SEARCH

SEARCH BY CITATION

Keywords:

  • Human adult hepatic progenitors;
  • Stem cell;
  • Oval cell;
  • Hepatobiliary cell;
  • Differentiation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

Activation and proliferation of human liver progenitor cells has been observed during acute and chronic liver diseases. Our goal was to investigate the presence of these putative progenitors in the liver of patients who underwent lobectomy for various reasons but did not show any hepatic insufficiency. Hepatic lesions were evaluated by histological analysis. Nonparenchymal epithelial (NPE) cells were isolated from samples of human liver resections located at a distance from the lesion that motivated the operation and were cultured and characterized. These cells exhibited a marked proliferative potential. They did not express the classic set of stem cell/progenitor markers (Oct-4, Rex-1, α-fetoprotein, CD90, c-kit, and CD34) and were faintly positive for albumin. When cultured at confluence in the presence of hepatocyte growth factor and either epidermal growth factor or fibroblast growth factor-4, they entered a differentiation process toward hepatocytes. Their phenotype was quantitatively compared with that of mature human hepatocytes in primary culture. Differentiated NPE cells expressed albumin; α1-antitrypsin; fibrinogen; hepatobiliary markers such as cytokeratins 7, 19, and 8/18; liver-enriched transcription factors; and genes characterized by either a fetal (cytochrome P4503A7 and glutathione S-transferase π) or a mature (tyrosine aminotransferase, tryptophan 2,3-dioxygenase, glutathione S-transferase α, and cytochrome P4503A4) expression pattern. NPE cells could be isolated from the liver of several patients, irrespective of the absence or presence of lesions, and differentiated toward hepatocyte-like cells with an intermediate hepatobiliary and mature/immature phenotype. These cells are likely to represent a resident progenitor population of the adult human liver, even in the absence of hepatic failure.

Disclosure of potential conflicts of interest is found at the end of this article.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

Liver cell transplantation is an attractive therapeutic approach to correct inborn errors of metabolism and to bridge patients with fulminant hepatic failure to transplantation or to spontaneous recovery [1, 2]. However, the supply of hepatocytes is limited by the shortage of liver donors and the modest expansion of these cells in culture. Thus, the development of alternative sources of hepatocytes, such as stem cells, is under current investigation in many laboratories.

The liver exhibits a tremendous proliferative response to loss of mass resulting from trauma or toxicity. In most cases, hepatocytes are able to undergo several rounds of replication [3]. However, during extensive liver injury and/or when proliferation of hepatocytes is impaired, the repair process relies on the emergence from the portal or periportal zones of a heterogeneous population of small, poorly differentiated precursors, the oval cells [4]. These cells invade the parenchyma, generally in the form of neoductules, and then differentiate into mature hepatocytes and cholangiocytes to reconstitute the architecture and function of the damaged liver. Oval cells coexpress hepatic markers, albumin, cytokeratins 8/18 (CK8/18), and biliary markers (CK7) and share some phenotypic characteristics with bipotent fetal hepatoblasts (such as albumin, α-fetoprotein [AFP], CK19, and CK8/18) and hematopoietic stem cells (such as c-kit and CD34) [5].

Although debated for a long time, the existence of a human counterpart to the rodent oval cells is now acknowledged; these cells are referred to as liver progenitor cells (LPC). Ductular reactions arising from activated and proliferating LPC have been observed after severe hepatocellular necrosis due to chronic viral hepatitis, chronic biliary hepatitis, and alcoholic and nonalcoholic fatty liver diseases and are particularly pronounced in the cirrhotic liver, a stage that most frequently precedes hepatocellular carcinoma [6, [7]8]. The number of LPC increases with the severity of the disease [9] and correlates with the degree of inflammatory infiltrate [10]. Electronic microscopy and three-dimensional studies of diseased livers support the hypothesis that LPC reside in or near the canals of Hering [11]. LPC share morphology and immunohistochemical characteristics with poorly to moderately differentiated hepatocellular carcinoma cells and with combined hepatocellular-cholangiocarcinoma cells of transitional type [12]. Thus, LPC may play a role in human tumor development, consistent with the role of oval cells in rodent hepatocarcinogenesis [13]. Understanding the intra- and extracellular signals that govern the proliferation, differentiation, and transformation potentials of LPC is therefore of major importance.

Although experimental rodent models have led to the establishment of several oval cell lines [14], few studies on adult LPC in humans have been reported. At present, the Hep RG cell line represents the unique human oval cell line [15]. Selden et al. [16] identified proliferating epithelial cell colonies from the liver nonparenchymal cell population of a patient with subacute hepatic necrosis. These cells express a combination of hepatocytic and biliary markers, as well as the embryonic stem-cell marker Oct-4, and secrete albumin and α1-antitrypsin (AAT). Interestingly, Crosby et al. [17] reported the isolation of rare c-kit+ and CD34+ cells from normal livers, albeit in smaller numbers than from cirrhotic livers. These cells may give rise to biliary cells in vitro. This last study supports the notion that LPC may be present in normal human liver.

In the present work, we isolated and characterized nonparenchymal epithelial (NPE) cells from the livers of patients who underwent partial hepatectomy for various pathologies but exhibited no sign of hepatic insufficiency. These cells exhibit a marked proliferative potential, and when cultured under appropriate conditions, they differentiate into hepatocyte-like cells that express intermediate hepatobiliary and fetal/mature phenotype.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

Chemicals

Culture media, additives, collagenase (type IV), and Histopaque were purchased from Sigma-Aldrich (Saint-Quentin Fallavier, France, http://www.sigmaaldrich.com), fetal calf serum from Invitrogen (Cergy-Pontoise France, http://www.invitrogen.com), and cytokines/growth factors from AbCys (Paris, http://www.abcysonline.com).

