Forkhead box A2 regulated biliary heterogeneity and senescence during cholestatic liver injury

Abstract

Background & Aims

Biliary-committed progenitor cells (small cholangiocytes, SMCCs) from small bile ducts are more resistant to hepatobiliary injury than large mouse cholangiocytes (LGCCs) from large bile ducts. The definitive endoderm marker, FoxA2 is the key transcriptional factor that regulates cell differentiation and tissue regeneration. Our aim was to characterize the translational role of FoxA2 during cholestatic liver injury.

Methods

mRNA expression in SMCCs and LGCCs was assessed by PCR array analysis. Liver tissues and hepatic stellate cells from PSC and PBC patients were tested by real-time PCR for methylation, senescence and fibrosis markers. Bile duct ligation (BDL) and MDR2 knockout mice (MDR2^−/−) were used as animal models of cholestatic liver injury with or without healthy transplanted large or small cholangiocytes.

Results

We demonstrated that FoxA2 was notably enhanced in murine liver progenitor cells and SMCCs, and was silenced in human PSC and PBC liver tissues relative to respective controls that are correlated with the epigenetic methylation enzymes DNMT1 and DNMT3B. Serum ALT and AST levels in NOD/SCID mice engrafted with SMCCs after BDL showed significant changes compared with vehicle-treated mice, along with improved liver fibrosis. Enhanced expression of FoxA2 was observed in BDL mouse liver after SMCC cell therapy. Furthermore, activation of fibrosis signaling pathways were observed in BDL/MDR2^−/− mouse liver as well as in isolated hepatic stellate cells by laser capture microdissection, and these signals were recovered along with reduced hepatic senescence and enhanced hepatic stellate cellular senescence after SMCC engraft.

Conclusions

The definitive endoderm marker and the positive regulator of biliary development, FoxA2, mediates the therapeutic effect of biliary-committed progenitor cells during cholestatic liver injury.

Introduction

The biliary epithelium is a complex network of interconnected ducts that increase in diameter from small to large bile ducts (1–2). The larger portion of the biliary epithelium is lined by mature, cAMP-dependent large cholangiocytes, whereas small (constitutively quiescent) cholangiocytes line small bile ducts (2). Cholangiocytes are supported on a basement membrane and surrounded by connective tissue, extracellular matrix and the peribiliary plexus (3–4). These cells are specialized to act as an interface within the harsh environment imposed by bile and are morphologically and functionally heterogeneous, but with phenotypic patterns implicating a single maturational lineage. Small and large cholangiocytes have been characterized in the intrahepatic biliary epithelium of rats and mice and this phenotype is similar in humans as well (2, 5–6). It has been proposed that small cholangiocytes contain a population of biliary committed progenitors, showing expression of various biliary progenitor markers, and incorporate into neo-bile ducts at the sites of injury . When large cholangiocytes are damaged, small, Ca²⁺-dependent cholangiocytes are activated, acquiring phenotypic and functional features of large cholangiocytes and resulting in the repopulation of the injured large bile ducts (2, 6–7).

The definitive endoderm markers FoxA2 and Sox17 are key transcriptional factors of cell differentiation (8–9). These transcriptional factors are essential for the establishment of developmental competence in the foregut endoderm and the initiation of liver specification. Studies have shown that hepatic deletion of FoxA2 causes cholestasis in mice fed a cholic acid–enriched diet (40). In addition, FoxA2 is downregulated in human subjects with primary sclerosing cholangitis (PSC) and biliary atresia (10). Other studies have demonstrated that FoxA2 and Sox17 are downregulated in the bile duct ligated (BDL) mouse model of obstructive cholestasis (11). Thus, it seems likely that FoxA2 and Sox17 play a major role in patients with chronic cholestatic disorders. The Notch pathway is necessary for the specification of the biliary epithelium, and Notch pathway ablation results in failure of hepatoblast specification into cholangiocytes, resulting in bile duct paucity, a characteristic of Alagilles syndrome (12). Furthermore, ectopic activation of the Notch pathway in fetal hepatoblasts by overexpression of the Notch intracellular domain (NICD) results in hyperarborization of biliary ductules (12). It has been shown that TNFα phosphorylates FoxA2 through IKKα. This activates NUMB transcription, which is able to inhibit NCID and prevent further downstream activation (13). Interestingly, activated Notch signaling can affect the expression of both FoxA2 and Sox17 , suggesting the functional link among these three multipotent molecules during the differentiation process.

Hepatocyte growth and proliferation may be blocked under the conditions that occur during severe liver injury . When this occurs, cholangiocytes originating from the portal ductules and canals of Herring are able to initiate phenotypical changes allowing them to express hepatocyte-associated transcription factors (14). Cholangiocytes can also acquire stem cell phenotypes and in turn, become hepatocytes, restoring liver regeneration when hepatocytes fail to proliferate (15). However, an alternative interpretation is that regeneration after hepatic injury/partial hepatectomy is derived from liver and biliary stem/progenitor cells. Biliary stem/progenitor cells are thought to be located within the canals of Hering and have common markers with biliary epithelia. Progenitor cells become more widespread in diseased conditions (16). Therefore, when hepatocytes fail to proliferate in response to injury, biliary progenitor cells are one of the most important regenerative alternatives. The current study evaluates the possible role of FoxA2 mediated biliary cell restorative therapy during cholestatic liver injury, especially recovery effects on biliary injury and liver fibrosis using BDL animal models (for chronic liver injury) and MDR2^−/− mice that develop periportal fibrosis similar to human PSC (17).

Materials and Methods

Reagents

All reagents were purchased from (Sigma-Aldrich, St. Louis, MO) unless otherwise indicated. Antibodies against DNA (cytosine-5)-methyltransferase 1 (DNMT1), DNA (cytosine-5)-methyltransferase 3B (DNMT3B), cadherin 1 (CDH1), cadherin 2 (CDH2) and β-actin were purchased from Cell Signaling Technologies (Danvers, MA). The S100A4 primary antibody was purchased from AbCam (Cambridge, MA). All real-time PCR primers were obtained from Qiagen (Valencia, CA).

