Robust protein-based engineering of hepatocyte-like cells from human mesenchymal stem cells

Background:

Cells of interest can be prepared from somatic cells by forced expression of lineage-specific transcription factors, but it is required to establish a vector-free system for their clinical use. Here, we report a protein-based artificial transcription system for engineering hepatocyte-like cells from human umbilical cord-derived mesenchymal stem cells (MSCs).

Methods:

MSCs were treated for 5 days with 4 artificial transcription factors (4F), which targeted hepatocyte nuclear factor (HNF)1α, HNF3γ, HNF4α, and GATA-binding protein 4 (GATA4). Then, engineered MSCs (4F-Heps) were subjected to epigenetic analysis, biochemical analysis and flow cytometry analysis with antibodies to marker proteins of mature hepatocytes and hepatic progenitors such as delta-like homolog 1 (DLK1) and trophoblast cell surface antigen 2 (TROP2). Functional properties of the cells were also examined by injecting them to mice with lethal hepatic failure.

Results:

Epigenetic analysis revealed that a 5-day treatment of 4F upregulated the expression of genes involved in hepatic differentiation, and repressed genes related to pluripotency of MSCs. Flow cytometry analysis detected that 4F-Heps were composed of small numbers of mature hepatocytes (at most 1%), bile duct cells (~19%) and hepatic progenitors (~50%). Interestingly, ~20% of 4F-Heps were positive for cytochrome P450 3A4, 80% of which were DLK1-positive. Injection of 4F-Heps significantly increased survival of mice with lethal hepatic failure, and transplanted 4F-Heps expanded to more than 50-fold of human albumin-positive cells in the mouse livers, well consistent with the observation that 4F-Heps contained DLK1-positive and/or TROP2-positive cells.

Conclusion:

Taken together with observations that 4F-Heps were not tumorigenic in immunocompromised mice for at least 2 years, we propose that this artificial transcription system is a versatile tool for cell therapy for hepatic failures.

INTRODUCTION

Liver transplantation is applied as standard treatment for patients with decompensated cirrhosis, but because of a shortage of liver donors and high medical costs, alternative therapeutic options are required.1 As a promising approach, clinical trials of cell therapy have been conducted, and lines of evidence suggest that the transplantation of mesenchymal stem cells (MSCs) provides benefits to patients.2,3 However, the efficacy of MSC transplantation has not been demonstrated.2 As a possible limitation of MSCs, it was suggested that the hepatic differentiation of MSCs is insufficient,4 implying that it is important to establish an efficient differentiation system. Owing to recent progress, hepatocyte-like cells can be engineered from fibroblasts by direct reprogramming, in which cells of interest are converted directly from somatic cells,5,6 suggesting that the efficacy of MSC-based therapy can be improved by adopting cell engineering methods for direct reprogramming.

For hepatocyte differentiation, several transcription factors that include hepatocyte nuclear factor (HNF) 1α, HNF3α/β/γ, HNF4α, HNF6/Onecut-1 (OC-1), OC-2, CCAAT/enhancer-binding protein alpha, hematopoietically expressed homeobox, and prospero homeobox 1 have been identified.7–14 By forced expression of some of these genes, for example, HNF1α, FOXA factors (HNF3α/β/γ), HNF4α, and GATA-binding protein 4 (GATA4),15 direct reprogramming for hepatocyte-like cells has been tried. Among these molecules, FOXA factors and GATA4/6 have been nominated as pioneer factors,16–18 the expression of which is renewed during embryonic endoderm formation and is pivotal for the subsequent expression of hepatocyte-specific molecules. Pioneer factors open the local nucleosomal domain, and facilitate the access of other factors to their target genes.19 These observations suggest that combining pioneer factors with hepatocyte lineage–specific genes would improve the reprogramming efficiency of hepatocyte-like cells.

Because number of reports on direct reprogramming were based on expression vectors, and the establishment of human hepatocyte-like cells takes 14–28 days,20–22 it is necessary to establish a novel cell engineering system that enables us to prepare clinically applicable hepatocyte-like cells safely and quickly. As a vector-free system for cell engineering, we have established a protein-based artificial transcription factor (ATF) system and generated mouse-induced pluripotent stem cells by using an ATF targeting the microRNA-302/367 cluster gene.23 An ATF requires 3 functional components: a nuclear trafficking peptide (NTP), a transcription activator-like effector as a DNA-binding module, and a transcriptional module. NTP is a cell-penetrating peptide composed of 10 amino acids (RIFIHFRIGC) that can transport cargo molecules into the nucleus when added to the culture medium of cells. Transcription activator-like effector typically contains 17 repetitive units, each of which comprises 34 amino acids with polymorphic 12th and 13th amino acids and recognizes a single nucleotide.24,25 As a transcriptional module, VP64, a tetramer of VP16 from herpes simplex virus,26,27 and VP64-p65-Rta, and histone demethylase and histone acetyl-transferase were utilized for better combination with transcription activator-like effector.28–31

In the present study, we applied the ATF system to umbilical cord-derived MSC (MSC-UCs) to engineer hepatocyte-like cells and demonstrated that combined treatment with 4 ATFs targeting HNF1α, HNF3γ, HNF4α, and GATA4 reproducibly committed MSC-UCs to hepatocyte-like cells (4F-Heps). Notably, just 5-day treatment with the ATFs was sufficient to generate 4F-Heps from MSC-UCs, and injection of mice with the obtained cells could rescue lethal hepatic failure. Taken together with data showing that the epigenetic profiles of MSC-UCs were changed in 5 days, we propose that this ATF system is a versatile tool for cell engineering, especially for the generation of MSCs for patients with cirrhosis.