Liver Sample Analysis

The use of human liver samples for scientific research has been approved by the French National Ethics Committee. Liver cells were prepared from liver lobectomy specimens resected for medical reasons unrelated to our research program (Table 1). The tissue encompassing the lesion was dissected by the surgeon and sent for histological studies, whereas the remaining encapsulated downstream tissue, which was located at distance from the lesion, was used for nonparenchymal cell and hepatocyte preparation. Before tissue dissociation, a small fragment was removed. The specimen was fixed in 10% neutral buffered formalin and embedded in paraffin. Serial sections (3 μm) were processed for routine staining, including hematoxylin and eosin and Masson's trichrome stainings. Immunohistochemical staining was carried out using an avidin-biotin-peroxidase complex method. The primary antibody used was directed against CK19 (dilution 1:50; DakoCytomation, Trappes, France, http://www.dakocytomation.com).

Table Table 1.. Clinical data of patients and liver biochemical and histological parameters
Thumbnail image of

Cell Isolation

Liver tissue dissociation and hepatocyte preparation were performed as previously described [18]. Hepatocytes were cultured on collagen I-coated plates (BD Biosciences, Le Pont de Claix, France, http://www.bdbiosciences.com) in either a short-term (ST) or long-term (LT) culture medium [19, 20].

The nonparenchymal cell population was enriched and cultured as described by Selden et al. [16], with minor modifications. The supernatant of the first low-speed centrifugation of the postcollagenase liver cell suspension was passed through a 40-μm filter, and the cells were collected by centrifugation at 400g for 10 minutes at 4°C. The pellet was resuspended in 20 ml of Hanks' balanced salt solution, 1% fetal calf serum (FCS), layered upon a Histopaque cushion, and centrifuged at 400g for 25 minutes at 4°C. Cells at the interface were collected and cultured on plastic dishes (Nunc, Roskilde, Denmark, http://www.nuncbrand.com) at a density of 125,000 cells per cm2 in minimal essential medium α supplemented with 10% FCS, 20 ng/ml hepatocyte growth factor (HGF), 10 ng/ml epidermal growth factor (EGF), 25 mM glucose, 1 μM thyrotropin-releasing hormone, 1 μM hydrocortisone, 10 μg/ml insulin, 50 μg/ml albumin-linoleic acid, 0.1 μM selenium acid, 0.5 mg/ml ferrous sulfate, 0.75 mg/ml zinc sulfate, 10 mM nicotinamide, streptomycin, and penicillin. This medium is referred to thereafter as the expansion medium (ExpM).

Cell Culture and Differentiation

Growing epithelial cell colonies were individually picked and amplified in the ExpM. Medium was changed twice a week, and cell passages were performed with 0.25% trypsin (Invitrogen). At confluence (day 0), differentiation was initiated by replacing the ExpM with differentiation medium 1 (DM1) consisting of the ExpM without serum, or with DM2 consisting of 60% low glucose Dulbecco's modified Eagle's medium, 40% MCDB-201, supplemented with insulin-transferrin-selenium + 1, 50 nM dexamethasone, 0.1 mM ascorbic acid 2-phosphate, 20 ng/ml HGF, 20 ng/ml fibroblast growth factor 4 (FGF4), penicillin, and streptomycin, as described [21]. Media were changed twice a week.

Albumin Quantification

Albumin secreted in culture media was measured with a sandwich enzyme-linked immunosorbent assay (ELISA) according to the manufacturer (Bethyl Laboratory, Montgomery, TX, http://www.bethyl.com). Species specificity of the anti-human albumin antibodies was verified using fetal bovine serum.

Western Blotting

Secretion of proteins in 4-day aliquots of culture media was assayed by Western blotting. Primary antibodies directed against human albumin, AAT (DakoCytomation,) or fibrinogen (Sigma-Aldrich) were used at a 1/1,000 dilution. Detection of the chemiluminescent signal of anti-mouse, anti-goat, and anti-rabbit horseradish peroxidase-conjugated secondary antibodies (1/10,000) was performed with the ECL Western blotting detection kit (Amersham Biosciences, Little Chalfont, Buchs, U.K., http://www4.gelifesciences.com).

Fluorescence-Activated Cell Sorting Analysis

Cells were detached with 0.25% trypsin (Invitrogen). Antibodies against c-kit (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com), CD90 (Neomarker, Fremont, CA, http://www.labvision.com), CD34 (Immunotech, Marseille, France, http://www.immunotech.com), and β2-microglobulin (Immunotech) were incubated for 1 hour at room temperature at 1 μg per 100,000 cells in phosphate-buffered saline (PBS) 1% FCS. After three washes in PBS, cells were incubated with anti-mouse IgGFab2 phycoerythrin secondary antibody (Immunotech). Then, after washes, cells were exposed to a mouse antibody directed against CD105 (Neomarker), coupled with Alexa 647 using the Zenon system (Invitrogen). Cells labeling was analyzed using FACSVantage and CellQuest software (BD Biosciences).

