Tight Junction Protein Claudin-1 Enhances the Invasive Activity of Oral Squamous Cell Carcinoma Cells by Promoting Cleavage of Laminin-5 γ2 Chain via Matrix Metalloproteinase (MMP)-2 and Membrane-Type MMP-1

The malignancy of cancer cells depends largely on their proliferative, invasive, and metastatic activities (1–5), and invasive and metastatic activities are most closely associated with cell-to-cell and cell-to-extracellular matrix adhesion (6–13). Cell-to-cell adhesion apparatuses, such as tight junctions, adherens junctions, desmosomes, and gap junction, strongly regulate the invasion activity of cancer cells (14–19). Tight junctions, one of the epithelial junctional complexes, are located at the most apical part of the complexes and consist of claudin and occludin. Tight junctions construct epithelial cellular sheets and participate in the polarity of epithelial cells through its own polymerization (20).

Among the above adhesion apparatuses and their components, E-cadherin has been extensively studied about its biological activity and association with cancer cell invasion. It was reported that the invasive activity and metastatic potential of cancer cells were promoted when the cellular adhesion by E-cadherin decreased (21, 22). In addition, it was clarified that E-cadherin has a crucial role in cellular cross-talk. E-cadherin-mediated cell-to-cell interaction is modulated by protein interactions on the cytoplasmic membrane surfaces. A recent study showed that phosphorylation of both p120 protein and β-catenin affects their interaction with E-cadherin, inducing their binding to the cytoskeleton (23). Their binding seems to be strongly associated with interaction between α-catenin and/or α-actinin, vinculin, ZO-1, and actin, but the details of the binding mechanism remain to be clarified.

The transcription of mouse E-cadherin is under the control of Snail, a strong repressor that specifically interacts with the mouse E-cadherin promoter (24). Overexpression of Snai1 leads to a dramatic conversion of epithelial cells to a fibroblastic phenotype accompanied by the complete suppression of E-cadherin expression and an increase of the invasive property of the cells. In addition, high expression of Snail mRNA with low expression of E-cadherin mRNA was reported in highly invasive and metastatic cancer cells. Including these reports, there are many studies on the relationship between Snail and E-cadherin (25–27). However, there are extremely few studies about the relationship between the expression of occludin and claudins in cancer cells and their relationship to invasion activity and metastatic potential (16, 28, 29), although the relationship between E-cadherin expression and cancer invasion has been deeply explored. One of the few studies on claudins indicated that a pancreatic carcinoma cell line having low metastatic and invasive potential exhibited strong claudin-4 expression, whereas claudin-4 expression was very weak in highly invasive cell lines (30). It was also reported that overexpression of claudin-4 was associated with strong reduction of the invasive potential in vitro, inhibiting the colony formation in soft agar, and that, in vivo, tail vein-injected claudin-4 overexpressing cells formed for fewer metastatic foci in the lungs of the mice compared with mock-transfected cells. Claudin recruits membrane-type matrix metalloproteinase-1 (MT1-MMP) and pro-MMP-2 on the cell surfaces to elevate the focal concentration of MT1-MMP and pro-MMP-2 and, consequently, activates pro-MMP-2 (31). However, it has not yet been finally concluded whether claudins regulate cancer invasion or not.

Laminin, as well as type IV collagen, is an important constitutional element of the basement membrane (32). Laminin-5, a member of the laminin family, forms hemidesmosomes with adhesion molecules called integrin (33). Laminin-5, a heterotrimer consisting of α, β, and γ chains, not only functions in cell adhesion but also plays a role in signal transduction in association with cytokines and growth factors (34). The laminin-5 γ2 chain is cleaved by MT1-MMP and active MMP-2 (35, 36). The cleaved γ2 chains bind epidermal growth factor receptors (EGFR) on cancer cell surfaces and transmit intracellular signals that promote cell growth and mobility (37). It was also reported that MT1-MMP cleaves the β3 chain of human laminin-5 and that the cleaved laminin-5 β3 chain enhances the migration of prostate carcinoma cells (38).

