Human fetal livers were obtained from elective abortions. Gestational age was determined by ultrasonic measurement of the diameter of the skull and ranged from 14 to 18 weeks. The use of this tissue was approved by the Medical Ethical Committee of the Academic Medical Center, Amsterdam, the Netherlands, subject to informed patient consent in compliance with the Helsinki Declaration. We isolated HFLCs on three independent occasions; in each case four fetal livers were pooled. Cells were isolated as described previously [14]. HFLCs were seeded in DMEM culture medium (Dulbecco's modified Eagle's medium, BioWhittaker) containing 10% heat-inactivated fetal bovine serum (HI-FBS, BioWhittaker), 2 mM L-glutamine (BioWhittaker), 1 μM dexamethason (Sigma), 10 μg/mL insulin, 5.5 μg/mL transferrin, 6.7 ng/mL selenium-X (ITS mix, Life Technology), 100 U/mL penicillin, 100 μg/mL streptomycin (penicillin/streptomycin mix, BioWhittaker) at a density of approximately 3*10⁵ cells/cm² in Primaria 6-well plates (BD Falcon). Clonal derivatives were obtained by limiting dilution. The selection procedure used and the functionality of the clonal derivatives are described elsewhere [14]. Near-confluent cultures were detached by 5 min incubation with 0.25% trypsin/0.03% EDTA (BioWhittaker) and split at 1:4 ratios. The number of population doublings (PD) was calculated as PD = log (N_f/N_i)/log 2, in which N_f is the final number of cells harvested and N_i is the number of cells initially seeded. No corrections were made for cells that did not re-attach after passaging, since their proportion was negligible. The period in which PD number progressed linear with culture time was used to calculate the PD time (T_PD).

Mature primary human hepatocytes were isolated from seven patients undergoing partial hepatectomy, because of metastatic carcinoma. The tumour free liver tissue used in each case for the hepatocyte isolation ranged between 2 to 10 grams. The procedure was approved by the Medical Ethical Committee of the Academic Medical Center subject to informed patient consent. The hepatocyte isolation method was adapted from the protocol described by Seglen [15] as previously described [14].

NKNT-3 cells were kindly donated by Prof. I. Fox, University of Nebraska, USA. The NKNT-3 cells were cultured on Primaria 6-well culture plates (BD Falcon) and in 75 cm2 culture flasks using CS-C complete serum free medium (Cell Systems Corporation) with 0.2 mg/ml hygromycin B (Invitrogen) and 1 U/ml penicillin/streptomycin (BioWhittaker). Cultures were passaged with a split ratio of 1:5 according to instructions for CS-C medium. Cre-mediated recombination to revert immortalization of NKNT-3 cells [4] was carried out by transduction of the adenoviral vector AxCANCre (Riken DNA Bank (Tsukuba Life Science Center, Japan) as described previously [5]. We analysed both reverted, hence transduced with AxCANCre and selected with G418, and unreverted i.e. untreated cells. HepG2 cells were obtained from ATCC (HB-8065) and cultured in Primaria tissue culture flasks in DMEM culture medium as described above for HFLCs.

All cultures were maintained at 37°C in a humidified atmosphere (95% air, 5% CO2) and the medium was changed every 2-3 days.

The cDNA of the human telomerase reverse transcriptase (hTERT) gene, kindly provided by R.L. Beijersbergen, Netherlands Cancer Institute, the Netherlands, was introduced in the cells by lentiviral transduction. The lentiviral vector backbone was described as LTRCMVR2 by Marcusic et al. [16]. The lentiviral vector contained a cytomegalovirus promoter controlling the expression of a reverse tetracycline (Tet) responsive transcriptional activator and a Tet responsive element controlling the expression of the hTERT gene. In this system hTERT transcription can be increased by adding 1 μg/mL doxycyclin to the medium.