MATERIALS AND METHODS

Cells

Human MSC-UCs, human bone marrow-derived MSCs, and HepG2 cells, a human HCC cell line, were provided by the JCRB Cell Bank (Ibaraki, Japan). MSCs and HepG2 cells were maintained in Cellartis MSC Xeno-Free Basal Medium (Takara Bio), or in Dulbecco’s modified Eagle medium (Gibco), respectively, under humidified conditions with 5% CO₂. Induced pluripotent stem cells derived hepatocytes (iPSC-Heps) were purchased from Takara Bio and maintained in the medium provided in the kit.

Expressing ATF proteins and purification

Identification of the optimal combination of ATFs targeting HNF1α, HNF3γ, HNF4α, and GATA4 and the purification methods were described in Supplemental Materials and Methods (http://links.lww.com/HC9/A147). Target sequences of synthesized TALEs were described in Supplementary Table S1 (http://links.lww.com/HC9/A147). As control, NTP-tagged enhanced green fluorescent protein was expressed as a fusion protein of glutathione S transferase, NTP, and enhanced green fluorescent protein by pEU-01 vector (CellFree Sciences, Ehime, Japan), and prepared by the same procedure as the ATF proteins.

Preparation of 4F-Heps

Onto 100 mm dish (NIPPON Genetics), 0.3×10⁶ MSC-UCs were plated and the culture medium was replaced with the same medium supplemented with 4 ATFs on the next day. The final concentration of each protein was 0.25 nM, and the same procedure was repeated twice a day for 5 days. On day 7, treated cells were harvested by using Accutase (Millipore), and subjected to analysis. No supplements for hepatocytic differentiation such as insulin or transferrin were used.

qRT-PCR and western blot analysis

Expression of RNA and protein was analyzed by qRT-PCR and western blot analysis, as described in Supplemental Materials and Methods (http://links.lww.com/HC9/A147). Primers used for qRT-PCR were summarized in Supplementary Table S2 (http://links.lww.com/HC9/A147).

Flow cytometry analysis and immunohistochemical analysis

For flow cytometry analysis, anti-albumin (ALB) (1:500; Takara Bio), cytokeratin 19 (CK19, 1 μg/mL, Cambridge, UK) (1:500; Abcam), cytochrome P450 (CYP) 3A4 (1:500; Abcam), delta-like homolog 1 (DLK1) (Abcam) and trophoblast cell surface antigen 2 (TROP2) (Abcam) were used as primary antibodies. After reacting with alexa 488 (Abcam) or PE-Cy5 (Santa cruz Biotechnology)-labeled antibodies, positive cells were detected by FACSVerse (Beckton Dickinson). For immunohistochemical analysis, anti-HNF1α, anti-HNF3γ, anti-HNF4α, anti-GATA4. anti-OTC (ornithine transcarbamylase), or anti-CPS1 (carbamoyl phosphate synthetase-1) antibodies (all at 1:1,000; Abcam, Cambridge, UK) were used as primary antibodies. Precise procedures were described in Supplemental Materials and Methods (http://links.lww.com/HC9/A147).

Functional properties of 4F-Heps

We performed Periodic acid-Schiff Staining, and analyzed LDL uptake, urea synthesis and activity of cytochrome P450. To show the activity of human CYP2D6, debrisoquine (DEB) was injected into mouse, and 4-hydroxydebrisoquine DEB (4-OH DEB), a product of DEB metabolized by CYP2D6, was measured by liquid chromatography-tandem mass spectrometry, according to a reported method with some modifications.32 Human ALB in the culture supernatant of 4F-Heps and serum samples of mice that were injected with 4F-Heps were analyzed by enzyme-linked immunosorbent assay (ELISA). The precise procedure of each assay was described in Supplemental Materials and Methods (http://links.lww.com/HC9/A147).

Microarray and DNA methylation analyses

Expression array analysis was performed by Takara Bio. The received data were analyzed using GeneSpring (Agilent) and R (Rstudio). DNA methylation analysis was performed by Takara Bio. The received data were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, and candidate signaling pathways were extracted by functional enrichment clustering analysis. Data were deposited in NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE205543 and GSE 205546.

Functional analysis of 4F-Heps with ALB-TRECK/SCID mice

All animal experiments were approved by the President of National Center for Global Health and Medicine (NCGM), following consideration by the Institutional Animal Care and Use Committee of NCGM (approval ID: no. 22037), and were carried out in accordance with institutional procedures, national guidelines and the relevant national laws of Japan on the protection of animals. Male ALB-TRECK/SCID/SCID mice (TRECK-ALB mice), which are immunodeficient mice in which the human diphtheria toxin (DT) receptor under the control of the ALB gene promoter has been introduced,33 were acquired from the Tokyo Metropolitan Institute of Medical Science (Tokyo, Japan) and maintained in the animal facility at the National Center for Global Health and Medicine. All mice used in this study were housed in an air-conditioned animal room at 23±2 °C and relative humidity of 40%–60% under specific-pathogen-free conditions, with a 12-hour light/dark cycle (8:00–20:00/20:00–8:00). All mice were fed a standard rodent CE-2 diet (CLEA Japan) and had ad libitum access to water.

To evaluate the function of 4F-Heps in vivo, 10-week-old TRECK-ALB female mice were injected with DT (1.9 ng/g body weight) (Fujifilm) on day 1, and serum alanine aminotransferase (ALT) was measured on day 3 by using SPOTCHEM (Arkray). Mice with ALT measurements of 8000–23,000 IU/mL were divided into 2 groups, so that equal numbers of mice with comparable levels of liver damage were distributed (see actual data in Supplemental Figure S1, http://links.lww.com/HC9/A138), and 3.0×10⁵ 4F-Heps were transplanted into one group of mice through the tail vein on day 4. To the other group of mice, the same number of nontreated MSC-UCs (NC) was injected. On day 11, anti-Fas antibody administration at 0.35 μg/g body weight (BD Bioscience), a sublethal dose,34 was started. The mice were injected with the antibody 3 times every other day, and their survival was carefully monitored. To investigate cellular dynamics of injected 4F-Heps, mouse livers were prepared at 16 weeks or 3 week after injections and subjected to immunohistochemical analysis with anti-human Ki-67, human ALB, and human CK19 antibodies.

