Hepatic progenitor cells play a major role in regenerating diseased liver. In rodents, progenitors forming hepatocytes or cholangiocytes are identified by the stem cell marker Thy-1. The aim of this study was to ascertain whether progenitor cells expressing Thy-1 could be identified in human fetal liver. Midtrimester human fetal liver was immunostained for Thy-1, cytokeratins 18 and 19, vimentin, CD34, CD45, and fibrinogen. Thy-1+ and Thy-1+CD34+ populations were purified using fluorescence-activated cell sorting (FACS). Immunofluorescence and mRNA expression were used to examine the bipotential nature of purified stem cells. We found that Thy-1+ cells were concentrated in portal tracts but were also scattered in parenchyma. In FACS-prepared cells, 0.18–3.08% (median 0.65%, n = 14) of cells were Thy-1+. Immunophenotyping revealed that some Thy-1+ cells coexpressed cytokeratins 18 and 19, others, fibrinogen and cytokeratin 19. RT-PCR demonstrated that Thy-1+ cells expressed mRNA for Thy-1, cytokeratin 18, and cytokeratin 19, and Thy-1+CD34+ cells expressed mRNA for α-fetoprotein, transferrin, and hepatocyte nuclear factor-4α. Thy-1+ cells were identified in fetal liver. These cells expressed several lineage markers, including coexpression of biliary and hepatocellular proteins and mRNA. These data suggest that Thy-1 is a marker of liver stem cells in human fetal liver.
- stem cells
in response to major liver injury, a population of putative stem cells in the periportal region is mobilized to contribute to liver repair (9, 23). These progenitor cells, which are positive for cytokeratin 19 (CK19) and α-fetoprotein (α-FP), can differentiate into either hepatocytes or biliary epithelium. Human hepatic stem cells are central to future therapies in liver disease. However, liver stem cells in humans remain uncharacterized. In rodents, liver stem cells have been termed “oval cells” because of their large nucleus to cytoplasm ratio and an ovoid nucleus (2, 9, 12). These oval cells have been shown to express the stem cell markers Thy-1 (CD90), CD34 and Sca-1, along with liver-specific markers, including α-FP, γ-glutamyl transpeptidase, and CK19 (25, 26). There is evidence that oval-type cells exist in rat fetal liver (11) and in human adult liver (4, 30), but there are no data in human fetal liver. Hepatic regeneration is postulated to involve liver stem cells (4, 30) and circulating bone marrow stem cells (3, 14, 18, 24, 31, 32). Notably, the fetal liver contains hepatic and hematopoietic stem cells (HSCs) in close proximity, both of which may be Thy-1+. Adult liver stem cells may, therefore, represent fetal hepatic or HSCs rendered dormant until major liver injury provides a stimulus to proliferation. Regardless of origin, there is undoubtedly a progenitor cell compartment capable of replenishing hepatocytes and biliary epithelium. This cell population could prove an invaluable therapeutic tool.
Human liver stem cell studies so far have identified CD34+ and c-kit+ (stem cell markers) cells in regenerating adult liver (6) and CD34+ cells in fetal liver (22), which coexpress liver-specific cytokeratin markers. Despite the volume of literature in animal studies relating to Thy-1 as a stem cell marker, no studies have investigated whether Thy-1+ cells exist in human fetal liver or whether such cells exhibit the multilineage characteristics of liver stem cells. The aim of this study was to ascertain whether stem cells expressing the oval cell marker Thy-1 could be identified in human fetal liver and, further, to assess whether these cells coexpress hepatocyte and biliary markers consistent with a stem cell phenotype.
MATERIALS AND METHODS
Human Fetal Liver Cell Isolation and Culture
Livers from midtrimester therapeutic terminations of pregnancy were obtained from the Royal Infirmary of Edinburgh with informed consent from patients and approval of the Local Research Ethics Committee. Livers were collected in ice-cold Williams' E medium (WME; all culture media, additives, and solutions obtained were from GIBCO Invitrogen, unless otherwise stated), dissected free of fibrous material, and digested with collagenase type II (Worthington Biochemical, Lakewood, NJ) in Hanks' balanced salt solution. Digestion was halted with DMEM/10% FCS. Cells were centrifuged and resuspended in culture medium (WME supplemented with 10% FCS, 50 U/ml penicillin, 50 μg/ml streptomycin, 2 mM glutamine and Insulin-Transferrin-Selenium-X). Viability was determined with 0.2% trypan blue using a Neubauer hemacytometer. Cells were cultured at 37°C in air-5% CO2 in either polystyrene flasks (Corning, Corning, NY) or eight-well chambered glass slides (Lab-Tek, Naperville, IL). Thy-1+ cells were cultured with 20 ng/ml stem cell factor, 10 ng/ml thrombopoietin, and 20 ng/ml Flt3-ligand (PeproTech, Rocky Hill, NJ).
