Tensin2 variant 3 is associated with aggressive tumor behavior in human hepatocellular carcinoma

Authors

  • Judy Wai Ping Yam,

    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Frankie Chi Fat Ko,

    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Chung-Yiu Chan,

    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Tai-On Yau,

    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Edmund Kwok Kwan Tung,

    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Thomas Ho-Yin Leung,

    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Dong-Yan Jin,

    1. Department of Biochemistry, LKS Faculty of Medicine, The University of Hong Kong, Pokfulam, Hong Kong
    Search for more papers by this author
  • Irene Oi-Lin Ng

    Corresponding author
    1. Department of Pathology, SH Ho Foundation Research Laboratories and Hong Kong Jockey Club Clinical Research Centre, The University of Hong Kong, Pokfulam, Hong Kong
    • Room 127B, University Pathology Building, Department of Pathology, University of Hong Kong, Queen Mary Hospital, Pokfulam, Hong Kong, China
    Search for more papers by this author
    • fax: (852) 2872-5197


  • Potential conflict of interest: Nothing to report.

Abstract

Tensins are a new family of proteins that act as an important link among extracellular matrix, actin cytoskeleton, and signal transduction and have been implicated in human cancers. Tensin2 was initially identified in a search for new tensin family members that share extensive sequence homology with tensin1. Tensin2 was highly expressed in liver tissues. A recent study reported that one of the splicing variants of tensin2, variant 3, promotes cell migration. In the present study, we aimed to elucidate the role of variant 3 in hepatocarcinogenesis by assessing the expression of variant 3 mRNA in hepatocellular carcinoma (HCC) tissue and ectopically expressing variant 3 in HCC cell lines. Analysis of variant 3 expression in human HCC tissue revealed it was overexpressed in 46% (23/50) of tumor tissues as compared with the corresponding nontumorous livers. High expression of variant 3 was significantly associated with venous invasion (P = .037), tumor microsatellite formation (P = .022), and tumor nonencapsulation (P = .049). Our ectopic expression study showed that variant 3 significantly promoted the cell growth and motility of HCC cells. The clonal transfectants of variant 3 were more closely packed and resulted in a higher saturation density than in the control vector transfectants. Variant 3 expression also enhanced the proliferation rate in culture and in vivo tumorigenicity in nude mice. In conclusion, we reveal a novel role for variant 3 in the progression of HCC and suggest the feasibility of elevated variant 3 expression as a tumor progression marker for HCC. (HEPATOLOGY 2006;44:881–90.)

Hepatocellular carcinoma (HCC) is one of the major malignancies worldwide and the second most common fatal cancer in Asia and Hong Kong. The molecular mechanisms underlying the development, invasion, and metastasis of HCC are areas of active research. A better understanding of these areas may lead to the development of more effective prognostic markers and potential therapies for HCC. Adhesion proteins including intracellular adhesion molecule 1,1 E-cadherin,2 CD44 variants,3 and extracellular matrix protein, laminin-54 have been reported to be biomarkers for an invasive and metastatic phenotype of HCC. Proteinases responsible for the degradation of extracellular matrix, for instance, matrix metalloproteinase,5 urokinase-type plasminogen activator,6 uPA receptor, and uPA inhibitor,7 are also important factors involved in the invasiveness and metastasis of HCC. Angiogenesis regulators critical to HCC metastasis has drawn a lot of attention in recent years. Vascular endothelial growth factor,8 angiopoietin,9 basic fibroblast growth factor,10 platelet-derived endothelial cell growth factor,11 and thrombospondin12 have been reported to be angiogenic factors crucial to HCC angiogenesis.