Immunocytochemistry

Cells were fixed with 3% formaldehyde and 0.05% glutaraldehyde in 60 mM piperazine-1,4-bis(2-ethanesulfonic acid), 25 mM HEPES, 10 mM EGTA, 10 mM MgCl2, pH 6.9, for 15 minutes at room temperature. Fixed cells were permeabilized with 0.2% Triton X-100 in Tris-buffered saline (TBS) for 5 minutes. Primary antibodies directed against OV-6 (kindly provided by Dr. Sell, Albany Medical College, Albany, NY) in a 1/50 dilution; CK8/18 (Neomarker), albumin (clone HSA-11; Sigma-Aldrich), CD105, c-kit, CD90, CD34, and β2-microglobulin at 1/100; and CK7 (Neomarker) and AFP (Immunotech) at 1/200 in TBS 1% FCS were applied to the cells for 30 minutes at room temperature. Staining was performed with Fast Red Envision kit (DakoCytomation). After counterstaining with hematoxylin (Sigma-Aldrich), labeled cells were examined under an inversed microscope Zeiss Axiovert 200 (Carl Zeiss, Le Pecq, France, http://www.zeiss.com), and images were captured using Axovision Software (Carl Zeiss).

Indirect Immunofluorescence

Mouse antibodies directed against CK8/18 and CK7, rabbit antibody directed against AAT (DakoCytomation), and goat antibody-fluorescein isothiocyanate directed against human albumin (Bethyl Laboratory) in PBS 3% FCS were applied to the cells for 1 hour at room temperature. After washes, cells were incubated with anti-mouse Alexa 568 or anti-rabbit Alexa 647 secondary antibodies (1/1,000 dilution; Invitrogen) for 45 minutes at room temperature. Nuclei were labeled with Hoechst 33258. Immunofluorescent labeling was examined under a fluorescent microscope (Leica Microsystem, Rueil Malmaison, France, http://www.leica.com), and images were analyzed using Metamorph software (Universal Imaging Corp., Downington, PA, http://www.moleculardevices.com).

Reverse Transcription-Polymerase Chain Reaction

After extraction with TRIzol reagent (Invitrogen), 1 μg of total RNA was reverse-transcribed using random hexaprimer and the Moloney murine leukemia virus reverse transcriptase kit (Invitrogen). Polymerase chain reaction (PCR) was performed using the Taq Polymerase kit (Invitrogen) in a mastercycler gradient apparatus (Eppendorf, Le Pecq, France, http://www.eppendorf.com). Sequence of primers and size of the amplification products are summarized in supplemental online Table 1. When possible, primer pairs were designed from different exons to avoid false positives due to DNA contamination. Quantitative PCR was performed using the Roche Molecular Biochemicals Light Cycler system (Roche Diagnostics, Meylan, France, http://www.roche-applied-science.com). The following program was used: one step at 95°C for 8 minutes; 40 cycles of denaturation at 95°C for 15 seconds, annealing at 70°C for 7 seconds, and elongation at 72°C for 18 seconds. Amplification specificity was evaluated by determining the product melting curve. Quantification of all target mRNAs was validated by the use of calibration curves showing a linear relationship between dilutions ranging from 1 to 1/500,000 (for albumin and CYP3A4), from 1 to 1/10,000 (for tryptophan 2,3-dioxygenase [TDO]), from 1 to 1/5,000 (for hepatocyte nuclear factor 4α [HNF4α]), or from 1 to 1/1,000 (for other genes) of a pool of mRNA isolated from human fetal (BioChain Institute, Hayward, CA, http://www.biochain.com) (for glutathione S-transferase π [GSTπ] and cytokeratins 7 and 19) or adult (for the other genes) hepatocytes. The expression of 18S RNA was used for relative quantification. Results are expressed as relative accumulation of mRNA with respect to levels at day 0 (taken arbitrarily as 1), which corresponds to confluent NPE cells after amplification in ExpM, and before differentiation. RNA from the nonhepatic cancerous HEK293 and HeLa cell lines and from ovarian tissue, referred to hereafter as control cells, was included in the study.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

We have been developing primary cultures of adult human hepatocytes for many years. Under our culture conditions, hepatocytes retain several phenotypic and functional markers for at least 35 days [19, 22] and are susceptible to HCV and HDV infections and permissive to viral RNA replication [23, 24]. During the course of these long-term cultures, we frequently observe the emergence of small colonies of epithelioid cells that proliferate within the hepatocyte monolayer. Immunocytochemistry analysis shows that these cells express markers of oval cells, such as albumin, AFP, CK7, CK19, and OV-6 (supplemental online Fig. 1). In the conditions defined for culturing hepatocytes, these cells survive only a few days. All our attempts to isolate them were unsuccessful. However, these observations suggest that LPC may be isolated along with hepatocytes during the liver perfusion process. This prompted us to determine whether these cells are present in the nonparenchymal fraction of the liver and could be cultured out of the hepatocyte context.

Human Liver Samples

Liver samples are fragments of lobectomies resected for various medical reasons. Clinical and biological and histological characteristics of the patients' livers are summarized in Table 1. In all cases, liver function was preserved as illustrated by the fact that biological parameters were within or close to their normal range. The injury of liver samples, such as steatosis, portal inflammatory infiltrate, and fibrosis, was evaluated by histological examination of serial tissue sections after H&E or Masson-trichrome stainings. In addition, semiquantitative evaluation of ductular reaction was performed by CK19 immunotyping [25] (supplemental online Fig. 2) and revealed that some of the tissue samples exhibited a normal architecture (liver samples from patients 3, 5, and 6), whereas others showed a mild ductular reaction (liver samples from patients 1, 4, and 7) or a marked architectural reorganization associated with cholestasis (liver sample from patient 2).

