Homophilic Interactions of Tetraspanin CD151 Up-regulate Motility and Matrix Metalloproteinase-9 Expression of Human Melanoma Cells through Adhesion-dependent c-Jun Activation Signaling Pathways*

The tetraspanin membrane protein CD151 has been suggested to regulate cancer invasion and metastasis by initiating signaling events. The CD151-mediated signaling pathways involved in this regulation remain to be revealed. In this study, we found that stable transfection of CD151 into MelJuSo human melanoma cells lacking CD151 expression significantly increased cell motility, matrix metalloproteinase-9 (MMP-9) expression, and invasiveness. The enhancement of cell motility and MMP-9 expression by CD151 overexpression was abrogated by inhibitors and small interfering RNAs targeted to focal adhesion kinase (FAK), Src, p38 MAPK, and JNK, suggesting an essential role of these signaling components in CD151 signaling pathways. Also, CD151-induced MMP-9 expression was shown to be mediated by c-Jun binding to AP-1 sites in the MMP-9 gene promoter, indicating AP-1 activation by CD151 signaling pathways. Meanwhile, CD151 was found to be associated with α3β1 and α6β1 integrins in MelJuSo cells, and activation of associated integrins was a prerequisite for CD151-stimulated MMP-9 expression and activation of FAK, Src, p38 MAPK, JNK, and c-Jun. Furthermore, CD151 on one cell was shown to bind to neighboring cells expressing CD151, suggesting that CD151 is a homophilic interacting protein. The homophilic interactions of CD151 increased motility and MMP-9 expression of CD151-transfected MelJuSo cells, along with FAK-, Src-, p38 MAPK-, and JNK-mediated activation of c-Jun in an adhesion-dependent manner. Furthermore, C8161 melanoma cells with endogenous CD151 were also shown to respond to homophilic CD151 interactions for the induction of adhesion-dependent activation of FAK, Src, and c-Jun. These results suggest that homophilic interactions of CD151 stimulate integrin-dependent signaling to c-Jun through FAK-Src-MAPKs pathways in human melanoma cells, leading to enhanced cell motility and MMP-9 expression.

that CD151 association increases the binding activity of integrin ␣ 3 ␤ 1 to laminin through stabilizing its activated conformation (20). It was also reported that CD151 regulates platelet function by modulating outside-in signaling events of the major platelet integrin ␣ IIb ␤ 3 (21). In addition to integrin association, CD151 associates with phosphatidylinositol 4-kinase and protein kinase C on the cytosolic surface, thereby linking integrins to these signaling molecules (8,22). CD151 has also been shown to regulate expression of a protein-tyrosine phosphatase, PTP, and its recruitment to cell-cell junctions (19) and to inhibit adhesion-dependent activation of Ras (23). It thus appears that the intracellular signaling pathways initiated by integrin binding to the extracellular matrix could be altered by the integrin-associated tetraspanin CD151. Taken together, CD151 is thought to participate in adhesion-dependent transmembrane signaling pathways by modulating integrin activity and modifying integrin-mediated outside-in signaling pathways as well. However, the modified integrin signaling pathways by which CD151 manifests its activity have not been established.
In this report, we investigated the functional effects of CD151 expression on cellular activities related to cancer invasion and metastasis and then attempted to identify CD151-mediated signaling pathways for the induction of such cellular functions. We showed that CD151 increases motility and MMP 2 -9 expression of human melanoma cells through adhesiondependent c-Jun activation-signaling pathways. Furthermore, we established that these signaling pathways are initiated not only by the matrix binding of integrin molecules but also by homophilic interactions between CD151 proteins on the surface of neighboring cells. Finally, detailed analysis of signaling events indicated that the CD151-␣ 3 ␤ 1 /␣ 6 ␤ 1 integrin complexes increase c-Jun activity through the activation of FAK, Src, p38 MAPK, and JNK.
Transfection of CD151 cDNA and Selection of Stable Clones-Full-length CD151 cDNA was subcloned into the EcoRI/KpnI sites of a pcDNA3 vector (Invitrogen), downstream of a cytomegalovirus promoter. The CD151 cDNA expression construct was transfected into MelJuSo human melanoma cells by using Lipofectamine (Invitrogen) according to the manufacturer's instructions. pcDNA3 vector only was also transfected as a control. Neomycin-resistant clones were isolated by growing the cells in DMEM/F-12 containing 10% fetal bovine serum and 0.5 mg/ml G418 (Invitrogen). Stable transfectant clones were characterized by immunoblotting and flow cytometric analyses for their expression levels of CD151 protein.
Reverse Transcription-PCR Analysis-Total cellular RNA was purified from the cultured cells using Trizol reagent (Invitrogen) according to the manufacturer's protocol. First strand cDNA synthesis was performed with 1 g of total RNA using a cDNA synthesis kit (Promega, Madison, WI). For PCR amplification, 5Ј-aaggtaccaggatgggtgagttcaacgag-3Ј was used as the sense primer, and 5Ј-atgaattcggtcagtagtgctccagcttg-3Ј was used as the antisense primer. This primer pair amplifies a 760-bp fragment of CD151 cDNA. The reaction mixture was subjected to 25 PCR amplification cycles of 60 s at 94°C, 90 s at 55°C, and 90 s at 72°C. ␤-Actin amplification was used as an internal PCR control (27) with 5Ј-gatatcgccgcgctcgtcgtcgac-3Ј as the sense primer and 5Ј-caggaaggaaggctggaagagtgc-3Ј as the antisense primer. The PCR products were visualized using ethidium bromide in 1% agarose gel. 2  Immunoblotting Analysis-Cells were washed, harvested, and lysed in lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 20 g/ml leupeptin, and 2 mM benzimidine) on ice for 10 min. For phosphoprotein analysis, cell lysis buffer was supplemented with phosphatase inhibitors (1 mM sodium orthovanadate, 1 mM NaF, and 10 mM ␤-glycerophosphate). After centrifugation at 15,000 ϫ g for 10 min, the supernatants were collected and quantified for protein concentration by Bradford assay. Equal amounts of protein per lane were separated onto 10% SDS-polyacrylamide gel and transferred to an Immobilon-P (Millipore Corp., Bedford, MA) membrane. The membrane was blocked in 5% skim milk for 2 h and then incubated with a specific antibody for 2 h. After washing, the membrane was incubated with a secondary antibody conjugated with horseradish peroxidase. After final washes, the membrane was developed using enhanced chemiluminescence reagents (Amersham Biosciences).