Tensin is a new family of focal adhesion proteins that act as an important link among extracellular matrix (ECM), actin cytoskeleton, and signal transduction.13 Recent studies have revealed that tensin proteins are also implicated in human cancers. Tensin1 is the most well-characterized member and best known for its activity in organizing the actin cytoskeleton and mediating signal transduction.13 Expression of human tensin1 was shown to be down-regulated in prostate and breast cancer cell lines.14 Induction of tensin1 by cancer chemopreventive agent contributes to the anti-invasive activity of cancer cells.15 Very recently, other structurally related tensin family members have been identified. Focal adhesion localization is the common feature of all members. Tensin2 was identified in a search for tensin family members that share extensive sequence homology with tensin1.16 Tensin2 has been reported to localize at the end of actin stress fibers and to colocalize with vinculin and tensin1 at the focal adhesions in NIH3T3 cells. Tensin2 was also found to interact with the Axl receptor tyrosine kinase.17 There are 3 alternative splicing variants of tensin2. The shortest splice variant of tensin2, variant 3, has been shown to promote cell migration.16 Another family member, tensin3, is a component of epidermal growth factor–mediated signaling pathway.18 Cten, a COOH-terminal tensinlike member with restricted expression in prostate and placenta was shown to be down-regulated in prostate cancer.19 In another study, cten expression correlated with tumor progression in thymoma.20 Differential tissue distribution suggests tensin members may possess different functions and properties in various tissues. Although the high sequence homology of N and C terminals among tensin family suggests its properties of binding to actin, integrin, and phosphotyrosine residues are well conserved, it is their middle region, which is the most divergent, that might confer the unique biological functions of tensins.

Tensin2 variant 3 (variant 3) has been shown to promote cell migration16 and possibly is related to cell invasion, in the present study, we focused on the study of variant 3 in HCC, particularly its role in metastasis of HCC. Clinicopathological analysis of variant 3 expression revealed that overexpression of variant 3 in tumor tissue was associated with a more aggressive tumor phenotype and metastasis-related pathological features. Moreover, we demonstrated the positive effects of variant 3 on the ability of HCC cells to proliferate, migrate, and invade and its tumorigenicity and invasive potential in nude mice. Our study illustrates a role for variant 3 in metastasis and the clinical implications of it as a tumor progression marker in HCC.

Abbreviations

HCC, hepatocellular carcinoma; PCR, polymerase chain reaction; T/NT, tumor/nontumor.

Materials and Methods

Clinical Samples.

Paired samples of HCC and the corresponding nontumorous liver tissues from 50 Chinese patients who had had surgical resection at Queen Mary Hospital, the University of Hong Kong, were randomly selected for the study. All specimens were obtained immediately after surgical resection, snap-frozen in liquid nitrogen, and kept at −80° C. Frozen sections were cut from the tumorous and nontumorous liver blocks separately and examined histologically to ensure the tissues were composed of homogeneous cell populations.

Clinicopathological Analysis.

The clinicopathological features of the patients analyzed were sex, age, tumor size, cellular differentiation according to the Edmondson classification, venous invasion without differentiation into portal or hepatic venules, direct liver invasion, tumor microsatellite formation, tumor encapsulation, tumor node metastasis stage (pTNM), serum hepatitis B surface antigen (HBsAg) status, and background liver disease in nontumorous liver tissue. They were analyzed as we previously described.21 The tumors of the 50 patients with HCC ranged from 1.8 to 27.0 cm in diameter (mean: 8.3 ± 5.4 cm). Twenty-nine tumors (58%) were at least 5 cm in diameter, and 21 (42%) were less than 5 cm in diameter. Twenty-one tumors (42%) had better cellular differentiation (Edmondson's grades I and II), and 29 (58%) had poorer cellular differentiation (Edmondson's grades III and IV). Tumor encapsulation was detected in 31 tumors (62%). Twenty-four patients (48%) had venous invasion. Tumor microsatellite formation was present in 24 of the 50 HCC cases (48%), and direct liver invasion was found in 17 cases (34%). Twenty patients (40%) had tumors at earlier stages (pTNM stages I and II), and 28 (56%) had tumors at more advanced stages (pTNM stages III and IV). The nontumorous tissue of 26 patients (52%) showed cirrhosis, and 26 (43.3%) had chronic hepatitis. Hepatitis B virus surface antigen (HBsAg) was detected in serum in 36 patients (78%).

Quantitative and Semiquantitative Reverse-Transcription Polymerase Chain Reaction.