Human Subjects

Human samples were obtained from Dr. Pietro Invernizzi (Liver Unit and Center for Autoimmune Liver Diseases, Humanitas Clinical and Research Center, Rozzano, Milan, Italy) under a protocol by the Ethics Committee of the Humanitas Research Hospital; the protocol was also reviewed by the Veterans’ Administration IRB and International Research Committee. The use of human tissue was also approved by the Texas A&M HSC College of Medicine Institutional Review Board.

Formalin-fixed, paraffin-embedded liver sections (4–5 μm thick) were obtained from 3 patients with PSC and 3 patients with primary biliary cholangitis (PBC). Five patients had advanced stage (3–4) biliary diseases (18) and one patient with PBC had an early stage of disease. The 6 control healthy livers were obtained from patients undergoing resection of liver metastasis (Please see Table 1 for patients’ information).

Table 1.

Characteristics of liver donors of normal controls, PSC and PBC patients.

	Age (Years)	Sex	Pathological Diagnosis	Cirrhosis
PSC Patient 1	94	Female	Advanced stage	Yes
PSC Patient 2	18	Female	Advanced stage	No
PSC Patient 3	88	Male	Advanced stage	No
PBC Patient 1	65	Female	Advanced stage	Yes
PBC Patient 2	63	Female	Early stage	No
PBC Patient 3	55	Female	Advanced stage	Yes
Normal Control 1	72	Female	Normal	No
Normal Control 2	34	Female	Normal	No
Normal Control 3	64	Male	Normal	No

Animal Models

The Animal Care and Use Committee of Baylor Scott & White approved all the animal protocols used in the study. Male NOD.CB17-Prkdc^scid/J (NOD/SCID) and MDR2^−/− (FVB.129P2) mice (25–30 gm, 6 to 12 wks of age), were obtained from Jackson Laboratories (Bar Harbor, ME).

Purified Cholangiocytes and Biliary Cell Lines

Small and large cholangiocytes were isolated by counterflow elutriation followed by immunoaffinity separation (6) using a monoclonal antibody (a gift of Dr. R Faris, Brown University, Providence, RI) against an unidentified antigen expressed by all immortalized normal intrahepatic human (H69, a gift from Dr. G. Gores, Mayo Clinic, Rochester, MN) (19) and murine cholangiocytes (20).

Isolation of Hepatic Stellate Cells by Laser Capture Microdissection

Frozen liver sections were sectioned with a cryostat and affixed to the membrane side of nuclease and human nucleic acid free 2.0mm PEN membrane slides (Leica Microsystems, Wetzlar, Germany). A laser capture microdissection (LCM) system LMD7000 (Leica, Buffalo Grove, IL) was used to capture desmin-positive stellate cells and collect them in a PCR tube. The collected cells were used to isolate RNA with the Arcturus Pico Pure RNA Isolation Kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions.

Statistical Analysis

All data are expressed as mean ± SE. The differences between groups were analyzed by Student's t-test when two groups were analyzed or ANOVA if more than two groups were analyzed. A P value <0.05 was used to indicate statistically significant differences.

Please see supplementary Materials and Methods for more detailed information of this section.

Results

Characterization of Biliary Progenitor Phenotypes in Human and Murine Small and Large Cholangiocytes

To demonstrate potential progenitor phenotypes in small and large cholangiocytes in culture, we performed real-time PCR-based superarray analysis. The Stem Cell PCR Array was chosen because it contains 84 key genes related to the identification, growth and differentiation of stem/progenitor cells. Some of these markers are seen frequently in the definitive endoderm (DE) of the developing embryo, and these DE markers could be useful in the identification of progenitor cells located in small cholangiocytes. We found that 7 of the 84 genes were markedly upregulated in small cholangiocyte lines (SMCC) relative to large cholangiocyte lines (LGCC) used as controls; we also detected several down-regulated genes. The two most up-regulated genes were the DE markers, FoxA2 and Sox17 (Fig. 1A). Up-regulation of these genes in mouse and human small and large cholangiocytes was also evaluated by real-time PCR, which confirmed the array data (Fig. 1B). Activated TGF-β dependent Notch1 and Notch ligand, Jagged 1 (JAG1), were also found in SMCCs compared to LGCCs accompanied by a reduced level of E1A Binding Protein P300 (EP300). Interestingly, several other markers for progenitor cells of the adult epithelium, such as BMP1 and ALDH1, were also overexpressed in mouse and human SMCCs.

Figure 1. Functional analysis of FoxA2, Sox17, Notch1 and BMP1 in small cholangiocytes/biliary committed progenitors.

Panel A: Relative gene expression profile between small cholangiocytes (SMCCs) versus large cholangiocytes (LGCCs) is shown (n=3). The expression of a panel of diverse stemness-associated genes was evaluated by real-time PCR using Mice Stem Cell PCR Array from SABioscience. FoxA2 and Sox17 are the most up-regulated genes among the six stem cell signaling pathways in biliary committed progenitors. Panel B: Total RNA was extracted from SMCC, LGCC, H69-SM and H69-LG cells, FoxA2 and Sox17 expressions were assessed by quantitative real-time PCR. The relative mRNA expression was normalized to expression of LGCC or H69-LG as % of control. *p < 0.05 when compared with mRNA expression of LGCC or H69-LG cells. Panel C: Immunocytochemistry for FoxA2 and Sox17 was performed in TGF-β and Activin A (both at 10 ng/ml) treated mice liver progenitor cells in William's E medium for 7 days. A significant increase in the percentage of FoxA2 and Sox17 positive cells is observed in LPC cell lines after TGF-β + Activin A treatment. *p < 0.05, compared with expression in the TGF-β or Activin -only group. Panel D: TGF-β regulated differentiation process (MET) in human small and large cholangiocytes. Proteins were isolated from human small and large cholangiocytes treated with TGF-β (10 ng/ml) or diluents control for 7 days, Western blot confirmed reduced MET process after TGF-β treatment. Western blots of H69 cell lysates (TGF-β treated groups and controls) were performed and sequentially probed with antibodies against mesenchymal markers S100A4, Vimentin, CDH2; epithelial markers CDH1, and β-actin as a loading control as indicated. Representative immunoblots are shown on the Panel along with quantitative data that show the mean ± S.E. from four separate blots of independent experiments on Suppl Fig. 1A. Panel E: H69 human cholangiocytes were separated by counterflow elutriation into small and large populations. RNA was extracted from these separate populations and Cytokeratin-19 (CK-19) expression was analyzed with qPCR. No significant differences were observed between these two groups.