Flow Cytometric Analysis-Cells were incubated with 10 g/ml anti-CD151 monoclonal antibody (mAb) for 30 min, washed with cold PBS, and then incubated with saturating concentrations of fluorescein isothiocyanate-conjugated goat antimouse IgG (PharMingen) for 30 min at 4°C. After washing with PBS, the cells were fixed with 2% formaldehyde in PBS. Cell surface immunofluorescence was analyzed by flow cytometry performed on a FACScan (BD Biosciences).
Immunoprecipitation-Cells were lysed in immunoprecipitation buffer (25 mM Hepes, pH 7.5, 150 mM NaCl, 5 mM MgCl 2 ) supplemented with 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 20 g/ml leupeptin, and 1% Brij 98 or 1% Triton X-100 for 2 h at 4°C. The lysate was centrifuged (16,000 ϫ g, 15 min), and the supernatant was precleared with a combination of protein A-and protein G-agarose (Amersham Biosciences) precoated with normal mouse IgG for 2 h at 4°C. After preclearing, the lysate was incubated with a specific antibody coupled to the protein A/G-agarose beads for 2 h at 4°C. Immune complexes collected on the beads were then washed four times with immunoprecipitation buffer and resolved by SDS-PAGE. Proteins were detected by immunoblotting analysis using specific antibodies.
Invasion Assay into Matrigel-24-well Transwell chamber inserts (Corning Costar, Cambridge, MA) with 8-m porosity polycarbonate filters were precoated with 80 g of basement membrane Matrigel (BD Biosciences) onto the upper surface and with 20 g of gelatin onto the lower surface. Culture supernatant of NIH3T3 fibroblasts in DMEM supplemented with 10% fetal bovine serum was placed in the lower well. MelJuSo cells suspended in DMEM/F-12 medium containing 0.1% fetal bovine serum were added to the upper chambers (2 ϫ 10 4 cells/well) and incubated for 24 h at 37°C in 5% CO 2 . Cells were fixed and stained with hematoxylin and eosin. Noninvading cells on the upper surface of the filter were removed by wiping out with a cotton swab, and the filter was excised and mounted on a microscope slide. Invasiveness was quantified by counting cells on the lower surface of the filter.
Wound-healing Migration Assay-For the measurement of cell migration during wound healing, cells (5 ϫ 10 5 ) were seeded in individual wells of a 24-well culture plate. When the cells reached a confluent state, cell layers were wounded with a plastic micropipette tip having a large orifice. The medium and debris were aspirated away and replaced by 2 ml of fresh serum-free medium. Cells were photographed every 12 h after wounding by phase-contrast microscopy. For evaluation of "wound closure," five randomly selected points along each wound were marked, and the horizontal distance of migrating cells from the initial wound was measured.
Gelatin Zymography-Type IV collagenase activities present in conditioned medium were visualized by electrophoresis on gelatin-containing polyacrylamide gel as previously described (28). Briefly, conditioned medium from cells cultured in serumfree medium was mixed 3:1 with substrate gel sample buffer (40% (v/v) glycerol, 0.25 M Tris-HCl, pH 6.8, and 0.1% bromphenol blue) and loaded without boiling onto 10% SDS-polyacrylamide gel containing type 1 gelatin (1.5 mg/ml). After electrophoresis at 4°C, the gel was soaked in 2.5% Triton X-100 with gentle shaking for 30 min with one change of detergent solution. The gel was rinsed and incubated for 24 h at 37°C in substrate buffer (50 mM Tris-HCl, pH 7.5, 5 mM CaCl 2 , and 0.02% NaN 3 ). Following incubation, the gel was stained with 0.05% Coomassie Brilliant Blue G-250 and destained in 10% acetic acid and 20% methanol.
Cell Aggregation Assay-L cells transfected with CD151 or vector alone were washed with PBS containing 2 mM EDTA and rendered into single cell suspension by seven gentle passes through a 22-gauge needle after scraping. After washing with Puck's saline (5 mM KCl, 140 mM NaCl, 8 mM NaHCO 3 , pH 7.4), suspensions of single cells (1 ϫ 10 5 cells/ ml) were seeded into individual wells of a 24-well culture plate and incubated in 5% CO 2 at 37°C with agitation at 70 -80 rpm using an orbital shaker. Photographs were taken every 15 min after incubation under a phase-contrast microscope on three predetermined fields, and both the total cell number (A) and the number of cells remaining as single cells (B) were counted. The results were expressed as the percentage of cells that formed aggregates as follows: (A Ϫ B)/A ϫ 100 (%). In some experiments, the transfectants were preincubated with antibody (20 g/ml) and then washed free of unbound antibody before incubation. In experiments to determine whether aggregation was homophilic, distinct populations of cells were prelabeled with 5-and 6-CFSE (carboxyfluorescein diacetate succinimidyl ester) (Molecular Probes, Inc., Eugene, OR) before suspension. For these experiments, phase and fluorescent images of the same field were photographed after a 30-min incubation with orbital shaking.
