The stem cell niche of human livers: Symmetry between development and regeneration


  • Lili Zhang,

    1. Department of Cell and Molecular Physiology, University of North Carolina School of Medicine, Chapel Hill, NC
    2. Department of Infectious Diseases, The First Affiliated Hospital of Nanjing Medical University, Nanjing, People's Republic of China
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Neil Theise,

    1. Departments of Pathology and of Medicine (Division of Digestive Diseases), Beth Israel Medical Center of Albert Einstein College of Medicine, New York, NY
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Michael Chua,

    1. Department of Cell and Molecular Physiology, University of North Carolina School of Medicine, Chapel Hill, NC
    2. Michael Hooker Microscopy Facility, University of North Carolina School of Medicine, Chapel Hill, NC
    Search for more papers by this author
  • Lola M. Reid

    Corresponding author
    1. Department of Cell and Molecular Physiology, University of North Carolina School of Medicine, Chapel Hill, NC
    2. Department of Biomedical Engineering, University of North Carolina School of Medicine, Chapel Hill, NC
    3. Program in Molecular Biology and Biotechnology, Cancer Center, and Center for Gastrointestinal and Biliary Disease Biology, University of North Carolina School of Medicine, Chapel Hill, NC
    • Department of Cell and Molecular Physiology, CB# 7038, UNC School of Medicine, Chapel Hill, NC 27599
    Search for more papers by this author
    • fax: 919-966-6112.

  • Potential conflict of interest: Nothing to report.


Human livers contain two pluripotent progenitors: hepatic stem cells and hepatoblasts. The hepatic stem cells uniquely express the combination of epithelial cell adhesion molecule (EpCAM), neural cell adhesion molecule (NCAM), cytokeratin (CK) 19, albumin ±, and are negative for α-fetoprotein (AFP). They are precursors to hepatoblasts, which differ from hepatic stem cells in size, morphology, and in expressing the combination of EpCAM, intercellular cell adhesion molecule (ICAM-1), CK19, albumin++, and AFP++. The hepatic stem cells are located in vivo in stem cell niches: the ductal plates in fetal and neonatal livers and canals of Hering in pediatric and adult livers. The hepatoblasts are contiguous to the niches, decline in numbers with age, wax and wane in numbers with injury responses, and are proposed to be the liver's transit-amplifying cells. In adult livers, intermediates between hepatic stem cells and hepatoblasts and between hepatoblasts and adult parenchyma are observed. Amplification of one or both pluripotent cell subpopulations can occur in diseases; for example, hepatic stem cell amplification occurs in mild forms of liver failure, and hepatoblast amplification occurs in forms of cirrhosis. Liver is, therefore, similar to other tissues in that regenerative processes in postnatal tissues parallel those occurring in development and involve populations of stem cells and progenitor cells that can be identified by anatomic, antigenic, and biochemical profiles. (HEPATOLOGY 2008;48:1598–1607.)

The existence of hepatic stem cells (hHpSCs) has been debated for many years because of an assumption that HpSCs are hepatoblasts (hHBs) with the signature feature of α-fetoprotein (AFP) expression.1 Recently, we reported that livers from humans of all donor ages contain stem cell populations that do not express AFP but are precursors of hHBs that do.2–7 In adult livers, the probable stem cell niche has been shown to reside in the most proximal biliary structures, the canals of Hering, the meeting point between hepatocyte canaliculi and the bile ducts.8, 9

Before studies that showed derivation of repopulating hepatocytes from these structures, demonstration of a progenitor cell population relied on immunohistochemical staining of hepatobiliary cells with a mixed phenotype, that is, cells co-expressing hepatocytic (such as albumin) and cholangiocytic markers [such as cytokeratin 19 (CK19)]. The rationale was that the presumptive stem cells in fetal livers, the hHBs, have this mixed phenotype and, thus, it was expected that an adult hepatic stem cell would have a similar one.1, 10 However, this interpretation proved difficult, because the cells in the canals of Hering are quite distinctive from hHBs, being strongly positive for cholangiocytic markers, but only rare cells weakly express albumin and a few other hepatocytic markers. Cells with the dual phenotype emerge in significant numbers only in the setting of injury, in “ductular reactions.”11–13 Either there is developmental asymmetry between adult and fetal livers, with the adult phenotypes of stem and progenitor cells differing from those of fetal development, or else one of these developmental models is incorrect.

