The rat HCC cell line, McA-RH7777 (MRH), was obtained from the American Type Culture Collection (ATCC; Rockville, MD, USA). Buffalo rats (6–8 weeks old) were obtained from the Charles River Laboratories (Davis, CA, USA). All animal studies were conducted in accordance with the animal care policy of the Fudan University and the Animal Research Committee.

Intrahepatic tumor implantation was carried out according to Yang et al(12) with minor modifications. MRH cells were injected into the flank of Buffalo rats. HCC tissue was recovered from the transplanted rats 1 month after tumor cell injection, and cut into cubes ~2×1×1 mm³. The rats were anesthetized with ketamine. A small subxiphoid midline incision was made to expose the left lateral lobe of the liver. A small superficial incision was made in the liver and a cube (size, 2×1×1 mm³) of tumor was transplanted. The liver incision was closed with 7–0 suture to avoid any possible early peritoneal seeding of the HCC. Tumour extensions were determined by ultrasound on days 7, 14, 21 and 28 after tumor establishment.

Whole lungs of the recipient rats were harvested on days 7, 14, 21 and 28 for quantitation of metastasis as follows. The trachea of the rat was catheterized with a 22-gauge Angiocath (Becton Dickinson Vascular Access, Sandy, UT, USA) and the lungs were insufflated with 15% India ink solution. The lungs were preserved in Fekette’s solution for 24 h, followed by 10% formalin. The tumor nodules on the surfaces of the lungs and within the lungs were counted under an operating microscope at ×10 magnification (13).

Cross-sections of liver were checked for the presence of tumor tissue on days 7, 14, 21 and 28 after intrahepatic tumor transplantation. Rats were randomly selected for sacrifice on days 7, 14, 21 and 28 (n=4 each), respectively. Livers were fixed in formalin, embedded in paraffin, and stained with H&E for histological examination.

Liver biopsies taken at all check-points were incubated overnight at room temperature with phosphate-buffered saline (PBS), 10% sucrose and 4% formaldehyde solution. Specimens were frozen at −70°C and 7-μm-thick frozen sections were prepared using a cryostat (Leica CM3000; Meyer Instruments, Houston, TX, USA). Cellular hyperpermeability was achieved using the 0.3% Triton (Sigma, Steinheim, Germany). Bovine serum albumin 1% was used to block non-specific background staining. Following washing with PBS, the slides were incubated with primary antibody to lymphocyte markers and anti-α-SMA rabbit primary monoclonal antibody (mAb) (Roche, Basel, Switzerland) at a dilution of 1:750 for 1 h at room temperature in the dark, then washed three times with PBS. The secondary antibody Cy-5 (BD Biosciences, Franklin Lakes, NJ, USA) was then added, followed by three washes with PBS. In each double-stained set, a single lymphocyte subset was stained using anti-α-SMA. Fluoresceine isothiocyanate (FITC)-conjugated mouse anti-rat CD4 and allophycocyanin (APC)-conjugated mouse anti-rat CD8 cell mAbs (BD Biosciences) were used, respectively. Sections were stacked and covered with Fluoromount-G (Birmingham, AL, USA) to maintain staining and prevent fluorescent bleaching. Sections were then stored at 4°C until analysis by confocal microscopy (14,15).

For the detection of apoptosis, periodate-lysine-paraformaldehyde-fixed cryosections were stained by the TUNEL technique, using an In Situ Apoptosis Detection kit (Roche), according to the manufacturer’s instructions. Sections were incubated at 37°C for 1 h with terminal deoxynucleotidyl transferase in the labeling-safe buffer including FITC, and observed under a confocal laser scanning microscope.

After TUNEL staining, cryosections were incubated with Cy5-conjugated anti-rat α-SMA mAb (1:500 dilution; Roche) or phycoerythrin (PE)-conjugated anti-rat CD3 mAb (1:100 dilution; Serotec) overnight at room temperature. They were observed under a confocal laser scanning microscope. FITC images and PE or Cy5 images were entered separately into the computer and superimposed using Adobe Photoshop software (Adobe Systems UK, Uxbridge, UK).

