Regulation of Integrin Activity by MIA*

MIA (melanoma inhibitory activity) has been identified as a small protein secreted from malignant melanoma cells, which interacts with extracellular matrix proteins including fibronectin. Here, we show that MIA negatively regulates the activity of the mitogen-activated protein kinase pathway in malignant melanoma. Using far Western blotting and co-immunoprecipitation we searched for MIA-binding cell surface proteins. We found that MIA interacts with integrin α4β1 and α5β1, leading to down-regulation of integrin activity and reduction of mitogen-activated protein kinase signaling. These findings also suggest that MIA may play a role in tumor progression and the spread of malignant melanomas via mediating detachment of cells from extracellular matrix molecules by modulating integrin activity. Inhibiting MIA functions in vivo may therefore provide a novel therapeutic strategy for metastatic melanoma disease.

We have previously identified MIA (melanoma inhibitory activity), an 11-kDa protein secreted into the tissue culture supernatant from malignant melanoma cells (1). MIA expression in vivo correlates with progressive malignancy of melanocytic tumors (2). Additionally, in recent studies we detected enhanced MIA protein levels specifically in the serum of patients with metastatic melanomas (3). In vitro studies revealed a role for MIA in supporting the invasive and migratory potential of melanoma cells. In vivo studies in two animal model systems confirmed the importance of MIA in melanoma metastasis. MIA expression levels parallel closely the capability of melanoma cells to form metastases in syngeneic animals (4,5).
Three-dimensional analyses of MIA by multidimensional NMR (6 -8) or x-ray crystallography (9) indicate that MIA defines a novel family of secreted proteins that adopt an SH3 domain-like fold in solution. Furthermore, NMR spectra revealed that MIA interacts with peptides matching to type III human fibronectin repeats that are closely related to ␣4␤1 integrin-binding sites (6). These data support a model in which MIA regulates attachment to specific components of the extracellular matrix. Based on these results and on the observation that MIA alters melanoma cell morphology, we determined that MIA treatment results in cell detachment by decreasing interactions between melanoma cells and extracellular matrix molecules (10). The study presented here was performed to find additional MIA-interacting proteins and to identify signaling pathways regulated by MIA.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-The melanoma cell lines Mel Im and Mel Ei, have been described in detail previously (11). The cell line Mel Ei was derived from a primary cutaneous melanoma, and the cell line Mel Im was derived from a metastasis of malignant melanoma. To establish fibronectin-deficient fibroblast-like cells, primary embryonic fibroblasts were isolated from E13.5 fibronectin (flox/flox) embryos and immortalized by retroviral transduction of the SV-40 large T antigen and cloned (12). Subsequently, two clonal lines were treated with a cre-transducing adenovirus to delete the floxed fibronectin genes. The deletion of fibronectin alleles in both cre-treated clones was confirmed by PCR, and the lack of Fn protein expression was confirmed by immunoprecipitation using metabolically labeled conditioned media of these cells. Integrin ␤1-deficient cells were generated as previously published (13).
Expression and Purification of Recombinant MIA Protein-Escherichia coli M15(pREP4) cells transfected with the expression plasmid pQE40-MIA expressing 108 residues of human MIA (Gly 25 -Gln 131 ) were grown to an absorbance A 600 nm ϭ 0.6, induced by 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 4 h and lysed by sonication. The protein was renatured from E. coli inclusion bodies as previously published (1). Refolded human MIA was applied to hydrophobic interaction chromatography and further purified on a S-Sepharose Fast Flow (Sigma). Finally, gel filtration was performed on a Superdex 200 Prep Grade (Sigma). Fractions containing human MIA were pooled and concentrated. Purified protein was checked by SDS-PAGE and high pressure liquid chromatography and was shown to be 95% pure (data not shown).
