Introduction

The liver is an organ with a tremendous capacity to regenerate. The understanding of this process has both progressed and changed over the last few decades. It was assumed that the liver is reconstituted primarily through the division of mature liver cells. However, over the last few years, there has been increasing evidence of committed hepatic stem cells participating in liver regeneration (Alison 2003). The stem cells, of which the exact anatomic site is considered to be located in the terminal branches of the biliary tree (canals of Hering), feed into a compartment of transient amplifying cells most widely termed as oval cells owing to their oval shaped nucleus. These intrahepatic progenitor cells are activated into proliferation under conditions of severe and chronic liver disease when the multiplication of mature hepatocytes is impaired. The 2-acetylaminofluorene (AAF)/partial hepatectomy (PH) rat model of oval cell activation has been used extensively (Evarts et al. 1989; Anilkumar et al. 1995; Golding et al. 1995). This method of stimulation results in the proliferation of progenitor cells emerging from the portal or periportal zones (biliary tree) towards the central vein of the liver parenchyma, whilst differentiating along the liver plate.

Although substantial knowledge of oval cell mediated liver regeneration has been accumulated over the last few years (Sigal et al. 1992; Oh et al. 2002; Fausto and Campbell 2003), the immunophenotypic features of these progenitors were not characterised consistently. Studies have shown that oval cells express a phenotype that is transitional between biliary cells and hepatocytes (Factor and Radaeve 1993; Yang et al. 1993; Alison 2003; He et al. 2004; Paku et al. 2004). Various markers have been described to identify oval cells in rat liver such as α-foetoprotein (AFP), connexin 43, gamma-glutamyltranspeptidase (GGT), CD49f, and various cytokeratins (7, 14 and 19). As a matter of fact, antigenic characterisation is very much reliant on antibodies, which were developed to detect selected features such as OV-6 and H-4 (Dunsford et al. 1989). Investigators also demonstrated a strong link to extrahepatic cell types displaying several haematopoietic markers such as Thy1, Sca-1 and CD34 (Petersen et al. 1998, 2003) and c-kit (Fujio et al. 1994).

One has to assume that distinct subpopulations of oval cells are to be defined, whose dynamic gene expression profiles depend on the extent of the underlying liver damage, the respective time point in the process of activation and relative spatial distance from the portal field. So far, only few research studies in the past emphasised the co-localisation of various differentiation markers at the cellular level in reference to the anatomical cell location in the injured liver tissue (Paku et al. 2004).

In the present study, we characterised cellular events during the proliferative response to high dose AAF/PH induced liver damage. Morphological and molecular biological analysis of the stimulated tissue and isolated progenitor cells revealed that the heterogeneous oval cell populations not only demonstrate hepatocytic and biliary features, but also display an extrahepatic link with a subset of nestin-positive cells involved in this process of liver regeneration. Morphological assessment further demonstrated that there appears to be a zonal hierarchy of the activated cell populations with ductular and non-ductular features.

Materials and methods

Materials

Medium, buffers and foetal calf serum were supplied by GIBCO BRL, Germany. Collagenase type IV and anaesthetic ether were supplied by Sigma-Aldrich (Munich, Germany). All further chemicals were reagent grade and unless specified otherwise supplied by Sigma-Aldrich.

Primary antibodies were used in this study as summarised in Table 1. Secondary species-specific peroxidase conjugated antibodies (EnVision Kit) were purchased from DAKO Diagnostica, Germany. Secondary fluorescence labelling antibodies (Alexa Fluor) were obtained from Molecular Probes (Goettingen, Germany).

Table 1 Antibodies used for immunohistochemical and immunofluorescence studies
Full size table

Animals and oval cell activation

Male Fischer rats (150–170 g) were purchased from Charles River, Germany. To activate oval cells, animals were treated according to a modified method of the experimental AAF/PH model (Tatematsu et al. 1984). The carcinogenic substance AAF (Sigma-Aldrich) was dissolved in corn oil at a concentration of 2 mg/ml. Rats were administered AAF at a dose of 10 mg/kg body mass daily for 4 days. On day 5, a 70% PH was performed followed by five additional AAF applications starting from day 7 onwards. The organ was harvested at the time of PH and 4, 7 and 10 days after PH. Surgery was performed under ether inhalation anaesthesia. Animal care and experimentation procedures were carried out in compliance with German national and regional legislation on animal protection.

