The present study was approved by the Ethics Committee of Central South University, Changsha, China. After Kunming white mice (n=6; Central South University Animal Studies) were sacrificed by cervical dislocation, the whole liver was completely resected and repeatedly washed with phosphate-buffered saline (PBS; Gibco, Carlsbad, CA, USA). In order to avoid any potential contamination by large vessel and biliary endothelial cells, identifiable vascular structures were excised from the liver specimens. The remaining liver tissue was sectioned into 5-mm³ cubes, and then transferred to a dish containing 2.0 U/ml of dispase and 1X penicillin-phytomycin (Sigma, St. Louis, MO, USA) and incubated at 4°C for 24 h. After terminating the digestion with 10% fetal bovine serum (FBS; Gibco) in MCDB 131 medium (Sigma), the liver cubes were mechanically disaggregated in MCDB 131 medium with a flat instrument to release the endothelial cells. The cell suspension was transferred to a 15-ml conical tube and centrifuged at 600 × g for 10 min. Following centrifugation, the supernatant was discarded and the pellet was resuspended in appropriate volumes of MCDB 131 medium. The cell suspension was then pipetted onto a density gradient of 35% Percoll (Sigma) and centrifuged at 12,000 × g, 4°C for 15 min. Following centrifugation, the band which was located on the red cell band of the gradient was transferred very carefully to a 15-ml conical tube containing PBS. After mixing gently, the sample was centrifuged at 600 × g, 4°C for 10 min and the pellet was resuspended in MCDB 131 medium. Following centrifugation at 100 × g for 5 min, the pellet was suspended in the liver endothelial cell culture medium and plated on 6-well tissue culture dishes pre-coated with fibronectin (Sigma). Non-adherent cells or debris were removed by washing steps after 5 h of culture at 37°C in 5% CO₂ in a humidified incubator. The adherent cells were further washed with complete endothelial cell selective medium and cultured in the same medium. The endothelial cell selective medium contained 40% MCDB 131, 40% endothelial cell growth medium (EGM)-2 (Lonza, Basel, Switzerland), 10% FBS and 10% endothelial cell conditioned medium (EC-CM, see below). The medium was also supplemented with the following growth factors: 1% L-glutamine (Gibco), 10 ng/ml vascular endothelial growth factor (VEGF; Invitrogen, Carlsbad, CA, USA), 10 ng/ml basic fibroblast growth factor (bFGF; Invitrogen) and 1 ng/ml dexamethasone (Sigma).

The preparation of the EC-CM was as follows: The mouse bone marrow endothelial cell line (a gift from Professor Qiru Wang, Central South University, China) was cultured in Iscove’s modified Dulbecco’s medium (IMDM) with 10% FBS until 80% confluent. The medium was replaced with 5 ml IMDM without serum in each 100-mm plate to collect the conditioned medium. Following incubation for 24 h, the culture medium was collected. The collected conditioned medium was centrifuged at 740 × g for 20 min. The supernatant was then filtered with a 0.22-μm filter and stored at −20°C until use. EC-CM was used within a week after being thawed.

All staining was performed according to the manufacturer’s instructions, using 10⁶ cells and the recommended amount of antibodies, followed by incubation for 30 min at room temperature. Briefly, the cells were incubated separately with the primary antibody rat anti-mouse CD31 (ab56299, 1:50; Abcam, Burlingame, CA, USA) for 30 min. After washing 3 times with PBS, the cells were incubated with the secondary antibody goat anti-rat Alexa Flour 488 (A-11006, 1:1,000; Invitrogen). For the negative controls, the cells were incubated only with the secondary antibody. The expression level of CD31 was determined using a FACScan flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA).

The LSECs were seeded in fibronectin-coated coverslips. The cells were fixed in 4% paraformaldehyde for 15 min. After washing 3 times with PBS, the cells were incubated for 1 h at room temperature in PBS containing 5% BSA (Sigma) to block non-specific binding. For determining the expression of von Willebrand Factor (vWF), the cells were premeabilized by incubation with 0.5% Triton X-100 (Sigma) in PBS for 10 min at room temperature and then incubated with mouse anti-mouse vWF antibody (555849, 1:200; BD Biosciences) followed by incubation with avidin-biotin complex (EastCoast Bio, Birmingham, UK) for 60 min at room temperature. The peroxidase reaction was developed with 0.3 mg/ml 3-3′-diaminobenzidine (EastCoast Bio) in buffer containing 0.05% hydrogen peroxide (EastCoast Bio) for 5–10 min at room temperature. Images were obtained under a phase contrast microscope (Nikon, Tokyo, Japan). For determining the expression of other CD genes, the cells were incubated with primary antibodies [rat anti-mouse CD31 (ab56299), rat anti-mouse CD105 (ab188488), mouse anti-mouse CD90 (ab226), goat anti-mouse CK19 and goat anti-mouse desmin (ab80503); Abcam] at a 1:50 concentration followed by incubation with corresponding secondary antibodies of goat anti-rat Alexa Flour 488 (A-11006, 1:1,000; Invitrogen), rat anti-mouse Ig-FITC (RMG101, 1:250; Invitrogen) and bovine anti-goat Ig-FITC (1:250; Invitrogen) for 30 min at room temperature. The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma), and the slides were mounted with vector-shield, covered and sealed with nail polish. Images were acquired under an inverted fluorescent microscope (Nikon).