Total RNA was extracted from HCC samples and cell lines by TRIzol reagent (Invitrogen, Gaithersburg, MD). First-strand cDNA was synthesized from total RNA using random hexamers with Taqman reverse transcription reagents (Applied Biosystems, Foster City, CA). The cDNA was used for quantitative real-time PCR performed in an ABI 7700 system with TaqMan Gene Expression Assays (Applied Biosystems) in triplicate. The quantitative real-time PCR was performed in order to determine expression of variant 3. Three isoforms of tensin2 were generated by alternative splicing at the 5′ end. Variant 3 has a distinct 5′ untranslated region, which resulted in the use of a downstream translational start codon. A variant-specific amplicon and a TaqMan probe (5′-CTCTCCCCTCGGCTCTTTCCTAGGA-3′) of variant 3 were designed to cover the unique 8-103 and 21-55 nucleotides, respectively, of variant 3 (GI:38787969). Absolute copies of variant 3 were determined by a standard curve with variant 3 plasmid DNA and normalized with human 18s rRNA (Applied Biosystems). The ratio of the paired tumorous to nontumorous liver tissues from the same patient was defined as the variant 3 tumor/nontumor (T/NT) ratio. A T/NT ratio greater or equal to 2.0 was defined as overexpression in tumor tissue.

Expression of variant 3 in HCC cell lines was assessed by semiquantitative PCR. PCR amplification of variant 3 cDNA was performed using a set of splice variant-specific primers (5′-GCTGGCCGTCTGCGCACCC-3′ and 5′-AGGCAAGGCCTGACAGGCTGAAGT-3′) located in the distinct 5′ untranslated region of variant 3 in order to produce a 212-bp fragment that covered nucleotides 1-212 of variant 3 (GI:38787969). Expression of variant 3 in BEL7402 stable clones was confirmed by PCR using GFP-specific primer (5′-GATCACATGGTCCTGCTGGAGTTCGTG-3′) and variant 3–specific primer (5′-GAAGATCTTGCTGTAGTGCATGTAG-3′).

Cell Culture and Transfection.

The human HCC cell lines used in this study (HepG2, Hep3B, Huh7, SMMC7721, BEL7402, SNU182, SNU 449, SNU475, PLC, HLE, CL48) were obtained from the Shanghai Institute of Cell Biology and the American Type Culture Collection. H2P and H2M were generous gifts from X.Y. Guan.22 The nontumorigenic immortalized liver cell line MIHA was donated by S.T. Fan.23 BEL7402 cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal bovine serum. The pEGFP expression vector (BD Biosciences Clontech, Palo Alto, CA) carrying green fluorescent protein (GFP)-tagged variant 3, pEGFP-sv3, was transfected into BEL7402 cells using Fugene 6 Transfection Reagent (Roche Diagnostics, Indianapolis, IN). Transfected cells were selected in DMEM containing 400 μg/mL G418 (Invitrogen) and expanded as stable clones of BEL7402 expressing variant 3, namely, sv3#1 and sv3#2. Control clones of BEL7402 stably transfected with pEGFP vector, vec#1 and vec#2, were also established.

Cell Proliferation Rates.

Each well of 6-well culture plates was seeded with 2 × 104 cells in triplicate and maintained in growth medium. Cell number was counted with a hematocytometer at 24-hour intervals for 8 consecutive days. Means and standard deviations of the triplicates were calculated and plotted against time.

Colony Formation Assay.

One day before transfection 2 × 105 BEL7402 and Hep3B cells were seeded onto 35-mm plates. Two micrograms of pCS2+MT-sv3 (Myc-tagged variant 3) or pCS2+MT expression vector was cotransfected with 0.2 μg of pBABE-puro vector into the cells. After 24 hours, the cells were seeded on 100-mm culture dishes in a 1:5 dilution and grown for 2 weeks in medium containing 1 μg/mL puromycin (Sigma Chemical, St. Louis, MO). The puromycin-resistant colonies were fixed with 3.7% formaldehyde, and colony formation efficiency was examined by Giemsa staining (Sigma).

Orthotopic Liver Implantation in Nude Mice.

To assess the metastatic potential of stable clones of variant 3, 1 × 106 cells in 0.1 mL of phosphate-buffered saline were injected subcutaneously into the flank of BALB/C nude mice. After 3 weeks, the subcutaneous tumors were removed and cut into 1-2 mm3 cubes, which were then implanted in liver lobes of the nude mice as previously described.24 Five implantations were performed per cell line. After 2 months, the animals were sacrificed and examined. This experiment was performed according to the Animals (Control of Experiments) Ordinance (Hong Kong) and following the institute's guidance on animal experimentation.

