifferentiated human ES and iPS cells were maintained as described previously (Shiraki et al., 2008b).

Synthetic nanofiber (sNF) matrices were purchased from Corning Coster (Ultra-Web Synthetic Polyamine Surface #3873XX1; Cambridge, MA). Plate surfaces were coated with electrospun polyamide nanofibers. sNF matrices consisted of two kinds of polyamide polymers, A (C₂₈O₄N₄H₄₇)_n and B (C₂₈O_4.4N₄H₄₇)_n, which were crosslinked in the presence of an acid catalyst, and were 200–400 nm in diameter (average 280 nm). Pore sizes, similar to those of the cell basement membrane, were ∼700 nm. For comparison, Corning 96-well plates were pretreated for 3 hours at 37°C with 0.1% gelatin (Sigma-Aldrich), Matrigel (BD, Franklin Lakes, NJ) or CellStart (Invitrogen). Collagen I (Nitta Gelatin, Japan) was diluted with Dulbecco's modified Eagle's medium (DMEM; Invitrogen) at a concentration of 1 mg/ml and plate surfaces were treated for 15 minutes, then dried until use.

Murine ES cells plated at a density of 1.5×10⁴ cells/cm² in culture plates described above were grown for 8 days in DMEM containing 4,500 mg/l glucose, s NEAA, L-Gln, PS, β-ME, 10 mg/ml insulin, 5.5 mg/ml transferrin, 6.7 pg/ml selenium (Insulin-Transferrin-Selenium-G Supplement; ITS, Invitrogen), 0.25% AlbuMAX II (Invitrogen), 10 ng/ml recombinant human activin-A (R&D Systems, Minneapolis, MN), 5 ng/ml; recombinant human bFGF, and cultured for 8 days. On day 9 (d9), the medium was changed to RPMI-1640 (Invitrogen) containing 10⁻⁶ M retinoic acid (RA; Stemolecule all-trans retinoic acid; Stemgent, Cambridge, MA) and B27 supplement (Invitrogen). On d10, medium was switched to 2000 mg/l glucose DMEM (Invitrogen), 10% KSR, 10 ng/ml recombinant human hepatocyte growth factor (Peprotech, Rocky Hill, NJ) and 10 µM dexamethasone (Sigma-Aldrich), and cultured until d14. Next, 1 mM nicotinamide (NA; Sigma-Aldrich) and 1% dimethylsufoxide (DMSO; Sigma-Aldrich), were added to medium and KSR was removed. Medium was replaced every 2 days with fresh medium and growth factors.

Human ES/iPS cells were pretreated with the ROCK inhibitor Y27632 (Wako, Japan) 1 day before trypsinization. Cells were plated at a density of 3×10⁵ cells/cm² on Matrigel-coated sNF matrices with Y27632. The following two procedures were subsequently used to induce hepatic differentiation of various human ES/iPS cells; simplified (two-step) protocol, KhES3 and 201B7 cells; or conventional (three-step) protocol, Toe cells. In the simplified protocol, medium used at first contained B27 and 100 ng/ml activin-A in RPMI-1640, which was then, switched to 10 ng/ml HGF, 10 µM dexamethasone, 0.5% DMSO, 0.5 mM NA. In the conventional protocol, medium used first was the same as that in the simplified protocol, followed by 1% DMSO and 20% KSR in knockout DMEM/F12 (Invitrogen) for 6 days and, then DMEM containing HGF, dexamethasone and 10% KSR. Finally, the above medium was added with 0.5 mM NA. Medium was replaced every 2 days with fresh medium and growth factors. KhES3 and 201B7 cells were induced hepatic differentiation with simplified protocol, and Toe cells were treated with conventional protocol.

For detection of glycogen storage in the differentiated cells, periodic-acid–Schiff's (PAS) staining kit (Muto Pure Chemicals, Tokyo, Japan) was used. Cells cultured for 9 and 26 days, and undifferentiated ES cells were fixed in 3.3% formalin for 10 minutes, and stained following the manufacturer's instructions, then nuclear counterstaining with hematoxylin (blue) was performed.

The culture medium was replaced with fresh medium every 2 days, and supernatants were collected 24 hours after replacing the medium. The mouse (human) albumin secreted in the supernatant was determined using a mouse (human) ELISA quantification kit (Bethyl, Montgomery, TX).

Indocyanine Green (Daiichi-Sankyo Pharm., Japan) was diluted with the above culture medium to a final concentration of 1 mg/dl. The ICG test solution was added to the differentiated ES cells after the appropriate culture periods and undifferentiated ES cells were used as controls, and incubated at 37°C for 30 minutes. Then, after three washes with phosphate-buffered saline (PBS), the cellular uptake of ICG was examined by microscopy. The percentage ICG-positive areas represent the proportion of ICG-positive area versus total cell area, which were determined using ImageJ software (US National Institutes of Health, Bethesda, MD).