Isolation of oval cells

To isolate oval cells from stimulated rat liver (10 days after PH), a multiple-step digestion with proteases was performed as described by Wang et al. (2003) with minor modifications. Under Rompun/Ketavet anaesthesia, the portal vein was cannulated and perfused for 15 min with Dulbecco’s Minimum Eagle’s Medium (DMEM) containing 0.045% collagenase IV. Pre-digested tissue was removed from the cadaver and minced. The obtained cell mass was suspended in 50 ml of Krebs–Ringer solution containing 0.05% pronase, 0.05% collagenase and 0.001% DNase I. Thereafter, the suspension was incubated for 30–40 min at 37°C while gently stirring. Thereafter, the suspension was filtered through a 125 μm nylon mesh and centrifuged (50 g, 5 min) to allow the hepatocytes to sediment. The supernatant was then pelleted using centrifugation at 400 g for 5 min. Two Nycodenz® stock solutions at 30% (wt./vol.) were prepared, one with phenol red, one without. The stock solutions were subsequently serially diluted to 26, 19 (red), 15 and 13% (red) in Eagle’s buffered saline solution (Gibco) and sequentially layered (volume of 1.5 ml each). The cells of the pellet were resuspended in 11% Nycodenz solution and loaded on the top of the gradient. Centrifugation was then performed at 8,000 g for 30 min (Beckman L5-50 ultracentrifuge with SW41 Ti rotor without break). Cells were located at the four gradient interphases F1-4 starting at the top. Cells from the interphases were collected and finally washed in DMEM supplemented with 10% foetal calf serum. The F2 interphase was determined to contain the highest number of oval cells.

Immunohistochemical detection

Tissues were snap frozen in 2-methylbutane at -70°C. Cryostat sections of 5 μm thickness were fixed in ice-cold acetone for 10 min. The slides were washed in Tris Buffer. Endogenous peroxidase was blocked for 20 min with 0.3% H2O2 in methanol. After further blocking with 5% goat or rabbit serum diluted in Tris Buffer (pH 7.6) for 20 min, the cryosections were incubated with primary antibodies (anti-AFP, anti-CD49f, anti-CD26, anti-nestin or anti-Thy1) at 4°C overnight. Following washing, the cells were incubated with the secondary antibodies (peroxidase-labelled goat anti-rabbit or goat anti-mouse) for 1 h at room temperature. Slides were then washed and incubated with AEC solution for 30 min or DAB solution (detection of anti-AFP) at room temperature. Finally, the sections were counterstained with haematoxylin and mounted in aqueous mounting medium.

Immunofluorescence colocalisation

Tissues or cytospins of F2 oval cell fractions (3×104 cells were centrifuged onto a glass slide at 28 g for 5 min) were immunostained for the first antigen [incubation with the first primary antibody (anti-CD49f, anti-AFP or anti-desmin), using Alexa 488 conjugated goat anti-mouse IgG or anti-rabbit for fluorescence detection (1:400, 1 h at RT)] and then further processed with the second immunostaining protocol. After rehydration in Tris Buffer, specimens were blocked and subsequently incubated with the second primary antibody (CD49f, Thy1, nestin or CD26), rinsed with Tris and exposed to the second Alexa Fluor 555 goat anti-mouse IgG or Alexa Fluor 568 goat anti-mouse IgG2a (detection of anti-CD26) (1:400, 1 h at RT). Slides were finally covered with Vectashield® mounting medium with DAPI (1 μl/ml) (Vector Laboratories, UK) to visualise cell nuclei. Negative controls were carried out for each antibody by omitting the primary antibody from the protocol.

Multiple immunofluorescence-labelled specimens were serially excited and observed with the TEXAS Red-, FITC- and UV-filter set on an inverted confocal microscope (LEICA DM IRE2, Bensheim, Germany). Pictures of each filter set were digitally merged using layering technology software (Leica FW 4000, Version 1.1).

