Implication of hepatic stem cells in functional liver repopulation


  • Bruno Christ,

    Corresponding author
    1. Translational Centre for Regenerative Medicine (TRM), University of Leipzig, Philipp-Rosenthal-Straße 55, D-04103 Leipzig, Germany
    2. Department of Visceral, Transplantation, Thoracic and Vascular Surgery, Applied Molecular Hepatology Laboratory, University Hospital Leipzig, Liebigstraße 21, D-04103 Leipzig, Germany
    • Department of Visceral, Transplantation, Thoracic and Vascular Surgery, Applied Molecular Hepatology Lab, University Hospital of Leipzig, Liebigstraße 21, D-04103 Leipzig, Germany
    Search for more papers by this author
  • Sandra Pelz

    1. Department of Visceral, Transplantation, Thoracic and Vascular Surgery, Applied Molecular Hepatology Laboratory, University Hospital Leipzig, Liebigstraße 21, D-04103 Leipzig, Germany
    Search for more papers by this author


The liver has an enormous potential to restore the parenchymal tissue loss due to injury. This is accomplished by the proliferation of either the hepatocytes or liver progenitor cells in cases where massive damage prohibits hepatocytes from entering the proliferative response. Under debate is still whether hepatic stem cells are involved in liver tissue maintenance and regeneration or even whether they exist at all. The definition of an adult tissue-resident stem cell comprises basic functional stem cell criteria like the potential of self-renewal, multipotent, i.e. at least bipotent differentiation capacity and serial transplantability featuring the ability of functional tissue repopulation. The relationship between a progenitor and its progeny should exemplify the lineage commitment from the putative stem cell to the differentiated cell. This is mainly assessed by lineage tracing and immunohistochemical identification of markers specific to progenitors and their descendants. Flow cytometry approaches revealed that the liver stem cell population in animals is likely to be heterogeneous giving rise to progeny with different molecular signatures, depending on the stimulus to activate the putative stem cell compartment. The stem cell criteria are met by a variety of cells identified in the fetal and adult liver both under normal and injury conditions. It is the purpose of this review to verify hepatic stem cell candidates in the light of the stem cell definition criteria mentioned. Also from this point of view adult stem cells from non-hepatic tissues such as bone marrow, umbilical cord blood or adipose tissue, have the potential to differentiate into cells featuring functional hepatocyte characteristics. This has great impact because it opens the possibility of generating hepatocyte-like cells from adult stem cells in a sufficient amount and quality for their therapeutical application to treat end-stage liver diseases by stem cell-based hepatocytes in place of whole organ transplantation. © 2012 International Society for Advancement of Cytometry

From a formal point of view a tissue stem cell gives rise to progeny, which are the progenitor cells of mature differentiated cells of this tissue. Under resting conditions a stem cell is quiescent residing in the stem cell niche. Upon activation proliferation of the stem cell yields both progeny maintaining stem cell features (self-renewal) and progeny capable of proliferation and differentiation along the lineage(s) of the tissue to be generated, which requires prior commitment of the progenitor cells in at least one or more lineages (Fig. 1). These two processes are balanced in tissues with high turnover to maintain the steady state of tissue regeneration on the one hand and preservation of the stem cell pool on the other, both in normal tissue homeostasis and repair after injury. In the normal liver, tissue turnover is very slow, ranging from 0.0012% to 0.01% mitotic cells to the total hepatocyte number (1) leading to renewal of the parenchyma in about one year (2). It might therefore, not be surprising that stem cells in the healthy liver are rare if present at all. Knowledge about the “hepatic stem cell” emerged through investigations of liver embryonic development assuming that progeny of or the stem cell itself might reside in the adult liver. True hepatic stem cells should have the ability to reconstitute liver tissue after injury, in addition to the abilities for self-renewal and multiple differentiation. Therefore, the following review will reflect on the actual debate on the existence of the liver stem cell and on candidate hepatic stem cells during liver development, liver regeneration after injury and the capacity to functionally repopulate the host liver after transplantation of the respective cell type.

Figure 1.

Generation of mature cells from adult tissue stem cells. Stem cells reside in the stem cell niche. Upon activation the stem cell proliferates and gives rise to progeny with either stem cell features (self renewal) or progeny, the progenitor cells, which are committed to differentiate into the cell type(s) of the tissue. In the liver both hepatocytes and cholangiocytes emerge from the progenitor cells.

Stem Cells and Progenitors in the Liver

In the liver, progeny of the putative liver stem cell appear after massive injury in the portal areas of the liver lobule, which have been identified as the Canals of Hering, the most proximal parts of the intrabiliary ductular system. In humans, tubular structures appear after huge injuries. This kind of stem cell response was termed “ductular reaction” (3, 4), which is equivalent to the appearance and proliferation of oval cells in rodents (see below). Cells forming the ductules may typically be identified by the expression of cytokeratins 7 and 19, EpCAM and NCAM (epithelial and neural cellular adhesion molecule) and CD133 (5–7). They are heterogeneous in size ranging from 6 to 40 μm, representing the size of mature small cholangiocytes and hepatocytes, respectively (4). Progeny of these cells emerge both in the acutely and the chronically injured liver, the bipotent transit amplifying cell population, giving rise to committed progenitors of mature cholangiocytes as well as hepatocytes. The phenotype of the transit amplifying progenitors resembles that of fetal hepatoblasts. However, under resting conditions they comprise only less than 0.01% of the adult liver parenchyma, they make up more than 80% of the fetal liver. There is some debate as to whether the progenitors of the transit amplifying compartment represent “the hepatic stem cell” (8, 9). Their surface marker phenotype has recently been described including expression of EpCAM, NCAM, CD133, and cytokeratins 8/18/19 but no or negligible amounts of albumin, a prominent marker of differentiated hepatocytes, or alpha-fetoprotein. They comprise 0.5–2.0% of the hepatic parenchyma and in contrast to the bipotent hepatoblasts display pluripotent differentiation capacity (9, 10).