Expression of Recombinant Biotinylated MIA Protein-A MIA prokaryotic expression vector with a 15-amino acid Avi tag peptide sequence including a FXa cleavage site was constructed by overlap extension PCR. The MIA cDNA construct was cloned into the vector pIVEX2.3-MCS (Roche Applied Science). The expression vector was used in the rapid translation system, a cell-free E. coli-based protein transcription/translation system (Roche Applied Science). By adding biotin, ATP, and the E. coli biotin protein ligase BirA during the procedure, the protein was biotinylated at the introduced Avi tag at the N terminus and called biotinylated MIA. The correct function and folding of the protein was tested by performing a well established MIA-ELISA 2 and functional assays. * This work was supported by a Deutsche Forschungsgemeinschaft grant (to A.-K. B.).
Assays for Protein-Protein Interaction-Purified integrins (␣4␤1 (1:100 in PBS), ␣5␤1 (1:500 in PBS) (14)), 5% BSA in PBS, as control, or biotinylated MIA (1 g/ml in PBS), respectively, were coated each in a volume of 100 l onto 96-well high binding protein plates (Costar) overnight at 4°C. After 17 h the coated wells were blocked with 200 l of blocking buffer (25 mM Tris-HCl, 150 mM NaCl, and 5% BSA) for 2 h, washed three times with 200 l of binding buffer (150 mM NaCl, 25 mM Tris-HCl, 1 mM MnCl 2 , and 0.1% BSA), and exposed to 100 l of 1 g/ml purified biotinylated MIA in binding buffer for 2 h at room temperature. The controls were exposed to 1 g/ml BSA. After three washing steps with binding buffer, MIA binding was quantified using a peroxidasecoupled monoclonal anti-MIA antibody and the substrate 2,2Ј-azino-di[3ethylbenzthiazoline sulfonate] (ABTS) (Roche Applied Science). The reactions were measured at A 405 nm . The experiments were repeated three times with similar results.
For co-immunoprecipitation 100 g of cell lysates dissolved in binding buffer (20 mM NaPO4, 150 mM NaCl, pH 7,5) were precleared with 25 l of protein streptavidin-coupled G-Sepharose (Amersham Biosciences) at 4°C overnight. After centrifugation the supernatant was transferred into a fresh vial and incubated with biotinylated MIA with shaking at 4°C overnight. Then 50 l of protein streptavidin-coupled G-Sepharose was added for 1 h, pelleted, washed three times with binding buffer, resuspended in 20 l of Laemmli's buffer, heated at 95°C for 5 min, and separated on 12% SDS-polyacrylamide gels. All of the proteins binding to MIA were stained by silver staining (Silver Express Staining; Invitrogen). The experiments were repeated three times with similar results. Additionally, specific proteins were identified by Western blot.
Protein Analysis in Vitro (Western Blotting)-For protein isolation 2 ϫ 10 6 cells were washed with 1ϫ PBS and lysed in 200 l of RIPA buffer (Roche Applied Science). The protein concentration was determined using the BCA protein assay reagent (Pierce). Balanced amounts of cell proteins (50 g) were denatured at 70°C for 10min after addition of Roti-load-buffer (Roth, Karlsruhe, Germany) and subsequently separated on NuPAGE-SDS gels (Invitrogen). After transferring the proteins onto PVDF membranes (Bio-Rad), the membranes were blocked in 3% BSA/PBS for 1 h and incubated with a 1:1000 dilution of primary polyclonal rabbit anti integrin ␣2, 3, 4, or 5 antibody or anti-ERK1/2, anti P-ERK1/2 (R & D, Richmond, VA; Chemicon, Hampshire, UK) overnight at 4°C. A 1:2000 dilution of anti-rabbit-AP (Chemicon) was used as a secondary antibody. Staining was performed using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate solution (Zytomed, Heidelberg, Germany). All of the experiments were repeated at least three times with similar results. The peptide spot analysis (Jerini, Berlin, Germany) was performed to assay kinase activity following the manufacturer's instructions.
