Immortalized p19ARF null hepatocytes restore liver injury and generate hepatic progenitors after transplantation



Primary hepatocytes are blocked in mitotic activity and well-defined culture conditions only allow the limited expansion of these cells. Various genetic modifications have therefore been employed to establish immortalized hepatic cell lines, but, unfortunately, proper hepatocyte cultures conducting a faithful hepatic gene expression program and lacking malignancy are hardly available. Here we report the immortalization of primary hepatocytes isolated from p19ARF null mice using the rationale that loss of p19ARF lowers growth-suppressive functions of p53 and bypasses cellular senescence without losing genetic stability. The established hepatocytes, called MIM, express liver-specific markers, show a nontumorigenic phenotype, and competence to undergo Fas-mediated apoptosis. Intrasplenic transplantation of GFP-expressing parental MIM cells into Fas-injured livers of SCID mice revealed liver-reconstituting activity. In the regenerated liver, MIM cells localized in small-sized clusters and showed presence in structures comparable to canals of Hering, the site of oval cells. Transplantation of MIM-Bcl-XL cells, which are protected against apoptosis, and successive Fas-induced liver damage, enhanced donor-derived liver repopulation by showing differentiation into cholangiocytes and cells expressing markers characteristic of both fetal hepatocytes and oval cells. In conclusion, these data indicate that long-term cultivated p19ARF null hepatocytes are capable of generating hepatic progenitor cells during liver restoration, and thus represent a highly valuable tool to study the differentiation repertoire of hepatocytes. (HEPATOLOGY 2004;39:628–634.)

The hepatocyte represents the most abundant and versatile cell type of the liver, which accomplishes a multitude of metabolic and detoxifying functions essential for the adult organism. Although highly differentiated and quiescent hepatocytes constitute the liver parenchyma under healthy conditions, hepatocytes themselves possess an astonishing proliferative capacity and are the major source for hepatocyte replacement upon liver regeneration in response to surgical or toxic injury.1, 2 Besides the regenerative potential of hepatocytes, oval cells provide a liver stem cell repository with the ability to develop into both hepatocytes and bile duct epithelial cells (cholangiocytes), thus giving a bipotent progenitor cell.3, 4 These oval cells reside periportally in terminal bile ductules of the liver, termed canals of Hering, and have been further described to be located in the pancreas.3 In addition, bone marrow-derived hematopoietic stem cells (HSCs) have been effectively proven to undergo metaplasia into cells with hepatic epithelial cell lineage capability.5, 6 Notably, this developmental plasticity of cells from the adult bone marrow has been found to be associated with the fusion of HSCs with hepatocytes.7, 8 For all these three different liver stem cell compartments involved in tissue homeostasis and organ repair, the potential value of corresponding ex vivo cultures has been indicated for the treatment of liver diseases.9, 10

The access to hepatocytes with liver reconstituting activity is hindered by multiple obstacles, among them the very limited potential to expand hepatocytes in culture. Innate surveillance mechanisms linked to tumor suppressor activities are responsible for blocking cell cycle and for rapidly evoking senescence of hepatocytes ex vivo.10 Inactivation of the p19ARF/MDM2/p53 pathway therefore offers a unique opportunity to obtain limitless lifespan by the ablation of both cell cycle arrest and senescence.11, 12 On the one hand, loss of p19ARF (p14ARF in human) decreases the growth-suppressive functions of p53 while maintaining the Arf-independent activities of this protein. On the other hand, disruption of p19ARF function is sufficient to circumvent senescence without losing genetic stability.

In this study we show that the use of p19ARF deficient mice is a suitable approach to establish immortalized hepatocyte cell lines which are capable to repopulate a Fas-damaged liver after intrasplenic transplantation. Furthermore, enhancement of donor-derived liver restoration by transplantation of immortalized hepatocytes and successive Fas-mediated injury provides strong evidence that p19ARF null hepatocytes are able to generate liver progenitor cells. Thus, the novel hepatic cell model derived from p19ARF deficiency facilitates the understanding on which route hepatocytes may progress towards epithelial lineage commitment.


Alb, albumin; AFP, α-fetoprotein; CK, cytokeratin; GFP, green fluorescent protein; HNF, hepatocyte nuclear factor; MTT, methyl-thiazol tetrazolium; PAH, phenylalanine hydroxylase; PK, pyruvate kinase; RT-PCR, reverse transcription polymerase chain reaction; SCID, severe combined immunodeficiency; TGF, transforming growth factor; TUNEL, terminal deoxynucleotidyl transferase-mediated nick end-labeling.