The existing dogma is that hHBs are the stem cells in fetal livers and that they give rise to ductal plates through hepatoblast–mesenchymal interactions at the edges of developing portal tracts.14–16 We provide evidence here and in our past studies2–4 that this interpretation is not correct.

The new interpretation is based on efforts to identify hepatic stem cells based on the defining features of stem cells: cells with pluripotency and self-renewal capacity. Subpopulations of liver cells were isolated using immunoselection technologies and then subjected to stringent tests for pluripotency and self-renewal capacity in ex vivo and in vivo assays.2–7 Two subpopulations proved to be pluripotent: the hHBs and another population found to be precursors to hHBs, the hHpSCs. Self-renewal capacity was observed for hHpSCs but not for hHBs, at least under the conditions assayed.2, 5, 17 The hHBs were able to expand clonogenically and could be passaged in the same medium but under distinct substratum or feeder conditions from those for the hHpSCs.6 Moreover, the hHpSCs proved more resistant to ischemia than the hHBs.18 The previous work2–4 (Fig. 1) shows that hHpSCs and hHBs, in vivo and in culture in Kubota's medium (KM), a serum-free medium designed for progenitors, and on culture plastic, have doubling times of approximately 36 hours until they reach confluency, when they slow to doubling times of more than 48 hours. They can be restored to the faster doubling times by passaging. The cells in the colonies co-express CD133/1, CD44H, telomerase, Sonic and Indian Hedgehog proteins, and their receptor Patched, claudin 3, but not hemopoietic, endothelial, or mesenchymal cell markers.2–8 If the hHpSCs are transferred from culture plastic to STO feeders, they slow in growth and within a day or 2 produce eruptions of cords of cells that have a phenotype that is overlapping with that of the stem cells but also with distinctions: the cord-like structures have cells with dramatically higher levels of albumin, AFP, P450 A7, and intercellular adhesion molecule 1 (ICAM-1) but with no expression of claudin 3, neural cell adhesion molecule (NCAM), or markers for hemopoietic, endothelial, or mesenchymal cell markers. The cells in the cords have been defined as hHBs.2–8 (Online Supporting Table 1 and Table 2 gives a summary of some of the findings from these prior studies.)

Figure 1.

Human hepatic stem cells (hHpSC) in vivo and in vitro. (A) An hHpSC colony on culture plastic and in KM. Such colonies will persist for months with doubling times of approximately 36 hours and express EpCAM (green) and NCAM (red) (nuclei counterstained blue with 4′,6-diamidino-2-phenylindole). The colony is relatively homogenous and densely packed. (B) An hHpSC colony transferred from culture plastic to STO feeders will erupt with cords of human hepatoblasts (hHBs) within a few hours of the transfer, corroborating prior findings.10 Although both the hHpSCs and the hHBs express albumin (red), the level of expression increases dramatically in the hHBs. (Prepared by Dr. Nicholas Moss.) (C) Cells with the same antigenic profile as the hHpSCs in culture are located in the ductal plates of human fetal liver, shown stained for EpCAM (green) and cytokeratin 19 (red).

In cell culture assays, self-renewal capacity was observed for hHpSCs but not for hHBs under the conditions used.2 Still, the hHBs were able to expand clonogenically and could be passaged but under distinct conditions from those for the hHpSCs.2, 5–7 These prior findings led to our hypothesis that the hHBs are probable transit-amplifying cells.2

Armed with the sets of markers distinguishing the two pluripotent hepatic progenitors, hHpSCs and hHBs, we show that cells of the canals of Hering share the antigenic profile with ductal plate cells, whereas the “intermediate hepatobiliary cells” of ductular reactions share the antigenic profile with hHBs. The new interpretation shows that both hHpSCs and hHBs are present in livers throughout life.2, 4 The data suggest an elegant symmetry between fetal development and adult regeneration. On this basis we would also suggest that the term hepatoblast should not be restricted for use in the setting of fetal development, but might be a simple and accurate alternative to the awkward but accepted term intermediate hepatobiliary cells currently recommended for descriptions of adult reactive lesions.