There is a debate about the existence of “the” liver stem cell in the adult. Quiescent hepatic progenitor cells (HPCs) can be localized along the stem cell niche of the Canals of Hering, potential relicts of the fetal ductal plate in the adult liver (8). They express CK7, NCAM, and CD133. Upon challenge such as toxic injury the progenitors become activated and start to proliferate building the reactive ductules still expressing CK7, NCAM, and CD133. Cells making up the ductules undergo commitment into either the cholangiocyte (CK7+, NCAM+, CD133−) or the hepatocyte (CK7+, NCAM−, CD133−) lineage, which then further differentiate into either mature cholangiocytes or hepatocytes (7, 11, 12). Whether or not the quiescent HPCs are identical to the liver stem cell as described above or are progeny of as yet unidentified adult liver stem cells remains to be clarified (12).

How to Find the Liver Stem Cell?

There is an obvious problem: The liver stem cell has not yet been found or at least defined and agreed upon. The progenitor cells both in the adult and embryonic liver feature stem cell characteristics, thus it is conceivable that they are ultimately progeny of a stem cell appearing during embryogenesis for example from the outgrowing embryonic foregut (see below). The question is whether this type of cell persists during organogenesis in the embryonic or even the adult liver. Which marker should then be applied to identify the liver stem cell unequivocally? Is a cell expressing progenitor cell markers under injury conditions “the liver stem cell” or is it an adult tissue cell expressing that specific markers as the result of the specific challenge under injury conditions? A discussion has emerged concerning methods to be employed for the identification and functional characterization of the stem cell: either isolation and in vitro expansion and clonal characterization or lineage tracing using genetic labeling in vivo (13). Classically, markers were defined to specify the stem cell and definition criteria including functional features as mentioned above were employed to unravel the nature of the putative stem cell. The oval cells were specified by the expression of markers including cytokeratins 7 and 19, OV6, α-fetoprotein and CD90 besides others. Yet, these markers are not exclusively expressed on oval cells thus questioning the specificity of both the marker and the cell identified by this marker (14, 15). However, markers expressed are both biliary and hepatocytic, thus describing the oval cells' bipotent precursor character of the adult parenchymal cells. In humans, hepatocyte precursor markers such as the cytokeratins 7 and 19 and NCAM, seem to be expressed on different cell types of the ductular reactions, indicating the heterogeneity of the stem cell response and thus perhaps pointing also to different stem cell populations in humans (3, 5, 8). This highlights the obvious problem whether the unequivocal identification of adult hepatic stem or progenitor cells is lacking specific markers. Indeed, a recent approach to isolate adult liver progenitor cells using antigenic cell surface marking and FACS analyses revealed that both the normal mouse liver and livers after activation of the oval cell response yielded a subpopulation of cells exclusively expressing the profile CD45−/CD11b−/CD31−/MIC1−1C3+/CD133+/CD26−, which was, however, highly enriched in injured livers. The cell fraction expressed Sox9 and lineage tracing experiments with stable expression of Sox9 demonstrated that the cell fraction both from normal livers and enriched after oval cell challenge represents a progeny population of Sox9-positive precursor cells, thus corroborating the existence of hepatic stem cells in the adult liver (16, 17). In the embryonic liver hepatoblasts are the bipotent precursor cells of the adult hepatocytes and cholangiocytes. A subset of hepatoblasts expresses Sox9, which in the developing liver marks the cholangiocyte lineage, whereas Sox9-negative hepatoblasts are prone to hepatocyte lineage differentiation (18–21). Recent work confirmed that the ductal plate in the embryo is lined by the Sox9-expressing biliary precursor cells giving rise to cholangiocytes, periportal hepatocytes, and liver progenitor cells in the adult liver (22, 23), thus again corroborating a precursor/progeny relationship of embryonic hepatoblast-derived precursor cells persisting in the adult liver (8, 24).

The functional characterization in vitro of putative stem cells in the liver is mandatory defining both their origin and fate unequivocally. This requires highly enriched cell populations, which might then be cultured and investigated in terms of clonal growth, maintenance of stemness or acquisition of differentiation under specified inducing conditions. In addition, as mentioned above, a liver stem cell should be able to functionally reconstitute the hepatic parenchyma efficiently after injury. Liver progenitor cells might be isolated from the adult liver by FACS sorting using progenitor cell markers such as CD133 and EpCAM or progenitor cell markers in combination with lineage markers such as cytokeratin 19 (CK19) delineating biliary commitment of the precursor cell population (6, 25). Because it is still not clear whether progenitor cells appearing in the adult liver emenate from a common stem cell the term “liver stem cell” will be used in the following only to delineate functional stem-like characteristics of a liver progenitor cell population. Generally, the term “progenitor cell” will be used to describe the precursor/product relationship between cell populations appearing early in liver tissue homeostasis under normal and injury conditions and the final differentiated cells, the hepatocytes and cholangiocytes, respectively.

Regulation of Embryonic Liver Development

The development of the embryonic liver and the generation of hepatocytes from embryonic progenitor cells during hepatic development might reflect the differentiation from hepatocyte progenitor cells to hepatocytes during liver regeneration at both the molecular and cellular level in the adult. Hence, it is worthwhile considering the emergence of the early liver from the ventral foregut and its growth factor regulation.