Migration and Wound-Healing Assays.

The migration assay was performed with Transwell inserts that had 6.5-mm polycarbonate membranes and pores 8.0 μm in size (Corning Inc., NY). Briefly, 2 × 105 cells were resuspended in serum-free medium and added to the upper chamber. Culture medium containing 10% FBS was used as a chemoattractant in the lower chamber. The cells were incubated for 24 hours in a humidified incubator at 37°C. The cells that migrated through the membrane pores to the lower surface of the membrane were fixed and stained with crystal violet. Stained cells in each field were photographed, and the number of stained cells per field was counted in 5 fields. Each experiment was performed in triplicate. The wound-healing assay was performed as previously described.25

Invasion Assay.

The invasive potential of the cells was examined with a BioCoat Matrigel Invasion Chamber (BD Biosciences Clontech) that contained 8.0-μm membrane pores coated with Matrigel matrix. In brief, 5 × 104 cells were seeded in the upper chamber and allowed to migrate through the extracellular matrix to the lower chamber for 22 hours. The migrated cells were fixed, stained, and counted.

Statistical Analysis.

The T/NT ratio of each pair of clinical samples was transformed by logarithm. The log values were analyzed with the Student t test, the unpaired Student t test, and the Mann-Whitney U test, as appropriate, using SPSS for Windows 13.0 (SPSS Inc., Chicago, IL). Tests results were considered significant when the P value was less than .05.

Results

Variant 3 mRNA Expression in HCC Cell Lines and Human Tissues.

Three isoforms of tensin2 generated by alternative splicing at the 5′ end were identified (Fig. 1A).16, 17 Variants 1 and 2 are 1419- and 1409-amino-acid proteins, respectively. Each comprises a C1 domain of protein kinase C at the N terminus followed by a PTEN (phosphatase and tensin homolog deleted from chromosome 10) homology region and SH2 (src homology 2) and PTB (phosphotyrosine binding) domains at the C terminus. Variant 3 is a 1285-amino-acid protein, and it differs from tensin2 at the N terminal. It does not have a C1 domain. The smallest isoform, variant 3 has been shown to promote migration of HEK293 cells.16 We examined the expression of variant 3 mRNA in paired human HCC and corresponding nontumorous liver samples using real-time quantitative PCR. The expression of variant 3 in the tumor samples ranged from 6.67-fold underexpression to 21.68-fold overexpression compared with the corresponding nontumorous tissue. Of the 50 paired samples, 23 (46%) showed at least 2-fold overexpression (T/NT ≥ 2) of variant 3 mRNA in tumor tissue. A comparison of overall variant 3 mRNA expression in HCC and nontumorous liver samples showed that variant 3 mRNA was significantly overexpressed in tumor samples (P < .005, Mann-Whitney U test; Fig. 1B). Variant 3 was expressed in most of the HCC cell lines examined (Fig. 1C). Compared with Hep3B which has a relatively low expression level of variant 3, the nontumorgenic immortalized liver cell line, MIHA, even has a lower expression level of variant 3 (Fig. 1D).

Figure 1.

Variant 3 was overexpressed in HCC tissues and cell lines. (A) Schematic representation of tensin2 splice variants. Variant 1 and variant 2 are 1419- and 1409-amino-acid proteins, respectively. Each comprises a C1 domain at the N terminus followed by a PTEN homology region, and SH2 and PTB domains at the C terminus. Variant 3 is a 1285-amino-acid protein and is different from tensin2 at the N terminal. There is no C1 domain in the variant 3 isoform. (B) Variant 3 mRNA expression in primary HCC tissues determined by real-time quantitative PCR. Variant 3 showed significantly higher expression in the tumor group than in the nontumorous group (P < .005, Mann-Whitney U test). (C) Expression of variant 3 in HCC cell lines determined by semiquantitative PCR. Amplification of variant 3 cDNA was performed using a pair of specific primers flanking 1-212 nucleotides of variant 3. The quantity of cDNA was assessed by amplification of β-actin. Variant 3 was expressed in all but 1 of the cell lines examined. (D) Variant 3 has a low expression level in the nontumorigenic immortalized liver cell line, MIHA.