RT-PCR

Total RNA was extracted by using the RNeasy Mini Kit (Qiagen, Hilden, Germany). Total RNA (400 ng) was reverse transcribed into cDNA in a reaction volume of 20 μl using the Oligo (dT) primer and 200U Superscript™ II, RNase H- Reverse Transcriptase (Fa. Invitrogen™ Life Technologies, Karlsruhe, Germany) at 42°C. The PCR was performed according to the manufacturer’s instructions (Taq PCR Core Kit, Qiagen) by using forward and reverse primers as listed in Table 2 for 35 cycles with cycle times of 3 min at 94°C (94°C 30 s, annealing temperature 30 s, 72°C 30 s) and 10 min at 72°C. The PCR primer pairs spanned different exons to avoid amplification products from contaminating genomic DNA comprising genomic sequences. The amplified products underwent electrophoresis in 2% agarose gels and were subsequently stained with ethidium bromide for documentation with a digital camera.

Table 2 Primer sequences, annealing temperature and expected size of amplified products for RT-PCR analysis
Full size table

Results

Kinetics of oval cell stimulation

To demonstrate the kinetics of oval cell proliferation, different points in time were investigated during the course of progenitor cell activation (Fig. 1). Normal liver served as reference liver tissue before stimulation (first column). Mature hepatocytes in the parenchyma displayed a low and just detectable degree of AFP expression, but there were no cells with distinct AFP expression in the portal fields (first row). In contrast, scattered cells in close proximity to the portal triad already demonstrated high levels of CD49f immunostaining (second row). The CD26 or dipeptidylpeptidase IV was homogeneously expressed in the untreated liver (third row). Few cells were immunoreactive to nestin (fourth row). Thy1 expression was readily found in cells surrounding the portal triads (fifth row). Initial histopathological changes were detectable 5 days after starting AAF treatment on the day of PH (second column). At this point in time, the formation of ductular structures was clearly detectable. These cells stained for AFP, CD49f and CD26, whereas cells expressing nestin were visible next to these duct-like cells. Thy1 expressing cells were either located in non-ductular cells within the portal fields or Thy1 was weakly expressed in cells to be found in the parenchyma. Four days later (4 days after PH, third column), a distinct number of small cells with scant cytoplasm had accumulated in the portal areas. Duct-like structures displaying AFP, CD49f and CD26 were evident within the dense cell conglomerates. The nestin and Thy1 immunoreactive cells appeared not to be bound to duct-like morphology. The clusters of activated oval cells gradually increased in size as seen after 7 days following PH (fourth column). At the end of the stimulation protocol (10 days after PH, last column), large clusters of progenies were visible, which had clearly invaded the hepatic parenchyma. As seen at the earlier time points, AFP, CD49f and CD26 were vastly expressed in the duct-forming cells. Abundant nestin-positive cells could be also detected within the clusters. In fact, cells strongly expressing nestin appeared to be located more distally, and in the outer zone of the oval cell cluster. In contrast, the staining pattern of cells highly expressing Thy1 was more compact, with the immunoreactivity starting in the inner zone of the cluster. In contrast to all other investigated markers, CD26 was expressed in both mature hepatocytes and progenitor cells. During the course of oval cell stimulation, CD26-positive ductular cells formed in close proximity to the portal fields and emerged into the parenchyma, linking both cell entities in places. This observation was further confirmed by close-up images of CD26-colocalisation (see Fig. 2b).

Fig. 1
figure 1

Oval cell proliferation at various points in time of AAF/PH stimulation (day of PH, 4, 7 and 10 days after PH), immunohistochemistry at original magnification ×100. Normal liver tissue served as control. The number of activated oval cells increased markedly during the course of the experiment. At the day of partial hepatectomy, only few progenitor cells could be detected in close proximity to the portal triads. Activated cells accumulated (4 days after PH) and emerged from the portal fields (7 days after PH), finally forming large clusters expanding into the liver parenchyma (10 days after PH). Oval cell recruitment was associated with ductular formations expressing the foetal marker AFP (first row), the bile duct marker CD49f (second row) and the bile canalicular antigen CD26 (third row), which is additionally expressed on mature hepatocytes. The neural stem cell marker nestin was clearly expressed in the stimulated oval cells (fourth row). Moreover, progenitor clusters stained positively for the haematopoietic marker Thy1 (fifth row)

