Overexpression of a disintegrin and metalloproteinase 21 is associated with motility, metastasis, and poor prognosis in hepatocellular carcinoma

Cell motility plays an important role in intrahepatic metastasis of hepatocellular carcinoma (HCC), and predicts poor prognosis in patients. The present study investigated the role of a disintegrin and metalloproteinases (ADAMs) in HCC, since these proteins are known to be associated with cell motility. We confirmed the expression of 12 ADAMs with putative metalloproteinase activity in HCC cells, and established a KYN-2 HCC cell line stably expressing short interfering RNA against ADAM21 to investigate the effect of ADAM21 deficiency on HCC cell motility and metastasis in vitro and in vivo. We also examined ADAM21 expression in a cohort of 119 HCC patients by immunohistochemistry. ADAM21 was overexpressed in KYN-2 cells, and its knockdown reduced invasion, migration, proliferation, and metastasis relative to controls. In clinical specimens, ADAM21 positivity was associated with vascular invasion, large tumor size, high histological grade, and lower overall and recurrence-free survival as compared to cases that were negative for ADAM21 expression. A multivariate analysis revealed that ADAM21 positivity was an independent risk factor for overall (P = 0.003) and recurrence-free (P = 0.001) survival. These results suggest that ADAM21 plays a role in HCC metastasis and can serve as a prognostic marker for disease progression.

Hepatocellular carcinoma (HCC) is a major cause of cancer-related mortality, with approximately 700,000 deaths worldwide and more than 30,000 in Japan each year 1,2 . Most HCC patients have poor prognosis due to the high incidence of intrahepatic metastasis and recurrence after curative therapeutic intervention 3 . The former results from the dispersal of cells around the tumor periphery to surrounding liver tissues through the portal vein 4,5 , although the underlying mechanisms are not well understood. We previously reported that HCC cell lines with high motility exhibit high rates of intrahepatic metastasis in vivo 6 . Inhibiting motility in highly metastatic HCC cell lines by overexpressing a dominant-negative Rho kinase or by treatment with Rho kinase inhibitor decreased metastatic potential, indicating that cell motility plays an important role in intrahepatic metastasis of HCC 6,7 .
A disintegrin and metalloproteinases (ADAMs) were originally discovered as a family of proteins with sequence similarity to the reprolysin family of snake venomases 8 . To date, 21 members of the human ADAM family have been identified 9 . ADAMs share an N-terminal secretory signal sequence, prodomain, and metalloprotease, disintegrin-like, cysteine-rich, epidermal growth factor-like, transmembrane, and cytoplasmic domains 10 . More than half of ADAMs have an active site in the metalloprotease domain with a common HEXXHXXGXXH sequence, similar to matrix metalloproteinases 11 . Protease-type ADAMs are closely associated with cell invasion, proliferation and metastasis. Upregulation of ADAM17 is a pathological feature of HCC 12 ; suppression of ADAM17 by the microRNA miR-122 inhibited in vitro migration, invasion, in vivo tumorigenesis, intrahepatic metastasis, and local invasion in the livers of nude mice 13 . However, there have been no systematic studies on the role of other protease-type ADAMs in HCC.
To address this issue, the present study analyzed the expression of 12 protease-type ADAMs in HCC cell lines and found that ADAM21 was overexpressed in the highly motile and metastatic KYN-2 HCC cell line. We also found that high ADAM21 expression predicted poor prognosis in HCC patients. These results provide the first evidence that ADAM21 plays an important role in the malignant progression of HCC.