The aim of the present study was to characterize the biological role of claudin-1 up-regulation in squamous cell carcinoma (SCC) cells. We investigated the correlation between the expression of claudin-1 and invasion activity of SCC cells, and we examined the regulation of claudin-1 expression and the roles of MMP-2, MT1-MMP, and laminin-5 γ2 chains in the invasion of SCC cells. We showed that claudin-1 induced the cleavage of laminin-5 through MT1-MMP and active MMP-2, and the cleaved laminin-5 γ2 chains promoted the invasion activity of cancer cells.

Reagents and antibodies. MMP-2 inhibitor I and EGFR tyrosine kinase inhibitors AG1478 and PD168393 were obtained from Calbiochem (Bad Soden, Germany). Mouse anti-laminin-5 γ2 chain monoclonal antibody (mAb; clone D4B5), rabbit anti-MT1-MMP function-blocking polyclonal antibody AB8102, mouse anti-integrin-β₁ function-blocking mAb (clone P5D2; diluted 1:200; Chemicon International, Hofheim, Germany), rabbit anti-claudin-1 polyclonal antibody 18-7362, mouse anti-claudin-4 mAb 18-7341, rabbit anti-occludin polyclonal antibody 71-1500 (diluted 1:100; Zymed, South San Francisco, CA), rabbit anti-E-cadherin polyclonal antibody sc-7870, mouse anti-actin mAb sc-8432, goat anti-Snail polyclonal antibody sc-10432 (diluted 1:200; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and mouse anti-EGFR neutralizing mAb (clone LA-1; Upstate Biotechnology, Lake Placid, NY) were used in the present study.

Cell lines and cell culture. Among the SCC cell lines established from human metastatic lymph nodes in our laboratory, OSC-4 (highly metastatic), NOS-2 (highly metastatic), and OSC-7 (weakly metastatic) were used in the present study (39). The cells were cultured in DMEM (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) with 10% fetal bovine serum (FBS), 10 mmol/L glutamine, 100 μg/mL streptomycin (Invitrogen, San Diego, CA), and 100 units/mL penicillin G (Invitrogen) at 37°C in a humidified 5% CO₂/95% air atmosphere. In each experiment, subconfluent cells in plastic dishes were used after trypsinization.

Invasion assay. The invasive potential of the tumor cells was examined using a BioCoat Matrigel Invasion Chamber kit (Becton Dickinson Biosciences, Bedford, MA), and the invasive activity was determined according to the manufacturer's instructions (40). Briefly, before seeding into the transwell inserts, cells were released using trypsin-EDTA and sequentially rinsed with DMEM containing 10% FBS. The rinsed cells were resuspended in serum-free DMEM (1.5 × 10⁵/mL), and 500 μL of the cell suspension (7.5 × 10⁴) were added to the transwell insert chamber with a filter that was coated with Matrigel. In the lower compartment, 750 μL DMEM containing 10% FBS was used as chemoattractant. The cells were incubated with or without mouse anti-integrin-β₁ mAb (clone P5D2; 1:200), MMP-2 inhibitor I (20 μmol/L), rabbit anti-MT1-MMP function-blocking polyclonal antibody AB8102 (4.5 μg/mL), mouse anti-EGFR neutralizing mAb (clone LA-1; 10 μg/mL), and the EGFR inhibitors AG1478 (3 μmol/L) and PD168393 (2 μmol/L) in DMEM. Both chambers were incubated at 37°C under 5% CO₂/95% air atmosphere for 12 hours. The inserts were removed, and noninvading cancer cells remaining on the upper side of the filter were scraped off. Cells that had invaded into the lower side of the filter were then stained with H&E and microscopically observed and counted in five fields of view at ×200 magnification. The invasive activity of cancer cells was expressed as the mean number of cells that invaded to the lower side of the filter, and results were presented as mean ± SD of cells per field of view. All assays were done in triplicate.

Xenotransplantation of tumor cells. Cells (2 × 10⁵/0.05 mL) were injected into the middle third of the tongue margin of 6-week-old severe combined immunodeficient (SCID) male mice. The mice were raised to form a tumor in the tongue, and they were killed when the tumor grew to ∼5 mm in diameter. Each tumor was excised, and the tumor tissues were used for histopathologic evaluation and immunohistochemical analyses. A histopathologic evaluation was carried out by the mode of tumor cell invasion according to Jakobsson's classification [i.e., well-defined border (grade 1), cords and less marked boarder (grade 2), groups of cells and no distinct border (grade 3), and diffuse invasion (grade 4); ref. 41].