The lentivirus was produced as previously described [17]. In brief, HEK293T cells were transiently transfected by calcium phosphate precipitation with a third generation lentiviral vector system. Virus containing supernatant was collected at 24 and 48 hours following transfection, filtered through 0.45 μm Millipore filters, then 100-fold concentrated by centrifugation and added to the culture medium of the HFLCs in 24-well culture plates in a 1:50 (lentivirus preparation:culture medium) ratio. The hTERT lentivirus was introduced in three independent HFLC cultures and in five clonal derivatives, i.e. cBAL08, cBAL09, cBAL20, cBAL21 and cBAL29. Cells were passaged twice before integration of the lentiviral vector was confirmed by PCR using genomic DNA of transduced cells as template.

In the transplantation experiment cBAL111 cells were marked with Green Fluorescent Protein (GFP) by transduction using lentiviral construct pRRLcpptPGKGFPpreSsin [17] carrying the GFP gene under control of a phosphoglycerate kinase promoter. By fluorescence-activated cell sorting it was demonstrated that ≥ 95% of the transduced cells were GFP positive. There were at least five passages between transduction with GFP and transplantation of the cells. The GFP positivity, ratio as well as expression level, of the cells remained unchanged during passaging.

The cBAL111 cells, cultured for 15 days, were analyzed for progenitor cell markers by flow cytometry. For comparison HepG2 cells were included in the experiment. The cells were detached using accutase (Innovative Cell Technologies, Inc., USA), washed twice in DMEM culture medium, resuspended in 100 μl DMEM and incubated for 30 minutes on ice in the dark with 10 μl of the following antibodies: 1. a combination of mouse-anti human CD34-FITC and mouse-anti human CD326 (EpCAM)-APC (both Miltenyi Biotech Inc.); 2. a combination of mouse-anti human CD146-FITC (Miltenyi Biotech Inc., USA) and mouse-anti human CD326 (EpCAM)-APC; 3. mouse-anti human CD133 (Miltenyi Biotech Inc., USA). The cells incubated with the CD133 antibodies were washed in culture medium and subsequently incubated with the secondary antibody goat-anti mouse IgG-Alexa fluor 488 (Molecular Probes Invitrogen, USA). As negative controls, cells were incubated with either a combination of mouse IgG2a-APC and mouse IgG1-FITC (isotype controls) (both eBioscience, USA) or with goat-anti mouse IgG-Alexa fluor 488. The cells were analyzed by flow cytometry using BD FACScalibur (BD Biosciences, USA) and data were analyzed with WinMDI 2.8 software.

Hepatocyte function tests were performed at confluence in 6-well plates. After washing the cells twice with phosphate buffered saline (PBS, NBPI International) culture medium was replaced by 2.5 mL of test medium (William's E medium with 4% HI-FBS, 2 mM L-glutamine, 1 μM dexamethason, 20 mU/mL insulin (Novo Nordisk), 2 mM ornithine (Sigma-Aldrich), 100 U/mL penicillin, 100 μg/mL streptomycin, 0.5 mM NH4Cl, 2.75 mM D-galactose (Sigma), 90 μM lidocaine HCl (Sigma-Aldrich). Medium samples were taken after 0 and 72 hours of incubation. The cells were then washed twice with PBS and stored at -20°C for protein determination.

Urea concentrations were determined using the blood urea nitrogen test (Sigma Chemical Co). Albumin concentrations were determined via enzyme linked immunosorbent assays using cross-absorbed goat-anti-human albumin antibodies (Bethyl). Lidocaine concentrations were measured by fluorescence polarization immunoassay using a TDxFLx analyzer (Abbot Laboratories). Galactose concentrations were determined by the absorbance of nicotinamide-adenine dinucleotide (NADH) at 340 nm after enzymatic reaction with galactose dehydrogenase (Roche). Total protein/well was quantified by spectrometry using Coomassie blue (Bio-Rad). Production rates were established by calculating the changes in concentration during time and corrected for protein content.