MIA Far Western-SDS-PAGE was performed with 40 g of RIPA lysates from melanoma cell lines and controls, respectively. The proteins were transferred to PVDF membrane by electroblotting. The membrane was blocked with 3% BSA for 30 min and incubated with biotinylated MIA (1 g/ml) for 3 h at 4°C. After three washing steps with PBS, the membrane was incubated with alkaline phosphatasecoupled streptavidin for 30 min. Staining was performed using nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate solution (Zytomed). The experiments were repeated three times with similar results.
MIA-ELISA-MIA produced recombinantly was measured using an ELISA system (Roche Applied Science). 1 ϫ 10 6 cells were cultured in 5 ml of serum-free Dulbecco's modified Eagle's medium, and the amount of MIA secreted into the supernatant was quantified by ELISA. Mono-clonal antibodies coupled with biotin or peroxidase, respectively, were used to quantify MIA in a 96-well plate coated with streptavidin. 2,2Ј-Azino-di[3-ethylbenzthiazoline sulfonate] (ABTS) was used as substrate and quantified at A 405 nm .
FACS Analysis-To analyze binding of MIA to the cell surface, the cells were detached from flasks using 5 mM EDTA and resuspended in PBS. 10 6 cells were incubated with 1 g/ml biotinylated MIA for 1 h at 4°C. For specific competition probes were incubated with 400 and 800 ng/ml recombinant MIA, respectively. After three washing steps with PBS, the secondary antibody (neutravidin/fluorescein isothiocyanate) was added for 1 h at 4°C. After additional washing with PBS, FACS analysis was performed.
To analyze whether MIA regulates integrin activity, the cells were detached from flasks using 5 mM EDTA in PBS. The cells were resuspended in PBS (10 7 cells/ml) containing 4.5 g/liter glucose and treated with biotinylated MIA (1 g/ml), with MnCl 2 or with EDTA. One mM MnCl 2 was used to activate all integrins, and 1 mM EDTA was used to inactivate integrin activity (15). 50 l of the treated cells were mixed with 50 l of the primary antibody 12G10 (20 g/ml) (specific for activated integrins) and incubated for 60 min on ice. Additionally, an antibody was used staining ␤1 independent on the activation status (Chemicon). After three washing steps with PBS with 1% fetal calf serum, the cells were stained with the secondary antibody streptavidin/cy5 in PBS with 10% serum for 30 min. After three washing steps the probes were fixed (100 l of PBS, 100 l of 2% formaldehyde), washed, and resuspended in a volume of 300 l of PBS. FACS data were analyzed using the CellQuest software (Becton Dickinson). Statistical significance was determined by Kolmogorow-Smirnov analysis. All of the experiments were repeated three times.
Statistical Analysis-The results are expressed as the means Ϯ S.D. (range) or percentages. Comparison between groups was made using the Student's paired t test. A p value Ͻ0.05 was considered statistically significant. All of the calculations were performed using the GraphPad Prism software (GraphPad Software Inc, San Diego, CA).

MIA Reduces MAP Kinase Activity in Malignant Melanoma-One
aim of this study was to identify signaling pathways regulated by MIA. To analyze the influence of MIA treatment on MAP kinase (MAPK) activity, we used an assay measuring phosphorylation of myosin-binding protein by MAPK. This activity assay revealed a strong reduction of MAPK activity with a maximum 1 h after the beginning of MIA treatment (Fig. 1A). After excluding the MAPKs p38 and c-Jun N-terminal kinase to be regulated via MIA (data not shown), we used antibodies against phosphorylated ERK1/2 to specifically analyze ERK activity after MIA treatment. MIA treatment induced a strong reduction of ERK1/2 activity in melanoma cell lines; maximum of inhibition was detected 45 min to 2 h after beginning of MIA treatment (Fig. 1B). Regulation of ERK activity by MIA was confirmed by peptide spot analysis (data not shown).