Promoter Assay-A 1305-bp DNA fragment (Ϫ1285 to ϩ20), corresponding to the promoter of the human MMP-9 gene (29,30), was generously provided by Dr. Seung-Taek Lee (Yonsei University, Korea) (31). For mt-AP-1 of the MMP-9 gene promoter, in which distal and proximal AP-1 binding sites (Ϫ533 to Ϫ527 and Ϫ79 to Ϫ73, respectively) were destroyed, 5Ј-TGAGTCA-3Ј was changed to 5Ј-TGAGTtg-3Ј (underlined lowercase letters indicate the mutated bases) by the QuikChange II site-directed mutagenesis kit (Stratagene, La Jolla, CA). For mt-NF-B of the MMP-9 promoter, in which a NF-B binding site (Ϫ600 to Ϫ590) was destroyed, 5Ј-GGAATTCCCC-3Ј was mutated to 5Ј-GatcgatCCC-3Ј. After subcloning the mutant MMP-9 promoters into a promoterless luciferase expression vector, pGL3 (Promega), the corresponding mutations in the constructs were verified by DNA sequencing. The pGL3 vector containing wild-type or mutant MMP-9 promoter was transfected into MelJuSo cells by using Lipofectamine. Luciferase activity in cell lysate was measured using the Promega luciferase assay system according to the instructions of the manufacturer. To normalize luciferase activity, each of the pGL3 vectors was co-transfected with a pRL-SV40⌬Enh, which expresses Renilla luciferase by an enhancerless SV40 promoter (31).
Electrophoretic Mobility Shift Assay-Cells were incubated with serum-free medium for 4 h, and nuclear extracts were prepared as previously described (32). Double-stranded oligonucleotide probes corresponding to the putative AP-1 binding site (Ϫ86 to Ϫ66; 5Ј-TGACCCCTGAGTCAGCACTTG-3Ј; the AP-1 recognition sequence is underlined) and the putative NF-B site (Ϫ607 to Ϫ582; 5Ј-GCCCCGTGGAATTCCCCC-AAATCCTG-3Ј; the NF-B recognition sequence is underlined) in the proximal MMP-9 promoter sequences were labeled with [␥-32 P]ATP using T4 polynucleotide kinase and purified by a G-50 Sephadex column. The 32 P-labeled probes (ϳ40,000 cpm) were then incubated with nuclear extracts (10 g of protein) for 20 min at room temperature. Samples were resolved on native 5% polyacrylamide gel, and the gel was dried and subjected to autoradiography. Specificity for binding of AP-1 factors and NF-B to the corresponding sequences of the MMP-9 promoter was confirmed by using cold competitors having typical AP-1 and NF-B binding sequences (Promega), respectively.
Detergent-free Purification of Membrane Fractions-Mock and CD151 transfectant cells were washed with ice-cold PBS and then scraped into buffer A (20 mM Tris-HCl, pH 7.5, 2 mM EDTA, 1 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 20 g/ml leupeptin, and 2 mM benzimidine). The cells were homogenized using a Dounce homogenizer (20 strokes). A postnuclear supernatant was obtained by centrifugation (2500 ϫ g, 10 min, 4°C), adjusted to 10% sucrose and loaded onto a 30% sucrose cushion in an ultracentrifuge tube. After centrifugation for 60 min at 150,000 ϫ g in a T-1270 rotor of a tabletop ultracentrifuge (Beckman Instruments), a lightscattering band confined to the 10 -30% sucrose interface was collected and stored at Ϫ70°C until use.
In Vitro Kinase Assays-Cellular proteins (200 g) were incubated with anti-FAK, anti-Src, or anti-paxillin Abs and immunoprecipitated using protein A/G-agarose beads. Immune complexes collected on the beads were washed three times with immunoprecipitation buffer and once with kinase buffer (20 mM PIPES, pH 7.2, 10 mM MgCl 2 , 1 mM dithiothreitol) and added to an in vitro kinase reaction mixture containing 5 g of acid-denatured enolase and 10 Ci of [␥-32 P]ATP (33). The reaction was incubated at 30°C for 30 min and then stopped by boiling with SDS-sample buffer. After electrophoresis on a 10% polyacrylamide gel, the radioactive proteins were visualized by autoradiography.

CD151 Increases Motility, MMP-9 Expression, and Invasiveness of Human Melanoma
Cells-First, we examined CD151 expression levels in two human metastatic melanoma cell lines, C8161 and MelJuSo, among which C8161 was illustrated to have a higher metastatic ability than MelJuSo (24,25). A 760-bp PCR product and a 29-kDa protein band were detected in C8161, but not in MelJuSo cells, by reverse transcription-PCR analysis using CD151 cDNA-specific primers and immunoblotting analysis using anti-CD151 mAb, respectively (Fig. 1, A and B), indicating that CD151 is differentially expressed in a cell line-specific manner among human melanoma cells. The tet- and MelJuSo cell lines were analyzed by reverse transcription-PCR using CD151 cDNA-specific primers and immunoblotting using mAbs specific to each protein, respectively. ␤-Actin mRNA and actin protein from each cell line were also analyzed to control for equal amounts of mRNAs and proteins, respectively. C, the stable clones of CD151 cDNA-transfected MelJuSo cells were examined for CD151 expression by immunoblotting analysis using anti-CD151 mAb. D, cell surface expression levels of CD151 protein in CD151 transfectant clones were analyzed by flow cytometry using anti-CD151 mAb. raspanins CD9 and CD63, which were previously reported to suppress melanoma metastasis (28, 34 -36), exhibited similar protein amounts between these two human melanoma cell lines (Fig. 1B). Since a positive role of CD151 in cancer metastasis has been shown in some other cancer cell types (14,15), we hypothesized that the higher metastatic ability of C8161 cells, as compared with MelJuSo cells, may be attributed to the upregulation of CD151 rather than the down-regulation of CD9 and CD63. To determine whether CD151 expression indeed elevates the metastatic potential of melanoma cells, MelJuSo cells devoid of endogenous CD151 expression were transfected with a CD151 cDNA expression vector. Among several trans-fectant clones displaying high CD151 protein expression, two clones, CD151/M-76 and CD151/ M-77, were selected for further functional analyses (Fig. 1, C and D). To investigate the functional effect of CD151 expression on cellular activities related to the cancer metastasis process, the in vitro invasive efficacy of each transfectant clone was determined by Boyden chamber assay using Matrigel. CD151 transfectant clones, CD151/M-76 and CD151/ M-77, showed a 3-fold higher invasiveness than the mock transfectant ( Fig. 2A), suggesting that CD151 expression increases the invasive ability of melanoma cells. We further examined the migrating ability of CD151 transfectant clones into wounded spaces on culture plates. Both CD151 transfectant clones exhibited a 3-fold higher migrating ability than the mock transfectant (Fig. 2B). The basement membrane-degrading ability of the transfectant clones was also examined by measuring gelatinase activity in the culture supernatant. Among the two types of gelatinases, MMP-2 and MMP-9, the enzyme activity and protein level of MMP-9 were much higher in CD151 transfectant clones than in the mock transfectant, whereas the activity and expression of MMP-2 were not affected by CD151 expression (Fig. 2, C and D). To examine whether increased MMP-9 activity by CD151 affects cell motility, siRNA targeted to pro-MMP-9 was transfected into CD151 transfectant cells. As a result, knockdown of MMP-9 suppressed the stimulating effect of CD151 on the motility of MelJuSo cells, indicating that cell motility induced by CD151 involves MMP-9 activity (Fig. 2, E and F ). Since cell migration and gelatinase secretion are essential for the invasion process, it seems likely that the CD151-induced invasiveness is in part due to the positive effect of CD151 expression on motility and MMP-9 expression of melanoma cells.