Morphology and Phenotype of Epithelial Cells Isolated from the Nonparenchymal Fraction of Human Liver

Nonparenchymal cells were seeded on plastic dishes in the ExpM (described in Materials and Methods). By days 5–7, small colonies of flattened epithelial cells with a clear cytoplasm were observed among a mixed population of stellate-like and endothelial cells, which we refer to as NPE cells. There was no evidence of human hepatocyte contamination as assessed by phase contrast microscopy. All the NPE cell cultures prepared from the different liver samples behaved similarly.

After 2 weeks of culture, the colonies were surrounded by fibroblastic cells (Fig. 1A, top). Immunocytochemistry analysis revealed that in individual colonies, NPE cells are clearly CD105 (a surface protein of mesenchymal cell type) and CK8/18+, whereas the surrounding fibroblast-like cells are clearly CD105+ and CK8/18 (Fig. 1A). Both cell types express β2-microglobulin but are CD90, c-kit, CD34, OV-6, and CD45 (last not shown). This was confirmed by fluorescence-activated cell sorting (FACS) analysis after double labeling on a whole-culture dish containing several different colonies (Fig. 1B). Individual colonies of NPE cells were isolated from livers 1, 2, and 3, expanded in ExpM (five passages), and then analyzed for the expression of markers of embryonic stem cells (Oct-4 and Rex-1) and hepatic progenitors (AFP and albumin). The reverse transcription (RT)-PCR data shown in Figure 1C indicated that these colonies do not express Oct-4, Rex-1, or AFP mRNA when growing in the ExpM. Albumin mRNA was barely detectable but always present in the progeny of colonies isolated from all livers. These observations indicate that NPE cells do not exhibit in culture the classic set of stem cell/progenitor markers and are different from the LPC observed in our classic hepatocyte cultures (supplemental online Fig. 1). As we did not perform cell cloning, we cannot address the question of a putative filiation between NPE cells and the fibroblastic cell types observed during establishment of the cultures.

thumbnail image

Figure Figure 1.. Phenotypic characterization of nonparenchymal epithelial (NPE) cells: Absence of stem cell/progenitor markers. (A): Phase contrast photomicrographs (magnification, ×100) showing mixed populations of fibroblastic and epithelial cells during establishment of the culture in the expansion medium (ExpM). Immunocytochemistry labeling on individual colonies (day 7–10) show that NPE cells are CD105, CD90, c-Kit, OV-6 and β2-microglobulin+, CK8/18+, whereas fibroblastic cells are CD105+, CK8/18+, and β2-microglobulin+. (B): Fluorescence-activated cell sorting analysis of a whole dish (several colonies) confirmed that CD105 NPE cells are CD90 c-Kit, CD34, and β2-microglobulin+. (C): Colonies from patients 1, 2, and 3 were isolated and amplified in ExpM. Expression of Oct-4, Rex-1 (25 cycles), albumin, and α-fetoprotein (30 cycles) was analyzed by reverse transcription-polymerase chain reaction after five passages. hES, HepG2, and adult liver mRNA were used as controls. Abbreviation: hES, human embryonic stem cell.

Download figure to PowerPoint

Growth Characteristics of NPE Cells

In ExpM, NPE cells grew (Fig. 2A) for 20–25 doublings before exhibiting signs of senescence and dying (Fig. 2B). Growth characteristics were measured on different colonies from livers 2 and 4 with similar results. The doubling time varied between 50 and 80 hours, depending on the liver. We evaluated that, on average, 6 × 1010 NPE cells could be obtained from an initial amount of 3 × 105 cells. At confluence, NPE cells exhibited contact inhibition and formed densely packed monolayers in which they took the cobblestone-like morphology characteristic of epithelial cells (Fig. 2C). Once they had reached confluence, they stopped dividing and could be further maintained for at least 1 month in culture.

thumbnail image

Figure Figure 2.. Growth characteristics and morphological aspect of nonparenchymal epithelial (NPE) cells. (A): Representative photomicrographs of NPE cells (1) and one cloned NPE colony during proliferation (2) and at confluence (3) in the expansion medium (ExpM). (B): Cell growth curve of isolated NPE cells. Initially, 300,000 cells were plated in a 60-mm plastic dish and expanded in ExpM. Cells were passaged and counted every 3–4 days. The total number of cells is reported on the graph. (C): Morphological aspect of NPE cells after 24 days in ExpM (1), in differentiation medium 1 (DM1) (2), and in DM2 (3). Scale bar = 50 μm.