AFP, alpha-fetoprotein; CK19, cytokeratin 19; EpCAM, epithelial cell adhesion molecule; hHB, human hepatoblast; hHpSC, human hepatic stem cell; ICAM, intercellular adhesion molecule; Ig, immunoglobulin; KM, Kubota's medium; NCAM, neural cell adhesion molecule.

Materials and Methods

Human Liver Sourcing.

Liver tissue was provided by an accredited agency (Advanced Biosciences Resources [ABR]) from fetuses between 18 and 22 weeks of gestational age obtained by elective terminations of pregnancy. The research protocol was reviewed and approved by the Institutional Review Board for Human Research Studies at the University of North Carolina–Chapel Hill. Intact livers from cadaveric neonatal, pediatric, and adult donors were obtained from the pool of livers obtained through organ procurement agencies and then rejected for transplantation. Informed consent was obtained from next of kin for use of the livers for research purposes. Protocols received Institutional Review Board approval and processing were compliant with Good Manufacturing Practice.

Diseased Adult Livers.

Paraffin sections of human livers that had been diagnosed with a particular liver disease were obtained through the Division of Surgical Pathology of the Department of Pathology and Laboratory Medicine at University of North Carolina–Chapel Hill. These included livers with massive hepatic necrosis and cirrhosis from chronic biliary cirrhosis, alcoholic cirrhosis, and chronic hepatitis.

Preparation of Frozen Sections.

Liver samples were fixed in buffered paraformaldehyde 4% overnight at 4°C, followed by incubation with 30% sucrose at 4°C for 18 to 24 hours, washed with phosphate-buffered saline, and placed in optimal cutting temperature compound (Sakura), frozen in cold 2-mythelbutane (Fisher), and stored at −80°C. The tissue was sectioned at 5-μm-thick to 10-μm-thick frozen sections, in serial sections of 50-μm sections for each liver. Routine examinations were made in four sections stained with hematoxylin-eosin and with the others for immunohistochemistry staining.


Antibodies used for staining are summarized in Table 1. Fetal livers (16–20 weeks' gestation) and adult human liver tissues each were fixed in 4% paraformaldehyde overnight and stored in 70% ethanol. Tissues were embedded in paraffin and cut into 4-μm sections. Sections were deparaffinized with xylene and rehydrated with decreasing alcohol series. Antigens were retrieved by boiling sections for 25 minutes in a pressure cooker in Retrieval Buffer (Dako). Endogenous peroxidases and protein blocking were accomplished using standard procedures. Staining was performed with the RTU Vectastain Kit (Vector Laboratories) using the manufacturer's guidelines. 3,3′-diaminobenzidine was used as substrate and sections counterstained with hematoxylin. Sections were analyzed using a Leica DMIRB inverted microscope, and pictures were taken with a MicroPublisher camera (Q-Imaging) controlled by SimplePCI (Compix) software.

Table 1. Antibodies
Cytokeratin (CK) 191:300NovaCastraNCL-CK19
ICAM-1 (CD54)1:200Pharmingen664970
NCAM (CD56)1:250Becton Dickinson340363

Confocal Studies.

Antigen retrieval was done as described above. The sections were then treated with 20% goat serum in phosphate-buffered saline for 30 minutes, followed by incubation with two primary antibodies of different subtypes immunoglobulin (Ig) G1 and IgG2a at 4°C overnight. Secondary antibodies were goat anti-IgG1 conjugated with Cy5 (Southern Biotech, #1070–15) and goat anti-IgG2a conjugated with Alexa 568 (Molecular Probes, #A21134), both incubated for 1 hour at room temperature. Immunofluorescence was observed with a confocal microscope within 24 hours using a Zeiss 510 Meta Laser Scanning Confocal Microscope (Zeiss) or Leica SP2 Laser Scanning Confocal Microscope (Leica).