Functional Involvement of FAK, Src, p38 MAPK, and JNK in CD151-stimulated Motility and MMP-9 Expression-To identify the signaling molecules involved in CD151-induced cell motility and MMP-9 expression, CD151 transfectants were analyzed in the presence of several inhibitors of signal transduction mediators. Among the inhibitors tested, the inhibitors FIGURE 2. CD151 expression enhances invasiveness, motility, and MMP-9 production of MelJuSo melanoma cells. A, each transfectant clone (2 ϫ 10 4 cells) was seeded into a Transwell chamber insert equipped with a Matrigel-coated filter. After 12 h of incubation, cells on the lower surface of the filter were stained with Gill's hematoxylin and counted. Results are means Ϯ S.E. of triplicate cultures. B, confluent cell cultures were wounded with plastic micropipette tips. Cells were photographed at 48 h after wounding by phase-contrast microscopy, and the measurement of cell migration during wound healing was performed as described under "Experimental Procedures." Results are means Ϯ S.D. of triplicate cultures. The asterisks indicate that the differences are statistically significant (* and **, p Ͻ 0.01 versus mock transfectant, Student's t test). C, conditioned media obtained from cells cultured in serum-free medium for 3 days were electrophoresed on a 10% SDS-polyacrylamide gel containing type I gelatin. After the removal of SDS, the gels were incubated in gelatinase substrate buffer and then visualized by Coomassie staining. D, MMP-2 and MMP-9 protein levels in the conditioned media of the cell cultures were assessed by immunoblotting analyses using anti-MMP-2 and anti-MMP-9 mAbs, respectively. E, a CD151 transfectant clone of MelJuSo cells, CD151/M-77, was transfected with siRNA targeted to pro-MMP-9, and pro-MMP protein levels in cell lysate were analyzed by immunoblotting using MMP-9 mAb at 48 h after transfection. F, following siRNA transfection, cell migration was measured at 48 h after wounding in a similar fashion as in B. A dagger indicates that the differences are statistically significant ( †, p Ͻ 0.03 versus control siRNA-transfected cells, Student's t test).
including PP1 (a Src kinase family inhibitor), SB203580 (a p38 MAPK inhibitor), and SP600125 (a JNK inhibitor) suppressed the motility and MMP-9 expression of CD151 transfectant clones close to the levels of the mock transfectant (data not shown). To verify the participation of FAK, Src, p38 MAPK, and JNK in CD151 signaling pathway(s) for the induction of cell motility and MMP-9 expression, siRNAs targeted to FAK, Src, p38 MAPK, and JNK were transfected into CD151 transfectant cells. Protein levels of FAK, Src, p38 MAPK, and JNK were effectively knocked down by each specific siRNA (Fig. 3, A, D, G, and J). All four siRNA types employed inhibited the migrating ability of CD151 transfectant cells in a dose-dependent manner (Fig. 3, B, E, H, and K). Also, all of the FAK, Src, p38 MAPK, and JNK siRNA-transfected cells exhibited significantly decreased activities and expression levels of MMP-9 compared with control siRNA-transfected cells retaining endogenous levels of these signaling molecules (Fig. 3, C, F, I, and L). It thus appears that FAK, Src, p38 MAPK, and JNK are functionally involved in CD151 signaling pathway(s) leading to increased motility and MMP-9 expression of MelJuSo melanoma cells.