Fluorescence-Activated Cell Sorting
Labeling of cells with anti-Thy-1 (BD Pharmingen) and mouse monoclonal IgG1 anti-CD34-FITC conjugate (BD Pharmingen) was carried out at a concentration of 20 × 106 cells/ml of medium. Mouse IgG1 and mouse IgG1-FITC conjugate were used as isotype controls. Secondary labeling for Thy-1 utilized goat anti-mouse R-phycoerythrin conjugate (DakoCytomation). Thy-1+ and Thy-1+CD34+ sorted cells were subsequently collected in medium on ice after fluorescence-activated cell sorter (FACS) analysis by FACSVantage SE (Becton-Dickinson, San Jose, CA).
Immunohistochemistry of Fetal Liver Sections
Tissue was-snap frozen (liquid N2), embedded in optimum cutting temperature (Bayer, Newbury, UK) at −20°C, and sectioned at 4 μm (Leica Cryostat CM1850; Milton Keynes). Sections were collected on superfrost slides (VWR, Glasgow, UK), fixed in acetone (−20°C; unless otherwise stated, all chemicals from Sigma, St. Louis, MO), and stored at −70°C. Slides were thawed, and mouse monoclonal IgG1 anti-Thy-1 antibody (BD Pharmingen, UK) or control mouse IgG1 was applied to tissue sections and localized with rabbit anti-mouse IgG horseradish peroxidase (HRP) conjugate and the DAB Substrate Chromagen System (Vector, UK). Sections were counterstained with hematoxylin, dehydrated and mounted in Pertex (CellPath, Powys, UK).
Formalin-fixed paraffin-embedded sections.
Livers were fixed in 4% buffered paraformaldehyde for 24–48 h, embedded in paraffin, and cut at 5 μm. After dewaxing and rehydration, endogenous peroxidase activity was blocked by 3% H2O2/H2O. Antigen retrieval was achieved by heating in citrate buffer. Slides were incubated with mouse monoclonal IgG1 anti-CK18, CK19, CD34, vimentin, leukocyte common antigen (CD45), or negative control antibody (mouse IgG1; all antibodies from DakoCytomation, unless otherwise stated). Rabbit anti-mouse IgG-HRP conjugate was added and visualized as above. Slides were counterstained, processed, and mounted as previously described.
Immunohistochemistry and Immunofluorescence of Cultured Cells
Immunohistochemistry for cell-surface Thy-1 was followed by intracellular immunofluorescence staining. Cells were fixed with buffered 0.5% paraformaldehyde, endogenous peroxidase activity was blocked with 3% H2O2/H2O for 5 min, and slides were then incubated with mouse IgG1 anti-Thy-1 primary antibody or control antibody (mouse IgG1). After washing, rabbit anti-mouse IgG-HRP conjugate was added and visualized as previously described. Having labeled surface Thy-1, cells were permeabilized with 0.1% Triton X-100, followed by incubation with one of the following primary mouse monoclonal antibodies: anti-cytokeratin 18 (CK18), anti-CK19, anti-vimentin, anti-leukocyte common antigen (CD45), anti-proliferating cell nuclear antigen, or rabbit polyclonal anti-fibrinogen. Blocking solution alone, mouse IgG1, and normal rabbit serum were used as negative controls. Positive primary antibody binding was determined by incubation with either goat anti-mouse IgG tetramethyl rhodamine isothiocyanate conjugate (Sigma), Alexa Fluor 488 and 647 F(ab′)2 fragment of goat anti-rabbit IgG, or Alexa Fluor 568 F(ab′)2 fragment of goat anti-mouse IgG (Molecular Probes, Eugene, OR). Nuclear counterstaining was achieved by 5-min incubation with Hoechst 33258 (10 μg/ml; Sigma) for fluorescent microscopy or TO-PRO-3 (Molecular Probes) for confocal microscopy. Cells were examined under fluorescence and confocal microscopy (Leica Microsystems, Wetzlar, Germany).