Rats were anesthetized by intraperitoneal injection of ketamine, 10 mg/100 g body weight. HSCs were prepared from the livers of normal Buffalo rats or HCC tissue, as previously reported (16). In brief, the abdominal cavity of the anesthetized animals was opened to canulate the portal vein. The liver was subsequently perfused at a flow rate of 10 ml/min with Gey’s balanced salt solution (GBSS) without Ca²⁺ and Mg²⁺ for 10 min, followed by 100 ml of 0.12% pronase E (Roche) dissolved in GBSS with Ca²⁺ and Mg²⁺ (GBSS⁺) for 10 min. Then 100 ml of 0.1% collagenase (Roche) dissolved in GBSS⁺ was recirculated for 20 min. The normal liver and HCC tissue were then excised, dissected and incubated for 30 min at 37°C with continuous shaking, with 0.04% pronase E, 0.05% collagenase and 0.002% DNase I (Sigma) in 100 ml GBSS⁺. After digestion, the cell suspension was passed through a 0.22-μm mesh and centrifuged at 480 × g for 10 min. Subsequently, cells were purified by 8% Nycodenz (Sigma) gradient centrifugation. HSCs were grown in DMEM medium (Gibco-BRL) containing 10% fetal calf serum (FCS) before the experiments were performed. Cell viability determined by trypan blue exclusion was >90%. The purity of quiescent (q) HSCs and tHSCs ranged from 90–95%, as determined by desmin immunostaining. tHSC activation was determined by α-SMA immunostaining. qHSCs and tHSCs were obtained by culture on uncoated plastic dishes for 2 days (17).

Bone marrow cells isolated from F344 rat (Davis, CA, USA) femurs were lysed of red blood cells using red blood cell lysis buffer (Sigma), cultured in RPMI-1640 medium containing 10% v/v heat-inactivated FCS, 20 mmol/l hydroxyethylpiperazine-N-2 ethanesulfonic acid, 2 mmol/l L-glutamine, 0.1 mmol/l non-essential amino acids, 1 mmol/l sodium pyruvate, 20 μmol/l 2-mercaptoethanol, and antibiotics (100 U/ml penicillin and 100 mg/ml streptomycin) (subsequently referred to as complete medium) in the presence of rat recombinant granulocyte-macrophage colony-stimulating factor (4 ng/ml) and interleukin-4 (1,000 U/ml) (both from BD Bioscences). Non-adherent cells were released spontaneously from the proliferating cell clusters, harvested, washed, and resuspended in complete medium, as previously described (18).

Proliferation of nylon-wool-eluted Buffalo rat spleen T cells (2×10⁵/well in 100 μl) was stimulated via allogeneic DCs in a one-way mixed leukocyte reaction. Non-stimulated (quiescent, qT) and DC-stimulated (activated, aT) splenic T cells were added to qHSCs or tHSCs (1×10⁶/well) and further incubated for 48 h. Three rats were used for each group. Apoptotic T cells were identified via double staining with PE-conjugated anti-CD3 mAb and TUNEL. Following surface CD3 staining, cells were fixed in 4% paraformaldehyde and permeabilized with 0.1% Triton X-100 and 0.1% sodium citrate. The TUNEL reaction mixture from the Cell Death Detection kit (BD Biosciences, San Diego, CA, USA) was then added according to the manufacturer’s instructions. Cells incubated with the labeling solution in the absence of terminal transferase were used as negative controls. The percentage of TUNEL-positive lymphocytes was then calculated.

A confocal laser scanning microscope (14900 Superfusion System; Radnoti, Monrovia, CA, USA) was used to analyze the stained sections. Fifty images were collected from each liver using a charge-coupled device camera (Hall 100) and analyzed using Zeiss LSM Image Browser software. Image processing was performed using Adobe Photoshop software (Adobe Systems UK) (19).

Livers taken on days 7, 14, 21 and 28 were examined. Four rats were used per group for each time period and 10 sections were prepared per animal. The relative areas of α-SMA-positive cells and numbers of CD4⁺, CD8⁺ and TUNEL⁺ cells were counted in the HCC area, and the total area in one microscopic field, at a magnification of ×100. One microscopic field at ×100 was 11.4 mm²; a total of 57 mm² of the liver sections were therefore examined per animal. Cell density was calculated and expressed as cell number per mm²(7).