Full size image
Fig. 2
figure 2

Multi-layer immunofluorescence staining to co-localise differentiation markers in oval cells, nuclear counterstaining with DAPI (blue), a–e cryosections from AAF/PH stimulated liver (10 days after PH), original magnification×200. Long white arrows indicate the route of differentiation from the portal field towards the liver plate, f–h cytospins of F2 interphase oval cell fraction, original magnification×400. a Oval cell ductules expressed CD49f (red) in the basal membrane sections and AFP (green) in the cytoplasm. The insert shows the magnification of an orthogonally cut duct, b CD26 (red) was expressed in the apical (inner) sections (white arrowheads) of the oval cell ductules expressing CD49f (green) in the basal (outer) cell membrane domain. The insert demonstrates that CD26 expression of the ductules was continuous with the membrane staining of developing hepatocytes (arrowheads) which did not express CD49f, c AFP-positive ductular cells (green) were surrounded by Thy1-positive cells (red) in close proximity to the portal field, d nestin-positive cells (red) encircled the AFP-positive ductules (green) and appeared to expand into the liver plate, e in general, nestin-positive cells (red) lacked expression of desmin (green), which was visible on surrounding hepatic stellate cells (Ito cells), f the F2 oval cell fraction demonstrated numerous CD26-positive cells (red), some of which also stained positively for CD49f (green). The overlay of the two image levels (green and red) results in a yellow–orange colour marking the overlapping membrane domains (arrows), g nestin-positive cells (red) did not display strong expression for AFP (green). h The majority of nestin-positive cells lacked the expression of desmin, whereas only some desmin-positive cells stained weakly with the nestin antibody (arrows)

Full size image

Marker expression and co-localisation

Immunofluorescence staining and co-localisation studies were performed on cryosections from oval cell stimulated liver to characterise the activated cells and their differentiation towards the liver plate at the end of the AAF/PH stimulation protocol (day 10 after PH). Duct-like oval cells expressed CD49f in the basal sections of their cell membranes (Fig. 2a, see also inlay). These ductular cells were distributed homogeneously, extending from the periportal field into the liver plate. The AFP was clearly present in the cytoplasm of these ductular cells. The bile canalicular cell surface molecule CD26 (McCaughan et al. 1990) was clearly visible in the apical (inner) sections of the ducts (Fig. 2b). The antigen was also expressed in the basolateral membrane domain facing the bile canaliculi of maturing hepatocytes. Therefore, both cell types were linked by a direct continuous arrangement of CD26, depicting the margins of bile canaliculi developing from the duct-like oval cells and extending further to the liver parenchyma (see inlay of Fig. 2b).

In normal liver tissue, a small number of cells expressing nestin and Thy1 were visible within the portal fields and could not be co-localised with AFP. During the course of progenitor cell activation, Thy1 and nestin were mostly absent from the forming ductular cells (data not shown). At the endpoint of the stimulation, Thy1 could be clearly detected on abundant small cells located in close proximity to or even surrounding the ducts (Fig. 2c). Nestin-positive cells encircled the AFP-positive ductules and appeared to expand more distally within and beyond the oval cell cluster (Fig. 2d). Conversely, we were unable to co-localise nestin and Thy1 expression with the hepatic gap junction protein Connexin 32 (data not shown), which is recognised as a specific differentiation marker of mature hepatocytes (Unwin and Zampighi 1980).

Desmin represents a marker for stellate cells (Mabuchi et al. 2004). The specific staining pattern of the desmin antibody evidently differed from the anti-nestin staining (Fig. 2e). Desmin-positive cells could be localised throughout the oval cell cluster and displayed strong staining of fine cellular processes between the parenchymal cells. Only very occasionally could the co-localisation of both antibodies be determined.

Enrichment and characterisation of isolated oval cells

To further characterise the progenitor cells induced by AAF/PH (10 days after PH), we first isolated oval cells from the tissue and then enriched them using a discontinuous Nycondenz® density gradient. A previous publication reported an oval cell size of 8–10 μm in diameter, found predominantly in the F2 interphase fraction (Dabeva et al. 1997). Occasionally, some larger cells with the typical morphology of hepatocytes were visible, but these cells were found to be dead in cell attachment studies using collagen coated culture dishes (data not shown). The isolated cell fraction was also contaminated by blood cells (approximately 40% CD45-positive cells, data not shown).