Materials and Methods

Isolation, Immortalization, and Genetic Manipulation of p19ARF Null Hepatocytes.

Ten–14-week-old male mice deficient in p19ARF were subjected to standard liver collagenase perfusion for purification of hepatocytes as described previously.13, 14 Isolated hepatocytes were seeded on rat tail collagen I-coated tissue culture plastic and cultured in RPMI 1640 either with 10% fetal calf serum (FCS) or with 10% FCS plus 40 ng/ml transforming growth factor (TGF)-α (Sigma, St. Louis, MO), 30 ng/ml insulin-like growth factor (IGF)-II (Sigma), 1.4 nM insulin (Sigma), and antibiotics. Facultatively, 1 μM dexamethasone (Sigma) was added to the culture medium plus growth factors. Propagation and immortalization of hepatocytes was performed by passaging cells at a ratio of 1:3 twice a week in FCS-containing RPMI medium plus TGF-α, IGF-II, and insulin. The pool of immortalized hepatocytes undergone more than 100 cell doublings, termed MIM, was further seeded at low cell density for clonogenic expansion and isolation of hepatocyte colonies. The clone MIM-1-4, showing the strongest response to TGF-β1-mediated growth arrest and apoptosis, was employed for further studies. MIM-1-4 cells were in a first round retrovirally transmitted with the vector pMSCV-GFP (ClonTech, Palo Alto, CA), and in a second round infected with the plasmid pBabe harboring Bcl-XL or Bcl-2 for selection with the resistance marker puromycin.15, 16 Pure populations of the respective stable cell lines MIM-GFP, MIM-Bcl-2, and MIM-Bcl-XL were obtained after sorting of GFP-expressing cells using the FACSVantage (Becton Dickinson, Sunnyvale, CA). Pools of each cell type were propagated in RPMI 1640, 10% FCS and growth factors as described for MIM hepatocytes. All cells were kept at 37°C and 5% CO2 and routinely screened for the absence of mycoplasma.

Assays for Proliferation and Cell Death.

Analysis of proliferation kinetics17 and analysis of terminal deoxynucleotidyl transferase-mediated nick end-labeling (TUNEL)-positive cells to detect cell death were performed essentially as described previously.18 For all assays, triplicate measurements were performed twice and statistically evaluated. Fas-mediated apoptosis of hepatocytes was induced by incubation in medium containing the hamster monoclonal anti-Fas antibody Jo-2 (400 ng/ml; Pharmingen, San Diego, CA), and 100 ng/ml actinomycin D (Sigma) for 24 hours.19

Tumor Formation In Vivo.

MIM hepatocytes were detached from the tissue culture plate, washed with PBS, and resuspended in Ringer solution. Subsequently, 1 × 106 cells in 100 μl Ringer solution were subcutaneously injected each into three individual immunodeficient SCID/BALB/c mice. The potential of MIM hepatocytes to form tumors was viewed over a period of 6 months. All experiments were performed twice and carried out according to the Austrian guidelines for animal care and protection.

Reverse Transcription Polymerase Chain Reaction (RT-PCR).

Extraction of poly(A)+ mRNA, reverse transcription, and PCR were performed as described recently.17 The sequences of the forward and reverse primers: albumin (Alb), 5′-GAAATGCTCATACGATGAGC-3′ and 5′-CTTCAACAGTGGTTT-ATCGC-3′; α-fetoprotein (AFP), 5′-TGGCGTCAAACCTGAAGGCTTATC-3′ and 5′-CCTGAGCTTGGCACAGATCCTTG-3′; hepatocyte nuclear factor (HNF)-1α, 5′-GGTGGCCCAGTACACGCACA-3′ and 5′-GGTGGCATGGCAGGCTCAGA-3′; HNF-4α, 5′-CCTGGTCGAGTGGGCCA-AGT-3′ and 5′-TGGCAGACCCTCCGAGAAGC-3′; phenylalanine hydroxylase (PAH), 5′-TACCGGGCGAGACGAAAGCA-3′ and 5′-AAGGCCAGGCCACCCAAGAA-3′; ApoAI, 5′-GCATGATGCCTGGGCTCG-TC-3′ and 5′-TGGCCCTGGTGTGGTACTCG-3′; ApoAII, 5′-TGCTCGCAATGGTCGCACTG-3′ and 5′-ACTGGGCCTGGCACATCTCA-3′; RhoA, 5′-GTGGAATT-CGCCTTGCATCT-GAGAAGT-3′ and 5′-CACGAATTCAATTAACCGCATGAGGCT-3′. The amplification products for Alb, AFP, HNF-1α, HNF-4α, PAH, ApoAI, ApoAII, and RhoA were 759, 667, 479, 547, 320, 497, 318, and 696 basepairs, respectively. The specific amplification reactions were analyzed by agarose gel electrophoresis after visualization with ethidium bromide.