Total RNA was isolated from Thy-1+ and Thy-1+CD34+ adherent cells in tissue culture flasks using the manufacturer's recommended protocol for TRIzol Reagent (Invitrogen Life Technologies, Paisley, UK). The concentration and purity of the extracted RNA was determined using an Ultrospec 2000 spectrophotometer (Pharmacia Biotech, Cambridge, UK). The conversion of RNA to cDNA was carried out using a Reverse Transcription Kit from Promega (Madison, WI). The resultant cDNA was stored at 4°C for subsequent use in PCR amplification. PCR was carried out on the synthesized cDNA using reagents obtained from Promega. Total RNA or nuclease-free water was substituted for cDNA to act as negative controls, and primers for cytochrome B and β-actin were used as positive controls. Primers investigated were Thy-1, CD34, c-kit, CK18, CK19, α-FP, albumin, transferrin, and hepatocyte nuclear factor-4α (HNF-4α; TAGN, Newcastle, UK). PCR products were separated by running in 1.6% agarose gel, which was analyzed using Quantity One Gel Doc 2000 (Bio-Rad, Hercules, CA).
Fetal Liver Cell Isolation
Fourteen human fetal livers with a median gestational age of 16 wk (range 14–19 wk) were studied. Cell yield ranged from 338.1 to 2,502 × 106 cells, and viability of the freshly isolated fetal liver cells ranged from 90.6 to 99.7% (Table 1).
Immunohistochemistry of Liver Sections
Thy-1 expression was detected in endothelial cells, portal tract mesenchymal tissue, and scattered cells throughout the parenchyma (Fig. 1A). Vimentin is also expressed in these cells and is strongly expressed in bile duct epithelium (Fig. 1B). CK18 is expressed in bile duct epithelium and in parenchymal hepatocytes (Fig. 1C), whereas CK19 staining is specific to the ductal plate and biliary epithelium (Fig. 1D). CD34 expression is principally confined to the endothelium (Fig. 1E) both in the portal tract and in parenchyma. A few CD45+ cells are identified in the portal tract, but most are scattered throughout the parenchyma. CD45 was not present in biliary epithelium or endothelium (Fig. 1F).
Immunohistochemistry of Nonsorted Human Fetal Liver Cells
At the time of primary cell isolation (day 0), Thy-1+ cells were identified by peroxidase immunohistochemistry of nonsorted cells (Fig. 2). After culture in chambered slides for 7 days, 77.7% of cells expressed Thy-1 (compared with 0.18–3.08% by FACS analysis on day 1l; Table 2). This could represent proliferation of fibroblasts, since 79.7% of the Thy-1+ cells coexpressed the mesenchymal marker vimentin (Fig. 3A). However, epithelial cells were noted to be vimentin positive in tissue sections (Fig. 1B), and Thy-1+ cells coexpressing the hepatocyte marker CK18 (Fig. 3B) and the biliary epithelial marker CK19 (Fig. 3C) were also identified in significant proportions (Table 2). This implies vimentin is not simply a marker of mesenchymal cells, since there must be coexpression of vimentin and CK18 and CK19 in many fetal liver epithelial cells. Tissue cultures showed putative bipotential progenitor cells in the nonsorted population, as demonstrated by the expression of the hepatocyte marker fibrinogen and the biliary marker CK19 in the same cell (Fig. 4).
FACS for the Oval Cell Marker Thy-1
FACS analysis of human fetal liver cells on day 1 (Fig. 5) consistently demonstrated the presence of Thy-1+ cells, from 0.18–3.08% (Table 1). There was no correlation between the percentage of Thy-1+ cells and fetal age, weight, cell yield, or viability. FACS-sorted Thy-1+ cells were placed in chambered slides. Initially, cells had a small, round mononuclear morphology (Fig. 6A) and subsequently underwent rapid clonal expansion (Fig. 6B) followed by differentiation, resulting in a variety of mature cell phenotypes (Fig. 6, C and D). Some cells in particular took on an oval cell appearance (Fig. 6D). FACS was used to investigate coexpression of Thy-1 with another stem cell marker, CD34. A subpopulation of Thy-1+CD34+ cells was identified (0.44–2.04%, median 1.22%, of total cells, Fig. 5) and characterized further by RT-PCR (see below).