Human hepatic stem cell, hHpSCs, in culture on plastic and in KM form densely packed colonies of cells that co-express epithelial cell adhesion molecule (EpCAM) and NCAM (Fig. 1A). If the hHpSCs are transferred from culture plastic to STO feeders, they give rise to cord-like eruptions of cells with greatly elevated levels of albumin (red = albumin; Fig. 1B). STO feeders are embryonic mouse stromal cells that are ouabain-resistant and found useful as feeders for hepatic stem cells and hepatoblasts.17 Cells co-expressing EpCAM, CK19, and NCAM are found in the ductal plates in fetal livers (Fig. 1B).

Cells co-expressing EpCAM (green) and AFP (red) are found throughout the parenchyma of fetal livers (Fig. 2A, B). We also found that these cells co-express ICAM-1 and P450 A7, but not NCAM or claudin 3 (data not shown), corroborating our prior findings2, 3 on cells that we interpreted as hHBs (see also Online Supporting Table 1 for a summary of the phenotypic profile of hHBs found in prior studies). The hHBs in culture were found not to survive on culture plastic or in KM for more than a few days; rather, their survival proved dependent on the use of certain embryonic mesenchymal feeders such as those used here: STO feeders (Fig. 2C, D). In culture, the hHBs express AFP in colonies with cord-like structures interspersed with clear channels, hypothesized to be bile canaliculi (Fig. 2D).

Figure 2.

Human hepatoblasts (hHBs) in vivo and in vitro. (A, B) In vivo: hHBs in fetal livers. The hHBs express both EpCAM (green) at the cell surface and AFP (red). The level of AFP expression is strongest in the cells near the portal triads (A) and less strong but still evident in cells near the central vein (A, B). (C) In vitro. HHBs do not survive on culture plastic for more than approximately 1 week; longer survival occurs only when on embryonic mesenchymal feeders, such as STO feeder layers, on which they grow very slowly and express AFP (green). (D) Phase contrast image of hHBs on STO feeders and in KM showing typical cord-like structures interspersed by clear channels, the presumptive canaliculi. (The images for C and D were prepared by N. Moss.)

Location of hHpSCs and hHBs in Fetal, Neonatal, Pediatric, and Adult Livers.

Using the sets of markers defining hHpSCs versus hHBs, we have ascertained the histological location and the relative numbers of them in livers from fetal to adult donors (Fig. 3–5). The combination of antigens that uniquely defines hHpSCs (EpCAM, NCAM, CK19, and albumin, but not AFP) is evident in the ductal plates in fetal and neonatal livers (Figs. 1, 3, 4) and in the canals of Hering in adults (Fig. 5).

Figure 3.

Identification of human hepatic stem cells and hepatoblasts in 18-week fetal liver. (A, B) Immunostaining with diaminobenzidine (brown) shows EpCAM expression in ductal plate cells (A), as well as in parenchymal cells throughout the developing hepatic lobule (A, B). EpCAM was expressed most abundantly in ductal plate cells, where membranous and cytoplasmic staining was evident. By contrast, expression of EpCAM was restricted in a membranous pattern in hHBs (B). (Nuclear counterstaining with hematoxylin; original magnifications ×10.) (C) AFP expression (red) found in the hHBs but not in the ductal plates, whereas both stain for EpCAM (green; membranous staining is faint in hHBs because of the high brightness of the hHpSC in the ductal plate; see A for comparison). (Original magnifications ×10.) (D, E) ICAM1 expression is not found on the ductal plate nor in bile duct cells (D) but rather is largely in cells lining the sinusoids (E) within the parenchyma, although hHBs can show a membranous pattern. (Nuclear counterstaining with hematoxylin; Original magnifications: D, ×10; E, ×4.) (F) The ductal plate (DP) cells stain intensely for CK19, whereas CK 19 in the hHBs is fainter and in a particulate staining pattern. (Nuclear counterstaining with hematoxylin; Original magnification ×40.)

Figure 4.