Results
Real-time quantitative PCR analysis of proteinase-type ADAM expression in HCC cell lines. We evaluated the mRNA levels of ADAM8, 9,10,12,15,17,19,20,21,28,30, and 33-which have putative metalloproteinase activity-in KYN-2, PLC/PRF/5, and HepG2 cells and normal liver tissue by real-time PCR. ADAM17, 19, and 21 were more highly expressed in KYN-2 cells than in PLC/PRF/5 and HepG2 cells and normal liver (Fig. 1A). ADAM17 and 19 expression has been linked to many human cancers 9 , and ADAM17 is a target of the micro (mi)RNA miR-122 and is involved in HCC tumorigenesis and metastasis 13 . Many studies have reported that ADAM19 is a direct target of several miRNAs, including miR-153, -30c, or -145 and plays an important role in cell migration, invasion, and proliferation in various cancers [14][15][16] . Moreover, dysregulation of transforming growth factor-β/SMAD4 signaling leads to epigenetic silencing of its downstream target ADAM19 in ovarian cancer cells 17 . Therefore, in this study, we evaluated the expression and function of ADAM21 in HCC tumorigenesis and metastasis. An immunoblot analysis revealed that ADAM21 protein was highly expressed in KYN-2 cells and expressed at low levels in HepG2 cells and normal liver (Fig. 1B).

ADAM21 knockdown inhibits HCC cell proliferation, migration, and invasion and induces cell-cycle arrest and apoptosis.
To investigate the in vitro function of ADAM21, KYN-2 cells were stably transfected with either a control short interfering (si) RNA or one targeting ADAM21 ( Fig. 2A). Similar to the parent cell line, cells expressing control siRNA had a ruffled membrane and fusiform shape without cell-to-cell contacts (Fig. 2B). In contrast, cell expressing the siRNA against ADAM21 had fewer membrane ruffles and formed relatively cohesive colonies (Fig. 2B). We investigated changes in the expression of epithelial-mesenchymal transition (EMT) markers such as E-cadherin, zona occludens (ZO)-1, N-cadherin, and vimentin by immunoblotting upon ADAM21 knockdown and found no changes in E-cadherin and ZO-1 levels, whereas N-cadherin and vimentin were not detected in KYN-2 cells expressing either control or ADAM21 siRNA (data not shown). The cell proliferation assay revealed that ADAM21 knockdown reduced the growth of KYN-2 cells (P < 0.05 at 48 and 72 h and P < 0.01 at 96 h vs. control siRNA-transfected cells) (Fig. 2C, upper panel). We performed a cell-cycle analysis to clarify the mechanism by which ADAM21 knockdown inhibits cell proliferation and found that 96 h after seeding, KYN-2 cells expressing ADAM21 siRNA were arrested in G0/G1, and a smaller proportion were in S phase as compared to KYN-2 cells expressing control siRNA (P < 0.01 and P = 0.01, respectively) (Fig. 2C, middle panel). Immunoblot analysis revealed detectable cleavage of poly (adenosine disphosphate)-ribose polymerase (PARP) (Fig. 2C, lower panel). In addition, migration and invasion assays revealed that ADAM21 siRNA-expressing KYN-2 cells showed significantly reduced migration (P < 0.001) and invasion (P < 0.01) (Fig. 2D). We also observed a similar effect in cells transfected with siRNA oligonucleotides (Supplemental Fig. 1A-C).

ADAM21 knockdown slows tumor growth in a murine orthotopic implantation model. To
determine whether ADAM21 is involved in tumorigenesis or intrahepatic metastasis in vivo, KYN-2 cells expressing siRNA against ADAM21 or control siRNA were orthotopically implanted into the livers of SCID mice. Tumors formed in all mice at the site of injection; however, the maximum diameter of the primary tumor was smaller in the ADAM21 siRNA group than in the control siRNA group (Fig. 3A); this corresponded to a lower incidence of intrahepatic metastasis in the former (Fig. 3B), although the difference was not statistically significant. These results indicate that ADAM21 knockdown may inhibit tumorigenesis and intrahepatic metastasis in vivo.
High ADAM21 expression in HCC tumors is associated with clinicopathological parameters of HCC. We examined ADAM21 protein expression in 119 patients with HCC by immunohistochemistry.

