Establishment of cell clones with different metastatic potential from the metastatic hepatocellular carcinoma cell line MHCC97

Abstract

AIM: To establish clone cells with different metastatic potential for the study of metastasis-related mechanisms.

METHODS: Cloning procedure was performed on parental hepatocellular carcinoma (HCC) cell line MHCC97, and biological characteristics of the target clones selected by in vivo screening were studied.

RESULTS: Two clones with high (MHCC97-H) and low (MHCC97-L) metastatic potential were isolated from the parent cell line. Compared with MHCC97-L, MHCC97-H had smaller cell size (average cell diameter 43 μm vs 50 μm) and faster in vitro and in vivo growth rate (tumor cell doubling time was 34.2 h vs 60.0 h). The main ranges of chromosomes were 55-58 in MHCC97-H and 57-62 in MHCC97-L. Boyden chamber in vitro invasion assay demonstrated that the number of penetrating cells through the artificial basement membrane was (37.5 ± 11.0) cells/field for MHCC97-H vs (17.7 ± 6.3)/field for MHCC97-L. The proportions of cells in G0-G1 phase, S phase, and G2-M phase for MHCC97-H/MHCC97-L were 0.56/0.65, 0.28/0.25 and 0.16/0.10, respectively, as measured by flow cytometry. The serum AFP levels in nude mice 5 wk after orthotopic implantation of tumor tissue were (246 ± 66) μg•L¯¹ for MHCC97-H and (91 ± 66) μg•L¯¹ for MHCC97-L. The pulmonary metastatic rate was 100% (10/10) vs 40% (4/10).

CONCLUSION: Two clones of the same genetic background but with different biological behaviors were established, which could be valuable models for investigation on HCC metastasis.

Key Words: hepatocellular carcinoma; clone cells; metastasis

INTRODUCTION

Cancer cell population, either as a solid tumor mass in vivo or as a continuous cell line in vitro, is an ever-changing entity due to their genetic instability and selective environmental pressure. A tumor mass consists of different cell clones, a phenomenon known as tumor heterogeneity[1,2]. Based on this phenomenon, tumor cell clones of different biological properties have beenisolated from a number of human and animal tumor cell lines. These differences include a variety of biological characteristics such as tumor cell morphology, karyotypes, in vitro and in vivo growth patterns[3-7], DNA ploidy[8,9] tumorigenicity and drug sensitivity[5], metastatic patterns [4] and metastatic potentials[10-12], albumin secretion [13] and hyaluronan production[14].

Liver cancer is the 4th most common cause of death from cancer and China alone accounts for 53% of all liver cancer death worldwide[15], and the incidence is on slow but steady rise in both developing and the developed countries[16-20] Primary liver cancer in China, of which more than 90% is hepatocellular carcinoma (HCC), remains the second leading cancer killer that mainly affects middle-aged people-those in the prime of their most productive years[21]. Although gratifying progress has been achieved in clinical treatment at some centers, the overall survival for the whole HCC population at large is still very poor. Another dismal problem is that HCC is more prone to recurrence and metastasis even after curative resection[16,22-24]. Therefore researches in the mechanism and intervention of liver cancer recurrence and metastasis have special priority in China’s anti-cancer campaign. For a better insight into the mechanisms of HCC metastasis and for the development of new treatment strategies, an ideal HCC model system is essential. To serve this purpose, animal model of metastatic human HCC LCI-D20[25] and metastatic human HCC cell line MHCC97[26] have been established at the authors’ institute.

Although several human and animal liver tumor cell clones[13,27-37] have been established, few of these were suitable for the study of human HCC metastasis-the most fundamental characteristics of cancer and the ultimate cause of most cancer mortality. Recently, we isolated two human HCC clones with different metastatic potential from the parent cell line MHCC97, and explored some of their differences.

MATERIALS AND METHODS

Animals

Male athymic BALB/c nu/nu mice, 4-6 wk old, were obtained from Shanghai Institute of Materia Medica, Chinese Academy of Science, and housed in laminar- flow cabinets under specific pathogen-free (SPF) condition. All studies on mice were conducted in accordance with the National Institutes of Health “Guide for the Care and Use of Laboratory Animals”. The study protocol was approved by Shanghai Medical Experimental Animal Care Committee.