Induction of MMP-9 Expression by CD151 Is Mediated by Activation of AP-1 Factors-Since MMP-9 appeared to be a target gene up-regulated by CD151 signaling pathway(s) in MelJuSo melanoma cells, we investigated the transcriptional regulation mode of the MMP-9 gene by using several mutants of its 5Ј-proximal promoter region. When a reporter vector containing a wild-type promoter of the MMP-9 gene was transiently transfected into MelJuSo cells, the CD151 transfectant cells showed about a 20-fold higher luciferase activity than the mock transfectant cells (Fig. 4A). In contrast to the wild-type promoter, the promoters having mutations at the AP-1 binding sites (mt-5Ј-AP-1 and mt-3Ј-AP-1) did not respond to CD151 for their activities for reporter gene expression. However, mutation of the NF-B binding site did not abolish the stimulating effect of CD151 on MMP-9 promoter activity. To determine whether CD151 expression increases DNA binding activity of AP-1 transcriptional factors, we compared the binding of nuclear proteins to a putative AP-1 binding site (Ϫ79 to Ϫ73) of the MMP-9 promoter between mock and CD151 transfectant cells. As shown in Fig. 4B, DNA binding activity of AP-1 factors in CD151 transfectant cells was more significant than that in mock transfectant cells. Moreover, incubation with anti-c-Jun antibody resulted in a partial supershift of the AP-1/DNA complex with the gel shift assay, indicating that c-Jun participates in the formation of the AP-1/DNA complex. Thus, these data indicate that CD151 increases MMP-9 gene transcription by activating AP-1 transcription factors, including c-Jun. CD151 Signaling Pathway(s) Depends on Activation of the Associated ␣ 3 ␤ 1 and ␣ 6 ␤ 1 Integrins-CD151 has been reported to associate with various types of integrins and, in particular, forms stable complexes with ␣ 3 ␤ 1 and ␣ 6 ␤ 1 integrins in many types of cells (4,6,8,17,18). To determine whether ␣ 3 ␤ 1 and ␣ 6 ␤ 1 integrins are also associated with CD151 in MelJuSo melanoma cells, the CD151transfected MelJuSo cells were lysed with the nonionic detergent Brij 98, a mild lysis condition preserving tetraspanin-integrin interactions, and the cell lysates were immunoprecipitated with anti-CD151 antibody. As a result, ␣ 3 , ␣ 6 , and ␤ 1 integrin subunits were detected in the CD151 immunoprecipitate of CD151 transfectant cells but not in mock transfectant cells (Fig. 5A), although the protein level of each integrin subunit was not different between CD151 and mock transfectant cells (Fig. 5B). This result indicates that CD151 can form complexes with ␣ 3 ␤ 1 and ␣ 6 ␤ 1 integrins in MelJuSo cells. Since ␣ 3 ␤ 1 and ␣ 6 ␤ 1 integrins are known to be receptors of laminin/fibronectin and laminin, respectively, we examined whether CD151-stimulated MMP-9 expression is dependent on cell adhesion to extracellular matrix components, such as laminin and fibronectin. As illustrated in Fig. 5C, the stimulating effect of CD151 on MMP-9 expression became more prominent when the cells were attached to laminin, fibronectin, and laminin-rich Matrigel. However, CD151 did not exert its inducing activity for MMP-9 expression when the cells were plated on poly-(L)-lysine, which does not activate integrins. In CD151-deficient mock transfectant cells, a slight increase in MMP-9 expression was observed by cell adhesion to laminin and fibronectin, suggesting that integrin activation alone is not sufficient to induce MMP-9 expression. Thus, CD151 appears to cooperate with associated integrins to induce MMP-9 expression in melanoma cells.
To examine how associated integrins activated by their binding to extracellular matrix modulate CD151-dependent signaling pathway(s), the activation status of the CD151 signaling mediators, which were demonstrated in Fig. 3, were compared between mock and CD151 transfectant cells a short time after plating the cells on laminin. CD151 expression significantly elevated phosphorylation-dependent activation of signaling components, such as FAK, Src, p38 MAPK, and JNK dependently of cell adhesion to laminin (Fig. 5D). CD151-mediated phosphorylation of MAPKAPK-2 and c-Jun, downstream effectors of p38 MAPK and JNK, respectively, was also found to be dependent on cell adhesion to laminin. However, adhesion events without integrin activation, such as cell binding to poly-(L)-lysine, did not increase the phosphorylation levels of these CD151 signaling components. On the other hand, phosphorylation of ERK1/2 appeared to be affected by neither integrin activation nor CD151 expression, implying that ERK1/2 does not participate in integrin-dependent CD151 signaling pathways in MelJuSo cells. Taken together, these data strongly suggest that CD151 cooperates with associated integrins to provoke outside-in signaling pathways leading to the activation of FAK, Src, p38 MAPK, JNK, MAPKAPK-2, and c-Jun.
CD151 Is a Homophilic Interacting Protein-Since some membrane proteins involved in cell adhesion and migration, such as E-cadherin and CD99, were found to be self-ligand molecules and their homophilic interactions regulate intracellular signaling pathways (37,38), we tested the possibility that CD151 is a homophilic interacting membrane protein. After transfecting a CD151 expression vector into murine L-cell fibroblast cells, which do not exhibit homotypic cell-to-cell adhesion, we compared the ability of stable CD151 transfectant L cells to adhere to each other or to empty vectortransfected control L cells by spontaneous cell aggregation assay using cells in suspension. We found that control L cells did not aggregate, but CD151-transfected L cells aggregated in a time-dependent manner (Fig. 6B). The aggregation of CD151-transfected cells was reduced to half after incubation with anti-CD151 antibody, suggesting that the L cell aggregation is mediated by CD151. To confirm whether this aggregation was homophilic, we mixed CD151-transfected L cells with an equal number of fluorescently labeled control L cells. As a result, no fluorescent cells were present in the aggregates, indicating that CD151-expressing L cells did not bind to control L cells lacking CD151 (Fig. 6C). However, when fluorescently labeled CD151 transfectant cells were mixed with unlabeled control L cells, every cell in the aggregates was labeled (Fig. 6D). These data illustrate that CD151-expressing L cells bind only to the same type of L cells having CD151 but not to L cells lacking CD151. It thus appears that CD151 is a homophilic interacting cell surface protein.