MIA is known to bind to several matrix proteins including fibronectin and to promote detachment of melanoma cells (10). To investigate whether the reduction of ERK1/2 phosphorylation depends on MIA binding to fibronectin and consecutive detachment of the cells, we treated melanoma cells with RGD peptides. The RGD sequence in protein domains is known to be a binding site for integrins in extracellular matrix proteins like fibronectin and to be crucial for cells to perceive and attach to their environment. Only a weak reduction of ERK1/2 phosphorylation was detectable after treating the cells with RGD peptides, which implies that MIA influences ERK1/2 activity at least additionally via a different cellular surface motif than the RGD sequence (Fig. 1C).
MIA Binds to Cell Surface-To identify potential cell surface receptors and modulators of cell signaling, we used biotinylated MIA to analyze binding of MIA to the cell surface. FACS analysis revealed specific binding of MIA to the cell surface of melanoma cells (Fig. 2). Fig. 2A presents the results of the negative control. Following incubation with MIA, a significant shift of the fluorescence signal was FIGURE 1. Influence of MIA on MAP kinase activity in melanoma cells. Melanoma cells were incubated with MIA for different periods of time, and subsequently, cell lysates were generated. A, equal amounts of cell lysates were incubated with myosin-binding protein (MBP) to analyze MAP kinase activity, and phosphorylated myosin-binding protein was subsequently detected by Western blot. Down-regulation of MAP kinase activity by MIA was detected. Maximum of inhibition was observed after 1 and 2 h. B, Western blot analysis of ERK phosphorylation in Mel Im cell lysates revealed a decrease in ERK phosphorylation with a maximum of inhibition after 45 min to 2 h. To ensure equal loading ERK1/2 was stained as a control. C, incubation of melanoma cells with RGD peptides for different periods of time revealed no significant changes in ERK phosphorylation. To ensure equal loading, ERK1/2 was stained as a control. The values in percentages give the results of the densitometric analysis. All of the assays were repeated at least three times, and representative results are presented. detectable ( p Ͻ 0.001; Fig. 2B) as compared with the negative control ( Fig. 2A). This shift could be competed by unlabeled MIA protein, indicating the specificity of the binding of MIA to the cell sur-face (400 ng/ml unlabeled MIA (Fig. 2C) and 800 ng/ml (Fig. 2D) compared with Fig. 2B). Quantitative measurements showed complete competition of the signals on the cell surface. Because MIA is known to bind to fibronectin, we used mouse embryonic fibroblasts with or without fibronectin expression in the same experiment. Here, FACS analysis revealed no differences in binding of MIA between the two cell lines (Fig. 2, F and G), indicating that binding of MIA to the cell surface was not mediated via fibronectin. Fig. 2E shows the negative control for the mouse fibroblasts.
Identification of MIA Binding Partners-To identify MIA-binding proteins, co-immunoprecipitation (Fig. 3A) and far Western blot analysis (Fig. 3B) were performed. Several interacting proteins were found. Interestingly, both assays revealed bands of ϳ70, 80, and 100 kDa, respectively (Fig. 3, A and B). In co-immunoprecipitation assays, binding of these proteins could be almost completely competed by an access The RIPA lysates were separated on SDS-PAGE, blotted onto a PVDF membrane, and incubated with biotinylated MIA followed by streptavidin/alkaline phosphatase. Staining was performed via nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate. All of the assays were repeated three times, and representative results are presented.  of unlabeled MIA (Fig. 3A). Isoforms of integrins of these sizes were reported in several publications (16 -18). Therefore, MIA co-immunoprecipitated melanoma cell lysates were analyzed in Western blots to determine whether the bands detected by far Western blot analysis (Fig.  3B) and co-immunoprecipitation (Fig. 3A) resemble integrins. Antibodies against ␣2 and ␣3 did not show cross-reactivity with the immunoprecipitated proteins (data not shown). However, anti-␣4 and -␣5 antibodies revealed specific staining (Fig. 4, A and B). Fig. 4 (A and B) shows specific immunolabeling for ␣4 and ␣5 after MIA co-immunoprecipitation of RIPA lysates of melanocytes and of the melanoma cell lines Mel Im and Mel Ei. In the control without biotinylated MIA, no staining was found. In accordance with the ␣4-stained immunoblot of cell lysates (Fig. 4C), a band occurred at ϳ70 kDa and at 80 kDa, respectively (Fig.  4B). In immunoblots labeled with the ␣5 antibody, the unprocessed form of ␣5 appeared at ϳ130 kDa in melanocytes, whereas in the melanoma cell line Mel Im smaller bands were detectable at ϳ80 kDa (Fig.  4C). All of the RIPA lysates that were immunoprecipitated with biotinylated MIA clearly showed bands at ϳ80 kDa.