Download figure to PowerPoint

In Vitro Differentiation of NPE Cells Toward the Hepatocyte Lineage

To assess the in vitro differentiation of these cells, the progeny of a colony of NPE cells (passage 5, liver 4) was allowed to reach confluence in the ExpM and was then exposed to DM1 and DM2 [21] or kept in ExpM for 24 days. When confluent and upon exposition to the three media, cells stopped growing and exhibited morphological changes, suggesting a more differentiated phenotype, that is, enlargement of the cells, dark cytosol, and more refractile borders. This effect was more pronounced in DM2 (Fig. 2C). As a first test of differentiation, we evaluated the production of plasma proteins such as albumin, AAT, and fibrinogen by Western blot analysis of sequential aliquots of 4-day culture medium. The data were compared with those obtained with cells cultured in ExpM for 24 days and with 8-day-old functional human hepatocytes in primary culture (HHPC) in ST and LT culture medium. As shown in Figure 3A, the secretion of albumin, AAT, and fibrinogen increased in a time-dependent manner in both DM1 and DM2. The secretion of AAT and of the three fibrinogen forms was detected earlier (i.e., in the first 4 days of culture) than that of albumin. The secretion of albumin, undetectable by immunoblotting in early cultures, increased more rapidly in DM2 than in DM1. This result was confirmed by ELISA (Fig. 3B). Note that a low level of albumin secretion was also detected by ELISA in the cells cultured in the ExpM. After 24 days in DM2, the production of albumin by NPE cells reached a level close to levels observed in HHPC. Analysis of albumin and AAT mRNA expression by real-time quantitative RT-PCR (Fig. 3C) confirmed that the increase in secretion of both proteins is of transcriptional origin and, again, that DM2 is more efficient than DM1. A low basal level of albumin mRNA was detected in confluent NPE cells before differentiation, in accordance with ELISA (Fig. 3B) and RT-PCR analysis, as shown in Figure 1C. It slightly increased in NPE cells cultured for 24 days in ExpM (Fig. 3C, inset). Albumin and AAT mRNAs were absent or barely detected in the control cells. To evaluate the robustness of this differentiation protocol, these experiments were repeated on NPE cells isolated from all the liver samples available (patients 1–3 and 5–7). Western blot analysis reported in Figure 3D shows that in all cases, isolated colonies of NPE cells exposed to DM2 produced albumin and thus appeared to be directed toward the hepatocyte lineage. Moreover, these experiments suggested that NPE cells isolated from livers with various histological and/or pathological features showed a similar differentiation potential.

thumbnail image

Figure Figure 3.. Expression of hepatocyte-specific plasma proteins during the differentiation of NPE cells and secretion in the extracellular medium. A colony of NPE cells isolated from the liver of patient 4 was expanded in ExpM (five passages). When confluent, cells were cultured for 24 days in ExpM, DM1, or DM2. Production of plasma proteins in the extracellular medium during NPE differentiation was analyzed and compared with HHPC in ST or LT. (A): Albumin, AAT, and fibrinogen secreted during 4-day periods were analyzed by Western blot. (B): Albumin level was quantified by enzyme-linked immunosorbent assay from the same samples. (C): Albumin and AAT mRNA expression was analyzed by quantitative reverse transcription-polymerase chain reaction during NPE cell differentiation and compared with HHPC. Insert is a magnified representation of the fold mRNA change in DM1 and ExpM. (D): NPE cell colony were isolated from livers with various histological and/or pathological features (patients 1–7; Table 1) and expanded in ExpM. When confluent (day 0), they were exposed to DM2 for 24 days. Secreted albumin accumulated for 4 days in the extracellular medium was analyzed by Western blot at the beginning (days 0–4 in DM2) and at the end of the experiment (days 20–24 in DM2) and compared with albumin secretion of HHPC. DM2 containing bovine serum albumin was loaded to evaluate to species specificity of the antibody (left lane). Abbreviations: AAT, α1-antitrypsin; DM, differentiation medium; ExpM, expansion medium; HHPC, human hepatocytes in primary culture; LT, long-term medium; ND, not detected; NPE, nonparenchymal epithelial; ST, short-term medium.

Download figure to PowerPoint

During organogenesis, the numerous functions of the liver are controlled primarily at the transcriptional level by the concerted actions of a limited number of transcription factors, including hepatocyte nuclear factors (HNF) and members of the CAAT box enhancer-binding protein (C/EBP) family [26, 27], whose expression is liver-enriched but not restricted to this tissue [28]. These factors function in unique combinations in a regulatory network to control liver-specific gene transcription in well-differentiated liver cells [29]. Among these transcription factors, only the expression of HNF1α and HNF4α is strictly correlated with the differentiated state of hepatoma-derived cells [30]. Since HNF4α is a positive regulator of HNF1α, HNF4α may be a central factor among the liver-enriched transcription factors [31]. Also, in addition to its crucial role in nutrient transport and metabolism, HNF4α is a dominant regulator of epithelial cell phenotype, liver morphogenesis, and hepatocyte commitment [32, 33]. The hepatic expression of both C/EBPα and C/EBPβ increases during development and is greater postnatally than during embryogenesis. Both factors are critical for prenatal and postnatal hepatic glycogen synthesis and storage and are important regulators of liver functions such as nutrient metabolism, its hormonal regulation, liver regeneration, and the acute phase response [34].

We therefore evaluated the expression of these four transcription factors in NPE cells submitted to differentiation media by quantitative RT-PCR analysis, using HHPC and nonhepatic tissue and cell lines as controls. In cells cultured in DM1, DM2, and ExpM, C/EBPβ and HNF1α mRNAs were expressed at a constant level during the 24 days of culture (data not shown), whereas C/EBPα mRNA was not detected. The level of C/EBPβ mRNA was similar to that observed in HHPC and control cells, and HNF1α mRNA level was one order of magnitude lower in NPE cells than in HHPC and similar to ovarian tissue content. HNF4α mRNA (Fig. 4) level was very low in confluent NPE cells before differentiation (day 0), was close to the level detected in the nonhepatic cells, and remained unchanged after 24 days in ExpM. In contrast to C/EBPβ and HNF1α, HNF4α expression increased significantly (12-fold induction) with time in both DM1 and DM2. However, the levels observed after 24 days of culture were 50 times lower than in HHPC. Since it has been established that expression of HNF4α is specifically restricted to the hepatocyte cell type [27], our data suggest that NPE cells submitted to DM1 and DM2 are indeed directed toward the hepatocyte lineage.