RNA was isolated from the cell lines by using TRIzol (Boehringer Mannheim). As a reference, human liver samples were included in the analyses. First strand cDNA was generated from 500 ng of total RNA using 20 pmol of gene-specific RT primers specific for the mRNA of Albumin, α-1-Antitrypsin (AAT), Transferrin, Hepatocyte Nuclear Factor 4 α (HNF4α), Alpha-fetoprotein (AFP), π class Glutathione S transferase (GST π) and hTERT in combination with 5 pmol of RT primer for 18S ribosomal RNA and 134 units Superscript III (Invitrogen). Real-time reverse transcription PCR (RT-PCR) using SYBR green I (Roche) was performed as described previously [18]. The sequences of the RT and PCR primers and PCR conditions are given in Table 1.

Starting levels of mRNA, except for hTERT, were calculated by analyzing linear regression on the Log (fluorescence) per cycle number data using LinRegPCR software [19]. Starting levels of hTERT mRNA were calculated by standard curve analysis, using serial dilutions of hTERT containing plasmid ranging from 10² to 10⁹ copies/reaction and LightCycler software (Roche). The mRNA starting levels of Albumin, AAT, Transferrin, AFP, GST π, and hTERT were normalised for the starting levels of 18S ribosomal RNA. Normalised mRNA levels, except for hTERT mRNA levels, are expressed as a percentage of the mean mRNA starting levels of the two liver samples normalised for 18S ribosomal RNA starting levels.

For the detection of glutamine synthetase (GS), cBAL111 cells were cultured on 8-wells culture-slides (BD Falcon) for two days. Then the cells were washed twice with PBS and fixed by a 10-minutes incubation with ice-cold methanol-aceton-water mixture (2:2:1). Cells were incubated with 70% ethanol for 5 minutes, washed with PBS and subsequently incubated overnight with monoclonal GS antibody (Transduction Laboratories, Lexington, KY, G45020) diluted 1:1000. Antibody binding was visualized with the indirect unlabelled antibody peroxidase anti-peroxidase (PAP) method [20].

For the detection of albumin, cytokeratin (CK) 18 and CK19, cBAL111 cells were seeded on Immunoslides (ICN, Aurora, Ohio, USA) and cultured for two and 15 days. Cells were washed once in PBS with 0.1% Tween-20 and fixated in 4% paraformaldehyde in PBS for 15 minutes at room temperature. Cells were washed as before and incubated with a blocking buffer (3% BSA, 0,2% Fish gelatin (Sigma), 2%FCS) for one hour at room temperature. After a further washing step, the cells were incubated with the primary antibody for one hour at room temperature. As primary antibodies we used anti-albumin antibody (Sigma), mouse-anti-human CK18 (sc-6259, Santa Cruz) and mouse-anti-human CK19 (Santa Cruz) for detection of albumin, CK18 and CK19, respectively. Cells were washed again as before and incubated with 28 μg/mL Cy2 conjugated goat-anti-mouse IgG and 1 μg/mL tetramethylrhodamine isothiocyanate (TRITC) conjugated phalloidin for 1 hour in a humidified chamber at room temperature. Forty minutes before the end of this incubation period, 20 ng/mL Diamidinophenylindoldiacetate (DAPI) was added. Cells were washed again as before and embedded in Polymount (Polyscience, Washington, USA) and covered with a coverslip. Slides were analysed using an Axiovert 200 fluorescence microscope.

Cells were added to 0.35% low-melting-temperature agarose (Seaplaque) containing DMEM culture medium as described above and transferred at a density of 5000 cells/well to 6-well plates previously lined with 0.5% agar DMEM culture medium. After 15 days, the colonies were stained with 0.005% Crystal violet and counted.

