Activation of Wnt/β-catenin signalling via GSK3 inhibitors direct differentiation of human adipose stem cells into functional hepatocytes

The generation of hepatocytes that are derived from human adipose stem cells (hASCs) represents an alternative to human hepatocytes for individualized therapeutic and pharmaceutical applications. However, the mechanisms facilitating hepatocyte differentiation from hASCs are not well understood. Here, we show that upon exposure to glycogen synthase kinase 3 (GSK3) inhibitors alone, the expression of definitive endoderm specific genes GATA4, FOXA2, and SOX17 in hASCs significantly increased in a manner with activation of Wnt/β-catenin signalling. Down regulation of the β-catenin expression attenuates the effect of GSK3 inhibitors on the induction of these specific genes. The cells induced using GSK3 inhibitors were directed to differentiate synchronously into hepatocyte-like cells (HLCs) after further combinations of soluble factors by a reproducible three-stage method. Moreover, hASC-HLCs induced using GSK3 inhibitors possess low-density lipoprotein uptake, albumin secretion, and glycogen synthesis ability, express important drug-metabolizing cytochrome P450 (CYP450) enzymes, and demonstrate CYP450 activity. Therefore, our findings suggest that activation of Wnt/β-catenin signalling via GSK3 inhibitors in definitive endoderm specification may represent an important mechanism mediating hASCs differentiated to functional hepatocyte. Furthermore, development of similar compounds may be useful for robust, potentially scalable and cost-effective generation of functional hepatocytes for drug screening and predictive toxicology platforms.

GSK3 inhibitors up-regulate the expression of definitive endoderm specific genes dependent on Wnt/β-catenin signalling. To investigate whether the activation of Wnt signalling by Wnt3a and GSK3 inhibitors affected definitive endoderm specification from hASCs, we compared firstly the effect of Wnt3a and activin A on the regulation of the definitive endoderm specific genes GATA4, FOXA2, and SOX17. The results showed that 24 hours of exposure, the expression levels of these definitive endoderm markers in Wnt3a-treated cells and in activin A-treated cells were significantly increased compared to the vehicle control (Fig. 2a). Furthermore, the mRNA levels of GATA4 and FOXA2 in Wnt3a-treated cells was significantly higher than in activin A-treated cells (Fig. 2a). Then, we examined whether the GSK3 inhibitors could up-regulate the expression of definitive endoderm specific genes in hASCs. We treated hASCs with 0.2 μ M Chir98014 and 2 μ M Chir99021 for 24 hours. Wnt3a was used as a positively control. Consistent with changes in Wnt3a-treated cells, a pulse exposure of Chir98014 and Chir99021 can significantly increase the expression levels of GATA4, FOXA2, and SOX17 compared to the vehicle control (Fig. 2b). Likewise, the mRNA levels of GATA4 in Chir99021-treated cells were significantly higher than in Wnt3a-treated cells (Fig. 2b). Time course analyses indicated that GSK3 inhibitors-based conditions induced a peak expression of these definitive endoderm markers on day 1 (Supplemental Fig. S1). The expression of the definitive endoderm markers in hASCs did not increase with longer exposure to GSK3 inhibitors.
To assess whether the effect of GSK3 inhibitors on induction definitive endoderm specification depends on the β -catenin signalling pathway, we disrupted the signalling pathway by delivering a siRNA to knock down the expression of β-catenin gene. As shown in Supplemental Figure S2, β -catenin expression was successfully reduced by 50% compared to the control siRNA group at the mRNA levels (Supplemental Fig. S2a). Immunofluorescence staining and high content screening (HCS) quantitative analyses showed that the protein level of β -catenin expression in β -catenin siRNA group was significantly lower than the cells in control siRNA group (Supplemental Fig. S2b,c).