MIA Binds Directly to Integrin ␣4␤1 and ␣5␤1-To confirm direct binding of MIA to ␣4 and ␣5 integrins an ELISA with plates coated with the respective integrins was applied. MIA binding to both ␣4 (Fig. 5A) and ␣5 (Fig. 5B), integrin was demonstrated. Soluble ␣4 (Fig. 5A) and ␣5 (Fig. 5B) proteins competed MIA binding in a dose-dependent manner, indicating the specificity of the interaction between MIA and ␣4 and ␣5, respectively. Furthermore, we denatured the integrins bound to the plate by incubating them with 1 and 5 mM EDTA, respectively. EDTA complexes divalent cations and thus hinders integrins to establish their active conformation (19). Under these experimental conditions, MIA  binding was completely inhibited, indicating that MIA does not bind to inactivated integrins (Fig. 5).
MIA Inhibits Integrin Activity-Because MIA can inhibit MAPK activity, and integrins are known to modulate MAPK signaling (19), we speculated whether MIA inhibits integrin signaling by binding to integrin heterodimers. To analyze a putative effect of MIA on activated integrins, we performed FACS assays using the antibody 12G10, which is specific for activated integrins. Preparation of the cells for the FACS experiment resulted in inactive integrins (Fig. 6A, dashed line). Therefore, integrin activation was induced by pretreatment with 1 mM MnCl 2 (Fig. 6A, dotted line). Subsequent MIA incubation significantly down-regulated integrin activity ( p Ͻ 0.001) (Fig. 6A, bold line), almost reaching the level before MnCl 2 activation. In contrast, when cells were pretreated with EDTA to inhibit integrin activity, no effect of MIA was detected (data not shown). As control the equivalent experiment was performed using an antibody detecting integrin ␤1 independent of the activation status (Fig. 6D). Here, no changes by incubation with MIA were seen.
Analysis of MIA and Integrin Interaction-To analyze whether biotinylated MIA competes with RGD peptides for ␣4␤1 or ␣5␤1 binding, we performed additional ELISA experiments. Consistent with our data on ERK activation, we did not find any competition of biotinylated MIA by RGD peptides for ␣4␤1 or ␣5␤1 (Fig. 7).
In the reversal of the experiment (Fig. 9C), where the fibronectin fragments were coated to the plates, ␣5␤1 competed with MIA for binding to the 50K fragment in a dose-dependent manner, whereas ␣4␤1 competed only weakly with MIA for binding to 50K. When the plates were coated with the H/120 fragment, ␣4␤1 but not ␣5␤1 competed with MIA for binding to H/120 (Fig. 9D).
Next, we wanted to analyze whether the regulation of MAPK activity by MIA is modulated by direct binding of MIA to integrins. The 50K fragment was used in the experiments because of its inhibitory effect on MIA integrin interaction. Western blots with antibodies against phosphorylated ERK1/2 to specifically analyze ERK activity after MIA and MIA plus 50K treatment were performed (Fig. 10A). MIA treatment showed inhibition of ERK activity, which was reversed dose-dependently by the 50K fragment. Additionally, ␤1-integrin deficient cells (GD25-␤1) were used in comparison with wild type cells to prove the regulation of MAPK activity by MIA via integrins. No change in ERK activity was observed after treating the ␤1 integrinnegative cells with MIA, whereas the wild type cells showed clear reduction as expected (Fig. 10B).