Immunohistochemistry. Immunohistochemical staining using the Dako Catalyzed Signal Amplification II System (code K1497, Dako, Kyoto, Japan) was done for formalin-fixed and paraffin-embedded specimens (40), which were prepared from the tissues obtained from the tumors formed in the tongues of mice. Each section was deparaffinized and incubated with diluted primary antibody. After incubation for 15 minutes, each section was washed thrice and incubated with the appropriate secondary antibody for 15 minutes. The staining was completed by incubation of each section with 3,3′-diaminobenzidine tetrahydrochloride/hydrogen peroxide for 5 minutes. For the control staining, the primary antibody was replaced by nonimmune serum.

Western blot analysis. Cultured and harvested cells were lysed in ice-cold TNE lysis buffer containing 1 mol/L Tris-HCl (pH 7.6), 0.5 mol/L EDTA, and 10% NP40 for 1 minute. Total proteins were extracted according to standard procedures, and extracted proteins (50 μg/lane) were separated by SDS-PAGE. Proteins were then electrophoretically transferred onto Immobilon-P membranes (Millipore Corp., Bedford, MA), and the membranes were probed with each antibody. Serum-free DMEM conditioned medium was concentrated to 20 μL by an Ultrafree-CL 50,000 nominal molecular weight limit filter unit (Millipore) and mixed with 10 μL SDS sample buffer. Detection was done with the enhanced chemiluminescence system (Amersham Pharmacia Biotech, Piscataway, NJ).

Zymography. The gelatinolytic activity of MMPs in the conditioned medium of OSC-4, NOS-2, and OSC-7 cell lines was determined using zymography. Subconfluent monolayered cells were cultured in serum-free DMEM for 24 hours, and 45 μL of the conditioned medium were mixed with 15 μL SDS sample buffer. Subsequently, the mixtures were electrophoresed on 7.5% PAGE containing 1 mg/mL gelatin (Sigma Chemical Co., St. Louis, MO). After electrophoresis, the gels were soaked in 2.5% Triton X-100 for 1 hour to remove SDS and incubated in 50 mmol/L Tris-HCl (pH 8) containing 0.5 mmol/L CaCl₂ and 1 μmol/L ZnCl₂ for 16 hours at 37°C. The gels were then stained with 1% Coomassie brilliant blue G250. After 10 minutes, the gels were destained with 10% methanol and 5% acetic acid for 1 hour. The proteolytic activity was detected as clear bands on a blue background of the Coomassie brilliant blue–stained gel. The experiment was done in triplicate.

Small interfering RNA targeting of claudin-1 and claudin-4. Synthetic small interfering RNAs (siRNA) were purchased from Ambion, Inc. (Austin, TX). The target DNA sequences of claudin-1 and claudin-4 were GAGGATTTACTCCTATGCCGG and AAGGCCAAGACCATGATCGTG, respectively, and siRNA for claudin-1 (5′-GGAUUUACUCCUAUGCCGGtt-3′ and 5′-CCGGCAUAGGAGUAAAUCCtc-3′) and siRNA for claudin-4 (5′-GGCCAAGACCAUGAUCGUGtt-3′ and 5′-CACGAUCAUGGUCUUGGCCtt-3′) were used as annealed oligonucleotides. OSC-4 cell lines were transfected with siRNA using Oligofectamine Reagent (Invitrogen) according to the manufacturer's instructions. For each experiment, 4 μL Oligofectamine was preincubated with 26 μL serum-free DMEM for 10 minutes at room temperature. The 20 μL of 20 μmol/L claudin-1 siRNA oligonucleotide were mixed with 350 μL serum-free DMEM, then combined with 30 μL preincubated Oligofectamine/DMEM, and incubated at room temperature for 20 minutes. The cells were washed once with 2 mL serum-free DMEM and incubated with a solution that contained 1.6 mL serum-free DMEM and 400 μL of the preincubated oligonucleotide/Oligofectamine mixture for 4 hours in a 37°C humidified 5% CO₂/95% air atmosphere. After 48 hours, the cells and medium were collected, and the levels of claudin-1, MMP-2, MT1-MMP, and laminin-5 γ2 chains were analyzed.