Homophilic CD151 Interactions Enhance Cell Motility and MMP-9 Expression-To assess the effect of homophilic CD151 interactions on the motility and MMP-9 expression of MelJuSo cells, we prepared membrane fractions of MelJuSo cells transfected with either a CD151 expression vector or empty vector. As expected, CD151 was present in the membrane fraction of the CD151 transfectant but not in that of the mock transfectant (Fig. 7A). The CD151 transfectant cells treated with membrane fraction containing CD151 exhibited increased cell motility compared with untreated cells (Fig. 7B). The CD151-containing membrane fraction also increased MMP-9 expression in   (Fig. 7C). However, pretreatment of CD151 transfectant cells with anti-CD151 Ab blocked the inducing effect of the CD151 membrane fraction on MMP-9 expression (Fig. 7D). Meanwhile, the CD151-deficient membrane fraction obtained from mock transfectant cells did not affect the motility and MMP-9 expression of CD151 transfectant cells. In addition, mock-transfectant cells lacking CD151 did not respond to the CD151-containing membrane fraction for the induction of cell motility and MMP-9 expression. Thus, these results indicate that homophilic interactions of CD151 increase cell motility and MMP-9 expression in MelJuSo cells.

Homophilic CD151 Interactions Stimulate the c-Jun Activation Signaling Pathways in an Integrin-dependent Manner-
We examined whether homophilic interactions of CD151 trigger the outside-in signaling pathway(s) leading to c-Jun activation. Treatment of CD151 transfectant cells, but not mock transfectant cells, with the CD151-containing membrane fraction increased the phosphorylation levels of CD151 signaling mediators, such as FAK, Src, and c-Jun, in a time-dependent manner (Fig. 8A). However, phosphorylation levels of these signaling molecules were not increased in CD151 transfectant cells incubated with the CD151-deficient membrane fraction. These data suggest that homophilic CD151 interactions between neighboring cells activate the signaling pathways for c-Jun activation in one another.
We next investigated the possible influence of integrin activation on signaling pathways provoked by homophilic CD151 interactions. When CD151 transfectant cells were seeded onto plates coated with poly-(L)-lysine, homophilic CD151 interactions resulted in a slight increases in the phosphorylation levels of Src and c-Jun, along with no increase in FAK phosphorylation (Fig. 8B). However, cell adhesion to laminin not only increased the phosphorylation level of FAK but also significantly augmented the positive effect of homophilic CD151 interactions on the phosphorylation of Src and c-Jun. Kinase activities of FAK and Src associated with paxillin in focal adhesion complexes were also found to be increased by homophilic CD151 interactions dependent on cell adhesion to laminin (Fig. 8C). These data indicate a stimulating role of integrins in CD151-mediated signaling pathways. Meanwhile, as illustrated in mock transfectant cells attached to laminin, simple activation of laminin-binding integrins without any homophilic CD151 interaction was not sufficient to induce phosphorylation of these signaling molecules. To assess the involvement of CD151-associated integrins, ␣ 3 ␤ 1 and ␣ 6 ␤ 1 , in up-regulating CD151 signaling to c-Jun, we incubated CD151 transfectant cells with anti-␤ 1 integrin antibody before seeding the cells on laminin-coated plates. The anti-␤ 1integrin antibody effectively suppressed the stimulating effect of homophilic CD151 interactions on c-Jun phosphorylation (Fig. 8D), indicating direct participation of ␤ 1 -type integrins in modulating CD151 signaling for c-Jun activation. The dependence of CD151 signaling on ␤ 1 -type integrins was also observed in MMP-9 expression (Fig. 8D). Taken together, these results strongly suggest that activation of the CD151-associated ␣ 3 ␤ 1 and ␣ 6 ␤ 1 integrins amplifies the c-Jun activation signaling pathways initiated by homophilic CD151 interactions. We next investigated the participation of MAPKs in CD151 signaling to c-Jun by using inhibitors specific for ERK, p38 MAPK, and JNK. c-Jun phosphorylation in CD151 transfectant cells was significantly blocked by the p38 MAPK inhibitor, SB203580, and the JNK inhibitor, SP600125, as well as by the Src kinase inhibitor, PP1, but not by the ERK inhibitor, PD98059 (Fig. 8E). Since previous results showed CD151-induced adhesion-dependent activation of Src, p38 MAPK, and JNK, but not ERK (Fig. 5D), it is very likely that Src-mediated activation of p38 MAPK and JNK may play an important role in transducing CD151 signals to c-Jun. We finally examined whether integrin-dependent CD151 signaling events also occur in another melanoma cell line, C8161, which possesses endogenous CD151 (Fig. 1A). Similar to CD151-transfected MelJuSo cells, C8161 cells responded to the CD151-containing membrane fraction for the phosphorylation of FAK, Src, and c-Jun in an adhesion-dependent manner (Fig. 8F ). However, the phosphorylation levels of these signaling molecules in C8161 cells were not increased in the absence of homophilic CD151 interaction and integrin activation. These results indicate that homophilic CD151 interactions between two contacting human melanoma cells with endogenous CD151 activate the intracellular signaling pathways in one another with the cooperation of associated integrins.

DISCUSSION
Tetraspanin CD151 has been implicated in the regulation of cell motility, cancer invasion, and metastasis (14,15,39). Anti-CD151 antibody has been shown to inhibit wound-healing migration of endothelial cells, chemotactic motility of neutrophils, and phagokinetic motility of cancer cells (14,39). CD151 overexpression enhanced invasive and metastatic abilities of several cancer cell lines, whereas treatment of cells with anti-CD151 antibody suppressed these abilities (14,15). In this report, we also demonstrated that transfection of CD151 cDNA into a CD151-deficient melanoma cell line up-regulates MMP-9 expression, resulting in the promotion of cancer cell motility and invasiveness (Fig. 2). Among the tetraspanin proteins, CD9 and CD63 have also been associated with the invasion metastasis of melanoma, but these associations have opposing effects to that of CD151 (28, 34 -36, 40 -42). Transfection with CD9 resulted in suppression of cell motility and metastasis of murine melanoma cells (34,36). CD63 expression has been shown to be inversely correlated with the malignant progression of human melanoma (40,41), and several transfection studies have demonstrated the suppressing role of CD63 in the invasion and metastasis of melanoma cells (28,35,42). Thus, tetraspanins CD9, CD63, and CD151 appear to play a role in the processes of melanoma invasion and metastasis, in which CD151 acts as a positive effector opposite to the role of CD9 and CD63.