Parent Cell Line

The parent cell line MHCC97[26] is a human HCC cell line established from the animal model of human HCC LCI-D20[25].

Cloning Procedure

Cells of the 25th passage of the parent cell line were used for the current work. The cells were cultured at 37 °C in a humidified atmosphere of 50 mL•L¯¹ CO₂ and 950 mL•L¯¹ air. The culture medium was high glucose Dulbecco’s modified Eagle medium (DMEM) (GibcoBRL, Grand Island, NY, USA) supplemented with 100 mL•L¯¹ fetal bovine serum (Hyclone, Utah, USA). Two days after cell passage, the medium was transferred into a sterile tube (Corning Incorporated, Corning, NY, USA), centrifuged at 2000 r•min¯¹ for 10 min, and the supernatant was stored at -20 °C as conditioned medium. When the cells grew to approximately 80% confluence, the culture flask was placed in 4 °C refrigerator for 4 h to synchronize the cells, followed by incubation overnight at 37 °C as usual. In the following morning, single cell suspension was prepared after trypsinization (2.5 g•L¯¹ trypsin, Difco, prepared in Ca²⁺ and Mg²⁺ free Hanks solution d-Hanks), cell viability was confirmed by trypan blue exclusion, and cloning procedure was performed using the limited dilution method[6]. The cloned cells were preserved in culture medium containing 100 g•L¯¹ dimethyl sulfoxide (DMSO, Sigma Chemical Co, St Louis, MO, USA), and stored frozen in liquid nitrogen till used for in vivo screening.

In vivo Screening

In vivo screening was conducted in nude mice, when the cloning process was complete. The stored cells were thawed and propagated. Approximately 1 × 10⁷ cells in 0.2 mL culture medium were injected sc into the right flank of the mice, which were then observed daily for signs of tumor development. Once the subcutaneous tumor reached 1-1.5 cm in diameter, it was removed and cut into pieces about 2 mm × 2 mm × 2 mm which were implanted into the liver of each of 6 nude mice, using the method as described previously[25]. Five weeks later, the animals were sacrificed and autopsied. Lungs and other organs suspected of tumor involvement were sampled for histopathological studies. This was the first round of in vivo selection. In order to identify clones with maximal and minimal metastatic potential, all the recovered clones still viable after thawing were tested for initial screening. The parent cell line was used as controls.

Confirmation Test

Once the clones with maximal and minimal metastatic potential were targeted. They were subjected to the second round of confirmation test, which was performed essentially in the same way as the first round of selection, but more animals were used. Target clone cells were propagated and 5 × 10⁶ cells in 0.2 mL culture medium were injected into the left lower flank region of each of 5 nude mice (4 wk-old, 13 g - 17 g). The animals were observed for latency period, defined in this study as the time interval from the day of injection till the day of definite tumor mass about 5 mm in diameter at the injection site. The growth of subcutaneous tumor was recorded for 30 d, then fresh tumor tissues were implanted into the liver exactly the same manner as the first round of selection, 10 mice for each clone. Animal care and pathological studies were the same.

The confirmed clones were subjected to the following studies.

Morphological Observations

The cells were cultured on culture chambers (Lab-Tek) Chamber Slide, Nunc Inc. Naperville, III, USA) for 2 d and stained in Giemsa solution. Cell morphology was viewed under light microscope and representative pictures were taken. Transmission electron microscopy was conducted as described previously[11]. For scanning electron microscopy, cells grown on cover slips were fixed with 25 g•L¯¹ glutaraldehyde fixative (pH7.2), and observed directly under scanning electron microscope (HITACHI S-520, Japan).

Chromosome analysis

Chromosome preparation was performed with the method described by Seabright[38], with slight modification. Briefly, cells after 60 h of subcultu re were used. Colchicine (Shanghai Chemical Reagents Co. Shanghai, China) was added to the culture flasks to yield the final concentration of 0.04 mg•L¯¹, and the flasks were incubated for 4 h before the cells were harve sted. The cells were treated in hypotonic solution consisting of 1∶1 mixture (in volume) of 4 g•L¯¹ potassium chloride and of 4 g•L¯¹ sodium citrate, and then fixed in ice-cold methanol: glacial acetic acid (2:1, volume ratio). The slides stained in Giemsa solution (1∶10 dilution in pH6.8 PBS). Metaphase chromosome spreads were analyzed with Cytovision Chromosome analysis system (CytoVision^TM Image Analysis Workstations, USA).