By reducing the β -catenin expression level using siRNA, the effects of Wnt3a, Chir98014 and Chir99021 on definitive endoderm specification were assessed. The results showed that the reduction of β -catenin did not impact the expression of definitive endoderm specific genes (Supplemental Fig. S3), but it significantly decreased the mRNA levels of GATA4, FOXA2, and SOX17 in cells following treatment with Wnt3a or GSK3 inhibitors for 24 hours (Fig. 2c). These data showed that GSK3 inhibitors induce definitive endoderm specification of hASCs dependent on the activity of canonical Wnt/β -catenin signalling.

Production of definitive endoderm cells using GSK3 inhibitors.
To determine the production of definitive endoderm cells using GSK3 inhibitors, the cells were differentiated using a three-stage differentiation protocol as previously with modification 10 . During definitive endoderm induction, hASCs were divided into four groups (Fig. 3a). Briefly, at day 2, hASCs were incubated separately with 100 ng/mL activin A for 72 hours, 50 ng/mL Wnt3a, or 0.2 μ M Chir98014, or 2 μ M Chir99021 for 24 hours, 1% insulin-transferrin selenium (ITS) was added to the medium beginning at day 3. At day 5, the properties of differentiated cells in four groups were analysed. The results showed that the mRNA levels of definitive endoderm cells markers, including GATA4, FOXA2, SOX17 and C-X-C motif chemokine receptor 4 (CXCR4) were significantly increased compared to hASCs (Fig. 3b). To verify the differentiation efficiency, the expression of FOXA2 and CXCR4 in cells was examined by flow cytometric analysis. The data showed that the percentage of FOXA2 positive cells in four treated group was 78.8% (activin A-treated cells), 71.5% (Wnt3a-treated cells), 77.1% (Chir98014-treated cells), and 72.2% (Chir99021-treated cells) (Fig. 3c), and the percentage of CXCR4 positive cells in four treated group was 72.1% (activin A-treated cells), 73.1% (Wnt3a-treated cells), 65.4% (Chir98014-treated cells), and 62.3% (Chir99021-treated cells) separately (Fig. 3d). Immunofluorescence staining showed that the cells in four groups exhibited similar nucleus patterns of expression of GATA4 (Fig. 3e). HCS quantitative analyses for the expression of GATA4 demonstrated that the level of GATA4 proteins in response to Chir98014 and Chir99021 treatment with a comparable efficiency as that of activin A treatment, which were significantly higher than the hASCs (Fig. 3f). These findings demonstrate that Chir98014 and Chir99021 treatment may switch hASCs developmental fate and committed to the definitive endoderm lineage.
Definitive endoderm cells induced by GSK3 inhibitors possess the potential to differentiate into hepatocyte-like cells. The properties of differentiated cells in four groups were analysed at day 10 and day 20. Immunofluorescence staining data verified that the cells in the four groups were positive for hepatocyte nuclear factor 4 alpha (HNF4α ) (Fig. 4a) and α -fetoprotein (AFP) (Fig. 4c), which are markers of hepatic progenitor cells. To verify the differentiation efficiency, the expression of HNF4α and AFP in cells at day 10 was examined by flow cytometric analysis. The data showed that the percentage of HNF4α positive cells in four treated group was 86.9% (activin A-treated cells), 91.8% (Wnt3a-treated cells), 92.5% (Chir98014-treated cells), and 96% (Chir99021-treated cells) (Fig. 4b), and the percentage of AFP positive cells in four treated group was 99.3% (activin A-treated cells), 98.9% (Wnt3a-treated cells), 98.8% (Chir98014-treated cells), and 99.2% (Chir99021-treated cells) separately (Fig. 4d). These results indicated that almost all of the cells induced by GSK3 inhibitors synchronously acquired the properties of hepatic progenitor cells following 5 days of hepatic induction.