CD151 (46). Furthermore, outside-in signaling through ␣ 6 ␤ 1 integrin was markedly influenced by its lateral association with CD151 (45). The short C-terminal cytoplasmic region of CD151 was found to be particularly important for determining the outside-in signaling functions of ␣ 6 ␤ 1 integrin (45). Thus, most studies of CD151 have focused on its role in modulating the activity and function of associated integrin molecules. Therefore, participation of CD151 in signal transduction has been confined to its regulatory activity toward integrin-mediated transmembrane signaling events.
In contrast to previously identified roles for CD151 as a modulator for integrin-mediated signaling, we here demonstrated that CD151 can transduce its own signals leading to increases in MMP-9 expression, cell motility, and invasiveness. Cross-linking of tetraspanins CD81 and CD82 at the cell surface with antibodies was reported to transduce activation signals, such as tyrosine phosphorylation, calcium fluxes, and inositol turnover (47)(48)(49)(50). Therefore, we postulated that the CD151-involved sig-naling pathways may be initiated by ligand binding to CD151 as well as by activation of associated integrins. As a result of searching for a ligand for CD151, we found that CD151 is a self-ligand molecule, implying that homophilic interactions of CD151 proteins take place between two neighboring cells (Fig.  6). We also found that the positive effect of CD151 expression on cell motility and MMP-9 expression is further elevated when CD151-expressing cells were treated with a CD151-containing membrane fraction but not with a CD151-deficient membrane fraction (Fig. 7). Furthermore, treatment with a CD151-containing membrane fraction activated signaling molecules, such as FAK, Src, and c-Jun, in CD151-expressing cells (Fig. 8A), suggesting that homophilic CD151 interactions between two contacting cells provoke intracellular signaling events in both cells. However, the CD151 signaling appears to be dependent on the activation of laminin-binding integrins. Adhesion-dependent activation of several signaling molecules, including FAK, Src, p38 MAPK, JNK, MAPKAPK-2, and c-Jun, became FIGURE 6. Homophilic interactions drive CD151-dependent cell adhesion. A, L cells were transfected with CD151 or vector alone (control), and the stable transfectant clones were examined for CD151 expression by immunoblotting analysis using anti-CD151 antibody. B, two independent CD151 transfectant clones and control transfectants of L cells were suspended in Puck's saline and treated with anti-CD151 mAb or normal mouse IgG, and the aggregation assay was carried out as described under "Experimental Procedures." Data are means Ϯ S.D. of three independent experiments. C, the CD151 transfectants suspended in Puck's saline were mixed with an equal number of CFSE-labeled control transfectant cells. After 30 min of orbital shaking, phase-contrast and fluorescent images of the same field were photographed. Magnified images of the aggregated cells are shown in insets in phase-contrast images. D, the CD151 transfectants labeled with CFSE were mixed with unlabeled control cells, and the images were taken as described in the legend for C. evident when CD151-expressing cells were cultured on laminin-coated plates, whereas little activation of these signaling molecules was observed in the cells plated on poly-(L)-lysine, which does not activate any integrins (Fig. 5D). In particular, integrin activation by cell binding to laminin clearly augmented the stimulating effect of homophilic CD151 interactions on the activation of FAK, Src, and c-Jun, although homophilic CD151 interactions without integrin activation could increase phosphorylation levels of Src and c-Jun to some extent (Fig. 8, B and C). The requirement of both homophilic CD151 interactions and integrin activation for full activation of FAK, Src, and c-Jun was observed not only in CD151-transfected MelJuSo melanoma cells but also in C8161 melanoma cells possessing endogenous CD151 (Fig.  8F). Pretreatment of CD151-expressing cells with anti-␤ 1 antibody abolished the positive effect of homophilic CD151 interactions on c-Jun activation even when the cells were bound to laminin, demonstrating a critical role of CD151associated ␣ 3 ␤ 1 and ␣ 6 ␤ 1 integrins in CD151 signaling (Fig.  8D). The dependence of CD151 signaling on integrin activation was also observed in MMP-9 gene expression (Figs. 5C and 8D). Taken together, it appears that CD151 and the associated integrins stimulate signaling-triggering activities reciprocally, suggesting cross-talk between CD151 and inte-grin signaling events. Furthermore, CD151-␣ 3 ␤ 1 /␣ 6 ␤ 1 integrin complex-mediated signaling is not only initiated by integrin-activating cell-to-laminin adhesion but also provoked by homophilic CD151 interactions generating0 cellto-cell adhesion.