Cell Growth Curves

Cells in exponential growth phase were trypsinized to give single-cell suspensi on. 4 × 10⁴ viable cells in 1 mL of medium were added to every well of the 24-well tissue culture plates, which were incubated at 37 °C with 50 mL•L¯¹ CO₂. Cell numbers in two wells were counted in a hemocytometer every 24 h for 7 consecutive days, and cell growth curves were plotted based on these results. The tumor cell doubling time was calculated according to the following formula: TD = Tlg2/lg (N/N₀) (TD: doubling time, T: time interval, N₀: initial cell number, N: end-point cell number)[39].

Plate Efficiency (PE)

1 × 10⁷ cells•L¯¹ of single-cell suspension were made from cells of exponential growth phase. 0.2 mL of cell suspension (containing 2000 viable cells) and 4 mL of culture medium were added to each well (3.5 cm in diameter) of 6-well culture plate, which was incubated at 37 °C with 50 mL•L¯¹ CO₂ for 12 d, washed twice with warm PBS, and stained with Giemsa solution. The number of colonies was counted under microscope (40 ×). PE was calculated using the following formula: PE = (number of colonies/number of cells inoculated) × 100%.

In vitro invasion assay

Matrigel invasion assay was performed using the method by Albini et al[40], with modification. Boyden chamber inserts (Nunclon^TM, Denmark) with filter membrane pore size of 8 μm were used in the assay. Fifty μg matrigel (from the Department of Biology, Medical Center of Beijing University, Beijing, China) was coated to each filter and the chamber was incubated at 37 °C for 2 h to produce the artificial basement membrane. Tumor cells in serum-free DMEM (200 μL containing 1 × 10⁵ cells) were added to the upper compartment of the chamber, and 800 μL of conditioned medium was added to the lower compartment. After 24 h incubation, the matrigel was removed, the filter was washed, fixed and stained in Giemsa solution. Cells that had migrated to the under side of the filter were counted under a light microscope (200 ×). The results were expressed as the number of migrated cells per field and presented as the [AKx-D] ± s of three assays.

Flow cytometry

Cells at exponential growth phase were harvested and single-cell suspensions containing 1 × 10⁶ cells were made. The cells were treated following the standardized protocol and cell cycle analyses were performed by flow cytometry as described previously[41].

Immunocytochemistry

Cells directly cultured on slides were washed two times with PBS and then fixed in acetone for 5 min at room temperature. Albumin, alpha-fetoprotein (AFP), cytokeratin 8 and hepatitis B surface antigen (HBsAg) were detected by immunocytochemistry using a two-step labeled avidin-biotin immunoperoxidase method, as recommended by the supplier. Primary detection was by either rabbit polyclonal or murine monoclonal antibodies. Biotinylated secondary envision antibodies were goat anti-rabbit IgG and rabbit anti-mouse IgG (Dako, Denmark). Negative controls consisted of omission of the primary antibody, and all cells were counterstained with hematoxylin. The slides were viewed under microscope and the degree of staining was recorded.

Detection of hepatitis B virus DNA

Cells were harvested by trypsinization when they were at 90% confluence, and was had twice in PBS. Total cellular genomic DNA was extracted using the Qiangen DNA extraction kit (Qiagen GmbH, Germany). Hepatitis B virus (HBV) DNA was examined by fluorescent primer polymerase chain reaction (PCR) (LightCycler, Roche, USA)using 2 μg cellular DNA, according to the instruction of the HBV DNA detection kit (Shenzhen Piji Biotechnology Development Co. Shenzhen, China). Positive and negative standards were tested at thesame time.

Alpha-fetoprotein determination

At the end of the second test, when the mice were sacrificed, blood was taken from each animal, and the serum AFP levels were determined automatically (ACS: 180 Automated Chemiluminescence System, Bayer Corporation, USA).