The enriched F2 oval cell fraction was characterised by multiple layer immunofluorescence staining. Significant numbers of cells stained positively for CD26 (Fig. 2f). An important proportion of isolated cells also expressed CD49f. Owing to the polarised membrane domain staining of the latter antigen, the co-localisation with vast expression of CD26 could be recognised as a semilunar overlay zone. Immunostaining with the nestin antibody displayed a large number of positive cells with clearly dotted cytoplasmic staining (Fig. 2g). Cells expressing nestin did not display any relevant labelling for the liver progenitor marker AFP. The fraction was contaminated by some Ito cells, which could be detected with the desmin antibody (Fig. 2h). Although some of these cells showed limited nestin expression, they could be clearly distinguished from desmin-negative oval cells.

To demonstrate that nestin-positive cells displayed hepatocytic and epithelial features, we performed immunofluorescence co-localisation studies with albumin, PanCK (pancytokeratin) and CD26. Significant numbers of cells expressing nestin co-stained for albumin (Fig. 3, first row) or PanCK (Fig. 3, second row). Nestin expression was also found on CD26-positive cells (Fig. 3, third row). As the F2 cell fraction was contaminated by haematopoietic cells including lymphocytes, of which a subpopulation expresses CD26, it was important to show that cells positive for CD26 extensively overlap with staining for the hepatic differentiation marker CK18 (Fig. 3, fourth row).

Fig. 3
figure 3

Multi-layer immunofluorescence staining to characterise the F2 cell fraction containing nestin-positive cells (cytospins, 10 days after PH). The FITC and TRITC-fluorescence channels are demonstrated separately as marked, nuclear counterstaining with DAPI (blue), the last column presents the overlay, original magnification ×1,000. The majority of nestin-positive cells (red) co-expressed the hepatocytic marker albumin (green) (first row) as well as the epithelial cell marker PanCK (green) (second row). Nestin (green) was frequently found on CD26-staining cells (red) (third row), which on their part mostly expressed the specific hepatocytic marker CK18 (green) (fourth row). The fine dotted pattern of nestin staining did not always produce the yellow–orange overlay with green/red fluorescence, but could be clearly located in cells expressing both antigens

Full size image

Comparison of gene expression profiles

We compared the RT-PCR data of isolated mature hepatocytes (control), the F2 enriched oval cell fraction, oval cell liver tissue (10 days after PH), and embryonic day (ED14) foetal rat liver. The hepatocytic lineage markers albumin, CK18 and AFP were expressed in all four samples. As expected, the progenitor cell marker AFP was expressed to a maximum in foetal liver. The RT-PCR data supported the immunohistological results, suggesting that mature hepatocytes express AFP, albeit to a low degree at the protein level. The CK7 could be demonstrated in oval cell liver, in the oval cell fraction, as well as in foetal liver, whereas isolated mature hepatocytes did not express the bile duct marker. Additionally, hepatocytes did not show transcription of the extrahepatic markers such as Thy1 and nestin, which were detectable exclusively in the samples containing adult and foetal progenitor cells. As a matter of course, transcript levels associated with hepatic progenitors were evident in foetal liver. Detectable levels of CK7, Thy1 and nestin were also found in both samples containing oval cells (stimulated liver tissue and isolated cell fraction) (Table 3).

Table 3 RT-PCR analysis to compare the gene expression profiles of freshly isolated hepatocytes, F2 enriched oval cell fraction, oval cell liver tissue (10 days after PH) and foetal liver (ED14). As seen in the samples containing progenitor cells (oval cell fraction, oval cell liver and foetal liver), all genes of markers representing the hepatocytic lineage (albumin, AFP and CK18), the bile duct lineage (CK7) and extrahepatic cell markers (Thy1, nestin) were expressed in contrast to the hepatocyte lane which only showed hepatocytic transcripts. β2-Microglobulin served as house-keeping gene
Full size table

Discussion

The analysis of marker expression suggests that the proliferating oval cells constitute a heterogeneous compartment containing cells that may differ in their proliferation and differentiation profile (Zheng and Taniguchi 2003). When investigating the proliferative response of the injured liver, ductular progenitor cells are largely described (Sarraf et al. 1994; Paku et al. 2001, 2004; Yin et al. 2002). Nevertheless, the issue of atypical ductular cells (Jensen et al. 2004), periductular liver progenitor cells (Sell 2001, 2003) or even individual progenies (Yoon et al. 2004) has also been raised to characterise the cellular pattern.