All statistical analyses were conducted using SPSS 15.0 software. The data were expressed as the means ± standard deviation (SD). Significant differences were evaluated by unpaired Student’s t-tests. Correlations among the density of tHSCs, apoptotic T cells and tumour nodules in the lungs were analyzed using Spearman’s rank correlation. P<0.05 (two-tailed) was considered to indicate a statistically significant result.

Orthotopic HCCs were established with no perioperative complications. All rats (100%) developed tumors during the observation period (Fig. 1).

A previous study demonstrated metastatic lung nodules at day 30 in an orthotopic MRH hepatoma model (13). This observation was confirmed in our experiments; 3 out of 6 (50%) animals had significant evidence of tumor nodules by day 21 (Fig. 2D), and 6 out of 6 (100%) by day 28 (Fig. 2E). However, tumor nodules were not obvious at days 7 and 14 (Fig. 2B and C). The number of tumor nodules ranged from 2 to 384 at day 21, (mean, 234.5 nodules per animal), and from 54 to 579 at day 28, (mean, 318.7 nodules per animal). There were no tumor nodules in the normal rats, and 0–2 nodules in HCC rats on days 7 and 14 (means, 0.23 and 0.42 nodules per animal, respectively). No animals had tumor nodules at extrapulmonary sites. These results demonstrate that the number of tumor nodules gradually increased with tumor progression (Fig. 2F).

The expression of α-SMA (a marker of stellate cell activation) was evaluated in HCC. None of the normal animals showed abnormal α-SMA staining (Fig. 3A), except for the normal light staining. In HCC, however, activated HSCs stained positively with Cy-5 anti-α-SMA and appeared in situ as red cells (Fig. 3B–E). Although α-SMA⁺ cells might include myofibroblasts and vascular cells as well as activated HSCs, this remains the gold standard for identification of HSCs (14). The α-SMA signal increased markedly with time (Fig. 3B–E). The relative α-SMA area was 0.19±0.12% in normal animals (Fig. 3F), which increased significantly following HCC progression, to 0.79±0.55 at day 7 (P=0.001), 1.81±1.6 at day 14 (P=0.03), 3.02±1.7 at day 21 (P=0.04) and 4.14±1.9 at day 28 (P=0.025).

Figs. 4–6 show representative stained samples from normal liver and from HCC livers at 7, 14, 21 and 28 days. Slight T-cell infiltration occurred at day 7 (Fig. 4B) after tumor implantation, and this infiltration increased at days 14 (Fig. 4C), 21 (Fig. 4D) and 28 (Fig. 4E), respectively. There were fewer T lymphocytes in normal rat livers (Fig. 4A). Immunofluorescence staining demonstrated that the number of CD8⁺ cells per unit square (Fig. 5A2, B2, C2, D2 and E2) gradually increased at each sequential check point. Following tumor progression, CD8⁺ cells were found very close to tHSCs. Cy-5-conjugated anti-α-SMA is shown red (Fig. 5A1, B1, C1, D1 and E1) and APC-conjugated anti-CD8 marker is shown blue in single and merged stains. Areas stained for both CD8 and α-SMA markers are purple (Fig. 5A3, B3, C3, D3 and E3). These results suggest the existence of direct cell to cell attachment. CD8⁺ cells were always seen adjacent to tHSCs. There were few CD8⁺ cells in normal animals (Fig. 5F), and the numbers increased significantly, reaching a plateau within 14 days. The number of CD8⁺ cells per unit square (Fig. 5F) peaked at 21 days, and decreased at 28 days (Fig. 5F). The number of CD4⁺ cells per unit square, like CD8⁺ cells, also gradually increased at each sequential check point. CD4⁺ cells first appeared in situ at day 7 (Fig. 5F). Cy-5-conjugated anti-α-SMA (Fig. 6A1, B1, C1, D1 and E1) and FITC-conjugated anti-CD4 markers (Fig. 6A2, B2, C2, D2 and E2) are illustrated with single (red and green) and merged stains (Fig. 6A3, B3, C3, D3 and E3, yellow-green cells). CD4⁺ cells were much less abundant than CD8⁺ cells. Like CD8⁺ cells, however, the CD4 subsets were also mainly found attached to α-SMA-positive tHSCs (Fig. 6A3, B3, C3, D3 and E3). There were fewer CD4⁺ cells in normal rats, and they increased significantly at 7 days of HCC (Fig. 6F). No marked increase in CD4 staining was seen at day 14, but CD4⁺ cell numbers were significantly increased at days 21 and 28 (Fig. 6F).