TGF-β and Activin A Promote Human Liver Progenitor Cells to Acquire Small Biliary Phenotypes

DE has been successfully derived in vitro from mouse embryonic stem cells (ESCs) using culture conditions of low serum and high concentrations of Activin A. Under these conditions, it has been shown that TGF-β improves the efficiency of Activin A-induced DE differentiation from human ESCs (21). We hypothesized that TGF-β improves the efficiency of Activin A-induced biliary differentiation in human liver progenitor cells (LPCs). To test this hypothesis, we treated human LPCs with Activin A in the presence or absence of TGF-β in William's E medium for 7 days on plates coated with Matrigel®, which has been shown to promote the differentiation of hepatic progenitor cells to cholangiocytes. After treatment in William's E medium for 5 days, cells were analyzed for the expression of FoxA2 and Sox17. Immunofluorescent staining (IF) revealed that approximately 80% of the cells treated with Activin A and TGF-β expressed FoxA2 or Sox17, whereas only about 40% of the cells were FoxA2 or Sox17 positive when treated with Activin A alone (Fig. 1C). TGF-β alone did not affect biliary differentiation. Real-time PCR confirmed that the expression of FoxA2 and Sox17 was significantly increased upon co-treatment with both Activin A and TGF-β, whereas the addition of TGF-β alone did not induce biliary differentiation from human LPCs. Together, these preliminary data indicate that TGF-β augments human LPC derived SMCC formation by acting synergistically with Activin A.

Since TGF-β is a well-known mesenchymal to epithelial transition (MET) inhibitor, we aimed to demonstrate TGF-β activation of the mesenchymal markers, S100A4 and Vimentin in small and large cholangiocytes in vitro. TGF-β (10 ng/ml) induced significant cell differentiation after 7 days of incubation, as measured by incorporation of S100A4 and Vimentin protein expression, as well as by the epithelial marker, E-Cadherin (CDH2), to the mesenchymal marker, N-Cadherin (CDH1), switch and anti-apoptotic properties in human small and large cholangiocytes (Fig. 1D, Suppl. Fig. 1A and Suppl Fig. 2A). Analysis of cytokeritin-19 (CK-19) levels by real-time PCR showed similar levels of CK-19 in H69 LGCCs compared to SMCCs (Fig. 1E).

FoxA2 and Sox17 Expression is Epigenetically Regulated

To evaluate the potential epigenetic mechanisms by which the expression of FoxA2 and Sox17 is deregulated in small cholangiocyte/biliary progenitors, we performed a Real-Time PCR-based Superarray analysis. The Epigenetic Chromatin Modification Enzymes PCR Array was chosen because it contains 84 key genes encoding enzymes known or predicted to modify genomic DNA and histones to regulate chromatin accessibility and gene expression. The two most down-regulated genes detected by PCR array in SMCCs relative to LGCCs were DNMT3B and DNMT3A (Fig. 2A). Down-regulation of these genes was confirmed by real-time PCR (Fig. 2B). Additionally, analysis of the promoter regions of FoxA2 and Sox17 by the MethPrimer program (22) revealed the presence of several CpG islands ~2000 basepairs upstream of the 5’-coding regions, which are commonly modified by epigenetic regulation (Fig. 2C). To evaluate the direct relationship between methylation and FoxA2 or Sox17 expression in cholangiocytes, the methylation inhibitor, 5-Aza-CdR, was used to treat SMCCs or LGCC in vitro before evaluating FoxA2 and Sox17 expression. The marked increases of FoxA2 and Sox17 were noted in LGCCs (but not SMCCs) compared to controls after 5-Aza-CdR treatment (Fig. 2D). Using a methylation-specific PCR (MSP) assay, we found that the FoxA2 promoter was hypomethylated in SMCCs in culture (Suppl Fig. 1B) as well as in isolated small cholangiocytes from livers from BDL mice (Suppl Fig. 1C). These results suggest that the expression of FoxA2 and Sox17 may be potentially regulated by modulation of promoter methylation. Of note, histone acetylation has been shown to activate the expression of FoxA2 and Sox17 during the early stage of hepatocyte differentiation (23). To evaluate the direct relationship between DNMT3A and cell size, DNMT3A was overexpressed in SMCCs. Overexpression of DNMT3A showed an increase in overall cell size (Fig. 2E, top) as well as increased levels of the LGCC markers, CFTR, secretin and secretin receptor (5) (Fig. 2E, bottom) indicating the ability of DNMT3A to influence cell size through possible downregulation of DE markers. Additionally, the direct relationship between FoxA2 and Sox17 was analyzed by real-time PCR. Overexpression of DNMT3A showed reduced levels of FoxA2, Sox17, Notch1 and BMP1 (Fig. 2F) indicating the relationship between DNMT3A and these endodermal markers.

Figure 2. Epigenetic regulation of FoxA2 and Sox17 expression in biliary committed progenitor cells.