Statistical analysis

Principal component analysis was performed using gene expression data, and volcano plots were generated with GeneSpring software (Agilent). A heatmap of the differentially expressed genes (DEGs) was drawn using R program, gplots package. A quadrant plot was drawn using GraphPad Prism v8.0 (GraphPad Software). All results are described as the mean±SD. Statistical comparisons were performed using Student t test for paired observations. The significance test in the Kaplan–Meier method was performed by the long-rank test. A p-value <0.05 was considered statistically significant.

RESULTS

Combined treatment with 4 ATFs induces definite mRNA expression of endogenous genes related to hepatic differentiation

According to reports on the direct reprogramming of hepatic cells from fibroblasts,20,21,35 we carried out preliminary experiments and identified the best combination of ATFs, which targeted HNF1α, HNF3γ, and HNF4α (3F) plus GATA4 (4F) (Supplemental Materials and Methods, http://links.lww.com/HC9/A147, and Supplemental Figure S2, http://links.lww.com/HC9/A139, S3, http://links.lww.com/HC9/A140). Treatment of MSC-UCs with combined 4F for 5 days (Figure 1A) induced definite mRNA expression of endogenous target genes (Figure 1B), along with alpha fetoprotein and ALB (Figure 1B). Immunohistochemical analysis revealed that cells positive for HNF1α, HNF3γ, HNF4α, and GATA4 proteins were detected at frequencies of 56%, 66%, 48%, and 54%, respectively (Figure 1C, D). Positive effects of 4F were also confirmed by flow cytometry analysis detecting that 58%, 91%, 82%, and 78% of analyzed cells were positive for HNF1α, HNF3γ, HNF4α, and GATA4, respectively (red) (Figure 1E). In iPSC-Heps, mature hepatocytes that were differentiated from human iPS cells, 75%, 58%, 71% and 25% of cells were positive for HNF1α, HNF3γ, HNF4α, and GATA4, respectively (green). In contrast, treatment of 3F for 5 days poorly induced mRNA expression of these target genes (Figure 1B): the frequencies of cells positive for HNF1α, HNF3γ, HNF4α, and GATA4 were 25%, 43%, 25%, and 20% detected by immunohistochemical analysis (Figure 1C, D) and 32%, 52%, 39%, and 12% detected by flow cytometry analysis (blue, Figure 1E). Notably, 4F treatment markedly induced the expression of endogenous HNF3α (FOXA1) and GATA6 (Figure 1F), both of which have been proposed as pioneer factors for the lineage-specific expression of genes related to liver differentiation. These data encouraged us to further characterize engineered hepatocyte-like cells (4F-Heps).

FIGURE 1.

Endogenous mRNA expression of hepatocyte lineage-specific genes induced by 4F treatment. (A) Experimental protocol of treatment of umbilical cord-derived mesenchymal stem cells (MSC-UCs) with ATFs. MSC-UCs were treated with ATFs (final concentration, 0.25 nM) twice a day for 5 days and subjected to analysis on day 7. All experiments were performed using this protocol. (B) Endogenous mRNA expression of genes related to hepatocyte differentiation. MSC-UCs that were treated with 3F (ATFs-HNF1α/3γ/4α) or 4F (3F and ATF-GATA4) were subjected to qRT-PCR analysis. Fold increase in the expression of each gene was compared with that of human liver-derived mRNA as a positive control (PC). NC, nontreated MSC-UCs used as negative control. Mean±SD, n=3. *p<0.05. (C) Protein expression of hepatocyte markers detected by immunohistochemical analysis. The cells were counterstained with Hoechst 33258. Merged signals are shown. PC, induced pluripotent stem cells derived hepatocytes positive control. Bar, 100 μm. (D) Integrated data of (C) are shown. The number of positive cells was counted using BZ-X Analyzer and data were subjected to statistical analysis. Mean±SD, n=3. *p<0.05. (E) Flow cytometry profiles of 4F-Heps. MSc-UCs, which were treated for 5 days with 3F or 4F, were analyzed on day 7. The cells were reacted with anti-HNF1α, anti-HNF3γ, anti-HNF4α, and anti-GAT;A4 antibodies followed by the fluorochrome-labeled secondary anti-antibody. Horizontal axis, fluorescence intensity; vertical axis, frequency. Black, NC; green, induced pluripotent stem cells derived hepatocytes; blue, 3F; red, 4F. (F) Endogenous HNF3β and GATA6 mRNA expression. The 4F-treated cells were analyzed by qRT-PCR. Mean±SD, n=3. *p<0.05. Abbreviations: AFP, alpha fetoprotein; ALB, albumin.

4F-Heps have functional properties of hepatocyte-like cells

Approximately 65% of 4F-Heps were positive for Periodic acid-Schiff staining with multinucleated cells (Figure 2A, right panels, 2B), and LDL uptake (Figure 2C, D). Consistent with data suggesting that 4F-Heps were positive for urea synthesis (Figure 2E), immunohistochemical and western blot analyses detected expressions of ornithine transcarbamylase and carbamoyl phosphate synthetase-1, key enzymes in the urea cycle (Figure 2F–H). Finally, we detected human albumin in the culture supernatant of 4F-Heps (Figure 2I), although the concentration was quite low. These observations suggested that 4F-Heps have the properties of functional hepatocytes.

FIGURE 2.