An incorporation analysis of DiI-Ac-LDL was performed. The cell monolayer was incubated with 10 μg/ml of DiI-Ac-LDL (Biomedical Technologies Inc., Stoughton, MA, USA) in medium for 4 h at 37°C, washed 3 times and examined for the uptake of DiI-Ac-LDL under an inverted fluorescence microscope (Nikon).

The cells (2×10⁴) suspended in 200 μl EGM-2 medium supplemented with 50 ng/ml VEGF and 1% FBS were plated onto 50 μl of Matrigel (BD Biosciences) in a 96-well plate. Endothelial progenitor cells (EPCs) were used as a positive control, which were derived from the cord blood as previously described (40). The degree of tube formation was observed and photographed every 2 h under an inverted phase contrast microscope (Nikon).

The cells were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (Sigma) pH 7.4 for 6 h. After rinsing with ddH₂0, the specimens were post-fixed with 1% OsO₄ (Sigma) in ddH₂0 for 1 h, then rinsed with ddH₂0 prior to dehydration in a series of graded ethanol solutions and subsequently embedded in a mixture of Epon-Araldite. Thin sections were cut using a diamond knife mounted on an LKB Ultratome and stained with aqueous uranylacetate. The specimens were examined under a JEOL 1200 EX electron microscope (Jeol, Tokyo, Japan).

Positive SA-β-gal staining has been reported to reflect the replicative senescence of cells, but not in quiescent or terminally differentiated cells, and the stained cells are representative of the enzymatic activity (17). SA-β-galactosidase staining was performed as previously described (16). Briefly, the cells were fixed with 0.5% glutaraldehyde (pH 7.2). After washing with PBS (pH 7.2) supplemented with 1 mM MgCl₂, the cells were stained in X-gal solution [1 mg/ml X-gal, 0.12 mM K₃Fe (CN)₆, 0.12 mM K₄Fe (CN)₆, 1 mM MgCl₂] in PBS at pH 6.0 overnight at 37°C. The senescent (stained) and non-senescent (unstained) cells were viewed under a microscope (Nikon).

The density of the immortalized LSEC suspension at passages 20 and 21 was adjusted to 3×10³ cells/ml and plated on 24-well tissue culture dishes. Every other day, the cells were digested and enumerated. Cell growth kinetic curves were generated and the doubling-time was cacluated using the following formula: T = 0.693 (T2 − T1)/ln (N2 − N1) where T2−T1 is the difference in the value of time of twice successive detections; and N2−N1 is the difference in the value of cell numbers of twice successive detections.

Cell extracts were prepared using radio-immunoprecipitation assay lysis buffer (Sigma) containing 1 mM phenylmethanesulfonyl fluoride (Sigma). Proteins (50 μg) in the cell extracts were separated on a 10% SDS-PAGE gradient (Invitrogen) and transferred onto PVDF membranes (Millipore, Billerica, MA, USA) in accordance to the manufacturer’s instructions (Invitrogen). The membranes, which had been blocked with 5% skim milk, were incubated with mouse anti-p53 antibody (sc-374087, 1:500; Santa Cruz Biotechnology, Santa Cruz, CA, USA) and mouse anti-β-actin antibody (A1978, 1:3,000; Sigma), followed by incubation with horseradish peroxidase-conjugated goat anti-mouse IgG secondary antibody (ab175740, 1:10,000; Abcam). After washing with PBS-T solution 4 times, the immunoreactive bands were detected by electrochemiluminescence (Sigma) by exposing the blots.

The cells were harvested and G-banded chromosome preparations were made using standard methods (18). The harvested cells were fixed in a methanol:acetic acid (3:1) solution. The chromosomes were stained with 4% Giemsa solution in Gurr buffer and the number and G-C band of the chromosomes in metaphase from each cell line were determined.

To measure cell anchorage independence, the cells (1×10⁴) were plated in 6-well soft agar dishes (0.5% bottom and 0.33% top agar, respectively) for 3 weeks. HPG-2 cells (human hepatocellular carcinoma cell line; Cell Bank of Central South University, Changsha, China) were used as positive controls. The ability for anchorage-independent growth shown by colony formation was analyzed after 20 days. Colonies containing >10 viable cells were scored as positive.