Clinicopathological Correlation of Variant 3 Expression in Human HCC.

To investigate the implications of overexpression of variant 3 in HCC patients, we correlated expression of variant 3 mRNA with clinicopathological features of the 50 patients (Table 1). Three parameters, the presence of venous invasion, the presence of tumor microsatellites, and the absence of tumor encapsulation, were found to be significantly associated with variant 3 mRNA overexpression (Fig. 2). Variant 3 mRNA was found to be 2.15- ± 2.43-fold overexpressed in tumor samples from the 24 patients with venous invasion, whereas the mean change in expression of variant 3 in the 26 patients without venous invasion was 1.26- ± 2.36-fold (P = .037, unpaired Student t test). In addition, expression of variant 3 in patients with tumor microsatellite formation was higher (2.20- ± 2.35-fold) than in those without formation of microsatellite tumors (1.23- ± 2.39-fold; P = .022, unpaired Student t test). Overexpression of variant 3 mRNA was more frequently observed in HCCs that were not encapsulated (1.96- ± 2.44-fold), whereas variant 3 mRNA was found to be underexpressed in HCC tumor samples that were encapsulated (1.15- ± 2.39-fold; P = .049, unpaired Student t test). Our results suggested that overexpression of variant 3 was associated with more aggressive tumor behavior and therefore might contribute to tumor progression and metastasis in hepatocarcinogenesis.

Table 1. Correlation Between Clinicopathological Parameters of Patients With Expression of Variant 3
ParameterCategoryCases (n)Mean T/NTP§
  • Abbreviation: HBsAg, hepatitis B surface antigen.

  • According to Edmondson's grading.

  • §

    Unpaired Student t test.

  • *

    P < .05.

Venous invasionAbsent261.26.37*
 Present242.15
Tumor microsatelliteAbsent261.23.022*
 Present242.20
Direct liver invasionAbsent291.41.192
 Present172.06
Tumor encapsulationPresent181.96.049*
 Absent311.15
HBsAgAbsent121.58.806
 Present371.70
Cellular differentiationI-II211.35.222
 III-IV291.86
Tumor size<5 cm211.49.571
 ≦5 cm291.73
No. of tumor nodules1 nodule391.54.395
 ≦2 nodules112.00
Background liver diseaseNo cirrhosis241.90.257
 Cirrhosis261.42
Tumor pTNM stageI and II201.28.102
 III and IV281.98
Figure 2.

Clinicopathological correlation of variant 3 expression in human HCC. Quantitative PCR was performed to detect the expression of the variant 3 transcript in 50 HCC samples. Expression of 18s rRNA was used as an internal control to standardize the total amount of input RNA. The Y axis shows the relative T/NT ratio of variant 3 expression after normalization by 18s rRNA. Each dot represents the mean of triplicates, and the horizontal line among the dots indicates the mean of each group. The statistical significance of variant 3 overexpression was associated with (A) the presence of venous invasion (P = .037), (B) the presence of tumor microsatellite (P = .022), and (C) the absence of tumor encapsulation (P = .049).

Ectopic Expression of Variant 3 in HCC Cells Promoted Cell Growth.

Next, we addressed the in vitro effects of variant 3 in HCC cells by ectopic expression of variant 3 in an HCC cell model. An expression vector driving the expression of GFP-variant 3 fusion protein was introduced into BEL7402 cells, whose expression of variant 3 was relatively low among the examined cell lines (Fig. 1C). Two stable clonal transfectants of variant 3, sv3#1 and sv3#2, and 2 stable clones of the GFP vector, vec#1 and vec#2, were established. The vec#1 cells displayed a nonspecific uniform GFP signal throughout the cells, whereas a dotted pattern characteristic of focal adhesion proteins was observed in the sv3#1 cells (Fig. 3A). Apart from subcellular localization, expression of variant 3 in these stable clones was confirmed by reverse-transcription polymerase chain reaction (RT-PCR), shown in Fig. 3B. We first studied the effect of variant 3 on HCC cell lines by monitoring the growth rate of the stable transfectants (Fig. 3C). The doubling times of sv3#1 and sv3#2 were 21.1 ± 3.9 and 21.6 ± 3.4 hours, respectively, whereas those of vec#1 and vec#2 were 29.6 ± 3.9 and 27.7 ± 4.9 hours, respectively. On day 7, the vector control cells reached confluence, and the number of cells started to level off. In contrast, variant 3 cells continued to proliferate exponentially at confluence and were more closely packed, which might have been a result of enhanced loss of contact inhibition due to the expression of variant 3. On day 4, the number of both variant clones started to be significantly higher than the 2 vector control clones (P < .05, Student t test), and on day 8, the difference in cell number was even more marked (P < .005, Student t test). This result showed that sv3#1 and sv3#2 cells grew at a faster rate and had a higher saturation density than the control vec#1 and vec#2 cells.