Biological functions of 4F-Heps. (A) Periodic acid-Schiff staining. Arrowhead indicates multinucleated cells. High-magnification image of multinucleated cell indicated by arrowhead was shown in the right panel. (B) Integrated data of (A). The number of Periodic acid-Schiff-positive cells was counted using BZ-X Analyzer. Mean±SD, n=3. *p<0.05. (C) LDL uptake. After culturing overnight in the presence of LDL-DyLight 550, 4F-Heps were fixed in 2% paraformaldehyde, counterstained with Hoechst 33258, and examined under a microscope. Merged signals are shown. Bar, 100 µm. PC, HepG2 cells. (D) Integrated data of cells positive for LDL uptake. The number of LDL-DyLight 550-positive cells was counted using BZ-X Analyzer. Mean±SD, n=3. *p<0.05. (E) Urea synthesis by 4F-Heps. Urea synthesized by 4F-Heps was measured in the culture supernatant. PC, culture supernatant of HepG2 cells; NC, nontreated MSc-UCs; 4F, 4F-Heps. Mean±SD, n=3. *p<0.05. (F) Expression of hepatocyte markers. Immunohistochemical analysis using anti-OTC and anti-CPS1 antibodies was performed on NC and 4F-Heps. The cells were counterstained with Hoechst 33258. Merged signals are shown. Red, OTC or CPS1; blue, Hoechst 33258. Bar, 100 μm. (G) Integrated data of (F). The number of positive cells was counted using BZ-X Analyzer. Mean±SD, n=3. *p<0.05. (H) Expression of OTC and CPS1 proteins. Cell lysates of NC and 4F-Heps were examined by western blot analysis using anti-OTC and anti-CPS1 antibodies. PC, lysate of HepG2 cells. (I) Measurement of human ALB protein in culture supernatant. PC, culture supernatant of induced pluripotent stem cells derived hepatocytes; NC, nontreated MSc-UCs. Mean±SD, n=3. *p<0.05. Abbreviations: 4F, 4F-Heps; CPS1, carbamoyl phosphate synthetase-1; OTC, ornithine transcarbamylase.

4F-Heps are mainly composed of immature hepatocytes

To further characterize 4F-Heps, we carried out flow cytometry analysis to characterize cellular components of 4F-Heps with antibodies to ALB, CK19, and DLK1, a marker of hepatic progenitor cells.36 As positive control of mature hepatocytes, we included iPSC-Heps to the analysis. As shown in Figure 3A, most of iPSC-Heps were ALB-positive (ALB+), and negative of both CK19 and DLK1 (CK19−/DLK1−), indicating that iPSC-Heps were composed of mature hepatocytes. In contrast, 4% and 12.8% of 4F-Heps were ALB+/DLK+ and ALB−/DLK+ cells, respectively, and ALB+/DLK− cells were negligible (0.2%) (Figure 3A, top right column). In addition, ~19%, 10%, and 2% of 4F-Heps were CK19+/DLK−, CK19+/DLK+, and CK19−/DLK+ cells, respectively (middle right panel). Interestingly, double immunostaining with antibodies to CYP3A4 and DLK1 detected that 4.1%, 15.8% and 0.3% of cells were CYP3A4+/DLK1−, CYP3A4+/DLK1+, and CYP3A4−/DLK1+, respectively. Data imply that 4F-Heps contained DLK1+ cells that have CYP activity (Figure 3A, bottom right panel). We further analyzed with an antibody to TROP2, a marker of oval cells that emerge after liver injury and can differentiate to both bile duct cells and mature hepatocytes,36 and observed that 1.3%, 12.1%, and 39.2% of 4F-Heps were ALB+/TROP2−, ALB+/TROP2+, ALB−/TROP2+ (Figure 3B, right panel). Moreover, 1.1%, 20.8%, and 21.8% were TROP2+/DLK1−, TROP2+/DLK1+, and TROP2−/DLK1+ cells (Figure 3C, right penal). These data suggest that 4F-Heps are composed of 50% hepatic progenitors and ~19% of bile duct cells with negligible numbers of mature hepatocytes, and that the reprogramming efficiency of the 4F-ATF system is ~70%. (Table 1).

FIGURE 3.

Flow cytometry analysis of 4F-Heps. (A) 4F-Heps were treated with antibodies to markers of mature liver cells and hepatic progenitors. Top panels, anti-ALB and anti-DLK1 antibodies; middle panels, anti-CK19 and anti-DLK1 antibodies; bottom panels, anti-CYP3A4 and anti-DLK1 antibodies. Left column, nontreated umbilical cord-derived mesenchymal stem cells (NC); middle column, induced pluripotent stem cells derived hepatocytes (PC); right column; 4F-Heps (4F). One set of representative results of 3 independent experiments was shown. Values indicate percentage of positive cells. (B) Analysis of cells positive for ALB and TROP2. Left panel, NC; middle panel, PC; right panel, 4F-Heps. Three independent experiments were done and 1 representative set of data was shown. Values indicate percentage of positive cells. (C) Analysis of cells positive for TROP2 and DLK1. Left panel, NC; middle panel, PC; right panel, 4F-Heps. Three independent experiments were performed, and 1 representative set of data was shown. Values indicate percentage of positive cells.

TABLE 1.

Subpopulations of mesenchymal stem cells treated with 4F

Cells	NC	iPSC-Hep	4F-Heps
DLK1	<1a	<1	14.8±2.5
ALB
−b	<1	83±4.2	0.2±0.06
+	<1	<1	4.2±1.5
CK19
−	<1	<1	19.4±2.6
+	<1	<1	10.1±2.9
CYP3A4
−	<1	76±6.3	4.3±0.7
+	<1	<1	15.3±4.3

Note: Each analysis was done in triplicate, and data were depicted by mean+SD.

^aFrequency <1%.

^bDLK1 negative (−) or positive (+).

Abbreviations: 4F, 4 artificial transcription factors; ALB, albumin; iPSC-Heps, induced pluripotent stem cells derived hepatocytes.

Dramatic changes in the gene expression profile of 4F-Heps

We then examined a gene expression profile of 4F-Heps and compared it with those of nontreated MSC-UCs (NC) or MSC-UCs treated with NTP-tagged enhanced green fluorescent protein as a control protein (NE). Principal component analysis of microarray analysis data indicated that the gene expression profile of 4F-Heps was quite different from those of NC and NE (Figure 4A). Volcano plots were used to depict the number of DEGs with a change in expression of >2.0-fold. A comparison of the 4F and NC data sets revealed 1263 upregulated genes and 2045 downregulated genes, while a comparison of the 4F and NE datasets identified 1813 upregulated genes and 1815 downregulated genes (Figure 4B, middle and right panels). In contrast, comparison of the NC data sets with NE identified only 2 upregulated genes with 5 downregulated genes (Figure 4B, left panel). Based on these results, we used NC for further experiments as control.