Following a 10-day maturation process, the morphology of the cells became the polygonal epithelial cells-like. Quantitative comparisons of gene expression revealed that the mRNA levels of hepatic functional markers, including apolipoprotein C1 (APOC1), phosphoenolpyruvate carboxykinase 2 (PCK2), and carbamoyl-phosphate synthase 1 (CPS1), in cells at day 20 were highly increased compared with the expression in hASCs. The expression levels of APOC1 in Chir98014-treated cells and Chir99021-treated cells at day 20 were relative higher than the levels in activin A-treated cells. The expression level of CPS1 in Chir98014-treated cells at day 20 was also higher than the levels in activin A-treated cells. Nevertheless, the expression levels of these genes in cells at day 20 were relative lower than the levels in human hepatocyte (Supplemental Fig. S4). Immunofluorescence staining analysis showed that at day 20 all of the cells in four groups expressed albumin (ALB) and glutathione S-transferase alpha 2 (GSTA2), which were similar with human hepatocytes (Fig. 5a). Meanwhile, the expression of AFP, marker of hepatic progenitor cells, was dramatically decreased in cells at day 20 with different factors (Supplemental Fig. S5).
Low-density lipoprotein (LDL) uptake examination showed that the internalized LDL-DyLight TM 550 particles presented in the cells of four groups. The percentage of cells with LDL particles in cells of Chir99021 at day 20 was relatively higher than other groups, which resembles the human hepatocytes (Fig. 5b). Immunofluorescence staining analysis showed that at day 20 all of the cells in four groups expressed LDL receptor (LDLR) as well as human hepatocyte (Fig. 5b). ALB secretion determination showed that the cells derived from Chir98014 and Chir99021 groups secreted ALB at a level of 11.2 ± 0.02 ng/mL and 14.8 ± 0.02 ng/mL per 10 5 cells, which was greater than the level of in cells derived from activin A or Wnt3a (Fig. 5c). The level of secreted ALB in the cells derived from Chir98014 and Chir99021 was still lower than the level of in human hepatocytes (Fig. 5c). Glycogen synthesization determination showed that the level of glycogen in cells derived from four groups significantly higher than the levels of in hASCs (Fig. 5d). The level of glycogen in the cells from different groups was also lower that the level of in human hepatocytes (Fig. 5d).
These data indicate that hASCs induced by Wnt3a, Chir98014 and Chir99021 at stage of endoderm induction can be differentiated into hepatocyte-like cells (HLCs) that exhibit mature hepatocyte-specific protein and function following the favour hepatic differentiation and maturation induction. Therefore, we named HLCs from four groups as hASC-HLCs-activin A, hASC-HLCs-Wnt3a, hASC-HLCs-Chir98014, and hASC-HLCs-Chir99021. hASC-HLCs express cytochrome P450 (CYP450) enzymes and have CYP450 activity. To further evaluate the expression and activity of CYP450 in hASC-HLCs induced using GSK3 inhibitors, quantitative comparisons of the mRNA levels of the CYP450 subfamily was examined firstly (Supplemental Fig. S6). The expression of CYP1A2 in hASC-HLCs-Wnt3a, hASC-HLCs-Chir98014, and hASC-HLCs-Chir99021 was significantly higher than the expression in hASC-HLCs-Activin A. The expression of CYP7A1 in hASC-HLCs-Wnt3a, and hASC-HLCs-Chir99021 was significantly higher than the expression in hASC-HLCs-Activin A. The expression of CYP1A2, CYP2B6, CYP3A4 and CYP7A1 in hASC-HLCs-Chir99021 was significantly higher than the expression in hASC-HLCs-activin A. The expression of CYP2B6, CYP3A4 and CYP7A1 in hASC-HLCs-Chir99021 was significantly higher than the expression in human hepatocytes. The expression of CYP2E1 in hASC-HLCs from four different groups was significantly higher than the expression in human hepatocyte.
Immunofluorescence analysis verified that hASC-HLCs from the four groups were positive for CYP2A6, CYP2B6, CYP2C9/19, and CYP3A4, which was similar to hepatocytes from human liver tissue (Fig. 6a).