Intrasplenic Transplantation of Hepatocytes and Liver Injury.

Five million of cultured cells resuspended in 50 μl Ringer solution were injected into the spleen of severe combined immunodeficiency (SCID) mice. Ten mice were each employed for injection of MIM-GFP and MIM-Bcl-XL hepatocytes. Sublethal liver damage was induced by intraperitoneal injection of Jo-2 (0.1 mg/kg) 6 hours before intrasplenic transplantation, causing about 40% dead hepatocytes as determined by TUNEL analysis of liver sections. To enhance engrafting, the administration of Jo-2 was repeated five times once a week. Three control mice obtained successive Jo-2 treatment without transplantation. Transplanted cells were assessed by counting immunohistochemically labeled GFP-positive cells in sections of the regenerated liver and the percentage of transplanted cells was calculated from the ratio of exogenous GFP-positive cells to endogenous hepatocytes. The success rate was about 1% and 5% for MIM-GFP and MIM-Bcl-XL hepatocytes, respectively. All experiments were performed according to the Austrian guidelines for animal care and protection.

Immunohistochemistry and Confocal Immunofluorescence Microscopy.

Liver and spleen fixed in 4% phosphate-buffered formalin were embedded in paraffin. Four μm thick sections were stained with hematoxylin and eosin for standard microscopy. For immunohistochemistry, a rabbit (Santa Cruz Biotechnology, Santa Cruz, CA) or mouse (Boehringer Mannheim, Germany) antibody raised against GFP, a rabbit anti-albumin antibody (Nordic Laboratories, The Netherlands), a goat anti-AFP antibody (Santa Cruz), a goat anti-M2-PK antibody (Rockland, Gilbertsville, PA), and a mouse anti-cytokeratin 19 antibody (DAKO, Carpinteria, CA) were each used at a dilution of 1:100. Corresponding biotinylated secondary antibodies were used and visualization was performed with the Vectastain ABC kit employing diaminobenzidine as substrate (Vector Laboratories, Burlingame, CA). For immunofluorescence analysis, cye-dye conjugated secondary antibodies (Jackson Laboratories, Bar Harbor, ME) were applied and fluorescence microscopy imaging was done with a TCS-SP confocal microscope (Leica, Heidelberg, Germany).

Results and Discussion

Immortalization of Hepatocytes Isolated From p19ARF Null Mice.

Using collagenase liver perfusion and a standard purification procedure,14 hepatocytes were isolated from male p19ARF −/− mice and plated on collagen-coated tissue culture plastic.13 One day after seeding the cells displayed a morphology typical for primary hepatocytes and started to proliferate at day 3 of cultivation (Fig. 1A). In contrast, control hepatocytes isolated from littermates showed mitotic inactivity, onset of senescence, and degeneration 2 weeks after isolation (data not shown). Depending on the supplementation of FCS-containing culture medium with TGF-α, IGF-II, and insulin, a morphologically homogenous population of p19ARF null hepatocytes could be rapidly expanded to mass culture without obvious growth crisis by doubling in number at about every 36 hours (Fig. 1B). Additional treatment with the synthetic glucocorticoid dexamethasone resulted in an increase in generation time, as characteristic for rodent hepatocytes.20, 21

Figure 1.

Immortalization and conditions for expansion of hepatocytes isolated from adult p19ARF deficient mice. (A) Phase contrast microscopy images of hepatocytes (FCS plus GF) in culture for the indicated times. (B) Growth factor-dependent proliferation of MIM hepatocytes. A starting population of 5 × 104 cells were plated each at low density to monitor cell numbers over several days without contact inhibition. On day 9, however, cells show saturation density accompanied by growth arrest. The values depicted represent the means of triplicate measurements which were performed twice. GF: TGF-α, IGF-II, and insulin; Dex, dexamethasone.

Properties of Immortalized MIM Hepatocytes in Culture.