FIGURE 4.

Gene expression profiles and DNA methylation status of 4F-Heps. (A) Principal component analysis of NC, NE, and 4F. (B) Volcano plot analysis of DEGs. DEGs were analyzed using GeneSpring. Comparisons of NC and NE (left panel), NC and 4F (middle panel), and NE and 4F (right panel) are shown. Cut-off values were p<0.05 and fold change ≥ 2.0. Green and red dots indicate upregulated and downregulated genes, respectively. Numerical values indicate the counts of green and red dots. (C) Heatmap analysis of NC and 4F. Data of genes important for hepatocyte differentiation (right upper panel) and pluripotency-related genes (right lower panel) were extracted. (D) Quadrant plot combining DEGs and differentially methylated targets. Quadrants I and III depict genes with increased DNA methylation, whereas Quadrants II and IV include genes with decreased DNA methylation. Quadrants I and II indicate genes with decreased gene expression, whereas Quadrants III and IV include genes with increased gene expression. Genes, the DNA methylation status of which were changed by delta beta >0.15, were picked up. (E) Validation data of (D). The prepared cells were analyzed by qRT-PCR. PC, human liver-derived mRNA. Mean±SD, n=3. *p<0.05. Abbreviations: 4F, 4F-Heps; NC, nontreated umbilical cord-derived mesenchymal stem cells; NE, NTP-tagged enhanced green fluorescent protein-treated umbilical cord-derived mesenchymal stem cells.

A heatmap of DEGs was used to highlight the upregulated and downregulated genes after 4F treatment. Well consistent with data of flow cytometry, the expression of DLK1 was up-regulated, and frizzled class receptor 4 (FZD4), Wnt family member 3A (WNT3A), jagged canonical notch ligand 2 (JAG2), and HNF3α, which play important roles in hepatic differentiation, were also up-regulated. In contrast, pluripotency-related genes that included LIF IL 6 family cytokine (LIF), Krüppel-like factor 4 (KLF4), and POU class 5 homeobox 1 (POU5F1) were downregulated (Figure 4C, right panel).

We then characterized the DEGs according to DNA methylation patterns. As shown in Figure 4D, DEGs were divided into 4 groups according to gene expression and DNA methylation by delta beta >1.5 (Quadrants I–IV). Quadrants I and III depict genes with increased DNA methylation, whereas Quadrants II and IV include genes with decreased DNA methylation. In contrast, Quadrants I and II possess genes with decreased expression, whereas Quadrants III and IV contain genes with increased expression. In Quadrants I and II, we observed cell division cycle 42 (CDC42), SRY-box transcription factor 5 (SOX5), neuropilin 2 (NRP2), tripartite motif-containing 62 (TRIM62), lamin A/C (LMNA), and HECT, C2, and WW domain-containing E3 ubiquitin protein ligase 2 (HECW2), which are signature genes of MSCs-UCs. In Quadrants III and IV, we found genes related to the WNT/β-catenin and BMP signaling pathways, which are related to hepatic differentiation. In addition, GATA6, which is reported to promote DNA demethylation as an upstream molecule during the early phase of hepatic differentiation,18 was found in Quadrant III. Moreover, forkhead box O1 (FOXO1), which is important for gluconeogenesis and glycolysis in hepatocytes, was also included in Quadrant IV (Figure 4D). Validation of gene expression by qRT-PCR revealed that CDC42, SOX5, NRP2, TRIM62, LMNA, and HECW2 expression was actually downregulated, whereas FGF2, Wnt family member 2 (WNT2), GATA6, FOXO1, bone morphogenetic protein 6 (BMP6), and FGF receptor 3 were upregulated (Figure 4E).

Profound effects of 4F treatment for 5 days on hepatic differentiation

To identify the chronological epigenetic changes induced by 4F treatment, we evaluated the DNA methylation patterns of cells that were treated with 4F for 3 or 5 days and collected at 2 days after the last treatment. In the cells treated for 3 and 5 days, we identified 809 and 895 genes, respectively, that were involved in signal transduction. These genes were subjected to KEGG pathway analysis, and candidate signaling pathways were extracted by functional enrichment clustering analysis (Supplemental Table S3, S4, http://links.lww.com/HC9/A147). Among the enriched top 25 pathways in the cells treated for 3 days, we identified genes related to pluripotency (Figure 5A, red bar in left panel). In striking contrast, the cells treated for 5 days expressed genes involved in the pathways downstream of PI3K-AKT, WNT, mTOR, and MAPK, all of which are involved in hepatic differentiation (Figure 5A, red bar in right panel). The DNA methylation rate of CDC42 and NRP2, which are markers of MSC-UCs, was increased by 4F treatment, whereas the methylation of TRIM62, which suppresses epithelial–mesenchymal transition by inhibiting c-JUN/SLUG signaling, HNF3γ, CYP1B1, annexin A2 (ANXA2), and BMP6, all of which are involved in promoting hepatic differentiation, decreased over time (Figure 5B). Validation of gene expression by qRT-PCR confirmed the downregulation of CDC42 and TRIM62 expression, while HNF3γ, CYP1B1, ANXA2, and BMP6 expression was upregulated according to the days of treatment (Figure 5C). These observations suggested that 5-day treatment with 4F dramatically changed the epigenetic profile and committed cells to hepatic-lineage cells differentiation.

FIGURE 5.