thumbnail image

Figure Figure 4.. Gene expression of HNF4α during differentiation of nonparenchymal epithelial (NPE) cells. A colony of NPE cells isolated from the liver of patient 4 was expanded in ExpM (five passages). When confluent, cells were cultured for 24 days in ExpM, DM1, or DM2. HNF4α mRNA levels were evaluated by quantitative reverse transcription-polymerase chain reaction and compared with human hepatocytes in primary culture, ovarian tissue, and nonhepatic cell lines. Abbreviations: DM, differentiation medium; ExpM, expansion medium; HNF, hepatocyte nuclear factor; LT, long-term medium; ST, short-term medium.

Download figure to PowerPoint

NPE Cells Coexpress Hepatocyte and Biliary Markers During In Vitro Differentiation

Expression of various cytokeratins previously shown to be markers of LPC (CK7, CK18, and CK19), cholangiocytes (CK7 and CK19), and hepatocytes (CK18) [35] was evaluated by real-time PCR. The data shown in Figure 5A revealed a continuous expression of these genes in all culture conditions. In contrast to CK18, CK7 and CK19 are not expressed at a significant level in HHPC, as expected. Next, indirect immunofluorescent double-labeling experiments were conducted on differentiating NPE cells at day 24 to evaluate the homogeneity and coexpression of these and other markers. In DM1, cells displayed uniform staining for CK8/18 and CK7 (Fig. 5B), whereas scarce isolated cells or small groups of cells expressed albumin strongly. In DM2, CK8/18 and CK7 were expressed in a more heterogeneous manner (Fig. 5C), and large clusters of cells strongly positive for albumin were observed in all fields of the dish. Albumin staining was localized mainly in the weakest CK7-positive cells. In contrast to albumin staining, cells expressing AAT were not found in clusters but were more homogeneously distributed throughout the dish in both media. Most of them coexpressed albumin. These differentiation properties were similarly observed in all the liver samples examined.

thumbnail image

Figure Figure 5.. Coexpression of hepatocyte and biliary markers in the differentiated nonparenchymal epithelial (NPE) cells. A colony of NPE cells isolated from the liver of patient 4 was expanded in ExpM (five passages). When confluent, cells were cultured for 24 days in ExpM, DM1, or DM2. (A): Cell-specific cytokeratins and actin mRNA levels were measured by quantitative reverse transcription-polymerase chain reaction and compared with human hepatocytes in primary culture. (B, C): Colocalization of albumin with AAT or with cytokeratins CK8/18 and CK7 in NPE cells cultured for 24 days in DM1 and DM2, respectively, was analyzed by immunofluorescent labeling. Abbreviations: AAT, α1-antitrypsin; DM, differentiation medium; ExpM, expansion medium; LT, long-term medium; ST, short-term medium.

Download figure to PowerPoint

NPE Cells Exhibit a Mixed Mature/Immature Phenotype During In Vitro Differentiation

Most hepatic functions expressed in the embryonic liver characterize the adult liver as well. However, at birth, a whole set of new functions accounting for the adult hepatic phenotype is activated. This is the case for individual isoforms of phase I and II drug metabolizing enzymes, which are submitted to temporal regulation during gestation and after birth, during the ongoing maturation of the liver. For example, CYP3A7 is the major cytochrome P450 (CYP) expressed in the fetal liver, but at birth, its expression decreases while expression of CYP3A4 increases, so that CYP3A4 becomes the major P450 expressed in the adult liver [36]. Similarly, GSTα and GSTπ are both expressed early in the embryo liver, but the level of the former rises during the first year of life, whereas the level of the second decreases during gestation, and it is no longer detected in adult hepatocytes [36]. Glucose-6-phosphatase (G6P) and carbamyl phosphate synthase-1 (CPS-1), which catalyze the terminal step of the two main pathways of liver glucose production and the first committed step of the hepatic urea cycle, respectively, are expressed early in the human embryo liver and increase after birth [37, 38]. Finally, TDO and tyrosine aminotransferase (TAT) are not expressed in the embryo but are considered markers of terminally differentiated adult hepatocytes [39, 40].

Expression of these genes in NPE cells submitted to differentiation media was therefore evaluated by real time quantitative RT-PCR, using HHPC as positive control and ovarian tissue and cell lines HEK and HeLa as negative controls. The results are shown in Figure 6. CYP3A7 mRNA exhibited a marked time-dependent induction in cells cultured in both DM1 and DM2. This mRNA was not expressed in cells cultured in ExpM; as expected, it was expressed at a low level in HHPC and was not detected in control cells. The adult isoform CYP3A4 mRNA was detectable and increased with time, as did CYP3A7 and HNF4α mRNAs in both media, but was expressed at a much lower level. These genes were not expressed in the nonhepatic control cells. GSTπ mRNA was expressed in NPE cells before and during differentiation; as expected, its level was lower in HHPC but as high as in the cancerous HEK and HeLa cell lines. Indeed, GSTπ has been identified in normal tissue and also in malignancies (e.g., uterus, ovary, kidney, gastrointestine, and lung) and was found to be the most common GST isoenzyme in the majority of established tumor cell lines [41, 42]. In contrast, GSTα mRNA was absent in the control cell lines and exhibited a time-dependent increase, especially in cells cultured in DM1, but its level at day 24 was much lower than that observed in HHPC. CPS-1 mRNA was expressed at a constant low level (60–100-fold less than in HHPC), irrespective of the culture conditions (data not shown), whereas G6P was significantly detected only in HHPC. Finally, both TDO and TAT mRNAs exhibited similar patterns of expression with a marked and time-dependent increase in DM1 and DM2, respectively, greater than the low expression detected in the cell lines. These data suggest that during in vitro differentiation, NPE cells acquire a mixed immature (CYP3A7 and GSTπ) and mature (TAT, TDO, GSTα, and CYP3A4) hepatocyte phenotype.

