Matrix Metalloproteinase 9 (MMP-9)-dependent Processing of βig-h3 Protein Regulates Cell Migration, Invasion, and Adhesion*

Background: Cell migration is involved in altering the cell and matrix interface on the cell surface. Results: βig-h3 is cleaved by MMP-9, and its cleavage results in changes in its binding properties, cell adhesion, cell migration, FAK/Src signals, and chemoattractant effects. Conclusion: MMP-9-cleaved βig-h3 modulates tumor cell and macrophage migration. Significance: The MMP-9-mediated βig-h3 processing mechanism is crucial for understanding cell migration. Cell migration is critically involved in inflammation, cancer, and development. In this study, transforming growth factor-β-induced protein (βig-h3) was identified as a substrate of matrix metalloproteinase-9 (MMP-9) by site-directed mutagenesis. βig-h3 has two cleavage sites with the consensus sequence Pro-Xaa-Xaa-Hy-(Ser/Thr) (Hy is a hydrophobic amino acid) (PGSFT beginning at amino acid 135 and PPMGT beginning at amino acid 501). Using recombinant human βig-h3 and MMP-9, βig-h3 from βig-h3-transfected HEK293F cells, and MMP-9 from MMP-9-transfected HEK293F cells, human macrophages, and neutrophils, we found that MMP-9 proteolytically cleaves βig-h3. Cleavage leads to the loss of its adhesive property and its release from extracellular matrix proteins, collagen IV, and fibronectin. Spheroids formed by increased cell-cell interactions were observed in βig-h3-transfected HEK293F cells but not in vehicle-transfected HEK293F cells. In human glioma U87MG cells, MMP-9 constitutive overexpression resulted in endogenous βig-h3 cleavage. βig-h3 cleavage by MMP-9 led to increased cell invasion, and βig-h3 knockdown also resulted in increased cell invasion. The βig-h3 fragment cleaved by MMP-9 could bind to the surface of macrophages, and it may play a role as a peptide chemoattractant by inducing macrophage migration via focal adhesion kinase/Src-mediated signal activation. Thus, intact βig-h3 is responsible for cell migration inhibition, cell-cell contact, and cell-extracellular matrix interaction. Experimental evidence indicates that MMP-9-cleaved βig-h3 plays a role in MMP-9-mediated tumor cell and macrophage migration.

The extracellular matrix (ECM) 3 is composed of various macromolecules such as collagen, laminin, vitronectin, secreted protein acidic and rich in cysteine (SPARC), and fibronectin that form a highly organized structure among cells. Changes in ECM composition alter the matrix structure and consequently modify matrix-matrix and matrix-cell interactions (1,2).
MMP-9, like other MMPs, belongs to a matrixin family of metalloendopeptidases and is synthesized as a 92-kDa zymogen, which is converted to an active enzyme with a molecular mass of 82 kDa (3,17). MMP-9 is induced by proinflammatory cytokines, and it plays a role in inflammation in various diseases (18 -20). It is inhibited by endogenous tissue inhibitors of metalloproteinases (TIMPs) (21,22). Until now, reported MMP-9 substrates included collagen IV or V, fibronectin, ICAM-1, plasminogen, and interleukin-2 (IL-2) receptor (17). MMP-9dependent release of vascular endothelial growth factor (VEGF) acts as a chemoattractant for osteoclast recruitment and invasion in development, although the actual substrate has not been identified (23). MMP-9-null mice have reduced development of carcinomas (24).
Transforming growth factor-␤-induced protein (also named ␤ig-h3, Bigh3, TGF-␤-induced protein h3, keratoepithelin, and RGD-CAP) is a secretory protein that is induced by transforming growth factor-␤ (⌻GF-␤) (25)(26)(27). The ␤ig-h3 protein has 683 amino acids and contains an N-terminal cysteine-rich domain and four internal repeat fasciclin (FAS1) domains. The FAS1 domains include YH and RGD motifs, both of which interact with integrins ␣1␤1, ␣3␤1, and ␣v␤5 (28 -32). ␤ig-h3 is ubiquitously expressed at high levels in various normal human tissues but at low levels in many tumor cell lines, including lung, breast, colon, prostate, leukemia, and kidney (33). ␤ig-h3 is produced in fibroblasts at significant levels but is undetectable in one kidney cell line, 293 cells (33). ␤ig-h3 protein suppresses human breast tumor progression and malignant transformation of cells (27, 34 -38). Mutation or altered expression of ␤ig-h3 has been linked to the pathogenesis of human corneal dystrophy and osteogenesis (39). However, the underlying molecular mechanism of ␤ig-h3 effects is not well understood. Although MMPs are responsible for extracellular matrix modulation, the mechanism of MMP-9 action is also not clear.
In this study, we found that MMP-9 induction coincided with ␤ig-h3 degradation in IL-1␤-or TNF-␣-stimulated glioma cells. We also found that ␤ig-h3 is a substrate of MMP-9 and that ␤ig-h3 processing by MMP-9 results in reduced binding affinity to ECM proteins such as fibronectin and collagen IV. ␤ig-h3 processing alters cell invasion and adhesion-related FAK/Src signals in glioma cells. Intact ␤ig-h3 is responsible for cell-cell contact, cell-ECM interaction, and cell migration inhibition. However, its cleavage by MMP-9 leads to increased cell migration. The cleaved fragment of ␤ig-h3 plays a chemoattractant role in macrophage migration via its regulation of adhesion-related FAK/Src signals.