DISCUSSION
In this study we aimed to analyze signaling pathways and putative receptor molecules of MIA. Previous studies revealed direct interaction of MIA with several matrix proteins such as fibronectin or laminin (19); however, cell surface receptors for MIA or MIA-regulated signaling pathways have not been characterized until now.
Analyzing the MAP kinase pathway, we found strong reduction of ERK1/2 but not of p38 and c-Jun N-terminal kinase activity after MIA treatment. To investigate whether this effect on ERK1/2 activity was a consequence of MIA binding to fibronectin inducing cellular detachment, we examined the effect of RGD peptides, simulating cellular detachment from fibronectin, on ERK1/2 phosphorylation. Incubation of melanoma cells with RGD peptides induced only a weak reduction of ERK1/2 phosphorylation. These results implied that MIA influences ERK1/2 activity either via a different binding motive on the integrins or by as yet unknown cell surface receptors.
FACS analysis revealed that MIA binds directly to the cell surface. Subsequent experiments with fibronectin positive and negative mouse embryonic fibroblasts showed no differences in MIA binding to the cell surface. Therefore, we hypothesized the existence of a cell surface molecule directly binding MIA and excluded the possibility of MIA binding to cells via matrix proteins bound to the cell surface.
To characterize MIA binding partners, we performed co-immunoprecipitation analysis with consecutive SDS-PAGE and silver staining and, additionally, far Western blots. Molecules of ϳ70, 80, and 100 kDa were found to bind to MIA in both experiments. Because several publications showed this pattern after staining for integrins ␣4 and ␣5 (16 -18), we analyzed whether these proteins were detectable in our co-immunoprecipitation experiments. Therefore, specific antibodies against integrin ␣4 and integrin ␣5, respectively, were used to analyze the MIA co-immunoprecipitate by Western blot. Both antibodies revealed specific staining at 70 and 80 kDa, whereas antibodies against integrins ␣2 and ␣3 did not show staining. It is known that the 150-kDa pro-protein forms of ␣4 and ␣5 are proteolytically cleaved by proprotein convertases. ␣4 is cleaved in the post-endoplasmatic reticulum into a C-terminal 70-kDa fragment and a N-terminal 80-kDa fragment that are noncovalently linked (21,22). For both ␣4 and ␣5, it has been shown that only the processed form is present on the membrane surface of cells (16,22,23). The antibodies used in our experiment were directed against the C termini of ␣4 and ␣5 integrins (16,18,24). In RIPA lysates of human melanocytes and Mel Im cells, the processed C-terminal fragment of ␣4 integrin was detected (Fig. 4C). This processed form was bound by MIA together with a 80-kDa form as seen by silver staining of the co-immunoprecipitates and in the far Western blot. For ␣5, however, we mainly found the pro-form of the protein in melanocyte RIPA lysates (130 -140 kDa). In Mel Im lysates lower bands at ϳ70 -80 kDa were detected, which presumably reflect the processed forms (see arrows in Fig. 4C). The staining further revealed that only the large form of ␣5 integrins is present in normal cells, whereas in melanoma cell lines both the unprocessed ␣5 and the processed forms, are detectable.
To verify the results of MIA binding to integrins ␣4 and ␣5, ELISA experiments were performed suggesting that MIA is able to directly bind to ␣4␤1 and ␣5␤1, respectively. These data were further supported by FACS analysis using the antibody 12G10 specifically detecting activated integrins. Our data indicate that MIA not only binds to integrins but also down-regulates their activity. Considering the fact that integrins are involved in ERK1/2 signaling (19), these results confirmed our data that ERK1/2 phosphorylation decreased in melanoma cells after incubation with MIA.