Immunohistochemistry of Thy-1+ Sorted Cells
Purified Thy-1+ cells were cultured in chambered slides. Many of the Thy-1+ cells coexpressed the mesenchymal marker vimentin (Fig. 7A). However, Thy-1+ sorted cells were also identified, which showed coexpression for CK18 (Fig. 7B) and CK19 (Fig. 7C). Bipotential cells, which stained positive for both CK19 (bile duct) and fibrinogen (hepatocyte), were observed (Fig. 8). Positive labeling of cell nuclei for proliferating cell nuclear antigen also demonstrated that these mature Thy-1+ cells were dividing in culture (Fig. 9).
A fraction within the Thy-1+ sorted population lost Thy-1 expression after 7 days in vitro. Figure 10A demonstrates Thy-1+ cells coexpressing CK19 among a group of smaller Thy-1 negative cells. Subsequent immunohistochemistry on the same population of cells revealed they expressed leukocyte common antigen (CD45), a marker of cells of hematopoietic origin (Fig. 11). This supports evidence that Thy-1+-derived cells in human fetal liver can have a hematopoietic lineage as well as hepatic lineage.
RT-PCR on Thy-1+ and Thy-1+CD34+ Sorted Cells
RT-PCR carried out on Thy-1+ sorted cells placed in culture for up to 41 days demonstrated mRNA expression for Thy-1, CK18, and CK19 (Fig. 12), substantiating the protein expression shown by immunohistochemistry. In the Thy-1+CD34+ population, mRNA expression for CD34, CK18, α-FP, transferrin, and HNF-4α was identified (Fig. 13).
The objective of this paper was to determine whether Thy-1+ cells were present in human fetal liver and whether these cells demonstrated multiple lineage markers consistent with a progenitor phenotype. The data show that human fetal liver contains cells that are positive for the stem cell marker Thy-1. These cells coexpressed hepatocellular and biliary proteins and mRNA. These findings demonstrate that Thy-1+ cells in human fetal liver correspond to a stem cell phenotype.
Hepatoblasts of the developing fetal liver give rise to parenchymal hepatocytes and biliary epithelial cells (28, 15). In this regard, they are analogous to oval cells, the bipotential progenitor cells of the canals of Hering. In adult human liver, progenitor cells coexpressing the stem cell marker c-kit, the biliary marker CK19, and the hepatocyte markers HepPar 1 and α-FP have been demonstrated in areas of massive necrosis within the canals of Hering (30). Thy-1 status has not been determined in adult human liver. However, published data show that purified human pediatric liver cells expressing CD34 may differentiate into biliary epithelial cells (6). Currently, the origin of stem cells contributing to human liver regeneration is debated, since the progeny of circulating stem cells from bone marrow may play a role (7). In fetal liver, stem cells corresponding to liver and hematopoietic compartments are both present, raising the possibility that stem cells are enriched in this preparation. Positive labeling within the population of Thy-1+ sorted cells for the hematopoietic marker leukocyte common antigen (Fig. 11) provides further evidence for the developmental link between future bone marrow stem cells and liver stem cells (12).
Thy-1+ cells were clearly visualized in portal tracts and parenchyma in this study (Fig. 1A), and these cells showed bilineage potential by their coexpression of a hepatocyte marker (fibrinogen) and biliary cell marker (CK19) in the same cell (Fig. 8). The identification of a Thy-1+ phenotype in the fetal liver indicates a possible relationship to oval cells (25). Thy-1 is a cell surface protein whose function is poorly understood (5, 16, 19, 20, 33). It is only expressed in liver by activated oval cells after induction by injury. It is most commonly recognized as a marker of HSCs, often in conjunction with CD34, being found on ∼1–4% of HSCs in human bone marrow (5). This study showed a fraction of Thy-1+CD34+ cells that expressed liver epithelial mRNA, suggesting that, in fetal liver, this dual-positive phenotype corresponds to a liver progenitor, rather than a hematopoietic cell. Interestingly, immunohistochemistry suggested that the endothelium of the portal vein was Thy-1+CD34+, as was the lining of vascular structures undergoing angiogenesis in the portal tracts. It may be that the widespread Thy-1+CD34+ staining in the portal tract corresponds to a primitive mesenchymal compartment capable of generating mesodermal (capillary) and epithelial (parenchymal) tissues, reinforcing the potential of these cells as a stem cell compartment.