Panel A: The expression of DNMT3B and DNMT3A is down-regulated in small cholangiocytes. Relative gene expression profile between SMCCs vs. LGCCs is shown (n=3). The expression of a panel of diverse epigenetic-associated genes was evaluated by real-time PCR using Mouse Epigenetic Chromatin Modification Enzymes PCR Array from SABioscience. Gene expression relative to GAPDH was plotted as the Volcano Plots, depicting the relative expression levels (Log 2) for selected genes in SMCCs versus control LGCC panels (Left). The relative expression levels and p values for each gene in the related samples were also plotted against each other in the scatter plot (Right). DNMT3B and DNMT3A are the most down-regulated genes among the five epigenetic signaling pathways in cholangiocytes. Panel B: Real-Time PCR confirmed the reduced mRNA expression of DNMT3B and DNMT3A in cultured SMCCs, isolated SMCCs from BDL mice liver and small human cholangiocytes (H69-SM) compare to respective controls. The percentages shown represent the mean value (relative to control) normalized with GAPDH from four independent experiments. Panel C: Analysis of the promoter region using MethPrimer software revealed the presence of several CpG islands ~2000 basepairs upstream of the 5’-region of the FoxA2 and Sox17 sequence. Panel D: Small and large cholangiocytes were treated with 10 μM 5-aza-2'deoxy-cytidine (5-Aza-CdR) or diluent control for 72 hr. The expressions of FoxA2, Sox17 and Notch1 were assessed by Western blot analysis. 5-Aza-CdR increased FoxA2 and Sox17 expression in LGCCs but not SMCCs. Representative immunoblots are shown on the top right along with quantitative data that show the mean ± S.E. from four separate blots of independent experiments on bottom panel. * p < 0.05 relative to controls, # p < 0.05 relative to SMCCs. Panel E: DNMT3A was overexpressed in SMCCs. Top Panel, cells were harvested with trypsin, stained with trypan blue and cell size was measured compared to control as a percentage with the control set at 0. *p < 0.05 compared to control. Bottom Panel, RNA was extracted and markers of small and large cholangiocytes were measured with qPCR. *p < 0.05 compared to control. Panel F: DNMT3A was overexpressed in SMCCs, RNA was extracted and DE markers were measured with qPCR. *p < 0.05 compared to control.

Altered TGF-β-induced Signaling and Remodeling and Fibrosis in Experimental Models of Cholestatic Liver Injury In Vivo

Consistent with the role of TGF-β induced cell differentiation in vitro, in rodent models of cholestasis, BDL or MDR2^−/−, secretion levels of soluble TGF-β were significantly up-regulated (Fig. 3A). To determine the mechanisms of TGF-β mRNA alterations and their effects on modulating target protein expression in vivo, we compared BMP1/FoxA2/Notch1/Sox17 expression in tissue homogenates and isolated cholangiocytes from mouse liver. The expression of BMP1/FoxA2/Notch1/Sox17 protein and mRNA was significantly decreased in BDL and MDR2^−/− mouse liver and isolated cholangiocytes compared to normal controls (Fig. 3B&C), which suggests alternate TGF-β regulated miRNA/mRNA expression patterns in different liver cell types. To explore this further, we overexpressed TGF-β in SMCCs and analyzed levels of DE markers. Unlike in total liver or isolated pooled (that include both small and large) cholangiocytes, FoxA2 was increased in SMCCs (Fig. 3D) indicating that upregulation of TGF-β leads to upregulation of FoxA2, which may indicate activation of the progenitor cells contained in SMCCs. Activated definitive endoderm differentiation markers FoxA2 and Sox17 were observed in small bile ducts after cholestatic liver injury in BDL and MDR2^−/−(Suppl. Fig. 2B). To further assess the extent of fibrosis/cirrhosis in BDL/MDR2^−/− mouse liver, Sirius red staining was performed on liver sections from both animal models (Figure 3E&F). Control mice had a normal distribution of collagen, whereas those with BDL and MDR2^−/− demonstrated obvious signs of fibrosis. Mice with BDL injury demonstrated bridging fibrosis and those with MDR2^−/− had diffuse collagen deposition suggesting that severe liver fibrosis in both models of cholestatic liver injury.

Figure 3. Altered cholangiocytes differentiation and remodeling process during cholestatic liver injury.

Panel A: Supernatants from liver tissue homogenates obtained from a surgical (Bile Duct Ligation) or transgenic (MDR2^−/−) mouse model of cholestatic liver injury were harvested with respective controls (normal or wild type), and soluble TGF-β levels were measured by ELISA (n=4). There were significant increases in soluble TGF-β levels in BDL or MDR2^−/− mice liver. Panel B&C: Verification of specific protein and mRNA expressions in BDL and MDR2^−/− mice liver (n=4). Proteins and total RNAs were isolated from BDL/MDR2^−/− mice liver tissue homogenates, the expressions of specific proteins and mRNAs were verified by Western blot (Panel B) and Taqman real-time PCR analysis (Panel C). In Panel B representative immunoblots are shown on top panel along with quantitative data that show the mean ± S.E. from four separate blots of independent experiments on bottom panel. Reduced expressions of FoxA2, BMP1, Sox17 and Notch1 after cholestatic liver injury were observed in both models. Panel D: TGF-β was overexpressed in SMCCs, RNA was extracted and levels of DE markers were measured with qPCR. *p < 0.05 compared to control. Panel E–F: Percentages of positive cells stained with Sirius red were counted in BDL and MDR2^−/− mice liver with the respective controls (Panel E&F). Significantly increased intensity of liver fibrosis was confirmed in both models by enhanced Sirius red expression (n = 5). *p < 0.05 versus normal or WT controls.