Statistical analysis

Fisher’s exact test and student’s t test were used, respectively, for comparisons of enumeration data (number of mice with lung metastases) and measurement data. The statistical analysis software package Stata 5.0 was used for the tests, and P ＜ 0.05 was considered as statistically significant.

RESULTS

Identification of Clones with Different Metastatic Potential

A total of 28 clones were isolated from the single-cell culturing of four 96-well plates. Among them, clones 2, 3, 12, 14 and 15 were discarded because of suspected microbial contamination; clones 5, 9, 16, 17, 18, 19, 20, 21, 22, 23, 26, 27, 28 were not viable when the cells in the first few passages were thawed from liquid nitrogen. The remaining ten clones (clones 1, 4, 6, 7, 8, 10, 11, 13, 24, and 25) were propagated for in vivo screening. The cells of each clone were first injected into the subcutis of nude mice. Then the subcutaneous tumors were implanted into nude mice liver, for evaluation of spontaneous pulmonary metastasis after 5 wk. The results were summarized in Table 1. It was found that most of these clones had similar metastatic potential. Lung metastases were 6/6 for clones 1, 7, 8, 10, 11 and 13, 5/6 for clone 4, 4/6 for clones 6 and 24, and 2/6 for clone 25.

Table 1 Lung metastases of 10 clones.

Clone number	n Tumor implanted	n Lung metastases after 5 wk
1, 7, 8, 10, 11, 13	6	6
4	6	5
6, 24	6	4
25	6	2
MHCC97	6	6

NOTE. Each clone was first sc inoculated into nude mice. Then small pieces of subcutaneous tumor were implanted into the liver of 6 nude mice for every clone. The number of mice with lung metastases was determined by histopathology 5 wk after orthotopic tumor implantation (see the in vivo screening for detail.).

Among the clones with greatest metastatic potential, clone 8 produced the most numerous lung metastases (median number of lung colonies 12/mouse). Therefore, clone 8 and clone 25 were selected as target clones and used for the second round in vivo confirmation test. For evaluation of subcutaneous tumor development, 5 nude mice for each clone were used, and every mouse wasin jected with 0.2 mL of cell suspensions containing 5 × 10⁶ cells. For evaluation of spontaneous metastasis after orthotopic implantation, 10 animals were used for each clone. The results were shown in Table 2 and Figure 1, Figure 2.

Figure 1 Subcutaneous tumor formation of the two clones. Thirty days after sc injection of 5 × 10⁶ tumor cells for each nude mouse, the average s.c tumor diameter was (1.94 ± 0.36) cm for MHCC97-H as against (0.84 ± 0.47) cm for MHCC97-L.

Table 2 Abdominal events and pulmonary metastases after liver implantation of subcutaneous tumor tissue.

	Clone number
	No. 8 (MHCC97-H)	No. 25 (HMCC97-L)
No of mice with tumor implantation	10	10
Tumor size by d 35/cm	1.42 ± 0.11	0.90 ± 0.26a
Abdominal events
Abdominal wall invasion	40% (4/10)	20% (2/10)
Bloody ascites	10% (1/10)	0% (0/10)
Intrahepatic metastases	80% (8/10)	0% (0/10)
Diaphragm invasion	10% (1/10)	0% (0/10)
Hepato-splenic/hepato-gastric	10% (1/10)	0% (0/10)
ligament invasion
Loco-regional lymph node	0% (0/10)	0% (0/10)
enlargement
Pulmonary metastases	100% (10/10)	40% (4/10)b

NOTE.

^aP ＜ 0.01, t test. The length (L), width (W) and height (H) of liver tumor was measured at autopsy using a caliper, and the tumor size was expressed as the geometric mean diameter (GMD) GMD = (L × W × H) 1/3;

^bP ＜ 0.05 Fisher’s exact test.

Figure 2 The liver tumor size of the two clones 5 wk after orthotopic inoculation. The tumor geometric mean diameter (GMD) for MHCC97-H was (1.42 ± 0.11) cm as against (0.90 ± 0.26) cm for MHCC97-L.