The pool of p19ARF null hepatocytes, termed MIM, strictly formed monolayers, and retained morphological features of liver parenchyma cells (Fig. 1A) as well as genetic stability as determined by flow cytometry after long-term cultivation (data not shown). Moreover, MIM cells failed to undergo anchorage-independent growth in semisolid medium and were not capable of forming tumors after subcutaneous injection into immunocompromised SCID mice (data not shown). A clone of MIM hepatocytes randomly isolated from immortalized MIM hepatocytes, showing strongest response to TGF-β1-mediated growth arrest, designated MIM-1-4, expressed an array of liver-specific markers, among them albumin, AFP, HNF-1α, HNF-4α, PAH, ApoAI, and ApoAII (Fig. 2). Typical for hepatocytes, GFP-expressing MIM-1-4 cells were sensitive to Fas-mediated apoptosis induced by the Fas agonistic antibody Jo-2 (Fig. 2B).22 By 3-fold overexpression of either antiapoptotic Bcl-2 or Bcl-XL (data not shown) in MIM-Bcl-2 and MIM-Bcl-XL hepatocytes, respectively, cell death was efficiently reduced (Fig. 2B). From these data we concluded that immortalized MIM cells cultivated from the liver of p19ARF −/− mice represent nontumorigenic hepatocytes which are sensitive to Fas-induced apoptosis.

Figure 2.

Expression of liver-specific markers and protection of MIM hepatocytes against Fas-induced apoptosis. (A) Monitoring of hepatocellular marker expression by linear RT-PCR of specific genes in total wildtype Balb/c liver (liver), primary wildtype Balb/c hepatocytes (prim. hep), and MIM-1-4 hepatocytes (MIM-1-4). PCR reactions with total brain (brain) and no cDNA input (no cDNA) are shown as negative controls. As a loading control, the constitutive expression of RhoA is shown. Alb, albumin; AFP, α-fetoprotein; HNF, hepatocyte nuclear factor; PAH, phenylalanine hydroxylase. The identity of each amplification product was verified by the predicted size. (B) Parental MIM-GFP, MIM-Bcl-2 or MIM-Bcl-XL hepatocytes were subjected to Fas-induced apoptosis and analyzed for cell death by quantification of TUNEL-positive cells, which are expressed in percent of untreated control cells.

Engrafted MIM Hepatocytes Show Liver-Repopulating Activity After Jo-2 Induced Injury and Display Hepatic Progenitor Cells in Ductular Structures.

To affirm that MIM cells represent functional hepatocytes, we further evaluated their ability to provide liver-reconstituting activity in vivo. To meet this task, MIM-GFP cells were injected into the spleen of SCID mice after an apoptotic liver damage induced by a single application of the Fas agonistic Jo-2 antibody. Donor hepatocytes, which have been accumulated in the spleen, contributed to liver restoration with an estimated abundance of about 1% in the parenchyma (Fig. 3A–C). This percentage of transplanted cells was calculated from the ratio of exogenous GFP-positive cells to endogenous hepatocytes in sections of the regenerated liver. The majority of transplanted MIM-GFP cells organized in ductular structures comparable to canals of Hering (Fig. 3C), whereas a lower portion of these hepatocytes localized in small-sized clusters as well as in isolated cells scattered throughout the entire liver. The emergence of such small-sized clusters of transplanted MIM-GFP hepatocytes might be either due to a sufficient resistance of cultured MIM-GFP cells against the cytotoxic effect of the Fas agonist, or might be an effect caused by the Jo-2-mediated destruction of the liver architecture. Donor cells organized in canals displayed a hepatocyte-like morphology, with a high cytoplasm to nucleus ratio as well as a cholangiocyte-like phenotype with poor cytoplasmic portions (Fig. 3C,D).3

Figure 3.

Intrasplenically transplanted MIM hepatocytes show liver reconstituting activity after Fas-induced liver injury. Immunohistochemistry employing an anti-GFP antibody (A,C,D) and hematoxylin/eosin staining (B) on tissue sections 16 days after transplantation. (A) The arrow indicates accumulated engrafted cells in the spleen. (B) Standard histology of the regenerated liver. (C) Repopulated MIM-GFP hepatocytes in the liver are organized in small-sized clusters (small arrows) or ductular structures (large arrows). (D) Magnification of a canal organized by donor cells. The left arrow indicates cells with a hepatocyte-like morphology, whereas the right one points to cells with a cholangiocyte-like phenotype.