During embryogenesis, the ventral foregut comes into close proximity to the cardiac mesoderm (Fig. 2). This apposition is important because the cardiac mesoderm liberates molecular signals, which specify the endoderm in this area to provide the prospective liver progenitor cells (26). By embryonic tissue explant experiments it has been shown that the main factors mediating these signals are fibroblast growth factors (FGFs) (27). Yet, the role of FGFs seemed not to be exclusive, because hepatic induction in the foregut endoderm required also BMPs (bone morphogenetic protein), growth factors of the TGFβ (transforming growth factor) superfamily provided by the septum transversum mesenchyme (28). The hepatic specification of the embryonic endoderm is paralleled by the crucial expression of transcription factors GATA4, liver-enriched HNF (hepatic nuclear factor) transcription factors and Foxa (forkhead box A) proteins driving the expression of hepatocyte-specific genes in differentiated hepatocytes (29–31). Liver development starts with budding of the foregut epithelium into the mesenchymal septum transversum. Proliferation of the endoderm epithelial cells finally results in migration into the septum transversum mesenchyme, which penetrates the budding endoderm and later contributes liver endothelial and perisinusoidal cells (Fig. 3). At this point in time invading cells, called hepatoblasts, grow into the mesenchymal tissue in cord-like structures, finally loosing contact with the endodermal layer and start to express hepatocyte-specific genes such as α-fetoprotein (AFP) and albumin. It is interesting to note that this process involves the transition from an epithelial to a mesenchymal character of the cells, which requires the rearrangement of the extracellular matrix fixing the hepatoblasts to the foregut endoderm basement membrane (26, 32–34). During this process specific homeodomain transcription factors seem to play a crucial role. Defective Hex (hematopoietically expressed homeobox) expression in the mouse embryo leads to budding in the region of the liver-specifying endoderm and expression of albumin indicating hepatoblast formation, though cells fail to proliferate and migrate into the septum transversum (35, 36). Similar consequences were observed in mice with defective Prox1 (prospero-related homeobox) expression. Mutant cells in the hepatic endoderm proliferated but failed to migrate out of the epithelium into the septum transversum. This might be attributed to the excess expression of matrix proteins collagen IV, laminin, and of E-cadherin favoring adherence to the basement membrane and close interactions between the hepatoblasts in the endodermal epithelium. Interestingly, in the mutant embryos liver lobes were to be found in the septum transversum but void of hepatoblasts (37). This indicates the strong implication of matrix remodelling at the site of hepatoblast migration into the septum transversum and mesenchyme-borne determinants in liver structural development. This is also corroborated by the finding that in Flk1−/− [encoding the vascular endothelial growth factor receptor 2 (VEGFR2)] mice, in which angiogenesis is deteriorated due to immature endothelial cells, liver gene expression is switched on in the hepatoblasts of the liver bud but no migration of the cells into the mesenchyme is observed. Because the budding hepatic endoderm at that time is surrounded by angioblasts in normal mice, which later become mature endothelial cells, it may be anticipated that the early endothelial cells provide crucial signals for hepatoblasts to proliferate and migrate into the septum transversum mesenchyme (38). Whereas FGFs and BMPs have an impact on the specification of the hepatic endoderm, HGF (hepatocyte growth factor) and TGFβ (transforming growth factor) seem to play a critical role in the maintenance of hepatoblast bipotency and lineage specification of hepatocyte or biliary differentiatiation [see below, (31)].

Figure 2.

Growth factor signalling during early liver morphogenesis. During liver development from the endoderm the action of fibroblast growth factor (FGF) and bone morphogenic protein (BMP) members derived from the cardiac mesoderm and the mesenchyme of the septum transversum promote competent cells of the endoderm to gain specification to hepatic lineage differentiation. The illustration shows a mouse embryo featuring the structures important for liver development (right). A higher magnification depicts the apposition of the cardiac mesoderm and the ventral endoderm of the foregut (left), which enables the paracrine delivery of hepatotropic morphogens and growth factors.

Figure 3.

Liver bud formation during embryogenesis. Cells (medium grey) from the endoderm grow out of the primordial endoderm and enter the mesenchyme of the septum transversum (arrows). This migration requires remodelling of the extracellular matrix allowing for the migration of epithelial-organised endodermal cells into the septum transversum. Stromal cells of the septum transversum provide non-hepatocyte cells such as the perisinusoidal cells. Outgrowing hepatoblasts are interspersed with primitive endothelial cells (light grey) later contributing the endothelial cells of the sinusoids.

Hepatoblasts express hepatocyte progenitor cell markers such as intermediate filament cytokeratins 14 (CK14), 8 (CK8), and 18 (CK18), gamma-glutamyltranspeptidase (GGT), and the placental form of glutathione-S-transferase (GST-P), some of which are also expressed in the oval cells deriving from the progenitor cell compartment in the adult rodent liver. In the following period of development, cells express markers specifying both the hepatocyte (AFP, albumin) and the biliary lineage (CK7, CK19) (early, bipotent progenitor cells). Commitment to either of the lineages is then featured by the expression of cell-specific markers of hepatocytes and biliary cells, respectively, although the cells still proliferate massively (late, unipotent progenitor cells) (39–41).

Hepatoblasts isolated from mouse embryos expressed albumin only after incubation with HGF, and the albumin-expressing cells differentiated into hepatocyte and biliary lineages in vitro. While promoting hepatocyte specification, HGF prevented biliary development of the bipotent hepatoblasts (42, 43). But, Jagged/Notch signaling is crucial for biliary differentiation of the hepatoblasts (44, 45). For comprehensive reviews of liver development during embryogenesis and summary of current knowledge of the sophisticated molecular network involved in commitment, specification, and differentiation of liver progenitor cells please refer to (23, 46–48).

During liver development, Notch signaling is required for biliary specification of embryonic hepatoblasts. Wnt signaling drives the hepatoblasts to hepatocyte differentiation (49, 50). Using laser-captured microdissection after staining of hepatic progenitor cells with a cytokeratin 7 antibody and immunohistochemical characterization of human liver tissue from patients suffering from acute hepatitis, viral-induced liver cirrhosis or primary biliary cirrhosis it was shown that Wnt and Notch signaling are also active in the adult human liver to drive proliferation and differentiation of the progenitor cells into the hepatocyte or cholangiocyte lineage (7). Analyses of the ductular reaction in human chronic liver disease specimens showed that the activation of the Notch pathway was triggered by expression of the Notch ligand Jagged1 by myofibroblasts thereby promoting biliary differentiation of the progenitor cells and that the enhancement of Wnt3a expression in macrophages after uptake of hepatocyte debris and paracrine activation of Wnt signaling in neighboring hepatic progenitor cells specified their hepatocyte differentiation (11, 12). This demonstrates that factors driving lineage differentiation in embryonic hepatoblasts are also involved in hepatic progenitor cells specification in the adult liver. Therefore, it is feasible to conclude that both cellular and molecular mechanisms of hepatocyte progenitor cell fate determination in the embryo persist in the adult liver.