Identification of human hepatic stem cells and hepatoblasts in neonatal and pediatric livers. (A) EpCAM expression in neonatal liver. EpCAM expression (diaminobenzidine, brown) is similar to that observed in fetal liver, in that there is a recognizable ductal plate around portal tracts (PT). (Nuclear counterstaining with hematoxylin; Original magnification ×4.) (B) Double immunofluorescence for alpha-fetoprotein (AFP) (red) and EpCAM (green) again highlight the EpCAM+/AFP− ductal plate, faint co-localization in hHBs, and absent staining of both in hepatocytes (H). The percentage of hHBs declines rapidly postnatally such that they constitute fewer than 0.01% of the parenchyma in pediatric and adult livers, though rare cells with faint AFP expression and membranous staining by EpCAM remain and can be found in livers of all donor ages. (Original magnification ×10.) (C) In pediatric livers, the ductal plate is no longer apparent. Rather, one observes canals of Hering (CoH) near to the portal triads (PT) and surrounded by mature hepatocytes (H); the antigenic and biochemical profile of the canals of Hering is identical to that of the ductal plate cells in fetal and neonatal livers. (Nuclear counterstaining with hematoxylin; Original magnification ×2.) (D) A confocal image of a pediatric liver section is shown, double stained for EpCAM (membranous, green) and cytokeratin 19 (cytoplasmic, red). (Original magnification ×40.) (E) Immunostain for EpCAM (brown) highlighting the canals of Hering (CoH), versus hHBs. The hHBs are tethered to the ends of the canals of Hering. (Nuclear counterstaining with hematoxylin; original magnification ×10.)

Figure 5.

Identification of human hepatic stem cells and hepatoblasts in adult livers. (A) Immunostaining for EpCAM shows positive cells (brown) only in bile ducts and in canal of Hering profiles near the portal triads (nuclear counterstaining with hematoxylin; original magnification ×2.) (B) The same staining for EpCAM (brown) showing cytoplasmic and membranous staining of interlobular bile duct, ductules and canals of Hering (CoH). Rare cells have membranous EpCAM staining are indicative of hHBs. (Nuclear counterstaining with hematoxylin; original magnification ×10.) (C,D) Serial sections showing cytokeratin 19 staining (C) and EpCAM staining (D). The canals of Hering (CoH) are strongly and intensely positive throughout for EpCAM and CK19. Near to the canals of Hering, and sometimes found tethered to the ends of them, are hHBs with membranous staining only of EpCAM. Hepatocytes (H) show no staining for either CK19 or EpCAM. In another set of serial sections (E-G) stained for CK19 (E) or for EpCAM (F-G), there is clear evidence of tethering of hHBs to the canals of Hering. Original magnifications of the enlarged images are 20×.

The combination of antigens uniquely defining hHBs (EpCAM, ICAM, CK19, albumin, AFP) is not in cells in the ductal plates (Figs. 2–4) but is present in cells throughout the parenchyma of fetal and neonatal livers and in individual cells or small groups of cells connecting to one end of, or adjacent to, a canal of Hering in pediatric and adult livers (Figs. 4, 5). Albumin was expressed weakly in some but not all cells of the ductal plates and canals of Hering; was found at significantly higher levels and in all the cells expressing AFP, that is, hHBs; and was found at the highest levels in mature hepatocytes (data not shown).

In mature liver tissue, bile duct cells but not hepatocytes were found to be positive for EpCAM. However, bile duct cells are readily distinguished from the progenitors in that they do not have the accompanying expression of either albumin or AFP (Fig. 5). In fetal livers, the cells that strongly co-express CK19 and EpCAM with double staining are in the ductal plates (Figs. 1, 3), whereas hHBs retain EpCAM expression at the membrane surface but not cytoplasmically and have CK19 as weak, particulate staining (Fig. 3F).

Age Effects on the Numbers of hHpSCs and hHBs and on the Histological Appearance of the Stem Cell Niche.