Our data contribute to the assumption that there are at least two main populations of cells within the activated oval cell compartment. First, we were able to detect cells that form duct-like structures and express bile duct (CD49f) as well as hepatocytic markers (AFP, CD26). This is consistent with previous results demonstrating the bipotent differentiation capacity of oval cells (Alison et al. 2004; Newsome et al. 2004). Furthermore, we identified a non-ductular population of cells, which was detected between and distally from these ductules. These cells expressed extrahepatic markers such as nestin and Thy1. Following oval cell isolation and enrichment, a subset of the nestin-positive cells was also shown to co-express hepatocytic and epithelial markers (Albumin, PanCK, CD26).

The CD49f is present on bile duct cells in normal liver and was recently described as a primitive hepatocytic endodermal marker. It was also found in foetal mouse liver cells with the capacity to differentiate into hepatocytes (Hoppo et al. 2004). In accordance with the studies by Paku et al. (2004), this antigen was a clear marker to detect the ductular progenitor cell population and to visualise their duct-like proliferation pattern extending out from the periportal areas of the injured liver. One has to point out that these activated ductular cells have to be carefully distinguished from duct-forming cholangiocytes which also express this marker. However, cholangiocytes are not known to express AFP, which was evidently present in the ductular progenitor cells.

The AFP is a widely accepted marker for liver progenitor cells indicating the early stage of hepatic differentiation as well as hepatocellular carcinogenesis (Abelev and Eraiser 1999; Nahon 1987). This serum glycoprotein is produced by hepatoblasts during the foetal development und by progenitor cells in the regenerating liver (Shiojiri et al. 1991; Dabeva et al. 1998). We could clearly detect AFP in the ductular progenitor cells. However, when viewing the different time points of oval cell activation, the non-ductular progenitor cells expressing nestin and Thy1 were negative for this early hepatocytic marker.

Thy1 is a highly conserved protein playing an important role in haematopoiesis (Craig et al. 1993). In the context of hepatic progenitor cells, the antigen was first described as a marker on oval cells by Petersen et al. (1998). In our studies, a subpopulation of the progenitor cells expressed this haematopoietic marker. However, we were not able to confirm that the Thy1-positive cells co-express AFP. This disparity may be explained by the different method of oval cell stimulation used by Petersen. Nevertheless, it has yet to be proved that Thy1-positive cells demonstrate the capacity to differentiate into hepatocytes in culture or following transplantation in vivo. It is worth mentioning that the co-culture with Thy1-positive mesenchymal cells induced the hepatocytic maturation of AFP-producing cells isolated from embryonic stem cell cultures (Ishii et al. 2005), suggesting a permissive effect of Thy1-positive cells during hepatic differentiation.

A close anatomic relationship between oval cells and non-parenchymal cells during the repair of liver injury has already been described (Yin et al. 1999). Both nestin and desmin serve as marker proteins, which have been recently demonstrated in stellate cells of the rodent liver (Messing 1999; Niki et al. 1999; Geerts 2004). In particular, desmin is recognised as one of the principal intermediate filament proteins expressed in hepatic stellate cells (Yokoi et al. 1984). The intermediate filament protein nestin was originally identified as a marker for neural progenitor cells (Cattaneo and McKay 1990; Bauer et al. 2005). In our study, desmin-positive and mainly nestin-negative cells with the morphological characteristics of stellate cells were located in close proximity to the stimulated progenitor cells. We may suggest that these cells represent hepatic stellate cells, which play an indispensable role in the development of progenitor cells (Matsusaka et al. 1999).