Although both clones were tumorigenic, the latency period differed considerably between them, being (5-10) (6.4 ± 2.2) d for clone 8 and (20-25) (21.3 ± 2.5) d for clone 25. By d30, the subcutaneous tumor sizes were (1.94 ± 0.36) cm for clone 8 and (0.84 ± 0.47) cm for clone 25 (P ＜ 0.01, t test). Both clones mainly exhibited expansive growth pattern, although clone 8 showed some tumor invasion in 2 of the 5 animals tested, one invading the hipbone and another invading the lumbar spine. The most obvious difference between the two clones regarding the subcutaneous tumor development was the growth rate.

Their differences became even more apparent after small bits of subcutaneous tum or tissues were implanted into the liver of nude mice for 5 wk, when it was found that clone 8 produced notable intrahepatic metastases in 8 of the 10 recipients examined, whereas clone 25 did not produce any observable nodules. Apart from intrahepatic metastases, clone 8 also caused abdominal wall invasion in 4/10 (40%), diaphragm invasion in 1/10 (10%), hepato-splenic and hepato-gastric ligaments invasion in 1/10 (10%), and bloody ascites in 1/10 (10%). These changes were not observed for clone 25, except 2/10 (20%) animals showed abdominal wall invasion. No enlargement of lymph nodes was observed for either clone.

Pathological studies of liver tumors from the two clones showed similar histology. The tumor cells were polygonal epithelial-like cells forming large tumor nests surrounded by thin fibroconnective tissues. Tumor necroses were prominent at the center of large tumor nests. Anaplastic tumor spindles aggressively infiltrating the adjacent tissues and tumor cells invading blood vessels were observed in clone 8, but not in clone 25. Moreover, the lung metastatic lesions formed by these two clones were also different. The lung metastases by clone 25 were usually small and located near the surface of the lung, while those formed by clone 8 were large and usually located in the lung parenchyma, pressing blood vessels and the bronchioles (Figure 3A, B). Thus it was established from these selections and observations that clone 8 was the most metastatic and clone 25 the least metastatic. They were designated as MHCC97-H and MHCC97-L, respectively.

Figure 3 Photomicroscopy of lung metastases of the two clones. MHCC97-H (3A) produced large metastatic lesion pressing the bronchioles while MHCC97-L (3B) only formed small metastasis. HE × 100.

Cell Morphology

Both clones showed polygonal epithelial-like morphology, with firm attachment to the culture flask. However, they were different in cell size (average cell diameter (50 ± 5) μm in MHCC97-L vs 43 ± 2 μm in MHCC97-H) and cell morphology (multiform in MHCC97-L vs uniform in MHCC97-H). Both cells had large conspicuous nucleus, with 1-3 big nucleoli scattered in the nucleus of MHCC97-L and 3-7 smaller nucleoli in MHCC97-H (Figure 4A, B). Electron microscopy revealed abundant microvilli and projections on the cell surface. Some of the projections on MHCC97-H extended far and formed bulges at the end, while those on MHCC97-L were short and compact. Both cells had many lysosomes in the cytoplasm, which were usually concentrated on one side of the cell in MHCC97-H and scattered around the cytoplasm in MHCC97-L. No obvious desmosomes, tight conjunction or other cell junction structures were observed, nor were virus particles or other particular particles.

Figure 4 Photomicroscopy of MHCC97-L (4A) and MHCC97-H (4B) illustrating the clear differences in the number of nucleoli. Giemsa × 400.

Chromosome analysis

Both clones were heteroploid. The chromosome number in MHCC97-H ranged from 37 to 68, with 68% of cells in the main range of 55-58 chromosomes. The range of chromosomes for MHCC97-L was from 44 to 105, and 58% in the main range of 57-62 chromosomes.

Cell growth curve

As shown in Figure 5, MHCC97-H grew much faster than MHCC97-L, their population doubling time being 34.2 h and 60.0 h, respectively.

Figure 5 Cell growth curve of two clones. 4 × 10⁴ viable cells were cultured in each well of the 24 well culture plate. Cell numbers were determined for 7 consecutive days. Each time point represents the mean of duplicate cell counts.

Plate efficiency

Colony formation rates were (22.2 ± 3.7)% in MHCC97-H and (18.6 ± 4.7)% in MHCC97-L, the difference being of no statistical significance (P ＞ 0.05, t test). However, the number of cells in each colony did differ between the two cell clones, with 8-15 cells in each colony of MHCC97-H and 3-5 cells in each colony of MHCC97-L.