In nonsorted cells, Thy-1+ cells appeared to predominate after 7 days in culture (Fig. 3). This could represent overgrowth of fibroblasts, since some cells stained for vimentin, and Thy-1 expression on fibroblasts has been identified previously (1). However, many Thy-1+ cells stained for CK18 or CK19 (Fig. 3, B and C, and Table 2), indicating an epithelial lineage. Within this nonsorted population, some cells were additionally dual-labeled with anti-fibrinogen and anti-CK19 (Fig. 4A). Previous studies have taken this pattern of staining of a hepatocyte and biliary marker within the same cell as evidence of bipotential progenitor cells (15, 25, 34).
The concept of multipotent adult progenitor cells from bone marrow capable of liver, lung, and gut differentiation was introduced by Jiang et al. (18). Injection of adult mouse bone marrow HSCs in a mouse model of fatal hereditary tyrosinemia rescued the mouse by regenerating normal liver cells (21). Human studies of bone marrow transplants from male donors to female recipients and liver transplants from female donors to male recipients detected the presence of Y chromosome-positive epithelial cells in the livers of recipients (8, 29). These data support the view that circulating stem cells can regenerate liver tissue in humans and mice. However, cell fusion may explain bone marrow regenerating liver, and this has introduced a note of caution to temper initial findings (31, 32).
To determine the ability of human stem cells to contribute to liver, Fiegel et al. (10) isolated CD34+ HSCs from human bone marrow, as did Jang and colleagues (17). Cells were incubated with hepatocyte growth factor or cocultured with injured liver, respectively, after which HSCs expressed albumin and CK19 mRNA (10) or liver-specific proteins and mRNA, including CK18, CK19, and albumin (17). These data confirm the capacity for human HSCs to differentiate along a liver pathway. Our data also strongly agree with these findings. For example, liver-derived Thy-1+ or Thy-1+CD34+ cells (phenotypes corresponding directly to bone marrow stem cells) expressed CK18, CK19, transferrin, α-FP, and HNF-4α mRNA. Cells were immunopositive for fibrinogen, CK18, and CK19.
Treatment of hepatic failure requires liver transplantation. The shortage of donor organs available is well documented. Hepatocyte transplantation has been carried out in selected cases with some clinical benefit (13), but this technique also relies on donor organs. Liver progenitor cells, capable of expansion and differentiation into hepatocytes or biliary epithelium, would therefore be an invaluable therapeutic tool. However, immunological rejection would still be an issue. The fetal liver is a rich source of stem cells (5, 27), which are shown in this work to express liver epithelial markers. Such a model system could help develop strategies to produce liver cells from a patient's own bone marrow, without the complication of immunological rejection.
The demonstration that Thy-1+ cells express hepatocytic and biliary markers correlates with rodent studies which show that oval cells express Thy-1 (11, 25). Oval cells are a reactive response to liver injury in adult animals, whereas these studies are directed to cells in the developing liver. However, the developmental origin of oval cells in adults is unclear. This paper provides significant evidence that oval cells may be present in the liver from the earliest stages of human development. It is possible that these cells, lying dormant in the adult liver, are the source of Thy-1+, oval, hepatic progenitor cells.
This study was supported by Peel Medical Research Trust, Mason Medical Research Foundation, Lothian University Hospitals National Health Service Trust Research and Development, Tenovus Scotland, and a British Journal of Surgery Research Bursary.
We are grateful for the assistance of Dr. R. A. L. Bayne, Dr. S. M. da Silva, and Dr. S. Coutts, Medical Research Council Centre for Reproductive Biology, University of Edinburgh, for the provision of liver samples, A. Shukla for assistance in cell isolation, and Dr. S. MacCall for cell sorting.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society