The majority of nestin-positive cells appeared to be compact with a round cell shape. After cell isolation, these cells could be distinguished from desmin-positive stellate cells even more clearly. Additionally, we were able to demonstrate the co-localisation of nestin-positive cells with epithelial and hepatocytic markers (PanCK, albumin, CD26), pointing to their differentiation into the hepatocytic lineage. These findings are supported by a most recent study in adult mice liver, demonstrating the expression of nestin to mark a subpopulation of small epithelial cells that meet the criteria for oval cells, including morphology, localisation, antigenic profile, and reactivity in response to injury (Gleiberman et al. 2005). Furthermore, nestin is considered to be associated with progenitor cell plasticity. It was found in embryonic and adult progenitor cells displaying multi-lineage potential (Wiese et al. 2004). More precisely, nestin has been proposed as a transient marker for potential hepatic progenitor cells derived from embryonic stem cells (Kania et al. 2004). The expression of nestin in adult hepatocytes proliferating in culture whilst adopting precursor characteristics was also demonstrated lately (Koenig et al. 2006). In the present study, nestin expression could be demonstrated in non-ductular progenitor cells from adult rat liver. However, it has to be proved that nestin-positive cells differentiate directly into hepatocytes or whether they in fact have some supplementary effect on hepatocytic differentiation. Moreover, we may have to consider the possibility that the nestin-positive cells represent dividing transitory cells, which not only give rise to hepatocytic cells but also demonstrate their ability to develop into the endothelial cell lineage, thereby supporting the new formation of hepatic tissue. The nestin protein was recently found in adult human angiogenic vasculature (Mokry et al. 2004). Accordingly, nestin-immunoreactive vessels were observed in capillaries of the corpus luteum, in neovascularisation of infarcted hearts as well as in solid tumours (Amoh et al. 2005). In addition, nestin was present in endothelial cells from neocapillaries during pancreas regeneration (Lardon et al. 2002). Whatever the nestin-positive cell represents or may develop into, one can state that the intermediate filament nestin is a reliable marker and indicator of tissue regeneration.

Oval cells represent a heterogeneous population of individual cell types, which may represent different stages of differentiation along the hepatocyte lineage (Sanchez et al. 2004). Overall, our morphological and immunophenotypical findings contribute to the assumption that the stimulated liver is organised according to a zonal hierarchy (Fig. 4), with a bipotent ductular stem cell population located in close association to the terminal branches of the biliary tree of the portal field. These cells are presumably linked to an amplifying population of non-ductular cells. We have to address the question as to whether there might be a direct switch from the ductular and hepatocytic markers to hepatocytic and extrahepatic markers. The zonal arrangement with a marker gradient from the inner to the outer zones of the stimulated oval cell cluster implies such a developmental transition. However, it is not yet fully clear as to how the different stem cell populations interact both with each other and with the mature liver cell population. The concept of at least two putative differentiation pathways of oval cells toward mature hepatocytes has to be considered, namely (1) the differentiation of ductular-like oval cells into hepatocytes, and (2) the differentiation of individual oval cells as/via transitional cells lacking duct-like features (Yoon et al. 2004). It is also conceivable that the cells expressing extrahepatic markers are not progenitor cells as such, but establish a heterogeneous cell compartment that has some auxiliary function towards the differentiation of progenitor cells into hepatocytes.

Fig. 4
figure 4

Potential zonal hierarchy of activated progenitor (oval) cells (a) and differentiation markers (b) found in the liver of AAF/PH treated animals

Full size image

In this study, as in other studies to date, the developmental hierarchy of progenitors still remains unclear. There is also no substantial evidence indicating the exact origin as well as the fate of progenitor cells at present. Transplantation studies may elucidate the fate of the different oval cell populations whilst differentiating into mature cells. However, there are no reports of successful oval cell transplantation with subsequent liver repopulation in the rat model. There is only one transgenic mouse model (FAH-/-) demonstrating the repopulation capacity of transplanted murine oval cells (Wang et al. 2003). The underlying genetic defect (homozygous deletion of the gene for fumaryl acetoacetate leading to lethal hepatic failure) provides an exceptionally selective growth environment for transplanted cells, which cannot be compared in any terms with any other non-transgenic experimental model or human liver disease.

In view of our observations, we would like to propose that at the very least two main populations of progenitor (oval) cells exhibiting hepatocytic and biliary (ductular cells) or hepatocytic and extrahepatic (non-ductular cells) markers can be identified. The different antigen profiles imply that the oval cell compartment contains a multitude of cells with different markers and expression levels. Therefore, the ‘representative’ oval cell cannot exist. Selecting one or even several markers to identify and isolate oval cells always comprises a choice of target cell, thus making comparative studies difficult. Looking for an alternative source of cells for transplantation purposes, the appropriate stimulus to prepare the host liver for oval cell repopulation still has to be determined. The question remains as to whether oval cells being activated under distinct highly toxic and carcinogenic stimuli really may have prospects for a clinically viable form of treatment. However, the analysis of the cellular pattern and various differentiation markers certainly gives us important insights into the regenerative capacity of the adult liver.