FoxA2 is Down-regulated in Human Cholestatic Livers

We have demonstrated that TGF-β levels are elevated in patients with PSC and PBC as well as the levels of fibrotic genes such as collagen A1, fibronectin and α-SMA (Suppl Fig. 1, Panel D-F). Next, we aimed to determine whether the expression of DE and epigenetic markers is altered under cholestatic conditions in humans. FoxA2 mRNA levels were virtually undetectable in patients with PSC (Fig. 4A, left), as well as in those with PBC with a reduction in 2 out of 3 livers measured (Fig. 4A, right). Meanwhile, the expression of Sox17 was significantly increased in PSC livers but reduced in PBC livers (Fig. 4B&C), suggesting different pathological mechanisms of this DE marker during human cholestatic liver disorders. In both PSC and PBC patients, Notch-1 and BMP-1 levels were increased compared to normal patients (Fig 4D&E). These results show that certain progenitor markers are indeed altered frequently in human PSC and PBC. To further verify the upstream mechanisms of FoxA2 regulation, we performed real-time PCR studies to detect the mRNA expression of the enzymes that catalyze the methyl-transfer reaction, the DNA methyltransferases. Both DNMT1 and DNMT3B were significantly up-regulated in PSC and PBC livers (Fig. 4B&C), which is consistent with our cell culture studies (Fig. 2B&D) and implies the potential epigenetic regulatory mechanisms of FoxA2 during the development and progression of human PSC and PBC. Thus, cholestatic liver injury of differing etiologies causes downregulation of FoxA2 in humans, which may occur through the epigenetic regulation mechanisms, and together with our findings, and suggests that low FoxA2 levels exacerbate fibrotic injury to the liver.

Figure 4. Enhanced fibrosis markers along with the reduced FoxA2 expression in human PSC/PBC liver.

Panel A–E: Altered FoxA2, Sox17, Notch1 and BMP-1 expression along with enhanced DNA methyltransferases in human PSC and PBC liver tissues. Total RNA was isolated from human liver from normal controls or patients with PSC/PBC. Real-time PCR analysis was performed, and the ratio of specific mRNAs to GAPDH mRNA expression in human liver samples was determined. The PCR products were also verified by 1.8% agarose gel electrophoresis. FoxA2 was significantly reduced in 5 among 6 PSC/PBC livers relative to normal controls (Panel A). Meanwhile, the expression of Sox17 was only silenced in PBC liver tissues but increased in PSCs (Panel B&C). DNA methylation enzymes DNMT1 and DNMT3B were also increased in both PSC and PBC livers, suggested the potential hypermethylation mechanisms of FoxA2 gene during PSC/PBC development. Additionally, both Notch-1 and BMP-1 were significantly reduced in PBC and PSC livers compare to controls (Panel D&E). Data represent mean ± SE from liver samples from three PSC or three PBC patients relative to three normal controls with three separate experiments. *p < 0.05 relative to normal controls.

Role of Small Cholangiocyte Therapy in Ameliorating BDL-induced Biliary Damage and Liver Fibrosis

Because small cholangiocytes may possibly contain a compartment of progenitor cells for repairing the biliary epithelium, we performed studies to determine whether suitable biliary support using transplanted small cholangiocytes would permit hepatic repair and regrowth of the damaged liver. Because cholangiocytes can be transplanted in large numbers in the peritoneal cavity, readily equaling or exceeding those in the liver, we labeled cultured SMCCs and LGCCs with a red fluorescent marker (PKH26), suspended them in extracellular matrix (ECM) and injected them into BDL mice 1 and 3 days post surgery. The translocation of engrafted biliary progenitors/cholangiocytes in the liver was detected in liver sections after cell therapy (Fig. 5A). Serum ALT and AST levels in NOD/SCID mice engrafted with SMCCs and liver stem cells (3 X 10⁶, i.p.) showed significant changes compared with vehicle treated mice (Suppl. Fig. 1D; n=5), along with significantly improved TGF-β levels in bile (Fig. 5B). The activated definitive endoderm differentiation marker, FoxA2, was observed in BDL mice liver after cholestatic injury coupled with SMCC cell therapy (Fig. 5C). Furthermore, decreased Sirius red staining (Fig. 5D&E), along with altered remodeling properties, such as activation of MMP-9/MMP-2 and silencing of TIMP-3, as well as enhanced expression of remodeling protein α-SMA (Suppl. Fig. 2 D&E), were observed in BDL mice liver recovered after SMCC engraft, suggesting that remodeling enzymes and proteins are important mediators for SMCC-mediated recovery of cholestatic liver injury and liver fibrosis. Additionally, we examined total liver tissue for markers of cholangiocytes and we observed upregulation of the large cholangiocyte marker in BDL mice treated with either small or large cholangiocytes. BDL mice treated with LGCCs showed upregulation of secretin and secretin receptor (Fig. 5F) which is expected due to the large influx of LGCCs. Additionally, we saw a decrease in the large cholangiocyte markers, AE2 and secretin receptor in SMCC-treated cholangiocytes most likely due to the significant increase in small cholangiocytes.

Figure 5. Effect of cell therapy on BDL-induced cholestatic liver injury.

Panel A: Localization of transplanted small cholangiocytes inside the mice liver: An additional group of mice was used for detecting transplanted small cholangiocytes location. Locally transplanted small cholangiocytes were labeled using PKH26 red fluorescent to detect small cholangiocytes which migrated to the liver in normal mice. A piece of mouse liver (~1 cm) was removed and fixed. The fluorescent signals from labeled SMCCs were predominantly detected in the mice liver 14 days after small cholangiocytes injection. Panel B: ELISA assay for TGF-β was carried out in the bile from small or large cholangiocytes transplanted mice after BDL. The TGF-β secretion in bile was significantly reduced after small cholangiocytes transplantation when compared to the large cholangiocytes and ECM control group after BDL. Panel C: Enhanced expressions of FoxA2 were detected in SMCC-BDL mice liver relative to ECM control mice liver by real-time PCR assay. Panel D&E: Reduced staining of Sirius Red was seen in SMCC-BDL mice liver when compared to the controls. Area of collagen present in liver sections stained with Sirius red were quantified in BDL mice liver with small or large cholangiocyte cell therapy relative to ECM control (Panel E). Significantly decreased intensity of liver fibrosis was detected in SMCC-BDL mice liver relative to ECM controls by enhanced Sirius red expression (n = 5). Original magnifications ×200. *p < 0.05 versus respective controls. Panel F: LGCC markers (AE2, Secretin (Sec), Secretin Receptor (SR)) and CK-19 were measured in cell therapy animals via qPCR (n=5) *p<0.05 vs ECM.