Figure 3.

Expression of variant 3 enhanced cell growth of BEL7402 HCC cell line. (A) Stable expression of control GFP (vec#1) and GFP-variant 3 (sv3#1) in BEL7402 cell line. A uniform GFP signal was observed in both the nuclei and the cytoplasm of the control vec#1 cells, whereas a scattered pattern of dots was seen in sv3#1 cells. (B) Expression of variant 3 in stable clones. RT-PCR analysis confirmed expression of variant 3 in both the sv3#1 and sv3#2 stable clones. (C) Growth rates of control GFP stable cells (vec#1 and vec#2, solid squares and circles, respectively) and variant 3 stable cells (sv3#1 and sv3#2, open squares and circles, respectively). Variant 3 stable clones exhibited faster growth than the control clones. (*P < .05, **P < .005, Student t test). (D and E) Colony formation assay. BEL7402 and Hep3B cells were transfected with variant 3 and grown for 2 weeks under antibiotic selection. Growth-resistant colonies were fixed and stained with Giemsa. Variant 3 significantly increased the number of colonies of BEL7402 and Hep3B cells formed compared with the vector control (*P < .005, Student t test).

We also performed the colony formation assay by transfecting variant 3 into the BEL7402 and Hep3B HCC cell lines, which had relatively low expression of variant 3 (Fig. 1C). Compared with the vector control, variant 3 significantly increased the number of colonies formed (P < .005, Student t test; Fig. 3D-E).

Variant 3 Increased Migration and Ability to Invade of HCC Cell Lines.

It was previously reported that variant 3 played a positive regulatory role in the migration of HEK293 cells.16 Notably, there was higher expression of variant 3 in H2M HCC cells than in H2P HCC cells (Fig. 1C). The H2P and H2M cell lines were established from a primary HCC tumor and its corresponding metastatic tumor, respectively.22 The paired H2P and H2M cell lines have been used as a model for studying HCC metastasis.26 The observations of differences in variant 3 expression in these paired cell lines prompted us to investigate the effects of variant 3 on the mobility and invasive potential of HCC cells.

In the present study we evaluated the effect of variant 3 expression on cell migration of HCC cell lines using a transwell migration chamber and the wound-healing closure assay. To eliminate the differential cell proliferation rates of the GFP-vector control cells and the stable clones of variant 3, all tested cells were treated with a cell division inhibitor, mitomycin C, before seeding. The transwell migration assay showed that significantly more sv3#1 and sv3#2 cells migrated than vec#1 and vec#2 (control) cells (P < .0001, Student t test; Fig. 4A -B). The wound-healing assay also consistently showed that the sv3 clones were able to close the linear wound in a shorter time than were the control cells (Fig. 4C).

Figure 4.

Expression of variant 3 promoted cell migration. (A) Transwell migration assay. Migrated control (vec#1 and vec#2) and stable variant 3 (sv3#1 and sv3#2) cells were stained, photographed, and counted. (B) Numerical data of migration assay. The number of migrated cells in 5 fields was counted for each clone. The number of migrated sv3 cells was significantly higher than that of the control cells (P < .0001, Student t test). (C) Wound-healing assay. The distances between the leading edges of the migrating cells were photographed and compared at 24-hour intervals. Variant 3 cells closed the wound within 96 hours, whereas the control cells did not.