Discussion
In the present study, we elucidated that pharmacological inhibition of GSK3 with the specific inhibitors Chir98014 or Chir99021 up-regulated the expression of definitive endoderm specific genes via activation of the Wnt/β -catenin signaling axis. hASCs from four different donors induced by GSK3 inhibitors possess the potential to differentiate into hepatocyte-like cells with mature function, including LDL uptake, ALB secretion and glycogen synthesis. Moreover, the hASC-HLCs-Chir98014 and hASC-HLCs-Chir99021 expressed the important drug-metabolizing CYP450 enzymes including CYP2A6, CYP2B6, CYP2C9/19, and CYP3A4 and the hASC-HLCs-Chir99021 had relatively higher CYP2C9 and CYP3A4 activity than other groups of hASC-HLCs. Our results demonstrate development of similar compounds may be useful for robust, potentially scalable and cost-effective generation of large numbers of functional hepatocytes for drug screening and predictive toxicology platforms. The induction and differentiation of definitive endoderm is an essential early step to generate of hepatocytes from human pluripotent stem cells (hPSCs) 25 . Higher efficient and reproducible definitive endoderm induction will contribute to the development of homogeneous and higher fidelity end-stage populations, such as hepatocytes. Evidence indicates that the Nodal and the Wnt signalling pathway plays a pivotal role in the establishment of the early primitive streak, which ultimately leads to endoderm specification, has been identified 26 . Various groups target these two pathways in order to generate definitive endoderm from hPSCs, but the signalling mechanisms facilitating definitive endoderm induction from hASCs are not well understood. Since it is simple, robust and lower cost than growth factors, the use of small molecules that modulate the generation of specific cell types from stem cells opens new pathways to optimize protocols in stem cell differentiation 27 . Rational design or screening small molecules to generate a specific cell linage differentiated from stem cells depends on our basic understanding of human development. Studies indicates that the Wnt signalling pathway plays a pivotal role in the initiation or generation of definitive endoderm from hPSCs 28 . Thus, directly inhibiting the signalling molecule GSK3, which mimics activation of the Wnt pathway, was used to induce the differentiation of hPSCs to definitive endoderm and then to the hepatic lineage [11][12][13] .
On the other hand, Wnt signalling has been identified to control the fate of mesenchymal stem cells (MSCs) 29 . However, little is known about the role of Wnt and its downstream intracellular pathways in the differentiation of definitive endoderm and hepatic progenitor cells from hASCs. Herein, we found GSK3 inhibitors increase the expression and nuclear distribution of β -catenin. GSK3 phosphorylated β -catenin, triggering its destabilization and degradation to maintain a very low level of β -catenin in the cytosol/nucleus. Accumulated nuclear levels of β -catenin in GSK3 inhibitors treated cells indicated the GSK3 activity was inhibited. Meanwhile, the levels of members of TCF/LEF transcription factors TCF7 and LEF1 in hASCs were increased upon GSK3 inhibitors stimulation. This implied that Wnt/β -catenin signalling pathway was activated in hASCs upon GSK3 inhibitors stimulation.
The data presented here demonstrate that the GSK3 inhibitors promoted the efficient up-regulation of definitive endoderm specific transcription factors GATA4, FOXA2, SOX17 and CXCR4. Down regulating of the expression β -catenin by siRNA caused the decreased expression of these definitive endoderm specific transcription factors in hASCs after treatment with Wnt3a, Chir98014, or Chir99021. The transcription factor FOXA2 is the earliest known marker of definitive endoderm. It's up-regulation at the onset of gastrulation is considered to represent a crucial step in commitment to definitive endoderm identity 30 . GATA4 is a zinc finger transcription factor that has important functions in several mesodermal and endoderm organs, including heart, liver and pancreas 31 . Study proposed that GATA4 is an important downstream effector of GSK3 and inhibition of GSK3 at least in part by removing its negative constraint upon GATA4 32 . In this study, Chir98014 and Chir99021 significantly increased the mRNA and protein levels of GATA4 and FOXA2 in hASCs after 24 hours exposure. At day 5, more than 75% of cells expressed the FOXA2 and CXCR4, the markers of definitive endoderm cells. The result suggested that the activity of Wnt/β -catenin signalling pathway in hASCs play a key role in definitive endoderm specification. Recent study using mouse ES cells showed that β -catenin and cMyc operate downstream of GSK3 inhibition to reduce Tcf7l1 transcription and activity, respectively. Tcf7l1 is a barrier to definitive endoderm specification, acting to directly block induction of the pioneer transcription factor, FOXA2. This explains the possible mechanism of GSK3 inhibition in promoting endoderm lineage specification 23 .