Rat fetal liver epithelial cells are capable of repopulation of the host liver after transplantation and to functionally reconstitute the host liver tissue (51, 52). Fetal hepatocytes expressing both hepatocyte and biliary markers have also been isolated from human fetal liver, which upon exposure to OSM differentiated into hepatocytes and biliary cells (53). Multipotent progenitor cells were derived from human fetal liver, capable of differentiation into hepatocytes but also into cells of other mesenchymal tissues such as bone and adipose tissue (54). Reports in literature demonstrate that the fetal liver harbors hepatic progenitor cells committed towards both hepatocyte and biliary cell differentiation. These cells may repopulate the liver after transplantation, thus fulfilling basic stem cell criteria.

Hepatic Progenitor Cells During Liver Regeneration After Partial Hepatectomy

Hepatocyte proliferation and differentiation during liver regeneration requires a plethora of regulatory events. HGF is mobilized during the first 3 h after partial hepatectomy by the remodeling of the extracellular matrix involving metalloproteinases (MMP) and their inhibitors. The increase in expression of the MMPs by hepatocytes is probably mediated by TNFα. The activation of HGF activates in turn its receptor c-Met within 30–60 min after partial hepatectomy, which is paralleled by the activation of the EGF receptor indicating a crosstalk between c-Met and EGF signaling and potential synergism. The increase in TNFα is accompanied by the increase in IL6, which both facilitate the activation of NFkB and Stat3, transcription factors important for entering the cell cycle and thus the proliferative response (55). Liver sinusoidal endothelial cells are obviously another source of HGF, which is released after stimulation with VEGF (vascular endothelial growth factor) (56). Despite the fact that hepatocytes express many growth factor receptors like those for PDGF, VEGF, or FGF the only mitogens for hepatocytes are HGF and the ligfands of the EGF receptor (HB-EGF, EGF, amphiregulin, TGFβ) (57, 58).

Interestingly, the proliferation of oval cells is also under the control of proinflammatory cyctokines. Overexpression of the TNF family member TWEAK (TNF-like weak inducer of apoptosis) stimulated oval cell, but not hepatocyte proliferation in the mouse liver through its receptor Fn14, a receptor of the TNF receptor family (59). Oval cell proliferation is markedly hampered in TNF receptor 1 knockout mice indicating the involvement of TNF signaling. IL6 and related cytokines such as leukemia inhibitory factor (LIF) and oncostatin M (OSM) seem to be involved in oval cell proliferation and hepatocyte differentiation (60, 61). Yet, replicative senescence is a critical parameter determining the regenerative capacity of hepatocytes, which is under the tight control of proinflammatory cytokines, growth factors, and hormones as well as regeneration in general. It has been argued that the coordinated action of priming signals, growth factors, and growth hormone are meeting at the point of FoxM1B gene expression, which was shown to be essential for hepatocyte growth (62–65).

After 2/3 partial hepatectomy the liver restores only the original size of the organ, which is accomplished by the proliferation of the hepatocytes requiring the individual hepatocyte to divide only 1.5-times. However, serial transplantation experiments in the albumin promotor-urokinase plasminogen activator (uPA) transgenic mouse revealed nearly unlimited regenerative capacity of hepatocytes. In this mouse model, the expression of uPA activates the precursor plasminogen to the active protease plasmin, which in turn causes severe hepatocyte damage and perinatal lethality (66). In the end, mice escaped death and analysis of the liver revealed repopulation of the liver by hepatocytes, which obviously displayed a survival advantage over hepatocytes bearing the transgene. Isolation of the surviving cells and transplantation into the livers of transgenic mice resulted in efficient repoulation of the diseased host organ by the transplanted hepatocytes, thus rescuing the lethal phenotype (67, 68). In the mouse model of human hereditary tyrosinemia type I, the genetic knockout of fumaroylacetoacetate hydrolase (FAH) caused the accumulation of toxic intermediates in tyrosine metabolism, which in turn destroyed hepatocytes thus representing a strong growth stimulus for transplanted healthy wildtype hepatocytes. In addition, transplanted hepatocytes had a proliferative advantage over diseased host hepatocytes, which finally led to a nearly complete replacement of the genuine liver cells by the transplanted cells. Using this model, the stem cell-like character of adult hepatocytes has been demonstrated by serial transplantation of hepatocytes derived from mutant livers colonized with transplanted wildtype cells. It has been calculated that for 6 rounds of liver repopulation a minimum of 69 cell divisions would have been necessary (69–71). These experiments imply that adult hepatocytes have a high replicative and repopulation capacity, i.e., they have the potential of self-renewal and functional tissue formation in vivo, ultimate features of a stem cell. In that sense, adult hepatocytes may be termed “unipotent stem cells”.

Animal models exist, which allow for the selective destruction of various liver cell subpopulations. These comprise intoxication by a single dose of allyl alcohol, which kills primarily hepatocytes surrounding the branches of the portal vein, carbon tetrachloride leading to destruction of the hepatocytes near the central vein, or ischemia/reperfusion injury causing randomly distributed hepatocyte necrosis and apoptosis in the liver parenchyma. Nevertheless, all of these injuries leave part of the hepatocytes intact, which take over the restoration of the damaged tissue by entering the mitotic response. This raises the question, whether the adult hepatocytes might be or feature characteristics of hepatic stem cells. Transplantation of hepatocytes into livers after partial hepatectomy yields a low repopulation rate by the transplanted cells if not a selective pressure for the transplanted cells creates a proliferative advantage over the host hepatocyte proliferation. Selective pressure to give transplanted cells a proliferative advantage over host cells may be achieved by treatment with retrosine (72), x-ray or ischemia and reperfusion (73–75). It has also been demonstrated that adult hepatocytes might differentiate into biliary cells reconstituting bile ducts in the liver after bile duct injury (76) and into pancreatic cells, both endocrine and exocrine (77, 78). Thus, adult hepatocytes may be regarded as hepatic stem cells in terms of self-renewal, multiple differentiation and repopulation potential.