Statistical analysis. Data are presented as mean ± SD. Statistical significance was determined by unpaired Student's t test. Comparisons with P <0.05 were considered statistically significant.

Invasion activity of SCC cells. We used an invasion assay to determine the levels of the invasive potentials of the cell lines OSC-4, NOS-2, and OSC-7. The histopathologic characteristics of these cell lines, such as differentiation degree and invasion mode of the tumor cells, are shown in Fig. 1A, to C. After cultivation of the cells for 12 hours, invasion of 139 ± 10 and 70 ± 5 cells per field of view through the porous transwells was observed for OSC-4 and NOS-2, respectively. For OSC-7, a smaller number of cells (10 ± 4 per field of view) invaded into the lower chamber (Fig. 1D).

Expression of claudin-1, claudin-4, occludin, and Snail in each cell line. Strong expression of claudin-1 was observed in the highly invasive cell lines OSC-4 and NOS-2, but its expression level was very low in the weakly invasive OSC-7 cell line (Fig. 2A). Conversely, claudin-4 and occludin were weakly expressed in both OSC-4 and NOS-2 and strongly expressed in OSC-7. The expression pattern of Snail was similar to that of claudin-1, that is, it was strongly expressed in OSC-4 and NOS-2 and weakly expressed in OSC-7.

Correlation between the expression of claudin-1 and the expression and activity of MMP. Strong expression of a 63-kDa molecule, which is an active form of MT1-MMP, was observed in the highly invasive cell lines OSC-4 and NOS-2, but the level of expression of this protein was very low in the weakly invasive OSC-7 cell line (Fig. 2B). Zymography revealed strong and slightly weaker MMP-2 activity in OSC-4 and NOS-2, respectively, but MMP-2 activity was not observed in OSC-7. Similarly, MMP-9 activity was high in OSC-4 and NOS-2 but not observed in OSC-7 (Fig. 2C).

Laminin-5 γ2 chains in the invading fronts of tongue tumors and in the conditioned medium of tumor cells. The expression of laminin-5 γ2 chains was examined immunohistochemically in the tumors formed in the tongues of SCID mice. Laminin-5 γ2 chains were not detected in the central portion of the tumors formed after the inoculation of OSC-4 (Fig. 3A) and NOS-2 (Fig. 3B) but were detected in tumor cells at the invading fronts of these tumors. The weak expression of laminin-5 γ2 chains was detected in the tumor formed in the murine tongue inoculated with OSC-7 (Fig. 3C). The cleaved laminin-5 γ2 chain (105 kDa) was detected in the conditioned medium from OSC-4 and NOS-2 (Fig. 3D). The level of laminin-5 γ2 chain in the conditioned medium from OSC-7 was considerably less than that detected in the conditioned medium from OSC-4 and NOS-2.

Influence of claudin-1 siRNA on the invasion activity of the tumor cells. After cultivation of claudin-1 siRNA-transfected and control siRNA-transfected OSC-4 for 48 hours, 12 ± 6 and 83 ± 5 cells per field of view moved into the lower side of the filters through the pores in the claudin-1 siRNA-transfected and control siRNA-transfected cells, respectively (Fig. 4). Control siRNA caused a statistically significant reduction in the invasion activity without any apparent effect on the expression level of claudin-1. The invasion activity of siRNA-transfected cells was ∼8.6% of that of the control cells.

Influence of claudin-1 siRNA on the MMP-2 activity, MT1-MMP expression, and levels of laminin-5 γ2 chains. The densitometric levels of active MMP-2 were 2,480 and 1,458 and those of MT1-MMP were 205 and 127 in control siRNA-transfected and claudin-1 siRNA-transfected cells, respectively. The densitometric levels of 105-kDa and 27-kDa laminin-5 γ2 chains were 221 and 217 in the control siRNA-transfected cells and 181 and 15 in the claudin-1 siRNA-transfected cells, respectively. The MMP-2 activity and the expression levels of MT1-MMP and 105-kDa and 27-kDa laminin-5 γ2 chains in claudin-1 siRNA-transfected OSC-4 were ∼59%, 62%, 82%, and 7% of their control levels, respectively. In good agreement with this, the levels of the 105-kDa and 27-kDa laminin-5 γ2 chains as well as MMP-2 and MT1-MMP in the conditioned medium were decreased by the transfection of claudin-1 siRNA (Fig. 5).