thumbnail image

Figure Figure 6.. Gene expression of some developmentally regulated liver-specific enzymes during differentiation of nonparenchymal epithelial (NPE) cells. A colony of NPE cells isolated from the liver of patient 4 was expanded in ExpM (five passages). When confluent, cells were cultured for 24 days in ExpM, DM1, or DM2. mRNA levels were evaluated by quantitative reverse transcription-polymerase chain reaction and compared with human hepatocytes in primary culture, ovarian tissue, and nonhepatic cell lines. Abbreviations: CYP, cytochrome P450; DM, differentiation medium; ExpM, expansion medium; GST, glutathione S-transferase; LT, long-term medium; ND, not detected; ST, short-term medium; TAT, tyrosine aminotransferase; TDO, tryptophan 2,3-dioxygenase.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

In this work, we isolated NPE cells from the livers of patients who underwent liver lobectomies without showing any sign of hepatic insufficiency. Characterization of these cells showed that they can be expanded in vitro and directed toward hepatobiliary cells with a mixed mature and immature phenotype, suggesting that they could comprise a liver progenitor population.

The progeny of NPE cell colonies was characterized in the current study and shown to share morphological and phenotypical characteristics with progenitors isolated by Selden et al. [16] and Parent et al. [15]. Notably, they express AAT, CK7, CK18, and CK19. However, neither Oct-4 nor AFP mRNA was detected during the expansion phase, and albumin expression was low. Immunocytochemistry and FACS analysis revealed further differences from other previously characterized progenitors and with the small oval cell-like colonies that spontaneously appear in our hepatocyte cultures (supplemental online Fig. 1). NPE cells do not express c-kit, CD34, and OV-6, which characterize hematopoietic stem cells and/or oval cells [5, 17]. In addition, they are CD90 and β2-microglobulin+, in contrast to the bone marrow- and liver-derived hepatic progenitors described by Avital et al. [43]. These phenotypic differences may be related to the transient expression of stem cell markers during activation and differentiation processes. For example, in rodents, the oval cell response is clearly heterogeneous and in response to allyl alcohol injury involves immature “nul” progenitors [4] that do not express the oval cell markers OV-6 or AFP. Moreover, the phenotype of cultured stem cells may be different from that of freshly isolated cells, as they are removed from their natural niche [44, 45]. Thus, the phenotype and cell type markers characterizing NPE cells isolated here is likely to result from the initial steps of in vitro selection and amplification used to isolate these cells. The phenotype of NPE cells may also be related to the nature of the initial cell sources used. Indeed, previous studies used diseased livers, whereas we used liver samples from patients whose liver function was considered normal in terms of biological parameters. Although a discrete reactive response could be observed in some of the tissues used here, we believe that the putative progenitors in these livers may exhibit an activated state that is different from the one they would adopt in a liver with acute failure. Moreover, we did not notice any differences between the NPE cell populations isolated from the different patients, regardless of the ductular reaction status of the liver they originated from. Our study confirmed in humans that oval cell proliferation is not a prerequisite for the isolation of cells behaving as hepatocyte progenitors, as previously demonstrated in mice [46]. Indeed, a recent study reported that the human hepatic stem cell population could persist in stable number in liver thorough life, with highly similar expression profiles at all developmental stages [47, 48].

While this report was being written, isolation and characterization of a stem cell population from normal adult human liver was reported. These cells share common features with ours: they were selected by stringent culture conditions, they did not express oval cell markers (c-kit and CD34) and they differentiated toward hepatocytes upon exposure to the same cytokines (FGF4 and HGF). They differ from the NPE cells by the absence of CK19 and by the presence of CD90 and AFP [49]. Whether these differences resulted from adaptation to culture conditions or from intrinsic cell properties remains to be determined.

Several possibilities can be proposed for the origin of the NPE cells characterized here. These cells could derive either from the hepatic epithelial lineage or from a resident dormant stem cell. Metaplasia of hepatocytes to biliary cells has been suspected to generate ductular hepatocytes in chronic cholestatic disease and shown to rescue biliary epithelium after bile duct ligation and toxic biliary injury [50]. In addition, mature rat hepatocytes were shown to dedifferentiate, expand in a clonal fashion, and re-enter hepatocytic and biliary differentiation pathways in a chemically defined medium containing EGF and HGF [51]. Recently, Fougere-Deschatrette et al. have shown that hepatocytes from normal adult mouse liver acquired in culture the expression of bile duct/oval cell markers [46]. It is therefore theoretically possible that the cells investigated here derived from mature hepatocytes that underwent a dedifferentiation and reprogramming process. However, differentiation was performed on the progeny of epithelial colonies, which exhibited marked proliferation, in sharp contrast to hepatocytes, which appear to be highly reluctant to proliferate in culture. Moreover, primary hepatocytes cultured in the ExpM rapidly degenerated but did not proliferate (data not shown). However, we cannot rule out the possibility that the proliferation conditions used here promoted the selection of a subpopulation of hepatocytes, such as small hepatocytes, rare and highly replicative hepatic progenitor-like cells that retain a normal differentiation potential [52]. On the other hand, the constant expression of CK7 mRNA and protein during the culture of the isolated NPE cells suggests that the culture conditions favored or amplified a biliary cell population. However, the fact that albumin expression (protein and RNA) was detected in undifferentiated NPE cells during the amplification phase, albeit at a low level, suggests that this cell population contains progenitors or stem cells. This is consistent with the observation [53] that cultured progenitors copurified with immunoselected biliary cells coexpress biliary cytokeratins and markers of liver progenitors, including albumin. The phenotype described here (i.e., CK19high, albuminlow, AFP) is in agreement with the gene profile expression of the human hepatic stem cell population described by Sicklick et al. [47] and Schmelzer et al. [48], which supposedly persists throughout life in constant numbers in the liver.