Potential Molecular Mechanisms by which Biliary Cell Therapy Improves BDL-induced Biliary Damage and Liver Fibrosis

To examine the molecular mechanisms by which biliary cell reversed BDL-induced hepatobiliary damages, we used the Mouse Fibrosis PCR array (PAMM-120ZA, Qiagen) and combined it with PCR of inflammation- and stem cell & development - related genes to evaluate mechanisms related to liver fibrosis, inflammation and stem cell signaling pathways. α-SMA, MMP-8 and MMP-9 were the most down-regulated genes among the fibrotic signaling pathways in cholestatic liver (BDL) coupled with SMCC therapy compared to those treated with LGCC therapy (Fig. 6A). Ingenuity Pathway analysis (IPA) was performed to ascertain the cellular context of the differentially expressed genes related to SMCC mediated liver repair. Pathway analysis indicated that the hepatic cellular senescence pathway was the most inhibited through FoxA2 related signaling mechanisms (Fig. 6B). Further analysis with IPA uncovered several differentially regulated cellular senescence gene alterations following small cholangiocyte cell therapy. Several of these genes are regulated by FoxA2 including p16, PAI-1, CCL2 and EGR1. To evaluate overall cellular senescence in SMCC therapy animals, liver sections were stained with SA-β-Gal (Fig. 6C) as described. SMCC therapy significantly reduced cellular senescence after BDL as quantified by SA-β-Gal staining (Fig. 6C) and fluorometric detection (Fig. 6E, left). Total liver isolates were used to evaluate levels of senescence markers, which revealed decreased levels of p16, PAI-1, EGR-1 and CCL2 in SMCC treated BDL mice compared to BDL mice treated with vehicle (ECM) (Fig. 6F, left). To evaluate the direct relationship between cellular senescence and FoxA2, the senescence gene, p16 was overexpressed in SMCCs, which led to a decrease in FoxA2 levels (Fig. 6D). This indicates a possible mechanism where biliary damage induces senescence and overexpression of senescence genes suppresses repair mechanisms.

Figure 6. Effect of cell therapy on BDL-induced cellular senescence during cholestatic liver injury.

Panel A: Relative gene expression profile between small cholangiocytes versus large cholangiocytes treated BDL liver is shown. The expression of a panel of diverse fibrosis-associated genes was evaluated by real-time PCR using Mice Fibrosis PCR Array from SABioscience. Gene expression relative to GAPDH was plotted as the Volcano Plots, depicting the relative expression levels (Log 10) for selected genes in BDL + Small versus BDL + Large control panels. α-SMA, MMP-2 and MMP-9 are the most down-regulated genes among the fibrotic signaling pathways in small cholangiocytes cell therapy after BDL injury. Panel B-C;E-F(left): Altered cellular senescence after small cholangiocytes cell therapy. Panel B presents Ingenuity Pathway analysis (IPA) of differentially regulated gene network after small cholangiocytes cell therapy. IPA was performed to understand the cellular context of the differentially expressed genes related to the recovery effects. Several genes implicated in cellular senescence are regulated by FoxA2, including p16 (CDKN2A), PAI-1 (SERPINE1), CCL2 and EGR1. Liver sections are stained with SA-β-Gal (Panel C) as described to reveal cellular senescence in BDL mice +small relative to +ECM and +Large controls. Small cholangiocytes cell therapy significantly reduced cellular senescence after BDL as quantified by SA-β-Gal staining (Panel C) and flurometric detections (Panel E, left), whereas the enhanced cellular senescence in isolated hepatic stellate cells by LCM was observed (Panel E. right). The alterations of cellular senescence markers p16, PAI-1, CCL2 and EGR1 after small cholangiocytes therapy in BDL mice liver were also verified (Panel F, left). *p < 0.05 versus normal controls. #p < 0.05 versus BDL+ECM controls. Panel D: p16 overexpression was performed in SMCCs, RNA was extracted and FoxA2 levels were measured with qPCR. *p < 0.05 versus control. Panel F (Right): Cholangiocyte supernatants derived from BDL mice treated with ECM, SMCC or LGCC were used to treat cultured stellate cells and inflammatory markers (CCL2 and IL-8) and a senescence marker (p16) were measured with qPCR. *p < 0.05 versus basal. #p < 0.05 versus BDL+ECM ECM.

We then aimed to evaluate the mechanisms by which SMCCs are able to suppress biliary fibrosis. Because stellate cells are implicated in liver fibrosis and cellular senescence is highly upregulated in total liver, we evaluated stellate cells from the livers of SMCC therapy mice for senescent markers. We found enhanced expression of the cellular senescence markers p16 and EGR1 in isolated mouse hepatic stellate cells by laser capture microdissection (LCM) (Fig. 7A&B), as well as upregulation of cellular senescence as measured by fluorometric analysis (Fig. 6E, right), suggesting that senescence of activated stellate cells limits liver fibrosis during FoxA2 mediated cell therapy process. We additionally isolated stellate cells from PSC patients with LCM and saw decreased levels of the senescence gene PAI-1 and increased levels of the inflammatory gene IL-8 (Fig. 7C) indicating that the stellate cells are active in this disease model.

Figure 7. Alterations of cellular senescence markers in mouse hepatic stellate cells isolated by laser capture microdissection.