To further specify MIA binding to ␣4␤1 and ␣5␤1 we performed competition experiments with the fibronectin fragments 50K and H/120 (a schematic model is depicted in Fig. 8). It is known that 50K contains an RGD site and binds specifically to ␣5␤1; H/120 binds specifically to ␣4␤1 via the connecting segment (IIICS) (14). We found that the 50K fibronectin fragment competed with MIA binding to ␣5␤1 dose-dependently. It also competed with MIA binding to ␣4␤1, although to a lesser extent. This might be due to the fact that MIA is able to directly bind to the 50K fragment at two different sites (FN6 and the RGD site) (10), possibly resulting in sequestration of MIA by the 50K fragment and inhibition of binding to the integrin. Additionally, this hinted to the possibility that the 50K fragment either binds to the same site of the integrins MIA bind to or it covers the MIA-integrin binding epitope. Otherwise, MIA bound by the 50K fragment would also be able to bind to the integrins.
Interestingly, when performing the experiments with low concentration of the FN fragment H/120, it did not compete with MIA on ␣4␤1 and ␣5␤1. At higher concentrations it even increased MIA binding mainly to ␣5␤1 integrin. It was shown in previous studies that MIA is able to bind to domains of the H/120 fragment (FN14 and the connecting segment IIICS) (10). Potentially, a complex was formed in which the H/120 fragment (bound by additional MIA molecules) was bound to the integrin ␣5␤1 via MIA. If H/120 binds to MIA without covering the MIA integrin-binding site, MIA would be able to support indirect binding of H/120 to ␣4␤1 integrins. We therefore speculated that the 50K fibronectin fragment blocked the MIA integrin-binding site, whereas the H/120 fibronectin fragment did not interfere with MIA-integrin binding. However, plating the fibronectin fragments first and, as a second step, incubating with MIA and subsequently competing with soluble integrins showed clear competition of MIA with soluble ␣5␤1 on 50K and ␣4␤1 on H/120. Only a slight reduction of MIA binding was found competing binding to 50K with ␣4␤1 or binding to H/120 with ␣5␤1. These results are explained by the fact that ␣4␤1 and 50K and ␣4␤1 and H/120, respectively, do not interact. However, a minor amount of MIA might be sequestered by ␣4␤1 or ␣5␤1. Consecutive assays revealed that the regulation of ERK activity by MIA is via direct binding of MIA to integrins. Integrins ␣4␤1 and ␣5␤1 have been shown to play an important role in melanoma development and progression (for review see Ref. 28).
Expression of both has been proven to be elevated in primary and metastatic melanoma (25)(26)(27). Furthermore, expression of the ␣4 subunit is associated with melanoma cell accumulation of disseminated cells in distant tissues (27). ␣4␤1 is usually expressed as a first step of extravasation to support binding to VCAM-1 expressed on activated endothelial cells. Binding of integrins to matrix proteins is a tightly regulated process. This is best exemplified in migrating cells where cells FIGURE 10. Influence of MIA on MAP kinase activity regulated via integrins. A, melanoma cells were incubated with MIA and the 50K fibronectin fragment, and subsequently, the cell lysates were generated. Western blot analysis of ERK phosphorylation in Mel Im cell lysates revealed a decrease in ERK phosphorylation. Additional treatment with 50K reversed this effect. To ensure equal loading ERK1/2 was stained as a control. B, incubation of GD25-␤1 integrin cells with MIA revealed no significant changes in ERK phosphorylation, whereas in the wild type cells (GD25) ERK activity was reduced. To ensure equal loading, ERK1/2 was stained as a control. The experiments were repeated three times, and representative results are presented.
form a zone of attachment in the front and detachment at the rear. We hypothesize that the control of integrin activity by MIA regulates migration of melanoma cells.
In summary, this study revealed that MIA directly interacts with integrin ␣4␤1 and ␣5␤1. This interaction modulates integrin activity and additionally leads to down-regulation of integrin signaling via ERK1/2.