Double staining for CD3 and TUNEL identified some HCC-infiltrating CD3⁺ T cells that were also positive for TUNEL staining (Fig. 7). Following tumor progression, PE-conjugated anti-CD3- (red; Fig. 7A1, B1, C1, D1 and E1) and TUNEL-stained cells (green; Fig. 7A2, B2, C2, D2 and E2) are illustrated as single and merged stains. Cells stained for both CD3 and TUNEL are yellow-green (Fig. 7B3, C3, D3 and E3). These results suggest the existence of CD3⁺ cell apoptosis. TUNEL⁺ cells were rare in normal livers (Fig. 7F), but were significantly increased at days 7 and 14 in HCC livers. The number per unit square of TUNEL⁺ cells increased and peaked at day 28 (Fig. 7F). Double staining for α-SMA (red) and TUNEL (yellow-green) revealed no obvious α-SMA⁺/TUNEL⁺ cells in normal rat livers (Fig. 8A), but double-stained cells were clearly found in HCC livers at each time-point (Fig. 8B–E). Direct α-SMA⁺ HSCs and TUNEL⁺ cell adhesion gradually increased on days 7, 14, 21 and 28. TUNEL⁺ cells were closely associated with α-SMA⁺ HSCs at each time point (Fig. 8A3, B3, C3, D3 and E3). Quantitative analysis demonstrated that the percentage of co-localized α-SMA⁺ HSCs and TUNEL⁺ cells per unit square gradually increased in the HCC area at each time point, while there was no appreciable increase in normal rat livers (Fig. 8F).

To confirm the results found in HCC sections, HSCs were isolated from normal and HCC livers and cultured. HSCs from the normal livers cultured for 2 days demonstrated quiescent features (qHSCs), with round or star shapes, abundant lipid droplets, and a lack of α-SMA expression. HCC HSCs cultured for 2 days (tHSCs) demonstrated activated features (data not shown). We investigated the adhesion of HSCs and T cells. Light microscopy clearly demonstrated adhesion of DC-stimulated T cells (aT) cells to tHSCs in aT/tHSCs co-culture (Fig. 9C). However, T cells were only slightly associated with HSCs in cultures of qT or aT cells alone (Fig. 9A and B), and in co-cultures of qT cells/tHSCs (Fig. 9D), aT cells/qHSCs (Fig. 9E) and qT/qHSCs (Fig. 9F). We then investigated the inducing role of tHSCs in T-cell apoptosis using co-cultured splenic T cells and tHSCs, and double staining for TUNEL and CD3. Splenic T cells, either non-stimulated (qT) or DC-stimulated (aT), were co-cultured with qHSCs or tHSCs for 24 h. Culture of qT cells alone (Fig. 9A) or aT cells alone (Fig. 9B) for 24 h showed a very low frequency of apoptosis (Fig. 10A and B). However, when aT cells were co-cultured with tHSCs (Fig. 9C), the former often became round when viewed under a phase-contrast microscope, and stained positive for TUNEL (Fig. 10C). Both cell rounding and TUNEL⁺ staining (Fig. 10D–F) were less frequent in co-cultures of qT cells/tHSCs (Fig. 9D), aT cells/qHSCs (Fig. 9E) and qT cells/qHSCs (Fig. 9F). While the percentage of TUNEL⁺ cells in total lymphocytes adhering to HSCs was low in co-cultures of aT cells/qHSCs and qT cells/tHSCs, it was significantly increased in co-cultures of aT cells/tHSCs (Fig. 10G).

tHSCs became increasingly prominent in HCC livers compared to normal livers, following HCC progression at each time point, accompanied by marked T-cell apoptosis and lung metastasis (Table I). Correlation analysis showed that the number of tHSCs was positively correlated with the percentage of T-cell apoptosis (r=0.861, P<0.05) (Fig. 11A), and the percentage of T-cell apoptosis was positively correlated with metastasis in the lungs (r=0.911, P<0.01) (Fig. 11B).