In vitro invasion assay

The numbers of cells that penetrated the artificial basement membrane were (37.5 ± 11.0) cells per high power field in MHCC97-H and (17.7 ± 6.3)/HP field in MHCC 97-L (P ＜ 0.05, t test).

Flow cytometry

Cell cycle analysis by flow cytometry revealed that the proportion of cells in G0-G1 phase, S phase, and G2-M phase for MHCC97-H/MHCC97-L were 0.56/0.65, 0.28%/0.25 and 0.16/0.10, respectively. MHCC97-H had more cells in S phase, and G2-M phase than MHCC97-L.

Immunocytochemistry

Both clones were positive for albumin, AFP and cytokeratin 8, attesting that they were indeed liver cancer cells. However, neither of the two had any positive stain for HBsAg.

Detection of HBV DNA Integration

Both clones were positive for HBV DNA.

AFP Production

Five weeks after orthotopic implantation of the tumor from MHCC97-H and MHCC97-L, AFP levels in nude mice serum were (246 ± 66) μg•L¯¹ and 91 ± 66 μg•L¯¹, respectively (P ＜ 0.01, t test).

DISCUSSION

Heterogeneous nature of tumor in terms of its in vitro and in vivo characteristics has been well recognized. Populations of human and animal tumors frequently demonstrate a great variation in a number of cellular and functional properties. Our study reported here confirmed that the metastatic human HCC cell line MHCC97 was also a heterogeneous population, consisting of subpopulations with different metastatic potentials. Immunocytochemical studies demonstrated that these two clones were positive for albumin, AFP and cytokeratin 8, demonstrating they were still HCC cells. HBV DNA integration into the cell genome also confirmed that they were related to their parent cell line MHCC97, which was HBV positive[26].

Systemic comparisons of these two clones revealed many differences between them. Heterogeneity of cell morphology is the most easily observed feature among different clones of the same origin, and it is usually the beginning of the study of tumor heterogeneity. This phenomenon has been well documented in various tumors, including lung cancer[10], colon cancer[42,43], breast cancer[6,44], squamous cell carcinoma of the skin[3] and the tongue[7,11]. In our study, the two clones identified here also showed differences in cell morphology. The highly metastatic variant MHCC97-H was small (average diameter 43 μm) and more uniform than the clone with low metastatic potential MHCC97-L (average diameter 50 μm). Cell size itself may be a mechanical factor influencing metastasis. The small size may facilitate cells to traverse through the blood vessels, evade the immune attacks in the circulation during tumor cell transport, and come up with less mechanical resistance during tumor cell penetration in the target tissue. Since our study was a spontaneous metastasis model, this difference in cell size between the two clones may have some impact on the ability to overcome host barriers. This is in keeping with a earlier finding by Suzuki et al[45], who used a mouse fibrosarcoma system to study the experimental metastasic ability (via tumor cell injection into the tail vein to observe the lung colony formation abilities) of various clones, and found that the clone with nearly 10-fold higher lung colony-forming ability was much smaller in cell volume than the low metastatic clone.

Another prominent difference in cell morphology is that MHCC97-H had more nucle oli (3-7 per cell) than MHCC97-L (1-3 per cell). More nucleoli are a marker of active cell proliferation. Derenzini et al[46] suggested that nucleoli perse could tell the metatatic potential of tumor cells. Their study indicated that larger nucleoli predicted more rapid tumor cell proliferation. However, our results seem to suggest that nucleolar number rather than the size is an indicator of fast tumor proliferation. Indeed, in our model system, the highly metastatic variant with more nucleoli grew much faster than the one with few nucleoli and low metastasis. This could be reflected in cell growth rate curve (tumor cell doubling time was 34.2 h in MHCC97-H and 60.0 h in MHCC97-L), fraction of cells in S phase and G2-M phase of the cell cycle, latency period in subcutaneous tumor development and liver tumor size at the endpoint of the study. Whether or not tumor cell growth rate is directly related to metastasis is not clear yet. Yasoshima et al[47] using metastatic gastric cancer cell line, and Samiei et al[48] using metastatic mammary clones found, that metastasis was independent of tumor cell growth; while other works[49,50] showed close association between tumor cell growth rate and metastasis. Our results suggest fast growing tumor is more prone to metastasis.