Hepatic Progenitor Cells During Liver Regeneration After Massive Injury

D-galactosamine causes massive hepatocellular injury so that regeneration by hepatocytes is restricted. In rodents, proliferation and differentiation of the oval cells occurs in the area of the canals of Hering (1, 79, 80). These areas are considered to be the hepatic stem cell niche, though the oval cells are probably not representing “the liver stem cell”. Today almost general agreement has been reached that oval cells are the progeny of adult hepatic stem cells in the liver, though the nature of this stem cell still remains unclear (81). The spatial and temporal expression of hepatocyte and cholangiocyte lineage markers demonstrated that oval cells in the rat model of 2-acetylaminofluorene/partial hepatectomy treatment gave rise to hepatocytes and biliary cells (82, 83). Marker expression by oval cells is inconsistent indicating that oval cells may not be a homogeneous cell population. They share marker expression with biliary epithelial cells such as CK7, CK8, CK18, CK19, with fetal hepatoblasts (AFP, γ-glutamyltranspeptidase), and with hematopoietic cells (Thy-1, c-kit, CD34), which might explain the differences in their differentiation capacity (39, 40, 84). As with adult hepatocytes, the repopulation capacity of oval cells is low in livers not challenged to provide the transplanted cells with a selective growth advantage (51, 85). However, when transplanted into FAH-deficient mice, liver repopulation by oval cells was as efficient as with adult hepatocytes (86). Hence, due to their potential of self-renewal, bipotent differentiation, and functional tissue replacement, although only after substantial challenge of the host liver, oval cells may be considered as hepatic stem cells.

Hepatic Progenitor Cells from Nonhepatic Origin

The liver is the hematopoietic organ of the fetus. It is therefore possible that hematopoietic stem cells may reside in the liver of the adult. This assumption is corroborated by the finding that oval cells express hematopoietic markers like CD34, CD45, Thy-1, c-kit, and others, which are also expressed by bone marrow-derived hematopoietic stem cells (87). Hence, oval cells might be candidate progeny of hematopoietic stem cells in the liver, which has been shown in cross-sex or cross-strain bone marrow and whole liver transplantation experiments using a mouse model of 2-acetylaminofluorene/CCl4 intoxication. Even if conversion of bone marrow-derived cells to oval cells was observed, the functional repopulation capacity of the hepatocytes derived thereof was rather low questioning the physiological relevance of this finding (88). The responsiveness to the stromal-derived factor 1α (SDF-1α) and its receptor CXCR4 on oval cells after substantial hepatic injury, and on invading hematopoietic cells from the bone marrow supported the assumption that oval cells may descend from bone marrow-derived hematopoietic stem cells (89). However, using three different models of oval cell activation in the rat, i.e., D-galactosamine, retrorsine/partial hepatectomy, and 2-acetylaminofluorene/partial hepatectomy, no transdifferentiation of bone marrow-derived cells to oval cells was observed (90). Hence, in most of the animal models investigated so far, functional repopulation of the host liver by transplanted hepatocytes differentiated from hematopoietic stem cells is marginal. In the FAH-deficient mouse model, repopulation of a host liver by transplanted bone marrow-derived hematopoietic stem cells was nearly as efficient as with adult hepatocytes (91). Yet, hepatocytes were not transdifferentiated from bone marrow-derived stem cells, but were a fusion product of host hepatocytes and marrow-derived cells (92, 93). Hematopoietic myelomonocytic cells have been identified in the donor bone marrow as the major source of the host hepatocyte fusion partners (94, 95).

Besides hematopoietic stem cells rodent and human bone marrow harbors CD34- and CD45-negative cells of mesenchymal origin, capable of multiple differentiation (96), which can be propagated in vitro and differentiate into cells with a mesodermal or an endodermal phenotype. Transplantation of these cells, termed mesenchymal stem cells (MSC), led to engraftment of different organs and differentiation into organ-specific cell types (97). Evidence exists that a subpopulation of MSCs, the multipotent adult progenitor cells (MAPCs), feature hepatogenic and biliary differentiation potential, though in vivo hepatic tissue repopulation still awaits proof (98, 99). Recently, multipotent progenitor cells capable of mesenchymal lineage differentiation have been isolated from human fetal liver, which after transplantation into immunotolerant Rag2−/−γc−/− mice formed clusters of albumin expressing human hepatocytes in the host liver (54). Published work from our own group demonstrated that human MSC from bone marrow gained in vitro the characteristic morphology and selected functions of hepatocytes such as glycogen storage, urea synthesis, and activation of hepatocyte-specific gene promoters. After transplantation into livers of immunodeficient mice, these cells engrafted predominantly in the periportal region of the liver lobule. In situ, the cells continued to store glycogen and expressed phosphoenolpyruvate carboxykinase, connexin 32, albumin, and the human hepatocyte-specific antigen HepPar1, indicating that the transplanted cells retained prominent qualities of hepatocytes after their regional integration (Fig. 4) (100). MSC have been isolated not only from human bone marrow (99) but also from cord blood (101, 102), and adipose tissue (103), which were capable of hepatocyte differentiation in vitro. After hepatic transplantation adipose tissue-derived MSC cells continued to feature hepatocyte-specific functions in vivo and cells pre-differentiated in vitro into hepatocyte-like cells displayed a better repopulation efficacy as compared with undifferentiated MSC (104, 105). A typical mesenchymal marker set is expressed in undifferentiated MSC comprising CD13, CD29, CD44, CD54, CD90, CD105, and CD166 but lacking hematopoietic markers such as CD14, CD34, and CD45 (Fig. 5A). The hepatocytic differentiation in vitro changed morphology from a fibroblastoid into a polygonal shape typical for hepatocytes (Fig. 5B) and increased expression of functional hepatocyte markers like the periportal and perivenous marker enzymes phosphoenolpyruvate carboxykinase and glutamine synthase, respectively, as well as expression of the plasma protein α1-antitrypsin (Fig. 5C).