Discussion
Recent advances in both diagnosis and therapeutics have allowed clinicians to identify HCC at earlier stages and recommend treatment in a timely manner. However, the prognosis of HCC remains poor, largely due to the high incidence of intrahepatic metastasis and recurrence after curative therapeutic intervention 3 . Additionally, although recent progress has advanced our knowledge and understanding of the molecular pathogenesis of HCC and the signaling pathways involved, the only mechanism-based pharmaceutical currently available is sorafenib 18 . There is therefore a need to identify biomarkers for predicting HCC recurrence and prognosis as well as new therapeutic targets.
We previously demonstrated that cell motility plays a critical role in intrahepatic metastasis of HCC 6,7 . In the present study, we examined protease-type ADAMs that are known to facilitate cell migration and invasion via proteolytic cleavage of extracellular matrix components, and found that ADAM21 was selectively overexpressed in KYN-2 cells. ADAM21 (also known as ADAM31) was first identified as a human testis-specific membrane metalloprotease with similarity to ADAM1 (fertilin-α), suggesting a role in sperm function 19 . In the developing and adult rodent central nervous system, ADAM21 is thought to be involved in axonal outgrowth and/or synapse formation 20 . There has been only one study on ADAM21 that identified this protein as a candidate factor involved in the invasion of HT29 colon cancer cells in vitro 21 . Our findings support this earlier report and provide further evidence that silencing ADAM21 in KYN-2 cells causes cell-cycle arrest and inhibits cell proliferation, migration, invasion, and metastasis in vitro and in vivo. Although ADAM21 knockdown resulted in morphological changes to an epithelial phenotype, there were no changes in EMT marker expression, suggesting that ADAM21 knockdown alone may be not sufficient to reverse EMT. Recently, two definitive in vivo studies demonstrated that EMT is not relevant to metastasis but is associated with chemoresistance in breast and pancreatic cancers 22,23 . Further study is needed to determine whether this applies to intrahepatic metastasis of HCC. In our animal model, ADAM21 knockdown did not completely abolish intrahepatic metastasis, indicating that it is a complex process mediated by many factors 24 .
Our analysis of tissue samples from HCC patients undergoing curative resection showed that ADAM21 expression varied, and was present not only in tumor cells but also in adjacent non-neoplastic hepatocytes. It has been reported that many types of tumor produce growth factors that stimulate the proliferation and invasion of tumor cells themselves (in an autocrine fashion) and of surrounding normal cells (in a paracrine fashion) 25 . To our knowledge, there are no previous reports on the regulatory mechanism of ADAM21. One possible mechanism is the paracrine stimulation of adjacent non-neoplastic hepatocytes by unidentified factors that upregulate ADAM21 production by tumor cells.
Our clinicopathological study showed that high ADAM21 expression level in HCC tumors is closely related to large tumor size, high histological grade, and presence of vascular invasion and is an independent risk factor for poor OS and RFS. This is consistent with the results of our in vitro and in vivo studies and suggests that ADAM21 is a potential biomarker for HCC progression. However, further studies are needed in order to identify the physiological substrate(s) of ADAM21 and clarify its mechanism of action in HCC.
In conclusion, ADAM21-mediated cell motility plays an important role in HCC metastasis and is a prognostic biomarker for HCC patient OS and RFS after hepatectomy. These results indicate that therapeutic strategies targeting ADAM21 may be an effective treatment for preventing HCC metastasis and recurrence.

Materials and Methods
Cell lines and culture. The PLC/PRF/5 cell line was obtained from the Japanese Collection of Research Bioresources (Osaka, Japan); HepG2 cells were obtained from the RIKEN BioResource Center (Tsukuba, Japan); and the KYN-2 cell line was provided by Dr. H. Yano (Kurume University, Kurume, Japan) 26 . We previously characterized KYN-2 cells as a highly motile and metastatic HCC cell line, whereas PLC/PRF/5 and HepG2 cells exhibit low motility and metastatic potential 6 . All cells were cultured in Roswell Park Memorial Institute (RPMI) 1640 medium containing 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin at 37 °C in a humidified atmosphere of 5% CO 2 . When required, the medium was supplemented with 0.25 μg/mL puromycin (Invitrogen, Carlsbad, CA, USA).