The previous finding that the hHpSCs remain relatively stable in terms of the percentage relative to the total parenchymal cell population, approximately 0.5% to 1.5% by flow cytometry, throughout life corresponds to similar findings by immunohistochemistry in intact tissue. The changes in the hHpSCs are in the histological appearance of the stem cell niche. In fetal and neonatal livers, the hHpSCs are concentrated in the ductal plates. In pediatric livers, and even more so in adult livers, the ductal plates are partially resorbed, leaving behind the canals of Hering, tethering terminal bile ducts to the parenchyma. One can trace these changes, progressing from fetal livers (Fig. 1–3), to neonatal and pediatric livers (Fig. 4), to adult livers (Fig. 5). The double-stained cells residing in the canals of Hering in postnatal livers are proposed to be the hHpSCs that give rise to hHBs found adjacent to the canals of Hering. Serial sections stained for CK19 and EpCAM show clearly the example of canals of Hering in adult liver (Fig. 5).

With donor age, sections of human livers were found to contain progressively fewer hHBs. These findings from immunohistochemistry and morphology of cells in liver sections complement those found previously in which the percentages of hHpSCs and hHBs in human livers were defined by flow cytometric analyses.2 Percentages of hHBs transition from being more than 80% of the parenchyma in fetal livers to less than 0.01% in adult livers. The percentages in neonatal livers changed rapidly, indeed, day by day of postnatal life, such that a neonate a few hours old might have more than 50% of the parenchyma being hHBs; however, by a month or two of postnatal life, the numbers were less than 1%.

Identification of hHpSCs and hHBs in Representative Disease States.

In conditions in which ductular reactions occur, the hepatobiliary cells of the ductular reaction share the phenotypic profile typical for hHBs from fetal livers (Figs. 6, 7). The more advanced the liver disease, the more widespread these cells are, and they are uniformly arranged around and connected to the ductular reactions. A lineage relationship is suggested between the small, ductular reaction hepatobiliary cells, with EpCAM and CK19 throughout the cells, and the larger cells with fainter CK19 staining and EpCAM expression confined to the membrane. The primary regenerative responses to liver necrosis involve expansion of the hHpSCs (EpCAM+, NCAM+, but AFP negative), whereas those in biliary cirrhosis involves presumptive hHBs (EpCAM+ at the plasma membrane and ICAM+, AFP+). In hepatic cirrhosis, both populations can be involved (Fig. 6D).

Figure 6.

Ductular reactions in massive hepatic necrosis represent expansion of hepatic stem cells. Immunostaining (brown) for EpCAM (A), NCAM (B), cytokeratin (CK) 19 (C), and NCAM (D); this panel of markers, co-expressed in the ductular reaction, are indicative of hHpSC expansion in this acute clinical situation (Counterstained with hematoxylin. Original magnifications: A, C, ×10; B, ×4; D, ×20).

Figure 7.

Biliary cirrhosis resulting in expansion of hHBs. (A) Large increases in hHBs (currently called “intermediate hepatobiliary cells”) occur in biliary cirrhosis, highlighted by membranous EpCAM staining. (Counterstained with hematoxylin. Original magnification ×2). (B) Higher magnification highlights the way hHBs are tethered to the canals of Hering (CoH) at the edges of the parenchymal/stromal interface. (Counterstained with hematoxylin. Original magnification ×20). (C) A serial section from the same liver shown in B indicates that HB are strongly positive for alpha-fetoprotein. (Counterstained with hematoxylin. Original magnification ×2). (D) NCAM staining of a liver section with severe biliary cirrhosis indicates increased numbers of canals of Hering (CoH) and of probable hHBs. (Counterstained with hematoxylin. Original magnification ×10).

The findings that define the distinctions between hHpSCs and hHBs and indicate their presumptive lineage relationship are summarized in a schematic figure (Fig. 8) and in tabular form (Online Supporting Table 1).

Figure 8.

Model of stem cells, hepatoblasts, and committed progenitor lineage stages in human livers. A schematic showing some of the known markers for hHpSC and hHB and the location of cells with those markers within the liver acinus.