Figure 4.

Integration of human mesenchymal stem cell-derived hepatocyte-like cells into the mouse liver in vivo. Mesenchymal stem cells were differentiated into hepatocyte-like cells in vitro under specified culture conditions (179). Cells were then delivered to the livers of Pfp/Rag2-/- immunodeficient mice via intrasplenic application. 12 weeks after transplantation, animals were sacrificed and transplanted human cells detected immunohistochemically by the expression of human hepatocyte-specific HepPar1 (dark staining). hBM-MSCs were isolated from human bone marrow (hBM) aspirates of voluntary donors as approved by the Institutional Ethics Review Board of the University Hospital Leipzig. Written informed consent was obtained from all donors of hBM-MSCs. Animal experiments were approved by the regulatory authorities of the animal protection commissary of the university. Scale bar - 100 μm; (pv – portal vein, cv – central vein)

Figure 5.

Expression of mesenchymal markers on undifferentiated MSC and changes in morphology and gene expression after hepatocyte differentiation. Human mesenchymal stem cells featured the typical mesenchymal expression profile lacking hematopoietic markers like CD14, CD34 and CD45 (A). Hepatocytic differentiation was demonstrated in mouse MSCs after incubation for 21 days in hepatocyte differentiation medium (179). Morphology changed from a typical fibroblastoid into the polygonal shape of differentiated hepatocytes (B). Expression of the periportal gluconeogenic marker enzyme phosphoenolpyruvate carboxykinase (PCK1) and of the perivenous glutamine synthase (GS) as well as of the major mouse acute phase plasma protein α1-antitrypsin (AAT) became obvious after hepatocytic differentiation as shown by sqRT-PCR (C; n=3, *p<0.05; n.s. – not significant). MSCs were isolated from human visceral adipose tissue obtained during elective surgery of donors as approved by the Institutional Ethics Review Board of the University Hospital Leipzig. Written informed consent was given by all donors of MSCs. Animal experiments were approved by the regulatory authorities of the animal protection commissary of the university. For further experimental details please refer to supplementary file 1. Scale bar in (B) 100 μm.

Mechanisms of Hepatic Stem Cell Differentiation

Hepatocyte-specific gene expression is regulated by the orchestrated action of sets of liver-enriched transcription factors (106–108). Hence, it is feasible that hepatic stem cell differentiation requires the onset of hepatocyte-specific transcription factor action. Again, as has been stated before, it may be anticipated that the differentiation of hepatic stem cells features the embryonic epithelial-mesenchymal transition of the primordial endoderm to the hepatoblasts in the fetal liver, which involves FGF and BMP signaling (33, 109). Very early events in hepatocyte-specific gene expression seem to be mediated by the action of transcription factors of the Foxa and GATA families. While in tissues, which do not express the albumin gene, the albumin enhancer is tightly packed and hence silent, it becomes occupied by Foxa2 and GATA4 in the embryonic endoderm. This is regarded to loosen and open up packed chromatin structures thereby facilitating binding of specific transcription factors such as C/EBPβ (CCAAT/enhancer binding protein β) and NF1 (nuclear factor 1), necessary for transcriptional activation of the albumin gene (110). The expression of Fox transcription factors is critical both for the induction of liver bud development by the proliferation of hepatoblasts, and for the induction of hepatocyte-specific gene expression. The Foxa transcription factor homologues, formerly termed HNF3, which are expressed both in the fetal and the adult liver, seem to be involved in expression of nearly all liver-specific genes, comprising amongst others tyrosine aminotransferase, phosphoenolpyruvate carboxykinase, albumin, transferrin, and transthyretin (26, 29, 109). FoxM1 is induced in proliferating and extinguished in terminally differentiated hepatocytes during embryonic development (62, 111). FoxM1 knockout mice exhibit defects in embryonic liver development, showing accumulation of polyploid hepatoblasts due to the lack of entering mitosis and failure of hepatoblast biliary differentiation (111, 112). In the adult liver, FoxM1 is expressed to only low levels in resting hepatocytes, but increases during hepatic regeneration after partial hepatectomy. This links FoxM1 to the regulation of S phase entry in proliferating cells, which is corroborated by the decreased expression of proteins critical for cell cycle progression (29, 113).

Recent evidence suggested that the maintenance of stemness is mediated by active Wnt signaling, which after inactivation of glycogensynthase kinase 3β (GSK3 β-) results in the stabilization of β-catenin, its nuclear translocation and activation of Wnt target genes, comprising cell cycle regulatory proteins such as myc and cyclin D1 (114–116). After partial hepatectomy, an immediate increase in β-catenin and its subsequent translocation into the nucleus was observed in rat liver, which coincided with the onset of the mitotic stimulus due to hepatectomy. Overexpression of β-catenin in transgenic mice resulted in a hepatotropic effect after partial hepatectomy yielding a 15–20% higher liver to body weight ratio as compared with basal ratios in nontransgenic mice, indicating a positive correlation between liver growth and Wnt signaling (117, 118). On the other hand, down-regulation of Wnt signaling coincided with hepatic specification and differentiation in a murine hepatic stem cell line (119), which is in contrast to the finding that the regulation of liver specification is mediated by Wnt2b in zebrafish (120). Similar results have been found in ex vivo mouse embryo cultures, in which Wnt3a and HGF promoted hepatic and biliary stem cell specification (121). The opposing role of Wnt and Notch signaling in hepatocyte and biliary fate determination in the embryonic and adult liver has already been described above.