Since Bcl-XL, in contrast to Bcl-2, has been shown to play a crucial role in hepatocyte physiology, and was efficiently protective against apoptosis (Fig. 2C),23 MIM-Bcl-XL hepatocytes were used for intrasplenic implantation and successive Jo-2-induced liver damage, aimed at enforcing donor-derived liver repopulation. This approach resulted in enhanced liver-reconstituting activity of donor cells, emerging with an abundance of about 5% in the liver parenchyma. As observed with MIM-GFP hepatocytes, most of the MIM-Bcl-XL donors localized in ductular structures, suggesting the potential of MIM hepatocytes to generate progenitor cells after successive liver regeneration. To determine whether MIM-Bcl-XL donors represent hepatic precursor cells, markers characteristic for the various liver cell types were used for colocalization with the transplantation marker GFP24: 1) α-fetoprotein (AFP) and albumin for oval cells and fetal hepatocytes; 2) M2-pyruvate kinase (M2-PK) for oval cells, fetal hepatocytes, and cholangiocytes; and 3) cytokeratin (CK) 19 for cholangiocytes. As compared to the Jo-2-treated control liver after regeneration (Fig. 4A,D,G,J), this analysis showed that engrafted MIM-Bcl-XL donor cells stained positive for AFP (Fig. 4B,C), M2-PK (Fig. 4E,F), albumin (Fig. 4H,I), and CK19 (Fig. 4K,L). Taken together, these findings provide first evidence that immortalized hepatocytes are capable of forming bile ducts and expressing markers specific for both oval cells and embryonic hepatocytes after transplantation in vivo.

Figure 4.

Transplanted MIM-Bcl-XL donors exhibit progenitor cell properties after successive Fas-induced liver damage. Confocal immunofluorescence images after staining sections of the regenerated liver (8 weeks after transplantation) with anti-GFP plus either anti-AFP (A–C), or anti-M2-PK (D–F), or anti-albumin (G–I) or cytokeratin (CK)19 (J–L). Arrows indicate GFP-positive donor cells (middle panel) localizing with respective marker proteins (right panel) in the regenerated liver with transplanted MIM-Bcl-XL cells (engrafted liver). Sections B, C, and E, F, and H, I are costained. Panels K and L show serial stains. As a control, the regenerated liver without transplanted cells is shown (left panel).

Here we show that p19ARF deficiency allows faithful immortalization of adult hepatocytes, as described for other cell types.11 The p19ARF null but p16INK4A positive hepatocytes employed for immortalization were obtained from the chemically unchallenged liver to avoid activation and isolation of oval cells. In addition, no enforced abundance of oval cells, which could be characteristic of p19ARF-deficient mice and could provide a bias for isolation of these cells, has been observed after comparing the expression of AFP and CK19 markers in liver sections of wildtype versus p19ARF null mice (data not shown). Accordingly, the hepatocytic morphology of immortalized MIM cells is indicated by granular-rich cytoplasm and large nuclei with many nucleoli, which is quite different from organelle-poor oval cells.3 In the context of the liver in vivo, however, the majority of MIM hepatocytes organize in ductular structures comparable with canals of Hering and display characteristics of oval cells which differentiate via alternative routes into different epithelial lineages. This observation indicates that hepatocytes are not narrowed down to generate unipotent progenies, but are still equipped to shift into a bipotential precursor compartment. Our data thus point to a broader plasticity of hepatocytic cell fate determination, as expected. Studies on the reversibility of hepatoblast-derived differentiation into cholangiocytes and hepatocytes in vitro and in vivo support this hypothesis.25, 26 To entirely rule out that the organization of transplanted MIM cells into ductular structures is a particular characteristic of p19ARF null hepatocytes, immortalized hepatocyte cell lines from other knock-out mice will be employed for transplantation and compared with the obtained findings. Yet the accessibility of the hepatic MIM model to genetic manipulation may provide novel insights into the differentiation repertoire of hepatocytes and specification of liver progenitor cells. In a therapeutic context, instructions for hepatocytes to yield liver progenitor cells facilitate new promising approaches superior to whole liver transplantation.5, 27


The authors thank Dr. Wilhelm Mosgoeller (Institute of Cancer Research, Vienna) for expertise in histological analyses and Dr. Dieter Printz (Children's Cancer Research Institute, St. Anna Kinderspital, Vienna) for cell sorting.