The cBAL111 cells overexpressing GFP were transplanted into four 6 week old Rag2-/-γc-/- mice [21]. The mice were anesthetized with an intraperitoneal injection of FFM mixture (2.5 mg Fluanisone/0.105 mg Fentanyl citrate/0.625 mg Midazolam HCl/kg in H₂O, 7 mL/kg). One million GFP-marked cBAL111 cells suspended in 100 μL HBSS were injected into the inferior tip of the spleen as described [22]. At both nine and 34 days after transplantation, two mice were sacrificed. The livers and spleens were harvested after in vivo fixation. For in vivo fixation, mice were anesthetized as described above and 25 mL PBS was flushed trough the circulation, followed by 25 mL 2% paraformaldehyde (PFA) in PBS. Subsequently, liver and spleen were harvested and cut into pieces of approximately 0.2 cm³. The tissue pieces were incubated in 4% PFA in PBS for four hours, followed by an overnight incubation in 30% sucrose solution. Tissues were snap frozen in liquid nitrogen and stored at -80°C.

Cryosections were 6 μm thick and were mounted on poly-L-lysine coated slides. Sections were incubated in Teng-T (10 mM Tris, 5 mM EDTA, 0.15 M NaCl, 0.25% gelatin and 0.05% Tween-20, pH 8.0) for 30 minutes before incubation with primary antibodies. Human mitochondria were visualized with a mouse-anti-human mitochondria antibody (Chemicon International) in a 150-fold dilution; vimentin was visualized with a mouse-anti-vimentin antibody, clone 9 (Boehringer Mannheim) in a 1000-fold dilution; carbamoylphosphate synthetase (CPS) was visualized with rabbit-anti-CPS antibody in a 1500-fold dilution and glutamine synthetase (GS) was visualized with monoclonal GS antibody (Transduction Laboratories) in a 500-fold dilution, human albumin was visualized with goat-anti-human albumin antibody in a 1:200 fold dilution. As a secondary antibody Alexa594 conjugated goat-anti-mouse IgG (Molecular Probes) was used in a 1000-fold dilution for the detection of vimentin and human mitochondria and in a 250-fold dilution for detection of GS. CPS antibodies were detected with Alexa594 conjugated goat-anti-rabbit IgG (Molecular Probes) in a 250-fold dilution. Goat-anti-human albumin immunoglobulins were detected using rabbit-anti-goat IgG conjugated with Alexa 594 (Molecular Probes). Slides were mounted in Vectashield containing 1 μg/mL 4,6-diaminidino-2-phenylindole (DAPI) to counterstain DNA.

Fluorescent in situ hybridization (FISH) analysis was performed on 6 μm sections of paraffin embedded liver tissue as described before [23]. Briefly, sections were treated to remove paraffin and the sections were denatured in 70% formamide for 2.5 minutes. 150 ng Biotin 11-dUTP labeled human genomic DNA, 200 ng digoxigenin 11-dUTP labeled murine genomic DNA and 5 μg salmon sperm DNA were denatured together and hybridized overnight at 37°C with the sections. After washing the slides, human DNA was visualized with FITC anti-avidin followed by biotinylated anti-avidin antibodies (Vector laboratories), whilst mouse genomic DNA was visualized with sheep rhodamine anti-digoxigenin followed by Texas red anti-sheep antibodies (Vector laboratories). Slides were mounted in Vectashield (Vector laboratories) containing 1 μg/mL DAPI to counterstain DNA.

Student's t tests were used to determine statistical differences. Significance was reached if P < 0.05. SPSS 12.0.1 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis. Average values (± standard error) are reported.

The maximum number of PDs between the HFLC preparations and clonal derivatives were different. Expression of hTERT was detected in some of the clonal derivatives analysed, however, all HFLCs and clonal derivatives eventually entered a state of terminal growth arrest (Table 2).