The immortalized LSECs were assayed for the ability to form tumors in SCID mice. Ten million viable cells were injected into the lower limb muscles of 6- to 8-week old SCID mice. Control mice were injected with HPG-2 cells. The animals were observed at regular intervals for tumor development for a period of 6 months. The mouse muscle tissue sections were stained with hematoxylin and eosin (H&E).

After the plating of the cells isolated from the mouse livers on fibronectin-coated culture dishes, small adherent clusters of cells were observed within 24 h. With the prolonged culture, these clusters slowly became enlarged until they formed a confluent monolayer. Replating of the cells after digestion revealed clonal growth and a typical cobblestone morphology, a characteristic of endothelial cells (Fig. 1A); the colonies were picked up, pooled and expanded for several weeks, and these culture showed a very homogenous morphology and strictly grew as a monolayer (Fig. 1B). The contaminating non-endothelial-like cells were successfully eliminated by the use of endothelial-selective medium and contrast-digestion after the second passage in serial culture.

At present, the only gold standard defining LSECs is the presence of fenestration with a diameter of 100–150 nm. We examined the monolayer of the cultured LSECs by transmission electron microscopy. As shown in Fig. 1C, these cells had typical transcellular fenestrations.

The most commonly used characteristics to identify endothelial cells include the expression of vWF and CD31, the uptake of acetylated low-density lipoprotein (Ac-LDL) and the formation of a capillary-like structure in vitro (19,20). The monolayers of the LSECs showed positive staining for vWF expression (Fig. 1D) and the uptake of Dil-Ac-LDL following 4 h of exposure to Dil-Ac-LDL (Fig. 1E). In addition, the LSECs formed typical capillary-like structures in Matrigel after 14 h of culture (Fig. 1F), and the expression of CD31, a marker of vascular endothelial cells, was not detected (Fig. 1G). To examine whether these LSECs were contaminated by other liver cells, we further examined the expression of CK19 (a marker of hepatocytes) and CD105, CD90 and desmin (markers of mesenchymal fibroblast-like cells which reside in perisinusoidal, portal and around the centrilobular vein; desmin is also a classical stellate cell maker). As shown in Fig. 1H and I, the expression of CK19/CD90/CD105/desmin was not detected in the monolayer of LSECs (CD105 and desmin; data not shown).

By passage 8, the majority of the LSECs appeared senescent as judged by the presence of numerous LSECs which exhibited a large and flattened morphology and SA-β-Gal activity (Fig. 2A). Following extended culture, a small population of round strong refractive proliferating cells emerged from the senescent cultures, and these cells showed clonal and multilayer growth (Fig. 2B), and their proliferation rate increased with the increase in the number of passages. As shown in Fig. 2C, at passages 20 and 21, these cells continued vigorous growth. The doubling time of the cells at passages 20 was 4.01 days, while the doubling time of the cells at passage 21 was 3.50 days. Thus, the cells at passage 21 displayed a higher proliferation potential than those at passage 20. All these characteristics were consistent with those of immortalized cells (13,14).

To clarify whether the immortalized LSECs had different properties compared to primary LSECs, we examined the characteristics of the immortalized cells. As shown in Fig. 3A, the immortalized LSECs maintained typical transcellular fenestrations, as well as the expression of vWF and the uptake of DiI-Ac-LDL; these characteristics were consistent with those of the primary cells. However, as shown in Fig. 3B, the immortalized LSECs lost the ability to form capillary-like structures, but obtained the expression of CD31 and desmin (Fig. 3C and D).

We then examined p53 protein expression and karyotype in the immortalized LSECs, which are both potentially important factors of immortalization. Compared with the primary LSECs, the protein expression level of p53 in the immortalized LSECs was significantly upregulated (Fig. 4A). However, the immortalized LSECs showed a normal karyotype (Fig. 4B).

Given that these immortalized LSECs showed some characteristics of transformed cells, such as growth characteristics, immunophenotypes and an extended lifespan, we futher analyzed their tumorigenic potential. As shown in Fig. 4C, no colony formation was observed in the immortalized LSECs in comparison to the HPG-2 cells, which served as a positive control. Therefore, the immortalized LSECs were not able to grow in an anchorage-independent manner, indicating no tumorigenic potential and a high differentiation status. This notion was further supported by the fact that all the SCID mice implanted with the immortalized LSECs did not develop any tumors within 6 months; by contrast, the SCID mice implanted with the HPG-2 cells developed tumors within 3 weeks (Fig. 4D).