Influence of MT1-MMP antibody, MMP-2 inhibitor I, EGFR antibody, and EGFR inhibitors on the invasion activity of the tumor cells. During 12 hours of cultivation, the number of cells that invaded into the lower side of the porous transwells was markedly decreased to 40 ± 8, 38 ± 6, 37 ± 5, 25 ± 3, and 20 ± 5 cells per field of view by the addition of MT1-MMP antibody, MMP-2 inhibitor I, EGFR antibody, and each of the EGFR inhibitors (AG1478 or PD168393), respectively. These reagents suppressed the invasion activity to ∼29% of the control level (Fig. 6). Antibody against integrin-β₁ slightly suppressed the invasion activity: 100 ± 20 cells per field of view invaded through the pores in the presence of antibody against integrin-β₁.

Influence of claudin-4 siRNA on the invasion activity of the tumor cells. After cultivation of the cells for 48 hours, the invasion of 100 ± 5 and 125 ± 10 cells per field of view through the pores was observed in control siRNA-transfected and claudin-4 siRNA-transfected OSC-4, respectively. Although claudin-4 siRNA significantly suppressed the expression level of claudin-4, the level of invasion activity was increased by the transfection of claudin-4 siRNA to ∼125% of the control level (Fig. 6).

Cancer invasion and metastasis are regulated by multiple molecules involved in multiple steps (1–5, 42). The first step of invasion as well as metastasis is loss of cell-to-cell adhesion (6–8, 10–12). Cell-to-cell adhesion is strongly regulated by apparatuses, such as tight junctions, adherens junctions, desmosomes, and gap junctions (14–19). Decreased expression of the component molecules of these adhesion apparatuses means a decrease of cell-to-cell adhesion, which would result in an increase of invasion activity of the cells. Of the components of these cell adhesion structures, E-cadherin, which is a major component of adherens junctions, has been extensively explored about its role in cancer cell invasion and metastasis (21, 22). However, the role of tight junctions in the invasion and metastasis of cancer cells has not been clarified.

Tight junctions consist of occludin (65 kDa), claudin-1 (∼23 kDa), and claudin-4 (∼23 kDa), each of which has four transmembrane domains (20). It became clear recently that tight junctions have physiologic roles, such as regulating ion and hydrophilic substance permeability and cellular polarity (43). In addition to these functions, the involvement of tight junctions in gene transcription, tumor suppression, and cell proliferation has recently been reported (44). These recent studies indicate that tight junctions are not a simple cell adhesion apparatus but a multifunctional apparatus associated with cell proliferation and invasion (16, 28–31, 44).

About the association of tight junction components with the invasion and metastasis of cancer cells, Michl et al. (30) reported that noninvasive pancreatic cancer cell lines exhibited strong claudin-4 expression, that claudin-4 was very weakly expressed in highly invasive cell lines, and that strong invasion potential was markedly suppressed by overexpression of claudin-4 in highly invasive cell lines. In addition, Dhawan et al. (28) reported that claudin-1 was localized in the normal mucous membrane of the large intestine, but a higher level of claudin-1 was localized in the cytoplasm and nuclei of colorectal cancer cells than in normal bowel epithelial cells. Furthermore, it was reported that induction of claudin-1 overexpression in colorectal cancer cells with a low invasive activity induced a highly invasive and metastatic potential. In the present study, the high invasive cell lines (OSC-4 and NOS-2) expressed claudin-1 more strongly than the weakly invasive cell line (OSC-7). However, claudin-4 and occludin were weakly expressed in both OSC-4 and NOS-2 and strongly expressed in OSC-7. These results are in accord with the report by Dhawan et al (28). When claudin-1 siRNA was introduced into the claudin-1-overexpressing OSC-4, their invasive activity was markedly suppressed, suggesting that claudin-1 positively regulates cancer cell invasion. According to others' reports, claudin-4 seems to repress the invasion of cancer cells (30). We do not know yet why claudin-1 and claudin-4, which belong to the same claudin family, function oppositely in the regulation of cancer cell invasion.