Real-time PCR analysis.
Total RNA was isolated using the RNeasy Mini kit (Qiagen, Valencia, CA, USA) according to the manufacturer's protocol. cDNA was synthesized from 1 μg of total RNA using a PrimeScript RT reagent kit (Takara Bio, Otsu, Japan) and used as a template for real-time PCR on a Light Cycler System (Roche Diagnostics, Indianapolis, IN, USA). Primer/probe sets were selected from the Taqman Gene Expression Assays collection (Applied Biosystems, Foster City, CA) ( Table 4). PCR conditions were as follows: 95 °C for 10 min, and 45 cycles of 95 °C for 10 s, 60 °C for 30 s, and 72 °C for 1 s. Results were normalized to the levels of human β-actin mRNA. Measurements were performed in triplicate.
Immunoblotting. Cell lysates were prepared and immunoblotting was carried out as previously described 27 .
Briefly, 30 μg of protein were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to an Immobilon membrane (Millipore, Bedford, MA, USA) that was blocked and probed with primary antibodies followed by the horseradish peroxidase (HRP)-conjugated secondary antibodies (GE Healthcare, Piscataway, NJ, USA). Peroxidase-labeled bands were visualized with enhanced chemiluminescence detection reagents (GE Healthcare) according to the manufacturer's instructions. To confirm equal loading, the membrane was reprobed with anti-β-actin antibody.
SiRNA-mediated knockdown of ADAM21. SiRNA against ADAM21 and control siRNA were generated using the piGENE vector (Nagase, Tokyo, Japan) and the following target sequences: ADAM21, GTAGAAACATTAGTACATC and control (T7STOP), CACCTTTTTTT. The siRNAs or control plasmid were transfected into KYN-2 cells with Lipofectamine 2000 (Invitrogen), and stable cell lines were generated from puromycin-resistant colonies.   Immunohistochemistry. Sections (5 μm thick) of formalin-fixed, paraffin-embedded tissue were deparaffinized in xylene and rehydrated in ethanol, treated with 0.3% hydrogen peroxide in methanol, and immersed in citrate buffer (10 mmol/L, pH 6.0). After autoclaving (121 °C, 10 min) and preincubating with normal swine serum, the sections were incubated overnight at 4 °C with primary antibody, which was detected with the Mouse/ HRP Envision + system (Dako, Carpinteria, CA, USA). Immunoreactivity was visualized using 0.05% diaminobenzidine tetrahydrochloride solution containing 0.01% hydrogen peroxide. Sections were counterstained with hematoxylin. Leydig cells in the testis tissue served as a positive control for ADAM21; negative controls were prepared by substituting normal mouse serum for primary antibody and showed no detectable labeling. ADAM21 expression in tumor specimens was evaluated by comparison with adjacent non-neoplastic hepatocytes and was classified as high expression, unchanged expression, or loss (i.e., signal intensity of the tumor specimen was higher than, similar to, or lower than that of non-neoplastic hepatocytes, respectively) or both negative expression (i.e., the tumor specimen and non-neoplastic hepatocytes both showed no signal). However, the final results were recorded only as negative (unchanged expression, loss, or both negative expression) or positive (high expression). Hematoxylin and eosin staining and immunohistochemical analyses were carried out by two independent observers (H.H. and N.K.) who were blinded to patients' clinicopathologic information.
Statistical analysis. Continuous variables are expressed as means ± standard deviation or median (range).
Associations among variables were evaluated with the Student's t test, Mann-Whitney U test, χ2 test, or Fisher's exact test as appropriate. Medical records and survival data were obtained for all patients. Deaths from other causes were treated as uncensored cases. The Kaplan-Meier method was used to estimate the cumulative incidences of events, and differences in these incidences were analyzed with the log-rank test. Univariate and backward stepwise multivariate survival analyses were performed using the Cox proportional hazards model. Statistical analyses were performed using SPSS v.17.0 for Windows (SPSS Inc., Chicago, IL, USA). All tests were two-sided, and P values < 0.05 were considered statistically significant.