Many genes that participate in tumor cell invasion and migration have been identified, including adhesion molecules, small GTPases, cytoskeletal components, and matrix metalloproteinases (51). However, there is little consensus on what controls the expression of these genes and how a program of gene expression is coordinated to manifest an invasive phenotype. In the present study, we demonstrated that CD151 functions as a positive regulator in the adhesion-dependent activation of c-Jun, a component of the AP-1 transcription complex. An increase in the phosphorylation level of c-Jun by cell adhesion to laminin was observed in CD151 transfectant cells but not in mock transfectant cells (Fig. 5D). Homophilic interactions of CD151 further increased c-Jun phosphorylation in an adhesion-dependent manner (Fig. 8, A, B, and F), demonstrating the marked effect of CD151 signaling on the activation of c-Jun upon integrin activation. Increased expression of AP-1 component proteins and AP-1 activity has been shown to enhance invasion and motility in various model systems (52). Overexpression of c-Jun induces the invasiveness of chick embryo fibroblasts and MCF-7 breast cancer cells (53,54). In contrast, expression of a c-Jun mutant in which Ser-63 and Ser-73 are mutated to alanine residues, so that the protein cannot be phosphorylated by JNK, inhibits the migration of fibroblasts (55). A dominant negative mutant of c-fos, one of the Jun subfamily partners in AP-1 dimers, inhibits the motility of fibrosarcoma cells, along with growth arrest at the G 1 phase of the cell cycle (56). Another Fos subfamily member, Fra-1 also modulates the invasiveness and motility of MCF-7 cells (57). We here showed that the stimulating effect of CD151 on cell motility was abolished when JNK was knocked down by siRNA (Fig. 3K). The functional involvement of AP-1 activity in invasion is more evident in the regulation of MMP-9 gene expression. Two AP-1 binding sites exist in the 5Ј-proximal promoter region of the MMP-9 gene (29), and both sites were found to be essential for CD151-induced MMP-9 gene transcription (Fig. 4A). Results from a gel mobility shift assay indicated that CD151 overexpression in MelJuSo cells increased the binding of nuclear proteins, such as c-Jun, to oligonucleotides containing AP-1 consensus sequences (Fig.   FIGURE 7. Homophilic CD151 interactions enhance motility and MMP-9 expression of MelJuSo cells. A, membrane fractions of mock and CD151 transfectant cells were prepared and subjected to immunoblotting analysis with anti-CD151 mAb. Total membrane protein levels were also examined by staining a parallel blot with Coomassie Blue. B, the cell motility of mock and CD151 transfectants treated with the membrane fractions of each transfectant (50 g of protein) was determined by wound migration assay. Results are means Ϯ S.D. of triplicate cultures. An asterisk indicates that the differences are statistically significant (*, p Ͻ 0.01, Student's t test). C, gelatin zymography and immunoblotting analysis using anti-MMP-9 mAb were carried out for the conditioned media obtained from the cells cultured in the absence or the presence of the membrane fractions of mock or CD151 transfectant cells for 3 days. D, mock and CD151 transfectant cells were pretreated with normal mouse IgG or anti-CD151 mAb (0.1 mg/ml) for 3 h and incubated with the membrane fractions of CD151 transfectant cells for 3 days. The conditioned media were subjected to immunoblotting analysis using anti-MMP-9 mAb. 4B). Thus, the CD151-intergrin complex-mediated signaling pathway(s) appears to activate AP-1 transcription factors, including c-Jun, which, in turn, elicits changes in the expression of genes involved in the invasion process of melanoma cells.
AP-1 transcription factors are subject to regulation by MAPK signaling pathways with respect to biochemical activity, gene expression, and protein stability (58 -60). Among three major MAPK pathways, the ERK1/2 pathway is activated by a variety of mitogens, including growth factors and tumor promoters, whereas the JNK and p38 MAPK pathways are mainly stimulated by environmental stress and inflammatory cytokines (61). Several lines of evidence indicate that functional interplay through pathway cross-talk exists between these MAPK signaling pathways (62)(63)(64)(65). We found here that the signaling events initiated by CD151-integrin complexes in MelJuSo human melanoma cells result in the activation of p38 MAPK and JNK, but not ERK1/2 activation, along with c-Jun phosphorylation (Fig. 5D). Inhibitors for p38 MAPK and JNK, SB203580 and SP600125, respectively, blocked the positive effect of homophilic CD151 interactions on c-Jun phosphorylation, whereas PD98059, an ERK1/2 inhibitor, did not (Fig. 8E). In addition, CD151-induced cell motility and MMP-9 expression were abrogated by transfection of siRNAs that knock down p38 MAPK and JNK (Fig. 3, G-L). These results indicate that both p38 MAPK and JNK play a role of upstream effectors for c-Jun activation in CD151-integrin complex-mediated signaling cascades in MelJuSo cells. Among several integrin-mediated signaling pathways, the FAK-Src-MAPKs pathway has been suggested to regulate cell movement by influencing adhesion turnover at the leading edge in migrating cells (66 -68). The results in this study indicate that CD151-␣ 3 ␤ 1 /␣ 6 ␤ 1 integrin complexes utilize the FAK-Src-MAPKs pathway to increase melanoma cell motility. Since CD151 increases the extracellular matrix binding activity of associated integrins (20,46) as well as their signaling activity, it is very likely that CD151 contributes to cell movement by strengthening integrin-mediated cell adhesion to the substratum and by activating the FAK-Src-MAPKs pathway. Thus, the role of CD151 in cell movement includes a signaling aspect as well as a structural aspect. Our present data show that the FAK-Src-MAPKs pathway leading to c-Jun activation also plays an essential role in CD151induced MMP-9 gene expression. The functional role of MAPKs signaling to AP-1 factors in the regulation of MMP-9 expression has been demonstrated in various cell types (64,69,70). Taken together, it appears that CD151-integrin complex activates MAPKs, such as p38 MAPK and JNK, through the FAK-Src signaling pathway in MelJuSo cells.
In summary, we have demonstrated for the first time that homophilic CD151 interactions activate integrin-dependent signaling events that lead to increases in c-Jun-mediated MMP-9 gene expression, cell motility, and invasiveness in MelJuSo human melanoma cells. The signaling pathways initiated by CD151-␣ 3 ␤ 1 /␣ 6 ␤ 1 integrin complexes increase c-Jun activity through the activation of FAK, Src, p38 MAPK, and JNK. Positive cross-talk between p38 MAPK and JNK pathways also contributes to c-Jun activation by CD151-integrin complexes. These findings may be useful in designing therapeutic interventions that block CD151-induced integrin-dependent AP-1 activation through FAK/Src-mediated activation of p38 MAPK and JNK, resulting in the reduction of MMP-9 expression and cell motility and the consequent blocking of invasion and metastatic spread of malignant melanoma.