After lentiviral introduction of the hTERT gene, the presence of the hTERT cDNA was confirmed in HFLCs and clonal derivatives by PCR on genomic DNA (results not shown). However, the HFLC cultures did not overcome the terminal growth arrest after the introduction of the hTERT gene; only one of the HFLC cultures' lifespan was extended by 30%. In addition, only one of the clonal derivatives was able to overcome the terminal growth arrest after the introduction of hTERT. This clone, cBAL08, previously showed the longest life span of 61 PDs and a relatively high endogenous hTERT expression (Table 2). Because the transduced cell line was capable of more than 120 PDs, which is twice the life span of the parental cell line cBAL08, and still did not show any sign of growth arrest, we considered this cell line to be immortalized and named it cBAL111. The hTERT mRNA levels of cBAL111 were 1 × 10⁵-fold higher as compared to its parental cell line cBAL08 (Table 2).

The organisation of the cBAL111 cell layer changed during culturing; the cells displayed a cubic shape at day 15 (Fig. 1C and 1D) instead of the more spindle shape at day 2 (Fig. 1A and 1B). The cBAL111 cultures were 100% positive for the hepatocyte markers GS, CK 18 and albumin as well as the cholangiocyte marker CK 19 at day 2 and day 15 after seeding (GS and albumin are shown at day 2 and CK18 and CK19 are shown at day 15 in Fig. 1). The CK18 staining was predominantly around the nucleus reaching into the cytoplasm, which is similar to the CK18 staining of primary human hepatocytes that are dedifferentiating in vitro [24]. These data suggest that under these conditions the cBAL111 cells resemble cells with progenitor characteristics rather than fully differentiated hepatocytes.

The 15 day-cultured cBAL111 cells and, for comparison, hepatoblastoma derived HepG2 cells were characterized for three progenitor cell markers by flow cytometry, i.e. CD146, a marker for various cells including mesenchymal progenitor cells and hepatocytes during liver regeneration [25,26], CD326 (EpCAM), an epithelial marker upregulated in gastrointestinal carcinomas [27] and CD133, a marker for progenitor cells of various origins, like endothelial cells, haematopoietic cells and neural cells [28]. The cBAL111 cells were positive for CD146, but negative for CD326 (EpCAM) (Fig. 2A and 2B) and CD133 (data not shown). HepG2 cells were also negative for CD133 (data not shown), but, in contrast to the cBAL111 cells, positive for CD326 and negative for CD146 (Fig. 2C and 2D).

Because the immunostainings suggested a transition in the organisation of the cell layer of cBAL111 cells, we tested their in vitro functionality at 2 days of culture, when the cells displayed a spindle like morphology and the culture was not confluent and at day 15, when cells were more cubic and confluence was reached. There was a trend of increased expression of markers for mature hepatocytes (production of albumin and urea, galactose elimination and transcript levels of transferrin, AAT and HNFα) with culture time and decreased expression of markers for immature hepatocytes (GSTπ and AFP) (Table 3). The in vitro functionality of cBAL08 after reaching confluence was comparable with that of cBAL111, particularly at day 15; no significant difference was detected between cBAL08 and cBAL111 at day 15 for urea and albumin production and the mRNA levels of Albumin, AAT, Transferrin, GSTπ and AFP differed less than 2.5-fold (Table 3). So, the immortalization of cBAL08 maintained the investigated hepatic functions. Further extension of the culture period did not clearly increase hepatic functionality (results not shown).

To investigate the putative tumorigenicity of cBAL111, the cells were seeded in soft agar. The cBAL111 cells were not able to form colonies in soft agar, where HepG2 as positive control, formed 61 ± 21 colonies from 5000 cells. The inability of cBAL111 cells to grow in an anchorage independent way, is an indication that cBAL111 cells are not tumorigenic.