Hepatic Stem Cells and Hepatocyte Transplantation

Transplantation of hepatocytes is considered a valuable alternative to whole liver transplantation (122–125). This therapeutical approach is based on the assumption that healthy donor-derived hepatocytes transplanted into the host liver engraft and functionally reconstitute the recipient organ in the long-term range. This concept has been proven, comprising transplantation of syngeneic, allogeneic, and xenogenic hepatocytes as well as hepatic and stem cells from nonhepatic sources in a variety of mammalian animal models (126, 127). Despite the promising outcome of laboratory animal experiments, there is only slow progress in translation of the method into clinical settings (128, 129). One principle reason for that is probably the poor quality of human hepatocytes isolated from marginal donor organs not suited for transplantation. Therefore, the search for alternative (stem) cell sources both from hepatic and nonhepatic origin as described above has called for much attention to generate stem cell-based hepatocytes for the treatment of human liver diseases (47, 130–132). The principles of hepatocyte transplantation are also valid for transplantation of stem cell-derived hepatocytes. These principles are therefore discussed in the following.

Hepatocytes infused via the portal vein, a technique applicable e.g., in rats, or injected into the spleen as in mice, are transported via the blood stream to the hepatic sinusoidal branches of the portal vein where ∼ 95% of the transplanted cells are entrapped in microemboli in the periportal region of the sinusoids (133, 134). Because of emerging hypertension followed by vasodilation, transplanted hepatocytes eventually travers the sinusoidal endothelium and integrate into the host liver parenchyma (135, 136). This process seems to require both host and donor cell–cell and cell–matrix communications, comprising cytokine and growth factor-mediated disruption of the sinusoidal endothelia at the site of hepatocyte entrapment (133, 135, 137). Only about 20% of transplanted cells integrate into the host liver tissue after 15–20 h. Hepatocytes remaining in the portal sinusoids are presumably cleared by a Kupffer cell response (138). Subsequently, integrated hepatocytes regain their polarity forming gap junctions and bile canaliculi with adjacent host cells and, compared with differentiated hepatocyte functions such as glycogen storage, glucose-6 phosphatase activity and albumin production, transplanted cells become indistinguishable from adjacent host hepatocytes (72, 139, 140). Attributable to the formation of substrate and hormone gradients in the sinusoids of the normal liver, hepatocytes surrounding the terminal branches of the portal vein (periportal hepatocytes) exhibit a different gene expression pattern than hepatocytes surrounding the distal branches of the hepatic vein (perivenous hepatocytes) (141–143). Transplanted hepatocytes engrafting in the periportal regions of the sinusoids acquire the gene expression pattern of periportal hepatocytes (140). However, shifting transplanted hepatocytes into the perivenous areas of the liver lobule by treatment with carbon tetrachloride resulted in switching the periportal to a perivenous hepatocyte phenotype (144). This indicates that the hepatic microenvironment also governs the differentiation state of a transplanted hepatocytes or stem cell-derived hepatocytes, directing position-specific gene expression.

In F344 rats bearing a spontaneous mutation in the gene encoding dipeptidylpeptidase type IV (DPPIV, CD26) wildtype donor hepatocytes could easily be detected in the recipient liver by histochemical or immunohistochemical staining of wildtype CD26 in the negative liver background (72, 145). After several months, donor hepatocytes nearly completely repopulated the host liver, providing functional metabolic compensation as demonstrated by colocalization of CD26 with hepatocyte markers such as glycogen content and glucose-6 phosphatase activity or expression of the canalicular marker ATPase and the hepatocyte-specific gap junction protein connexin 32 (72, 146–148). The capacity of hepatic progenitor cells to repopulate a recipient liver has been studied in this animal model. While bipotent fetal hepatoblasts repopulated around 10% of a healthy liver, no significant repopulation was observed using adult or embryonic hepatic stem cells (51, 149, 150). Recent data demonstrated the repopulation of a CD26-negative adult host liver by transplanted wildtype fetal liver progenitor cells isolated from ED14 fetal livers without any toxic insult (52). These fetal liver progenitor cells obviously featured a growth advantage over adult hepatocytes, probably by a mechanism, which induces apoptosis in the normal host liver hepatocytes surrounding clusters of transplanted fetal liver-derived progenitor cells (151).

Bcl-2 transgenic hepatocytes are resistant to Fas-mediated apoptosis. To provide a selective growth advantage for these cells after transplantation into a mouse liver, apoptosis was induced in host hepatocytes by administration of the agonistic Fas antibody Jo2. Thus, 1.5% transplanted transgenic hepatocytes could be expanded to 85% in the recipient liver (152–154). Similarly, Bcl-xL transgenic hepatocytes were protected in Fas-induced apoptotic livers with a comparable rate of amplification and host liver repopulation (155). As with adult hepatocytes, bone marrow-derived hepatocytes from Bcl-2 transgenic mice could be expanded in a wildtype mouse liver challenged with an anti-Fas antibody (156). Immortalised p19ARF-deficient hepatocytes, which had been retrovirally transduced with the Bcl-2 transgene, repopulated a Fas-induced apoptotic host liver to about 5%. Interestingly, the repopulated cells displayed expression of cholangiocyte, fetal hepatocyte and oval cell markers, indicating that these immortalized hepatocytes generated hepatic progenitor cells during liver regeneration (157). Inhibition of host hepatocyte proliferation by whole liver irradiation and induction of the appropriate growth stimulus by FasL-induced apoptosis favored massive hepatic repopulation by transplanted normal hepatocytes in livers of UDP-glucuronosyltransferase-deficient Gunn rats, a model for Crigler-Najjar syndrome type I, thus correcting the enzymatic defect (158).