Schematic representation of laser capture microdissection (LCM) procedures was displayed in Panel A. Frozen liver sections from control, BDL, BDL-SMCC and BDL-LGCC mice were sectioned with a cryostat and affixed the membrane side of nuclease and human nucleic acid free 2.0mm PEN membrane slides. A LCM system LMD7000 (Leica, Buffalo Grove, IL) was then used to capture desmin positive hepatic stellate cells and collect them in a thin walled PCR tube. The collected cells were then used to isolate RNA with the Arcturus Pico Pure RNA Isolation Kit, and real-time PCR analysis for cellular senescence markers p16, PAI-1, EGR1 and CCL2 were carried out. Enhanced expressions of p16 and EGR1 were discovered in isolated hepatic stellate cells from BDL-SMCC mice liver when compared to control BDL groups (Panel B), suggested SMCC cell therapy inhibited liver fibrosis through the induction of cellular senescence in hepatic stellate cells. *p < 0.05 versus normal or WT controls. #p < 0.05 versus Control-BDL mice liver. Panel C: Formalin fixed-paraffin embedded liver sections from PSC patients were sectioned at 6 μm and affixed to the membrane side of PEN membrane slides. Sections were stained for desmin and a LCM system LMD7000 was used to capture desmin positive cells. The collected cells were used to extract RNA as described above and the senescence marker PAI-1 and the inflammation marker IL-8 were quantified with qPCR. *p < 0.05 versus normal.

To evaluate the interactions between cholangiocytes and stellate cells, we used isolated cholangiocytes from our cell therapy BDL-treated animals to extract supernatants which contain the cytokines, extracellular vesicles and other molecules which are secreted by cholangiocytes and treated cultured stellate cells with these supernatants. After a 48 hour incubation, the stellate cells treated with SMCC-BDL cholangiocyte supernatants showed decreased levels of the inflammatory markers, CCL2 and IL-8 and increased levels of p16 (Fig. 6F, right) indicating the ability of molecules or vesicles secreted by cholangiocytes are able to affect inflammation and senescence in hepatic stellate cells. This ability to influence these cells most likely alters the activity of hepatic stellate cells and thus ameliorates the fibrosis observed in SMCC-treated BDL mice.

Discussion

The maintenance of homeostasis of the biliary epithelium is critical for the prevention and recovery from hepatobiliary damage that occurs during the pathogenesis of chronic cholestatic liver diseases such as PBC and PSC. Although the functional role for biliary progenitor cells in liver regeneration has been proposed, the molecular mechanisms by which transcriptional factors such as FoxA2 modulate hepatobiliary regrowth and hepatic repair are unknown. We identified definitive endoderm markers including FoxA2, Sox17 as well as BMP1 that are upregulated in SMCCs compared to control LGCCs by PCR array analysis. FoxA2 was also more enhanced in murine liver progenitor cells compared to SMCCs and LGCCs, and reduced in the liver tissues from human PSC/PBC patients. As in PBC and PSC, reduced FoxA2 expression was observed in murine small bile ducts in BDL and MDR2^−/− mouse liver, suggesting that it is an important mediator of biliary injury and recovery. The translocation of engrafted biliary progenitors/cholangiocytes into the liver after IP injection was confirmed by PKH26 red fluorescent labeling detected within the liver following cell therapy. Serum ALT and AST levels in NOD/SCID mice engrafted with SMCCs showed significant changes compared to vehicle treated mice, along with significantly improved liver fibrosis. Enhanced expression of the DE differentiation marker FoxA2 was observed in BDL mice liver after SMCC cell therapy, along with the reduced hepatic/enhanced hepatic stellate cellular senescence. Overexpression of FoxA2 in LGCCs and subsequent IP injection into BDL mice showed similar reductions in biliary injury to that seen with SMCC cell therapy. The identification of FoxA2 as an important transcriptional regulator of recovering events in vivo emphasizes an essential role of this DE marker in mediating biliary regeneration and repair. This discovery also provides insight into the contribution of altered remodeling, epigenetic regulation and senescence in recovering the severe liver injury.

Cholangiopathies cause morbidity and mortality and are the principal reason for liver transplantation. They are characterized by spotty rather than diffuse proliferation and loss of cholangiocytes lining different sized bile ducts. The elucidation of the intracellular mechanisms regulating the differential regenerative responses of small and large cholangiocytes to cholestasis and liver injury/toxins will play a pivotal role in the development of therapeutic strategies for the treatment of liver diseases (e.g., PBC, PSC and Cholangiocarcinoma). These cholangiopathies represent a serious public health concern due to the lack of novel therapeutic approaches and this subsequently results in liver transplantation or mortality (2, 24). During chronic hepatobiliary injury, a population of bipotent liver progenitor cells (LPCs) becomes activated to regenerate both cholangiocytes and hepatocytes (25–26). Cholangiocytes and hepatocytes share embryologic origins and this common heritage contributes to traits carried into adulthood (27–28). If small cholangiocytes with multipotential capacity exist within human and rodent bile ducts, these cells should also possess the ability to differentiate into either large cholangiocytes or hepatocytes during liver damage, such as diseased conditions in which large cholangiocytes or hepatocytes are lost or regenerative mechanisms are hampered. The plasticity of intrahepatic cholangiocytes and hepatocytes has been postulated that terminally differentiated cells of one lineage may directly differentiate into another lineage or undergo trans-differentiation . However, the data are interpretable also as expansion and differentiation of a progenitor’s population (29). Recent studies have demonstrated that meticulously isolated and rigorously characterized gallbladder epithelial cells cultured under defined in vitro conditions acquire hepatocyte-like properties, such as the ability to synthesize bile acids and take up low-density lipoprotein, without expression of markers of oval cells or hematopoietic progenitor cells . The recent discovery of biliary pluripotent cell populations provides for an alternative mechanism for the expansion and differentiation of certain subpopulations of cholangiocytes as opposed to transdifferentiation (1). Therefore, specific subpopulations of cells, such as small cholangiocytes that express known biliary progenitor cell markers, can be hypothesized to contain a multipotent cell population when exposed to certain pathological conditions. Our studies have suggested that such cells could attain functional pluripotent characteristics under the condition that large cholangiocytes are damaged or lost, and subsequently they could be used to repopulate damaged bile ducts and livers (7, 30).