The novel in vitro immortalized fetal hepatocyte cell line cBAL111 was subsequently compared with HepG2 cells, the conditionally immortalized hepatocyte cell line NKNT-3 and primary (mature) human hepatocytes two days after seeding. The primary human hepatocytes did not exhibit any decline in the tested functions within two days (results not shown). HepG2, cBAL111 at day 15 and NKNT-3 cells, both un-reverted and reverted for immortalization, synthesized urea, at a level 11-32 fold lower than mature hepatocytes (Table 3). The HepG2 cells and to a lesser extent the cBAL111 cells produced albumin in contrast to NKNT-3 cells. The galactose elimination of the cBAL111 cells at day 15 reached to 19% of the levels of the mature hepatocytes. The cBAL111 cells, like fetal hepatocytes, did not eliminate lidocaine, a marker for cytochrome P450 1A2 and 3A4 activity, whereas the mature hepatocytes eliminated 9.8 ± 9.3 nmol/h/mg protein.

In cBAL111 and NKNT-3 cells, the mRNA levels of albumin, AAT and transferrin, (markers of hepatocyte differentiation) were less than 1% of the corresponding levels in human liver. However, the transcript levels of HNF4α, a master regulator of liver development, increased up to 17% of the human liver levels in the cBAL111 cells at day 15. HepG2 cells showed mRNA levels for albumin, AAT and transferrin that were comparable to human liver. As markers associated with immature hepatocytes, AFP mRNA levels of cBAL111 and NKNT-3 cells were comparable to the in vivo level, while the GST π mRNA levels were 10-44 times higher. In HepG2 cells, mRNA levels for AFP were 1000-fold higher than the in vivo levels and GST π levels were undetectable. In summary cBAL111 at day 15 was positive for all tested hepatocyte parameters, whereas HepG2 and NKNT-3 cells both lacked one function.

To determine whether cBAL111 cells have the potential for hepatic differentiation, the cells were marked with GFP by lentiviral transduction and transplanted in the spleen of 4 immunodeficient mice. Nine and 34 days after transplantation, GFP expressing cells were detected in the murine spleen (results not shown) and liver (Fig. 3A and 3B). The presence of GFP positive cells in the murine livers was relatively low (~1%), since no liver repopulation model was applied, and remained stable between nine and 34 days. The cells appeared as single cells and no cell clusters were detected, which again suggested a lack of tumorigenicity.

The majority of these cells exhibited an elongated morphology, however a small number of cells (~1%) had the morphological characteristics of hepatocytes, given their cuboid appearance. No differences were observed between the livers harvested at nine and 34 days after transplantation. The GFP positive cells were confirmed to be from human origin by immunohistochemistry using a human specific antibody binding to mitochondria (Fig. 3C and 3D). Immunohistochemistry using an antibody against vimentin, a marker for dedifferentiated and mesenchymal cells [29], indicated a high expression in cBAL111 in vitro (data not shown) and in all the elongated cBAL111 cells found in the mouse liver, whereas the cBAL111 cells with cuboid appearance did not or hardly expressed vimentin (Fig. 3E). Furthermore these GFP positive hepatocyte-looking cells with cuboid morphology were indistinguishable from the surrounding mice hepatocytes with regards to CPS expression, which is expressed periportally [30] (Fig. 3F). In contrast, elongated cBAL111 cells did not express CPS. GS expression, which is present in vitro (Fig. 1), was absent in the transplanted cells in the periportal areas (Fig. 3G). No GFP positive cells were detected in pericentral areas, the site of GS expression in normal liver. A possible explanation for this is that the cells may have entered the liver via the portal vein and engrafted before reaching the pericentral area. Finally, the cuboid GFP positive cells stained positive for human albumin (Fig. 3H-J), whereas the elongated GFP positive cells were negative (not shown).

We then tested whether the GFP positive hepatocytes were the result of fusion between human cBAL111 cells and murine hepatocytes [31], by using FISH analysis. The results showed that nuclei reacting to the human probe (Fig. 4A and 4D) were negative for the murine probe (Fig. 4C and 4F). From the shape of the nuclei and the localization of cells, we concluded that these were cBAL111 derived hepatocytes.