Suboptimal hepatic induction is one likely cause of the variability that can influence hepatic maturation and contribute to the development of heterogeneous end-stage populations. The efficiency of hepatic induction was monitored through changes in protein expression patterns as assessed by immunofluorescence or flow cytometry. The latter approach provides a rapid quantitative read-out but is dependent on the availability of antibodies against ideal markers of on interest on the cells. We show here that we equally successfully acquired the homogenous hepatic progenitor cells when using Chir98014 or Chir99021 induction as with activin A and Wnt3a, and then treatment with BMP2 and FGF4 for 5 days as shown previously 10 . Nearly all cells in the four groups highly expressed the hepatic progenitor markers HNF4α and AFP. At the end of the final phase, the cells induced by Chir98014 or Chir99021 treatment expressed a number of hepatic markers at the transcriptional, and protein levels. Functional determination showed that hASC-HLCs-Chir98014 and hASC-HLCs-Chir99021 exhibited mature functions of hepatocyte, including LDL uptake, ALB protein production, and glycogen synthesis.
Detoxification of exogenous compounds is one of the liver's major metabolic functions and is mediated by a large family of proteins, including CYP450. Thus, the expression and activity of CYP450 are key markers for the HLC maturity. A recently study shows the phenotypic and functional properties of hESC-derived HLCs better mimic fetal hepatocytes with weak expression of CYP2A6, a hallmark of adult hepatocytes 16 . Our data confirmed that CYP2A6, CYP2B6, CYP2C9/19 and CYP3A4 were expressed as expected in hASC-HLCs. The abundant CYP2A6 and CYP3A4 in hASC-HLCs-Chir98014 or hASC-HLCs-Chir99021 were similar with the adult hepatocytes in a human liver tissue section. Similarly, the levels of CYP2C9 and CYP3A4 enzyme activity in hASC-HLCs-Chir99021 were relatively higher than that in hASC-HLCs-activin A. These data imply that hASC-HLCs in this study more closely resemble human hepatocytes for pre-clinical assessment of new candidate compounds in drug development 33 . Of course many efforts are still needed for promoting hASC-HLCs further maturation under the hepatic microenvironment, until they have equal metabolic abilities compared with human adult hepatocytes.
In conclusion, we demonstrate here a single chemical entity to direct differentiation of hASCs to definitive endoderm cells and towards functional hepatocytes via activation of the Wnt/β -catenin pathway at early stage. The generation of hASC-HLCs using the chemically directed differentiation may provide a foundation for scalable hepatocyte production for drug screening, predictive toxicology platforms and cell therapy.

Methods
Human subjects. Human tissues were obtained with informed patient consent and under the approval of the Ethics Committee of Capital Medical University (Beijing, China) for this project. All experimental protocols involving human tissues were reviewed and approved by the Ethics Committee of Capital Medical University. All methods involving human cells and tissue were carried out in accordance with relevant guidelines and regulations of the Ethics Committee of Capital Medical University.
Cell culture and human tissue. hASCs from four different donors were developed as our previously reported 10 . hASCs were maintained in DMEM/F-12 (Invitrogen, Grand Island, NY, USA) supplemented with 10% fetal bovine serum mesenchyme stem cell screened (FBS-MSCS, HyClone, Logan, UT, USA), 100 U/mL penicillin, and 100 μ g/mL streptomycin. At 80% confluence, the cells were passaged using 0.05% trypsin-0.02% EDTA (Sigma-Aldrich, St. Louis, MO, USA) and plated at a density of 5000 cells/cm 2 . Cells from passages six to nine were used in this study.