NPE cells differentiate into hepatocyte-like cells over a 24-day period when cultured in the absence of serum in two different media. Removal of serum from the ExpM (i.e., DM1 condition) was sufficient to induce the hepatic program. DM1 contains EGF and HGF. EGF stimulates hepatocyte proliferation in vitro and in our hands seems to play a role in the maintenance of the differentiated phenotype of HHPC [19, 22]. HGF plays a major role in the proliferation of mature hepatocytes during liver regeneration [54, 55] and in the maturation of differentiating hepatocytes in the postnatal liver [56, 57], and it also stimulates the differentiation of dormant stem cells in developing mouse livers [58]. DM2 contains FGF4 and HGF. This combination has previously been shown to promote hepatocyte differentiation in a large set of extrahepatic cells, such as mouse, rat, and human multipotent adult progenitor cells [59]; limbal cells [60]; and, more recently, human embryonic stem cells [61] and human adult hepatic stem cells [49]. DM2 was more efficient than DM1, as illustrated by albumin production, which reached a level close to that of HHPC. These results were invariably observed in the cultures from all liver donors. Immunofluorescence analysis revealed a population of cells that coexpressed both the hepatocyte and biliary cell markers.

As emphasized by Hengstler et al., observation of albumin and/or cytokeratins expressions is not an absolute and sufficient marker of the hepatocyte phenotype [62]. To more precisely define the hepatic phenotype of the differentiated NPE cells, we quantitatively compared the gene expression profile of several developmentally regulated hepatic markers in these cells to that of mature adult human hepatocytes in primary culture. The fact that only a subset of NPE cells underwent differentiation into hepatobiliary cells partly explains the low level of expression of some hepatocytic markers compared with HHPC. However, since the expression of other fetal and mature liver markers was evaluated by RT-PCR and not at the cellular level, it is not known whether the same cells express both types of markers or whether the differentiation protocol generated a mixed population of hepatobiliary cells at different maturation stages. The absence (C/EBPα and G6P), the low induced expression (CYP3A4), or the low basal and constant expression (HNF1α and CPS-1) of some markers of mature hepatocytes could be explained by the rather short period of differentiation. In preliminary experiments, we observed a slight induction of HNF1α while extending the differentiation duration, revealing the described hierarchical regulation between this factor and HNF4α [31]. Alternatively, it is possible that additional differentiation signals are required for upregulation of mature hepatic markers. Indeed, in vitro maturation of fetal hepatocytes constitutes a limiting step in the production of functional hepatocytes from human embryonic stem cells [63]. We conclude that a subset of NPE cells underwent significant differentiation through the hepatocyte lineage but could not complete the differentiation program to a mature phenotype within the culture conditions defined here. This partial response to differentiation stimuli could reflect a phenotypic heterogeneity in the NPE cell population isolated on morphological criteria and submitted to several rounds of amplification before the differentiation evaluation.

Our experimental protocol for isolation and culture of NPE cells is simple and reproducible (similar observations were made with all the livers tested irrespective of the pathology), and it allows the generation of a large number of cells. In addition to providing clear evidence for the existence of progenitors in the liver of patients exhibiting no sign of hepatic insufficiency, the isolation and characterization of these NPE cells could have important fundamental and therapeutic applications. Available information from this work does not permit us to determine whether the differentiation potential is inherent to the NPE cells and thus physiologically relevant or was reprogrammed in culture. Moreover, these experiments were carried out on cells resulting from short-term cultures (five generations). The stability and reliability of these differentiation features in aging cultures after serial passages should be analyzed.

Current investigations in our laboratory are aimed at defining more precisely the phenotype of NPE cells, improving the yield and efficiency of differentiation to a mature hepatic phenotype, and, most importantly, evaluating the ability of these cells to contribute to tissue repair in vivo.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

This work was supported in part by the European Community (STREP Predictomics, contract LSHB-CT-2004-504761). C.D. was supported by an INSERM-Région Languedoc Roussillon fellowship. We are very grateful to Dr. S. Sell (Albany Medical College, Albany, NY) for the gift of anti-OV-6 antibodies and to Dr. M.J. Vilarem (U632, Montpellier, France) for the gift of CYP3A primers. C. Duret thanks sanofi-aventis for financial support during the last part of the study.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosure of Potential Conflicts of Interest
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
supdatafigure1.pdf839KSupplemental Figure 1
supdatafigure2.pdf836KSupplemental Figure 2
SupdataTable1.pdf84KSupplemental Table

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.