Cytogenetic studies also revealed the differences in the chromosome number between the two clones. This is in keeping with a recent findings by Takeuchi et al[11], who used similar method to have isolated cloned cancer cells with different metastatic potentials, and found marked difference in modal chromosome number between the highly metastatic clone and the non-metastatic clone.

As HCC is a special health issue in China, basic and clinical researches in this field have been intensive. As early as in the 1960s, Chen[51] established the first human HCC cell line in the world. Later on several HCC cell lines were established[52-58]. Although most of these showed tumorigenicity when inoculated into experimental animals, rarely did they demonstrate the full potential for loco-regional and distant metastases, as seen so frequently in clinical patients. The metastatic HCC cell line MHCC97 was established in order to meet the urgent need for suitable models to study the mechanisms of and interventions on HCC metastasis. And now we took one step further to have isolated clones of different metastatic potential from the same cell line. These new models could be valuable for the study of HCC metastasis.

Cancer metastasis is the ultimate display of complex interactions between the malignant cells and the host defense mechanism. The process of metastasis consists of selection and sequential steps that include angiogenesis, detachment, motility, invasion of the extracellular matrix, intravasation, circulation, adhesion, extravasation into the organ parenchyma and growth[2]. The ability of cancer cells to form metastasis depends on a set of unique biological properties that enable the malignant cells to complete all those steps of metastatic cascade. For HCC invasion and metastasis, extensive studies have unveiled many molecular mechanisms involved in these processes, including P53[59,60]/CDKN₂ mutation, overexpression of H-ras/EGFR, nm23/TIMP[60], over-expression of CD44v6 and under-expression of nm23-H1[61], over-expression of metalloproteinase-9 and CD34[62], high level of laminin in the blood and tumor[63], intercellular adhesion molecule-1 (ICAM-1)[64], N-Actylglucosaminyltransferase V (GnT V) activity[65], high expression of urokinase-type plasminogen activator (uPA), its receptor (uPAR) and inhibitor (PAI-1)[66], chromosome 8p deletion[67,68]. But it is likely that other genes or gene locus in addition to these genes are also involved in the process of meta stasis in HCC, since most of these studies were focused only on a few genes or their products. It is reasonable to assume that there could be a group of relevant genes rather than a single or a few genes to account for tumor metastasis. The identification of those unknown genes related to metastasis is important in order to gain a complete picture of the molecular biology of HCC metastasis. For this end, a dependable model system that consists clones with high and low metastatic potential from the same origin should be the ideal study material. It is based on this rationale that some recent works seeking metastasis-related genes were conducted on the cloned cells from the same biological background. Reichner et al[35] isolated two clonal cell lines with different metastatic potentials from a rat hepatocellular carcinoma model induced by chemical carcinogens, and studied the differences in metastasis related mechanisms. They found no differences in the expression of several antigens noted to correlate with metastatic potential, including CD44 variant glycoprotein, p53, transferrin receptor, and E-cadherin. The only notable difference in the parameters studied was the level of IL-6. The highly metastatic clone released much more IL-6 than the low metastatic clone. Other studies use metastatic and non-metastatic human mammary carcinoma clone lines as comparing materials to seek metastasis-associated genes, and provided a vast amount of information on gene expression and metastatic phenotype[69].

In summary, our study confirmed that the metastatic HCC cell line MHCC97 is a he terogeneous population, consisting of cells with divergent biological properties. The two clones isolated from the parent cell line differed not only in in vitro characteristics like cell morphology and growth kinetics, but also in the most fundamental biological behavior-tumor metastasis. Since these two clones are from the same parent cell line, thus having the same genetic background, the differences in their phenotypes must have some underlying molecular mechanisms. In -depth study on their differences might help us gain new insight into the mechanisms of liver cancer metastasis.

ACKNOWLEDGEMENTS

The authors thank Professor Jin Gao, Ph.D. Chair of Pathology, Institute of Basic Medicine, Chinese Academy of Medical Science, for his help to this work. We also thank Dr. Xiaowu Huang for his help in immunocytochemical analysis and professor. Lun Xiu Qin for his critical reading of the manuscript.