Human hepatocytes (ScienCell Research Laboratories, Carlsbad, CA, USA) were cultured on collagen type I-coated 24-well plates at a density of 150,000 cells/cm 2 with hepatocyte medium (ScienCell Research Laboratories).
Human liver tissue sections were obtained from the department of pathology of beijing chaoyang hospital, and patients were informed with consent under ethical approval.
For subsequent hepatic differentiation, the medium was changed to MEM/NEAA (Invitrogen), supplemented with 0.5 mg/mL BSA, 1% ITS, 20 ng/mL BMP2 (Peprotech) and 30 ng/mL FGF4 (Peprotech) for 5 days. To allow for hepatocytes maturation, the cells were further treated with 20 ng/mL Hepatocyte Growth Factor (HGF) for 5 days, and 20 ng/mL HGF, 10 ng/mL oncostatin M (OSM) (Peprotech) plus 10 −6 M Dexamethasone (DEX, Sigma-Aldrich) treatment for another 5 days. The differentiation media were changed every 2 days. siRNA transfection. hASCs were plated at 20,000 cells/cm 2 in antibiotic-free basal medium 24 hours prior to transfection. siRNA transfection was performed following the manufacturer's protocol as previously described 34 . Briefly, ON-TARGET SMARTpool siRNAs directed against β-catenin (L-003482-00-0005, Dharmacon, Lafayette, USA) or Non-targeting siRNAs (D-001810-10-05, Dharmacon) were mixed with Transfection DharmaFECT 1 (Dharmacon), respectively. After a 20 minutes incubation at room temperature, the complexes were added to the cells at a final siRNA concentration of 25 nM. The medium was replenished with medium containing antibiotic 24 hours post-transfection. The culture medium was changed every 2 days for the duration of the experiment.

Real-time RT-PCR.
Real-time RT-PCR was performed as previously described 10,34 . Total cellular RNA was extracted from 2.5 × 10 5 cells with the RNeasy Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. For PCR analysis, 1 μ g of RNA was reverse-transcribed to cDNA using Superscript III reverse transcriptase and random hexamer primers (Invitrogen). Real-time PCR analysis was performed on an ABI Prism 7300 Sequence Detection System using the SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA). The reaction consisted of 10 μ L of SYBR Green PCR Master Mix, 1 μ L of a 5 μ M mix of forward and reverse primers, 8 μ L water, and 1 μ L template cDNA in a total volume of 20 μ L. Cycling was performed using the default conditions of ABI 7300 SDS Software 1.3.1. The relative expression of each gene was normalized against 18 S rRNA. The data are presented as the mean ± s.d. The primers used are listed in Supplemental Table S1.
LDL uptake cell-based assay. For LDL uptake cell-based assay, the cells were treated with LDL-DyLight TM 550 (Cayman chemical, Ann Arbor, Michigan, USA) working solution in serum-free medium. Incubate the cells at 37 °C with 5% CO 2 for an additional 18 hours. At the end of LDL uptake incubation, the culture medium was replaced with fresh culture medium. The degree of LDL uptake was examined under a Leica TCS SP8 confocal microscope (Leica, Wetzlar, Germany) with filters capable of measuring excitation and emission wavelengths 540 and 570 nm, respectively.
For LDL receptors examination, the cells were washed with TBS briefly, and fixed with cell-based assay fixative solution (Cayman chemical) for 10 minutes. Wash the cells with TBST three times for five minutes each. Incubate the cells for one hour with rabbit anti-LDL receptor antibody (Cayman chemical). Wash the cells with TBST three times for five minutes each. Incubate the cells for one hour with Dylight TM 488-conjugated secondry antibody (Cayman chemical). Wash the cells with TBST three times for five minutes each. The nuclei were counter-stained with DAPI (Sigma-Aldrich). The stained cells were examined under a Leica TCS SP8 confocal microscope (Leica, Wetzlar, Germany).