The expression of MT1-MMP and active MMP-2 decreased when claudin-1 siRNA was introduced into the cells, suggesting that claudin-1 induces increased expression of MT1-MMP through some signaling mechanism. It is well known that cell lines with overexpression of MT1-MMP and active MMP-2 are highly invasive (45). Claudin-1, claudin-2, claudin-3, and claudin-5 recruit MT1-MMPs and pro-MMP-2, which results in elevated focal concentrations of pro-MMP-2 (31). Our present result is compatible with that report: in OSC-4 with high claudin-1 expression, high levels of MT1-MMP and active MMP-2 were observed. Conversely, in OSC-7 (a weakly invasive cell line) with low expression of claudin-1, the expression of MT1-MMP was very weak and active MMP-2 was not detected. Furthermore, the invasion activity was inhibited by the introduction of claudin-1 siRNA into OSC-4. This suggests that enhancement of the invasion activity by claudin-1 was mediated via the enhanced expression of MT1-MMP and activation of MMP-2.

The mechanism of the enhancement of invasion activity by MT1-MMP and active MMP-2 is not yet known. It was recently reported that laminin-5 γ2 chain was overexpressed in the invasion fronts of a variety of carcinoma tissues as well as in cancer cells with a high invasion activity (34) and that such highly invasive cancer cells secreted a large amount of the cleaved form of laminin-5 γ2 chains in the conditioned medium (35). According to recent reports, laminin-5 γ2 chains are cleaved by MT1-MMP and active MMP-2, and the cleaved laminin-5 γ2 chains then bind EGFR of cancer cell surfaces, like soluble cytokines and growth factors, and thereby induce cell invasion activating signals (34). In accord with this notion, we observed high expression levels of laminin-5 γ2 chains in the invasion fronts of highly invasive cell lines in tumors formed in mouse tongues. The present study also showed that a large quantity of cleaved form laminin-5 γ2 chains was secreted in the conditioned medium of these highly invasive cell lines compared with that in the weakly invasive cell line. In addition, the highly invasive activities of the cells were strongly inhibited when the cells were treated with anti-MT1-MMP antibody or MMP-2 inhibitor I. However, their invasion activities were not inhibited by treatment with anti-integrin-β₁ antibody. In parallel with other researchers' reports (35, 36), these present results support the cleavage and activation of laminin-5 by MT1-MMP and MMP-2. As described above, the expression of MT1-MMP and active MMP-2 was enhanced by claudin-1. Therefore, a cascade of induction of MT1-MMP and active MMP-2 by claudin-1 is followed by the cleavage of laminin-5 γ2 chains. EGFR seems to be involved in this cascade because EGFR antibody as well as EGFR inhibitors largely suppressed the invasion activity.

It has been clarified that the expression of claudin-1 is negatively regulated by Snail (46). However, the expression pattern of Snail was similar to that of claudin-1. The present study seems to indicate that the expression of claudin-1 is regulated by some transcription factors other than Snail in SCC cells. Recently, Miwa et al. (47) reported that T-cell factor 4 (TCF4) and β-catenin complexes bound TCF4-binding elements at two sites in the 5′-flanking region of the claudin-1 gene and that the binding promoted transcription of claudin-1. That report suggested that claudin-1 is a new targeting gene of β-catenin. In addition, Hlubek et al. (48) recently reported that β-catenin promoted the transcription of MT1-MMP and laminin-5 γ2 chain. Taken together, the results of the present study, together with other reports, suggest that multiple pathways exist in the activation of MMPs and laminin-5 γ2 chains. A thorough exploration of the regulatory mechanism of the claudin-1 expression, enhancement of MT1-MMP expression, activation of pro-MMP-2, and cleavage of laminin-5 γ2 chains by claudin-1 will be necessary to understand how the invasion of cancer cells, including SCC cells, is controlled.

Grant support: Ministry of Education, Culture, Sports, Science, and Technology, Japan grant 16591886.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.