Chronological changes of signal cascades induced by 4F treatment. Cells treated by 4F for 3 or 5 days were subjected to microarray analysis of DNA methylation. (A) Top 25 signal cascades extracted according to the changes of DNA methylation. Based on KEGG pathway analysis, candidate signaling pathways were extracted by functional enrichment clustering analysis. Left panel, comparison of NC and MCS-UCs treated with 4F for 3 days; right panel, comparison of NC with 4F-Heps for 5 days. The vertical axis shows the detection frequency of genes in each pathway. The horizontal axis indicates the enriched pathway (details are shown in Supplemental Table S3, S4, http://links.lww.com/HC9/A147). (B) Chronological changes of DNA methylation of the promoter region after 4F treatment. DNA methylation status of representative genes shown in the quadrant plot (Figure 4D) was depicted. The percentage of DNA methylation observed on days 3 and 5 are shown. (C) Validation data of (B). Endogenous mRNA expression was analyzed by qRT-PCR. PC, human liver-derived mRNA. Mean±SD, n=3. *p<0.05. Abbreviations: 4F, 4F-Heps.

4F-Heps can rescue mice with lethal hepatic failure

To examine the in vivo function of 4F-Heps, we assessed whether they could rescue mice with lethal hepatic failure. After confirming that hepatic injury is crucial for accumulation of exogenously injected MSC-UCs to liver (Supplemental Figure S4, http://links.lww.com/HC9/A141 and Supplemental Materials and Methods, http://links.lww.com/HC9/A147), we designed an experimental protocol for cell transplantation (Figure 6A). Mice were injected with DT on day 1, and serum ALT levels were measured after 2 days (day 3). On day 4, mice with ALT levels at 8,000–23,000 IU/L were divided into two groups (Supplemental Figure S1, http://links.lww.com/HC9/A138). Then, one group of mice was injected with 3.0×10⁵ of 4F-Heps through the tail vein. As control, the other group of mice was injected with the same number of nontreated MSC-UCs (NC). To induce lethal hepatic damage, repetitive injections of a sub-lethal dose of an anti-Fas antibody (0.35 µg/g body weight) was initiated from day 7 after cell transplantation (days 11, 13, and 15 in Figure 6A).34 After 3 injections of the anti-Fas antibody, 83% of mice that were transplanted with 4F-Heps survived (10 of 12 mice tested). In contrast, 33% of mice in the control group survived (4 of 12 mice, Figure 6B, p<0.05).

FIGURE 6.

Functional evaluation of 4F-Heps in vivo. (A) Experimental protocol. TRECK-ALB mice were treated with 1.9 ng/g DT on day 1 and serum ALT was measured on day 3. Mice with ALT measurements of 8,000–23,000 IU/mL were divided into 2 groups and 3.0×10⁵ cells were injected via the tail vein on day 4 (see actual data in Supplemental Figure S1, http://links.lww.com/HC9/A138). On day 11, administration of 0.35 μg/g anti-Fas antibody, a sub-lethal dose, was started, and the mice were injected with the antibody 3 times every other day. (B) Survival curves of mice implanted with NC or 4F-Heps. Mean±SD, n=12. *p<0.05. (C) Measurement of serum ALB. Blood samples were collected from mice at 16 weeks after transplantation, and human ALB in the isolated serum was analyzed by ELISA. Mean±SD, n=3. *p<0.05. (D) Metabolism of DEB. DEB is selectively metabolized by human CYP2D6 and converted to 4-OH DEB. (E) Analysis of 4-OH DEB in serum. DEB was administered to mice at 16 weeks after transplantation, and blood samples were collected at 1, 2, 4, and 6 hours after the injection. Then, 4-OH DEB in serum was analyzed by LC/MS-MS. Mean±SD, n=3. *p<0.05. The vertical axis indicates the ratio of 4-OH DEB to DEB. (F) Expression of human CYP2D6 mRNA. RNA samples were extracted from liver collected from mice at 16 weeks after transplantation and analyzed by qRT-PCR. PC, human liver-derived mRNA. Mean±SD, n=3. *p<0.05. (G) Expression of mature hepatocyte markers. Whole liver tissue was harvested at 16 weeks after transplantation and paraffin-embedded tissue was prepared for each lobe. Immunohistochemical analysis was performed by using anti-ALB, anti-CK18, anti-CYP3A4, and anti-OTC antibodies. The cells were counterstained with hematoxylin. Bar, 100 µm. High-magnification images are shown in the framed area of each image. Immunostaining analysis of other lobes is shown in Supplemental Figure S5 (http://links.lww.com/HC9/A142). (H) Analysis of cellular dynamics of 4F-Heps in vivo. Expression of human specific Ki-67 and human ALB were analyzed at 3 weeks (3 wk) and 16 weeks (16 wK) after transplantation. Immunohistochemical analysis was performed by using anti-human specific Ki-67 and human ALB antibodies. The cells were counterstained with hematoxylin. For each sample, 5 different views were taken and statistically processed. Percentage of positive cells is shown below the pictures. Human ALB+ cells were indicated by red circle. Hu, human specific Ki-67. Bar, 100 µm. Mean±SD, n=5. (I) Detection of human CK19+ cells at 16 weeks after transplantation. After treating with anti-human CK19 antibody, cells were counterstained with hematoxylin. For each sample, 5 different views were taken and statistically analyzed. Percentage of positive cells is shown to the right of the pictures. Bar, 100 µm. Mean±SD, n=5. Abbreviations: 4F, 4F-Heps; 4-OH DEB, 4-hydroxydebrisoquine; ALB, albumin; ALT, alanine aminotransferase; DT, diphtheria toxin; OTC, ornithine transcarbamylase.

To prove hepatic function of the injected 4F-Heps, we first measured human ALB in mouse serum in the mice that were transplanted with 4F-Heps. Although the concentration was low, we detected human ALB at 16 weeks after injection (Figure 6C). We next assessed human cytochrome P450 activity by administration DEB, which is metabolized to 4-OH DEB selectively by human CYP 2D6 (Figure 6D).32 This experiment was performed on mice at 16 weeks after 4F-Heps transplantation. Liquid chromatography-tandem mass spectrometry analysis clearly detected 4-OH DEB at 4 and 6 hours after injection, whereas no 4-OH DEB was detectable in the control group at any timepoints (Figure 6E). Consistently, human CYP2D6 mRNA expression was detected in mouse liver at a level comparable to that of the human liver positive control (Figure 6F).