DNA methylation is an important epigenetic modification that can regulate gene expression and is tightly regulated by at least three DNA methyltransferases (DNMT1, DNMT3A and DNMT3B). Aberrant DNA methylation has been implicated in many human diseases including primary biliary disorders (31). Intrahepatic accumulation of toxic bile acids and excretion products causes cellular injury. Recent developments indicate that accumulation of toxic bile acids induces epigenetic alterations, particularly acetylation, methylation of histones, and hypo- and hypermethylation of DNA . This has created a renewed interest in biliary pathogenesis and is providing novel insight into actions of bile acids at the nucleosomal level in relation to gene expression and pathophysiological consequences. Although DNA methylation has been tightly linked to liver injuries and poor disease outcome in many hepatic disorders, including human PBC, its application to progenitor phenotype related mRNA expression is novel. A better understanding of how specific DNA methyltransferases contribute to aberrant mRNA expression will clearly advance the field and increase the knowledge of the mechanisms regulating the pathogenesis of cholestasis.

DNA methylation regulated by TGF-β activation has been demonstrated during chronic liver injury and malignant transformation. There is ample evidence for increased TGF-β activation in cholestatic liver injury (11, 32). Analysis of the transcriptional changes that accompany the process of biliary differentiation has identified several interrelated signaling pathways that are critical during bile duct differentiation in vitro (33). The molecule at the apex of these signaling pathways is BMP1, an astacin metalloproteinase. We demonstrated that BMP1 and Notch1 are upregulated in human PSC and PBC samples. BMP1 cleaves procollagen and allows the triple helix of collagen to form in the extracellular matrix. Therefore, BMP1 may play a prominent role in the remodeling of the extracellular matrix surrounding the developing bile ducts . Interestingly, BMP1 is down regulated when biliary progenitor cells are cued to differentiate towards mature cholangiocytes (34). Transient inhibition of the constitutive BMP pathway, either alone or in combination with TGF-β inhibition, is critical to the downstream effector FoxA2 signaling in stem cell cultures . The observation that a number of BMP and TGF-β responsive genes (such as FoxA2) are activated during bile duct differentiation in vitro suggests their essential roles in the formation of biliary structures.

Hepatobiliary regeneration is critical for the recovery of cholestatic liver diseases such as PSC and PBC. Accumulation of senescent hepatobiliary cells, which are insensitive to mitotic stimuli, may impair the reserve for hepatobiliary regeneration. In addition, the association between increased hepatobiliary nuclear area and hepatobiliary dysfunction suggests that senescent cholangiocytes/hepatocytes may not function as normal mature cholangiocytes/hepatocytes. That is why our cultured small cholangiocytes behave differently than total liver shown in the MDR2−/− and PSC patient samples. There is also an association between senescence measured as increased expression of senescent markers in cholangiocytes and hepatocytes and impaired liver synthetic function (increasing prothrombin time and decreasing serum albumin level) . Thus, accumulation of senescent cholangiocytes/hepatocytes may contribute to loss of functional hepatobiliary mass with sufficient accumulation of such cells eventually leading to hepatobiliary decompensation and liver related death, accounting for the strong link between cholangiocyte/hepatocyte senescent markers and adverse liver-related outcome. During clinical and biochemical dysfunction with progressive cholestatic liver disease, up to 87% of hepatobiliary cells in cholestatic liver diseases (CLDs) expressed senescence markers and may have impaired function. However, senescence of activated stellate cells limits liver fibrosis through reduced secretion of extracellular matrix components, enhanced secretion of extracellular matrix degrading enzymes, and enhanced immune surveillance. The discovery of an association between reduced hepatic/enhanced hepatic stellate cellular senescence and FoxA2, a DE maker of liver development and recovery in CLD in particular, suggests an important translational mechanism in CLD development, progression and recovery.

In conclusion, we have demonstrated the epigenetic regulation of FoxA2 expression during cholestatic liver injury. Additionally, we have shown that the expression of downstream anti-cellular senescence/fibrosis signaling of biliary regrowth/repair could be modulated by FoxA2. This indicates that therapeutic strategies to increase FoxA2 may be potentially useful to rebuild the hepatobiliary system after liver injury. Further work is warranted to evaluate the functional role of FoxA2 and the identified downstream targets and to develop therapeutic strategies by taking advantage of FoxA2 overexpression in vivo. The ability to therapeutically manipulate mRNA expression is also feasible, and recent proof-of-concept studies have shown that mRNA agonists targeted to the liver can modulate expression of downstream genes. Moreover, aberrantly expressed mRNAs of FoxA2 and other DE markers may be useful to establish a diagnosis and for assessing prognosis in liver injury. Knowledge of specific processes such as biliary proliferation, senescence, remodeling and mesenchymal transition that are regulated by FoxA2, and the identification of critical targets for such DE markers, provides novel insights into the mechanisms in the development and recovery of the intrahepatic biliary epithelium and hepatic function after chronic liver injury.

Abbreviations

BDL

bile duct ligation

FoxA2

forkhead box A2

LGCC

large mouse cholangiocytes

SMCC

small mouse cholangiocytes

PBC

primary biliary cholangitis

PSC

primary sclerosing cholangitis

S100A4

S100 calcium binding protein A4

TIMP-3

tissue inhibitor of metalloproteinases-3

EP300

E1A Binding Protein P300

p16

cyclin-dependent kinase inhibitor 2A

PAI-1

plasminogen activator inhibitor-1

EGR1

Early growth response protein 1

MDR2

Multidrug resistance protein 2

LCM

Laser capture microdissection