To check the inducibilities of cytochrome P450 activities in response to inducers, we used the P450-Glo CYP Assay Kit (Promega, Madison, WI). The differentiated ES cells were treated with 5 µM 3-methylcholantrene as inducers of CYP1A. The medium containing the inducers was changed every 24 hours. 48 hours after treatment, we changed the medium and used the appropriate luminogenic CYP substrates (Luciferine-CEE for CYP1A). The cells were incubated at 37°C for 3 hours, and then the supernatants were mixed with equal amount of detection reagent, according to the manufacturer's instructions. The luminescence was measured using a GloMax 96 microplate luminometer (Promega), and luminometer settings were as in the manufacturer's instructions. Cell numbers were calculated using CellTiter-Glo luminescent cell viability assays (Promega) to normalize P450-Glo assay values to cell number.

After culture for the appropriate times, cells were fixed in 4% paraformaldehyde in PBS for 30 minutes at room temperature. After removal of paraformaldehyde solution, the fixed cells were permeabilized with 0.1% Triton X-100 for 10 minutes. The permeabilized cells were rinsed several times with PBS and were then incubated with 20% Blocking One (Nacalai Tesque, Japan) in PBST (0.1% Tween-20 in PBS) for blocking. After blocking, the cells were incubated with the diluted antibody in 20% Blocking One in PBST (0.1% Tween-20 in PBS) in a humidified chamber overnight at 4°C. After washing the cells in PBST, cells were incubated with the secondary antibody in 20% Blocking One for 2 hours at room temperature in the dark. After washing off the secondary antibody in PBST, cells were counterstained with 6-diamidino-2-phenylindole (DAPI) (Roche Diagnostics, Switzerland). The following antibodies were used as primary antibodies: rabbit anti-alphafeto protein (Dako, Denmark), goat anti-albumin (Sigma-Aldrich), goat anti-Sox17, mouse anti-FoxA2 (R&D systems); secondary antibodies used were conjugated to Alexa Fluor 568, Alexa Fluor 488 and Alexa Fluor 633 (Invitrogen). For human ES cell cultures, goat antibodies against human albumin (Bethyl) were used as primary antibodies.

Cell proliferation was evaluated using Click-iT EdU assay kit (invitrogen). The cells cultured with or without NSC23766 were exposed to 10 µM of 5-ethnyl-2′-deoxyuridine (EdU) for 1 hour at 37°C before fixation. The fixed cells were processed for immunocytochemistry as described above, with an additional step for EdU detection. Before incubation with secondary antibodies, the cells were incubated with EdU in the Click-iT reaction cocktail and Alexa Fluor 488 for 30 minutes at room temperature, following the manufacturer's instructions. Images were collected using ImageXpress Micro (Molecular Devices) and EdU-positive nuclei per total number of nuclei were counted.

RNA was extracted from ES cells or mouse liver using an RNeasy mini-kit (Qiagen, Germany) and then treated with DNase (Qiagen). For reverse transcription reactions, 3 µg RNA was reverse-transcribed using ReverTra Ace (Toyobo, Japan) and oligo dT primers (Toyobo). One µl of fivefold-diluted cDNA (1% of the RT product) was used for PCR analyses. The primer sequences for each primer set are shown in supplementary material Table S1. For real-time PCR analysis, mRNA expression was quantified with SyberGreen on an ABI 7500 thermal cycler (Applied Biosystems, Foster City, CA). The PCR conditions were as follows: denaturation at 95°C for 15 seconds, annealing and extention at 60°C for 60 seconds, for up to 40 cycles. Each measurement was normalized to β-actin (mouse) and GAPDH (human) for each sample by subtracting the average β-actin (mouse) and GAPDH (human) C_t values (Threshold Cycle) from the average C_t for each gene. Target mRNA levels, expressed as arbitrary units, were determined using a standard curve method.

Murine ES cells were trypsinized and suspended at a density of 5×10⁴ cells/ml. Cells were then plated onto sNF either with or without 0.1% gelatin pretreatment; control plates were pretreated with 0.1% gelatin. Undifferentiated cells were harvested 48 hours after incubation under ES cell maintenance culture conditions at 37°C, whereas differentiated cells were harvested 9 days after hepatic differentiation started. The activation of Rac was determined using a Rac1 Activation Assay Kit purchased from Millipore. Briefly, cells were washed with PBS and suspended in lysis buffer provided by the supplier. Aliquots were taken from each cell lysate, and the amount of GAPDH proteins present in the lysates was determined and used for normalization. GTP-bound forms of Rac were then pulled down from lysates using reagents provided by the supplier, following the recommended instructions. Proteins present in total cell lysates or Rac pull-down samples were separated by SDS-PAGE (12%) and transferred onto a nylon membrane. Western blotting was performed using antibodies against Rac1, according to the ECL protocol provided by the suppliers. Luminescence of Rac1 bands was quantified using the GE ImageQuant LAS 4000 (GE Healthcare Life Science, Sweden).

We thank members of Gene Technology Center in Kumamoto University for their technical assistance.