Liver growth is initiated by an endodermal stem cell population in the embryonic foregut15, 19 and by processes leading to the subsequent formation of mature hepatocytes and cholangiocytes in response to paracrine signals from mesenchymal tissue surrounding the portal vasculature.14, 20, 21 Recently, this paracrine signaling has been shown to derive from angioblasts and their progeny, endothelia, which provide multiple signals, including a subset of fibroblast growth factors.22, 23 HNF1 and HNF6b signaling in a highly localized response are hypothesized also to be critical regulators of the process.24, 25 Ductal plates, comprising bands of cells around portal triads in fetal and neonatal livers, have intense EpCAM, CK19, and NCAM expression, weak albumin expression, and are devoid of AFP.8, 26–28 Because AFP has been considered a defining feature of hepatic stem cells, it has been assumed that hHBs are the stem cells, and that the ductal plates are an intermediate lineage stage towards biliary fates. Our data provide an alternate interpretation: that the ductal plates constitute the niche for the hHpSCs subpopulation. Proof that hHpSCs are precursors to hHBs derives from culture studies of immunoselected cells under defined culture conditions, and from transplantation of them into immunocompromised hosts in which they form mature human liver tissue.2–8

Ductal plates emerge during liver development as part of an angiogenesis/vasculogenesis process involving hedgehog signaling.4, 23, 27, 29 In separate studies we demonstrated that ductal plates are the location for intense expression of Indian and Sonic hedgehog proteins; hedgehog proteins and signaling was shown also in cultures of hHpSCs; Inhibitors of hedgehog signaling resulted in rapid death of the hHpSCs in culture.4

The phenotypic profile found for freshly isolated and cultured hHpSCs from livers of all donor ages parallels that found on ductal plate cells and on canals of Hering; in parallel, the phenotypic profile found for freshly isolated and cultured hHB livers of all donor ages parallels that found on hHBs in fetal and neonatal livers and on small numbers of cells tethered to the ends of the canals of Hering. The percentage of parenchymal cells that are hHpSCs was found to be stable throughout life.2 These hHpSCs are recognizable in these histological and immunohistochemical studies on sections of livers from varying age donors complementing the prior findings using flow cytometric and molecular analyses of hHpSCs versus hHBs isolated from fetal versus adult livers.2, 3

The hHpSCs of the postnatal liver have been hypothesized to exist in the canals of Hering and the smallest ductules located periportally14, 30 and are shown now to contain label-retaining cells, properties known for stem cell populations.9 It has been assumed that hHpSCs are not involved in normal maintenance of liver parenchyma or in repair after mild injury or after partial hepatectomy. In these events, it is assumed that hepatocytes accomplish reconstitution of cell mass. In chronic injury, in which replicative senescence prevents hepatocytes from participating in renewal, or in severe acute injury in which the hepatocytes have been largely eliminated or toxically injured, ductular reactions deriving from the proximal biliary tree appear.11, 12 These hepatobiliary cells are rapidly proliferative and display immunophenotypes with mixed hepatocyte (HepPar1, CK8 and CK18, rarely AFP) and cholangiocyte (CK7 and CK19) profiles.

Based on these immunophenotypic studies, conclusions about the nature of adult stem/progenitor cell populations when compared with the standard paradigm of fetal development have led to a difficult asymmetry between fetal and postnatal life. The presumptive stem cells in injured postnatal livers are in canals of Hering that generally have shown only cholangiocytic features. Adult cells with phenotype similar to hHBs are observed only in severe injury conditions. However, the standard paradigm of fetal liver development holds that hHBs are the stem cells that give rise to ductal plates as a subsequently derived population. Our data suggest an alternative interpretation: that the ductal plates contain the hHpSCs that secondarily give rise to hHBs, the presumptive transit amplifying cells. The ductal plates found in fetal and neonatal livers transition to become the canals of Hering in pediatric and adult livers. Thus, the canals of Hering9, 14 are shown to be the adult remnants of the ductal plates containing the hHpSCs that give rise to biphenotypic, rapidly proliferative, transit amplifying cells—the hHBs. This accords precisely with the current understanding of stem/progenitor cell repopulation in postnatal liver after severe acute or chronic injury. This symmetry clarifies the nature of the adult reparative process, indicating that it is, indeed, a recapitulation of fetal development. It suggests a simpler approach to the nomenclature of reactive hepatobiliary lesions in adults. The intermediate (that is, biphenotypic) hepatobiliary cells of the ductular reaction are assigned a complex, confusing name because of lack of a clear understanding of fetal versus adult processes. The word hepatoblast has been reserved for fetal tissue where liver is concerned. However, the suffix -blast is used in other adult tissues to indicate tissue-committed progenitors in which the parallels with fetal development are more straightforward: for example, myoblasts in skeletal muscle, erythroblasts, lymphoblasts, and myeloblasts in the bone marrow, and angioblasts in blood vessels. Given our revised understanding of fetal to adult development, we suggest that the intermediate hepatobiliary cells of ductular reactions are the functional and immunophenotypic correlates of fetal liver-derived hHBs and are therefore deserving of that simple name in adults.