The invasive potential of variant 3 stable clones was also examined using an invasion chamber coated with extracellular matrix. Significantly more of both variant 3 clones invaded than did the vector controls (P < .0001, Student t test; Fig. 5A -B). These assays clearly demonstrated that variant 3 enhanced the cell mobility and invasiveness of HCC cells.

Figure 5.

Expression of variant 3 enhanced the potential of HCC cells to invade. (A) Invasion assay. Invaded cells of control clones (vec#1 and vec#2) and variant 3 stable clones (sv3#1 and sv3#2) in the lower chamber were fixed and stained. (B) Numerical data of invasion assay. The number of invaded sv3 cells was significantly greater than the number of invaded control cells (P < .0001, Student t test). Data represent the mean and standard deviations of triplicates.

Ectopic Expression of Variant 3 in HCC Cell Line Enhanced Tumorigenicity in Nude Mice.

We further evaluated the growth and metastatic potentials of variant 3 stable clones by orthotopic implantation of tumor xenografts into the livers of the mice. Tumor xenografts that originated in sv3#2 and vec#2 stable clones were implanted in the left lobes of the mice livers. The diameter of each tumor was measured at the time of sacrifice as a measure of tumor size and volume 3 weeks after implantation. Over the course of the experiments, no mice showed signs of weight loss. The sv3#2 clone formed primary tumors in all livers in which sv3#2-originated xenografts were implanted (5 of 5), whereas there was a lower incidence of primary tumor formation in livers that received vec#2-originated xenografts (3 of 5), and 1 of the 3 tumors formed was found in the peritoneum instead of the liver (Table 2). Use of the sv3#2 clone also resulted in significantly larger primary tumors, 0.48 ± 0.11 cm, than did use of the vec#2 clone, 0.26 ± 0.04 cm (P < .005, Student t test; Fig. 6A). This result further suggested that variant 3 enhances tumorigenicity. Although no intrahepatic or extrahepatic metastases were observed in the sv3#2-originated tumor, 3 of the 5 tumors of the sv3#2 clone displayed local invasion, with invasive tumor fronts entering adjacent liver tissue (Fig. 6B). One of these 3 tumors also showed venous invasion. None of these invasive features were seen in the tumors of the vec#2 control clone.

Table 2. Summary of Orthotopic Implantation
ClonePrimary TumorInvasion
IncidenceSize (diameter, cm)
  • One of the 3 tumors formed was found in the peritoneum attached to the surface of the liver.

VEC#23/50.26 ± 0.040/5
SV3#25/50.48 ± 0.113/5
Figure 6.

Variant 3 enhanced tumorigenicity in vivo. (A) Variant 3 stable clone, sv3#2, resulted in a significantly larger primary tumor than the control stable clone, vec#2 (P < .005, Student t test). Arrows indicate locations of primary tumors. (B) Tumors of the sv3#2 clone displayed features of local invasion, with an invasive tumor front invading adjacent liver tissue (arrows), and 1 tumor showed venous invasion (arrowhead). None of these invasive features were seen in the tumors of the vec#2 control clone.

Discussion

Tensin family members have been shown to be down-regulated in various cancer cells14, 19; however, variant 3, the smallest isoform of tensin2 and lacking the characteristic C1 domain in the N terminal, was capable of promoting cell migration.16 In the present study, we found that variant 3 was overexpressed in human HCC tissue compared with the corresponding nontumorous liver tissue (Fig. 1B). Furthermore, overexpression of variant 3 in HCCs was significantly associated with more aggressive tumor behavior—venous invasion (P = .037), tumor microsatellites (P = .022), and tumor nonencapsulation (P = .049), as shown in Fig. 2 and Table 1. In HCC, venous invasion and formation of tumor microsatellite are established features of tumor metastasis. Also, we previously showed that tumor encapsulation in HCC was an independent favorable prognostic parameter for tumor recurrence.21 We used an HCC cell line model to show variant 3 had a role in HCC progression and metastasis. Ectopic expression of variant 3 increased the migration and invasion potentials of the HCC cells (Figs. 4 and 5). Tumor xenografts that had originated from sv3#2 also displayed invasive potential in orthotopic implantation in liver (Fig. 6).