Propagation of 4F-Heps and repopulation of human ALB+ cells in mouse liver

Immunohistochemical analysis done on tissues of livers that were prepared at 16 weeks after transplantation of 4F-Heps detected tremendous numbers of mature human hepatocytes positive for ALB, CK18, CYP3A4, and ornithine transcarbamylase (Figure 6G, one representative result of lobe is depicted; Supplemental Figure S5, http://links.lww.com/HC9/A142 shows results of other three lobes). By counting human ALB+ cells, we found that ~40% (35.8±9.8%) of cells were positive. It has been reported that numbers of total mouse liver-derived cells are ~5×10⁷,37 implying that injected 3×10⁵ of 4F-Heps expanded to ~2×10⁷ cells (~66-fold). Dramatic expansion of human ALB+ cells was well explained because 4F-Heps contained hepatic progenitors, as shown in Figure 3C. We further analyzed cells positive for ALB and Ki-67, a marker of cell proliferation, in the livers at early phase (3 wk) and late phase (16 wk) after cell transplantation. As shown in Figure 6H, human Ki-67+ cells and human ALB+ cells were detected at approximately 9% (8.7%+4.0%) and 4% (3.6%+1.1%), respectively, in mouse livers at 3 weeks after cell transplantation. In contrast, human Ki-67+ cells at 16 weeks were negligible (0.2%+0.3%). Additional analysis using an antibody to human CK19, a marker of bile duct cells, detected only small numbers of CK19+ cells in the mouse liver even at 16 weeks after cell transplantation (~0.2%, Figure 6I), suggesting that 4F-Heps were selectively differentiated to hepatocytes in the liver.

4F-Heps differentiate into Zones I and III hepatocytes

Hepatic lobules are composed of heterogeneous cells aligned between the portal and central veins.38 In Zone I, hepatocytes, which are provided with oxygen-rich and nutrient-rich blood, are active for gluconeogenesis and ammonia metabolism. In contrast, hepatocytes in perivenous Zone III are supplied with oxygen- and nutrient-deficient blood and express glutamine synthetase (GS), Which catalyzes glutamine from glutamate and ammonia (Figure 7A). In addition, it has been proposed that such a functional gradient in lobules is regulated by WNT/β-catenin signaling. To determine whether 4F-Heps have the potential to differentiate into these functionally different hepatocytes, we stained liver specimens with antibodies to glucose-6-phosphatase (G6P), which is a key enzyme of gluconeogenesis in hepatocytes present in Zone I,39 and to GS, which is expressed by cells in Zone III.39 Immunohistochemical analysis with these antibodies detected positive signals (Figure 7B). In addition, the microsomal fraction prepared from liver tissue was also positive for CYP1A2 and CYP2C19 activity, which are both active in Zone III (Figure 7C).40 Consistently, qRT-PCR analysis indicated that the expression of phosphoenolpyruvate carboxykinase (PCK), G6P, and cyclin D1 (CCND1), which are expressed in Zone I, and CYP1A2 and GS, which are characteristic of Zone III, was up-regulated (Figure 7D). Together with data showing that the expression of WNT2, an important gene for the functional gradient of hepatocytes, was induced by 4F treatment (Figure 4D, Quadrant III, E), data suggest that 4F-Heps can promote differentiation into cells in Zones I and III.

FIGURE 7.

Zonation of 4F-Heps in the liver. (A) Characteristic features of Zones I and III. Differential areas in liver lobules close to the portal vein (PV) and central vein (CV) with distinct gene expression. (B) Expression of markers for Zones I and III. Immunohistochemical analysis using anti-G6P and anti-GS antibodies was performed. The cells were counterstained with hematoxylin. Bar, 100 µm. High-magnification images are shown in the upper frame of each image. (C) CYP activity of mice transplanted with 4F-Heps. Liver tissue was harvested at 16 weeks after cell transplantation and the liver microsomal fraction was extracted. The specific activity of CYP1A2 and CYP2C19 was measured. Recombinant CYP1A2 and CYP2C19 given in the kit were used as PC. Mean±SD, n=3. *p<0.05. (D) Endogenous mRNA expression of Zone I and III markers. RNA was extracted from mouse liver at 16 weeks after transplantation and analyzed by qRT-PCR. PCK, G6P and CCND1 are genes expressed in Zone I, whereas CYP1A2 and GS are expressed in Zone III, PC, human liver-derived mRNA. Mean±SD, n=3. *p<0.05. Abbreviations: 4F, 4F-Heps; G6P, glucose-6-phosphatase; GS, glutamine synthetase; PCK, phosphoenolpyruvate carboxykinase.

DISCUSSION

Here, we developed a protein-based ATF system by which hepatocyte-like cells were generated from human MSC-UCs. By treating MSC-UCs with a mixture of 4 ATFs targeting HNF1α, HNF3γ, HNF4α, and GATA4, hepatocyte-like cells (4F-Heps) were obtained in 5 days. Both in vitro and in vivo experiments suggested that 4F-Heps possessed functional properties of mature hepatocytes (Figures 2, 3, 6), although the activity of ALB production by 4F-Heps was quite weak (Figures 2I, 6C). Consistently, flow cytometry analysis detected that numbers of ALB+ cells were small. However, ~20% of 4F-Heps were CYP3A4-positive and western blot analysis detected that 4F-Heps expressed definite amount of ornithine transcarbamylase and carbamoyl phosphate synthetase-1 proteins, compared to control human liver (Figure 2H). These observations suggest that 4F-Heps could compensate the deficiency of mouse liver functions by factors different from ALB, well consistent with a report that the ability of ALB production is not a limiting factor of reprogrammed cells for the activity of rescuing mice with hepatic failure.35