The combination of morphology and these new immunophenotypic data furthermore support the lineage concept in adult tissue: that hHpSCs reside in the proximal biliary tree, giving rise to proliferating transit amplifying cells with hepatoblastic features that retain membranous EpCAM staining (Figs. 5, 6). When differentiated in culture or transplanted into immunocompromised hosts, these cells give rise to hepatocytes that are devoid of EpCAM expression.2, 7, 31 These observations correlate with those of EpCAM expression in normal tissue as summarized in a recent review.32

The data presented here and in our prior reports support a model of the liver comprising of a classic stem cell niche with progeny that give rise to maturational lineages (Fig. 8). Lineage stages for mature hepatocytes correlate well with the large body of literature on zonation of functions within the liver acinus.33–35 In zone 1 (periportal) genes include those associated with gluconeogenesis (such as phosphoenolpyruvate carboxykinase) and connexin 26; zone 2 (midacinar) genes include transferrin and tyrosine aminotransferase; and zone 3 genes (pericentral) include those such as heparin proteoglycan36 and P450-3A1.37

Few studies have been done to identify any distinct traits among subpopulations of adult biliary epithelial cells. However, there are intriguing, recent findings by Alpini et al.38 that lend themselves to interpretation of a maturational lineage among biliary epithelia. The antigenic and biochemical profiles of the small cholangiocytes includes expression of BCL-2 but not cytochrome P450-2E1. By contrast, the large ones express the secretin receptor, cytochrome P450 2E1 and the Cl−/HC03− exchanger, but do not express BCL-2. We propose that the findings of La Russo39, 40 and those of Alpini and associates38 are indicative of a maturational lineage of biliary cells with subpopulations of small intrahepatic biliary epithelia representing early lineage stage(s), and with the extrahepatic biliary cells representing later lineage stages. Our working model of the early stages of the liver's maturational lineages is presented in Fig. 8, a schematic showing the stem cell niche giving rise to distinctive maturational lineage stages of parenchyma. Strategies for identifying, isolating and managing the various lineage stages of parenchyma are given in a recent review.41

This working model also has bearing on interpretation of liver regeneration in response to specific diseases. Massive liver necrosis is shown to involve expansion of progenitors with an antigenic profile equivalent to hHpSCs and minimal response from hHBs, perhaps indicative of quite rapid or of relatively complete maturation of hHBs directly to hepatocytes during an acute, that is, comparatively short injury/recovery process. By contrast, advanced chronic disease, in particular cirrhosis, involves expansion of hHBs; contributions by hHpSCs are assumed and are probable but unknown. It is likely that different, specific liver diseases will be selective for regenerative responses by subpopulations of the hHpSCs, the hHBs, the committed progenitors, or combinations of them, depending on the various complex cell–cell and cell–matrix signaling present in different disease states.42


Technical and administrative support were provided by Lucindea English and Victoria Morgan. Dr. Claire Barbier provided assistance with some of the figures, and Dr. Moss was generous in letting us use some of the images within Figure 1 and 2. We thank Mara Gabriel, Dr. Eric Lagasse, and especially Dr. Milton Finegold for helpful suggestions in the editing of the manuscript. The microscopy was done in the Michael Hooker Confocal Microscope Facility at University of North Carolina–Chapel Hill.