Focal adhesions are structural links between ECM and the actin cytoskeleton and are important sites of signal transduction where dynamic alterations of proteins in the focal contacts are involved during cell movement.27 Focal adhesion molecules have been shown to play critical roles in HCC. Paxillin, syndecan-1, and extracellular matrix metalloproteinase inducer have been reported to have clinical significance in HCC.28 Overexpression of focal adhesion kinase has also been shown to be a prognostic factor of HCC.29 Our current findings suggest tensin2 variant 3 could be another focal adhesion protein that has biological and clinical significance in HCC. Tensin members are actin-binding proteins at focal adhesions. Actin-binding domains have been localized at the N terminus of tensin.30 Tensin2 was found to be localized at the end of actin stress fibers and colocalized with vinculin at focal adhesions.16 Variant 3, by interacting and anchoring actin fibers at focal contacts, might play a crucial role in the attachment and detachment of cells in the process of cell migration. Indeed, the positive effect of variant 3 on migration of HEK293 cells has been demonstrated.16 Our current findings also substantiate the positive regulatory role of variant 3 in migration and invasion HCC cells.

Two structural domains at the C terminus of variant 3 facilitate its interaction with other signaling proteins. The presence of the SH2 domain in variant 3 enables its interaction with tyrosine phosphorylated proteins such as phosphoinositide 3-kinase.31, 32 Another structural domain of tensin proteins, PTB, is of particular interest because of its interaction with the NPXY motifs of integrin cytoplasmic tails.33 The PTB domain of Drosophila tensin is responsible for its recruitment to the sites of integrin.34 Tensin has also been found to colocalize with integrin α5 in human kidney.35 Integrins belong to a family of receptors that form the transmembrane linkages between ECM and the cytoskeleton. The extracellular domain of integrins interacts with ECM, whereas the cytoplasmic tail binds to cytoskeletal proteins. Aside from its structural role, integrin signaling is also involved in the regulation of various biological processes, such as cell adhesion, migration, and tumorigenesis.36, 37 Rho GTPases play an important role in integrin-mediated reorganization of the actin cytoskeleton, which underlies the cell migration process.38 Activities of alpha3beta1 integrin and matrix metalloproteinases have been implicated in HCC migration and invasion.39 Activation of the Rac signaling pathway has been demonstrated in HCC metastasis.26 The Rho/Rho-kinase pathway also facilitates the invasiveness of HCC.40 Overexpression of RhoA results in the induction of invasion by rat hepatoma cells both in vitro and in vivo.41 Other small GTPases that function as important molecular regulators in the rearrangement of the actin cytoskeleton during metastasis and tumorigenesis have been reported.42–44 The interaction of tensin proteins with the cytoplasmic tail of integrin and stress fibers of actin suggests the possibility that variant 3 may play a role in integrin-mediated organization of the cytoskeleton in the regulation of cell migration and invasion by HCC. Further studies are therefore warranted to elucidate the molecular targets of variant 3. A better understanding of these molecular targets may reveal a novel mechanism of regulation of cell metastasis in HCC.

In this study, we also investigated the positive effect of variant 3 on the growth of HCC cells. The results of the colony formation assay showed that variant 3 significantly increased the number of colonies formed when introduced into HCC cell lines (Fig. 3D-E). In addition, stable transfectants of variant 3 exhibited faster cell proliferation and reached a higher saturation density, probably because the defect in contact inhibition of HCC cells was magnified (Fig. 3C). This implied that variant 3 might play a role in cell-cell contact. In addition, variant 3 expression in HCC cells enhanced in vivo tumorigenicity in nude mice (Fig. 6). All these results showed that variant 3 might possess oncogenic properties.

The high recurrence rate of HCC because of metastasis has led to there being a poor prognosis for HCC. Clinical and basic research on metastasis of HCC has become an active area of investigation. Biomarkers including adhesion proteins, metalloproteinases, and angiogenesis regulators have been identified and shown to have the potential to be significant indicators of tumor progression and recurrence. The biological effect and clinical significance of variant 3 raise the possibility that it has the potential to act as a tumor progression marker for the prognosis of HCC and a therapeutic target in its treatment.

Acknowledgements

The authors thank Dr. Kwan Man for her technical assistance in orthotopic liver implantation and Dr. Chun-Ming Wong for assisting statistical analysis.

Ancillary