To our surprise, flow cytometry analysis detected that 4F-Heps were composed of 50% cells positive for a marker of hepatic progenitors (DLK1+ and/or TROP2+) and ~19% of bile duct cells (Figure 3, Table 1). To understand cellular dynamics of 4F-Heps in vivo, we examined cellular status in the mouse liver at early phase (3 wk) and late phase (16 wk) after cell transplantation. As shown in Figure 6H, we detected that human ALB+ cells at the frequency of 4% at 3 weeks, but they dramatically increased after 16 weeks, giving mature hepatocytes comprising ~40% of cells in the mouse liver. Human hepatocytes in the mouse liver were positive for various molecules that are expressed in mature hepatocytes present in Zones I and III (Figure 7C, D). In contrast, ~9% of cells were positive for human Ki-67 at 3 weeks after cell transplantation, whist human Ki-67+ cells were negligible at 16 weeks. Data suggest that 4F-Heps efficiently proliferated and fully differentiated into mature hepatocytes in the mouse liver (Figure 6G), In contrast, however, we observed quite small numbers of human CK19+ cells at 16 weeks (Figure 6I). Although DLK1+ or TROP2+ cells can differentiate to both human ALB+ cells and CK19+ cells,36 human cells in the mouse liver were mostly ALB+, implying that the microenvironment of mouse liver with injury determines the fate of the injected 4F-Heps to differentiate into hepatocytes.

We initially utilized 3 ATFs targeting HNF1α, HNF3γ, and HNF4α (3F), but these 3 ATFs could not induce effective mRNA expression of endogenous genes, especially HNF3γ and GATA4 (Supplemental Figure S6A, http://links.lww.com/HC9/A143). Because these molecules are proposed to be pioneer factors that are essential for liver differentiation,16–18 we included GATA4 as ATF target gene, and added ATF-GATA4 to 3F (4F). Fortunately, we found that the expression of endogenous HNF3γ, HNF4α, GATA6, and ALB mRNA was induced by 4F treatment (Figure 1B–D). Notably, GATA6 functions as an upstream molecule of GATA4 and HNF441 and induces DNA demethylation of lineage-specific genes during hepatic differentiation.18 Our data suggest the presence of a functional link in which HNF4α and GATA4 expressed by ATFs modify the function of GATA6, further implying that 4F treatment for 5 days formed a bi-directional circuit connecting GATA6 and its downstream molecules. Huang et al.35 screened 8 factors that were important for hepatocyte conversion from human fetal fibroblasts and proposed that GATA4 was not required. These discrepant observations might be due to the different roles of GATA4 in human fibroblasts and MSC-UCs or the differential effects of viral vectors and a protein-based ATF system on the cells.

Gene expression analysis revealed that 4F treatment dramatically altered the expression profile of MSC-UCs; 1200–1800 genes were upregulated and ~2000 genes were downregulated by 4F treatment of MSC-UCs for just 5 days (Figure 4B). The up-regulated genes included FZD4 and WNT3A, which are involved in the WNT signaling pathway, as well as matrix metallopeptidase 1, HNF3α, and HNF1α, which are characteristic of hepatocytes. In contrast, the downregulated genes included LIF, KLF4, and POU5F1, which are markers of pluripotent stem cells (Figure 4C). Combined analysis of gene expression and DNA methylation revealed that the mRNA expression of genes related to the pluripotency of MSC-UCs was decreased with concomitant increase of DNA methylation. Importantly, gene expression and DNA methylation were both analyzed at 2 days after the last treatment with 4F (day 7 in Figure 1A), implying that the epigenetic changes caused by 4F treatment have a relatively stable influence on gene expression. In addition, DNA demethylation in the PI3K-AKT, WNT, mTOR, and MAPK pathways was found in cells treated with 4F for 5 days, but not in the cells treated for 3 days (Figure 5A). In contrast, the mRNA expression of HNF3β and GATA6, which are involved in the differentiation of definitive endoderm,17,18 increased on day 5 (Supplemental Figure S7A, http://links.lww.com/HC9/A144), but changes of DNA methylation of these genes had initiated on day 3 (Figure 1F, Supplemental Figure S7B, http://links.lww.com/HC9/A144). These data implicate that 4F treatment for the initial 3 days promoted the differentiation of MSC-UCs to the stage of definitive endoderm, and the additional 2 days of treatment completed their commitment to hepatocyte-lineage cells.

Although further study is required how 4F treatment generates cells that are positive for both markers of hepatic progenitors and mature hepatocytes, our data suggest that the protein-based ATF system can quickly commit MSC-UCs to hepatocyte-lineage cells in 5 days. Taken together with observations that hepatocyte-like cells can be engineered also from human bone marrow-derived MSCs by 4F (Supplemental Figure S8, http://links.lww.com/HC9/A145) and 4F-Heps did not induce tumor formation in SCID mice for at least 2 years after transplantation (Supplemental Figure S9, http://links.lww.com/HC9/A146), we propose that the ATF system would be effective for engineering various types of MSCs applicable as a regenerative cell therapy to liver failures.

Supplementary Material

hc9-7-e0051-s001.tiff ^{(529KB, tiff)}

hc9-7-e0051-s002.tiff ^{(529.6KB, tiff)}

hc9-7-e0051-s003.tiff ^{(2MB, tiff)}

hc9-7-e0051-s004.tiff ^{(529.4KB, tiff)}

hc9-7-e0051-s005.tiff ^{(532.6KB, tiff)}

hc9-7-e0051-s006.tiff ^{(530.6KB, tiff)}

hc9-7-e0051-s007.tiff ^{(530.1KB, tiff)}

hc9-7-e0051-s008.tiff ^{(530.7KB, tiff)}

hc9-7-e0051-s009.tiff ^{(529.4KB, tiff)}

hc9-7-e0051-s010.docx ^{(38.7KB, docx)}