Xenograft human hepatocyte transplantation has been performed mainly in immunodeficient mouse models. Originally designed to study the biology of hepatitis virus infections, an immunodeficient Rag2−/− mouse expressing the urokinase-type plasminogen activator transgene (uPA+/+) was transplanted with woodchuck, human, or tupaia hepatocytes. Because of the depletion of host hepatocytes as a consequence of transgene expression, severe liver injury developed, which provided the transplanted hepatocytes with both an appropriate growth stimulus and proliferation advantage over the host hepatocytes. Thus, the system was suited to investigate liver viral infections (HBV, HCV) as well as liver repopulation by transplanted xenogeneic hepatocytes (159–162). This model has also been used to describe the differentiation transition from fetal liver progenitor cells to mature hepatocytes after transplantation of syngeneic mouse hepatocytes by means of quantitative RT-PCR analysis using cells excised from regenerative tissue by laser-assisted microdissection (163). Using severe combined immunodeficient (SCID) mice it had been shown that transplanted human progenitor liver epithelial cells effectively repopulated the host liver and differentiated into mature hepatocytes after induction of liver injury by carbon tetrachloride (164). Transplantation of human adult hepatocytes into livers of uPA/SCID mice resulted in a more than 90% repopulation of the host liver. The chimeric livers expressed various human cytochrome P450 subtypes to levels comparable with those in the respective human donor livers. Inducibility of P450 activity by selected compounds revealed a normal pharmacological response in the “humanized” mouse liver (165). Human umbilical cord blood-derived stem cells engrafted into livers of nonobese diabetic-severe combined immunodeficient (NOD/SCID) mice and differentiated into hepatocyte-like cells, which is however still under debate to cell fusion (166–168).

Antigenic exposition during rat preimmune fetal stages seemed to induce immunotolerance against human hepatocyte xenotransplants in the adult animal (169). Taking advantage of the immunotolerance during fetal development, the injection of human hematopoietic stem cells into fetal sheep or of human hepatocytes into fetal mice, chimeric livers were generated, haboring functional human hepatocytes in the sheep or mouse liver background without any immunological complications (170, 171). This noninjury animal model may provide a valuable tool for studying human hepatic stem cell differentiation and liver repopulation with the perspective of adaptation to clinical cell and gene therapy approaches for the treatment of human hereditary metabolic liver diseases.

Together these models exemplify the proof of the principle of hepatocyte and hepatic stem cell transplantation by functional repopulation of the host liver by cell transplants, thus opening a clinical perspective of cell therapy for the treatment of human liver diseases.

Clinical Implications of Hepatic Stem Cells

So far, the clinical translation of hepatocyte transplantation as an alternative to whole liver transplantation is hampered by the limited availability of suitable donor organs for the isolation of transplantable hepatocytes (122, 129, 172, 173). Hepatocyte transplantation without prior pre-conditioning of the host liver does not result in an effective repopulation of the recipient liver by the donor cells, thus limiting the number of engrafting cells necessary for the treatment of the diseased liver. The experimental protocols such as the use of toxic compounds like retrorsine or carbon tetrachloride in combination with partial hepatectomy are of course not applicable to clinical use. Therefore, attempts have been made in order to increase engraftment of the host liver by transplanted hepatocytes by enhancing the endothelial permeability using cyclophosphamide (174) or by enhancement of donor cell enrichment in the sinusoidal space using vasodilators (136). The principle of inhibition of host hepatocyte proliferation in order to create a growth advantage for transplanted hepatocytes has recently been proven by irradiation of the host liver prior to hepatocyte transplantation. Combined with a strong growth stimulus such as partial hepatectomy or high doses of thyroid hormone, irradiation resulted in nearly complete repopulation through transplanted hepatocytes (74, 175, 176). This approach has been employed to transplant normal rat hepatocytes into jaundiced Gunn rats (defective of bilirubin glucuronosyltransferase), which in the long-term range corrected the metabolic defect indicating functional integration of transplanted hepatocytes into the host parenchyma (75). The priming of liver regeneration could also be achieved by ischemia-reperfusion (73), which in combination with irradiation of defined parts of the host liver prior to cell transplantation may well be suited to clinical feasibility. This had been corroborated by the partial rescue of the LDL receptor deficiency in the Watanabe rabbit after transplantation of allogeneic hepatocytes into livers after regional irradiation and transient ischemia/reperfusion injury (177).

The plasticity, the differentiation potential and the proliferative capacity of hepatic stem cells makes them ideal candidates as alternative source of transplantable hepatocytes (178). The accumulating understanding of both the molecular mechanisms during liver regeneration and hepatocyte differentiation during embryogenesis and liver regeneration will create the basis for the generation of stem cell-derived hepatocytes in the future. Further experimental insight into the mechanisms of the integration of transplanted stem cell-derived hepatocytes into the host liver may help to improve the present unfavorable outcome of clinical hepatocyte transplantation and may require different procedures to precondition the host liver as well as the donor (stem cell-derived) hepatocytes in order to provide specified, optimised conditions in relation to the situation in the diseased liver of the individual patient.


To sum up there is increasing evidence for the existence of liver stem cells. The progeny of these cells constitute bipotent committed progenitor cells finally giving rise to both cholangiocytes and hepatocytes. Thus, there is a clear precursor/progeny relationship between the liver progenitor cell and the mature cholangiocytes and hepatocytes, respectively, linked by all possible intermediate developmental stages. It might be speculated whether the heterogeneity of the ductular reactions appearing in the human liver after different kinds of injury like major tissue loss after intoxication, viral infections or biliary obstructions is the differential answer of one kind of stem cell to different challenges or the response of different kinds of stem cells or even different kinds of stem cell niches (11, 12, 26).

However, even if basic understanding of the cellular and molecular mechanisms of hepatocyte stem cell differentiation is emerging, it will be a major goal in the future to investigate the basic principles of tissue regeneration by hepatocytes derived from stem cells both of hepatic and nonhepatic origin, in order to promote cellular integration and functional hepatic repair in therapeutic approaches. It might be worthwhile considering that different underlying liver diseases, which create different cellular and molecular hepatic microenvironments, require hepatocytes to be transplanted which meet these diverse specifications best, thus enabling the transplanted cells to integrate into, to proliferate at the site of their tissue integration and to rebuild the damaged host liver optimally.


The service of the “Interdisciplinary Centre of Clinical Research (IZKF) – Core Unit Fluorescence Technologies Cytometry” of the University of Leipzig is greatly acknowledged. The authors thank Madlen Hempel for her careful technical assistance and Marie-Luise von Hindte and Dieter Winkler for critical reading and style editing.