Cell Lines-Human embryonic kidney (HEK) 293F cells and human glioblastoma cell line U87MG were cultured in DMEM (Invitrogen) supplemented with 10% fetal bovine serum (FBS) (Invitrogen). The human astroglioma cell line CRT-MG was cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS. The cells were maintained at 37°C in a humidified incubator with a 5% CO 2 atmosphere. Cells were confirmed to be mycoplasma-free by PCR. The human monocyte cell line U937 was cultured in RPMI 1640 medium supplemented with 10% FBS and differentiated into macrophages with phorbol 12-myristate 13-acetate (PMA; Calbiochem) treatment (100 M) for 24 h. The neutrophils were purified from heparinized human peripheral blood by density sedimentation with Ficoll-Paque (GE Healthcare) followed by red blood cell (RBC) lysis with RBC lysis buffer. Isolated neutrophils were cultured in DMEM supplemented with 10% FBS, 50 units/ml penicillin, and 50 g/ml streptomycin. Purity of neutrophils was confirmed by Hema 3 staining.
Site-directed Mutagenesis-The P135E, P501E, and P135E/ P501E mutations in the pcDNA3.1-␤ig-h3-myc were generated by PCR using the following primers (the mutated codon is underlined) and by using a site-directed mutagenesis kit (iNtRON, Daejon, Korea): P135E forward 5Ј-GAG ATG GAG GGG GAG GGC AGC TTC ACC-3Ј and the P135E reverse 5Ј-GGT GAA GCT GCC CTC CCC CTC CAT CTC-3Ј; P501E forward 5Ј-CGG GTG CTG ACC GAG CCA ATG GGG ACT-3Ј and P501E reverse 5Ј-AGT CCC CAT TGG CTC GGT CAG CAC CCG-3Ј. The correct sequence and orientation of all cloned genes were verified by sequencing.
MMP-9 Purification-MMP-9 purification was performed using gelatin-Sepharose chromatography. Briefly, MMP-9 activated by APMA was applied to a column of gelatin-Sepharose equilibrated with buffer containing 50 mM Tris-HCl (pH 7.5), 0.4 M NaCl, and 5 mM EDTA. After washing, elution was performed using a buffer containing 50 mM Tris-HCl (pH 7.5), 1 M ␤ig-h3 Is a Substrate of MMP-9 NaCl, 5 mM EDTA, and 5% (v/v) DMSO. The eluted fraction was collected, concentrated, and dialyzed against PBS supplemented with MMP inhibitor-free protease inhibitor mixture (Sigma). The purified samples were analyzed by gelatin zymography and immunoblotted to confirm the depletion of other proteins.
Gelatin Zymography-The samples were loaded under nonreducing conditions onto a 7% SDS-PAGE containing 1 mg/ml gelatin. After electrophoresis and washing with a buffer containing 20 mM Tris-HCl (pH 7.5) and 2% Triton X-100 for 1 h, the gel was incubated in MMP reaction buffer containing 20 mM Tris-HCl (pH 7.5) and 10 mM CaCl 2 at 37°C for 16 h. Gelatinolytic activity was detected by staining with Coomassie Brilliant Blue G-250 staining solution.
LC-MS/MS Analysis-Protein bands (Fig. 1E, *) were excised from the stained SDS-polyacrylamide gels and de-stained with destaining solution (25 mM ammonium bicarbonate, 50% acetonitrile). In-gel digestion of dried gel pieces was performed with sequencing grade trypsin (Promega) in 25 mM ammonium bicarbonate buffer overnight at 37°C. The tryptic peptides were desalted using a GELoader tip (Eppendorf) packed with 1.5 g of Poros 20 R2 resin (PerSpective Biosystems) and applied onto a C-18 RP-HPLC column (75 m ϫ 150 mm). An Agilent 1100 Series LC system (Agilent Technologies) was used to separate tryptic peptides, which were eluted with a 0 -40% acetonitrile gradient for 60 min. The eluant was analyzed with a Finnigan LCQ Deca (ThermoQuest) equipped with a nanoelectrospray ion source. Spray voltage and tube lens voltage was 1.9 kV and 40 V, respectively. The temperature of the capillary was kept at 210°C, and capillary voltage was 30 V. The individual spectra from LC-MS/MS were processed using Turbo-SEQUEST software (ThermoQuest) and searched with NCBI databases using MASCOT software (Matrix Science Ltd.). LC-MS/MS analysis was conducted by ProteomeTech.
Solid-phase Binding Assay-Tissue culture plates (96-well) were coated with 10 g/ml collagen IV or fibronectin and stored at 4°C overnight. After saturation with 1% bovine serum albumin (BSA control, Sigma) for 1 h at room temperature, the plates were incubated with recombinant human ␤ig-h3 (1 M, Sino Biological Inc.), untreated, or pretreated with recombinant human MMP-9 (0.25 M, ProSpec) at 4°C for 24 h and then washed in PBS with 0.1% Tween 20. After washing, the ␤ig-h3 Is a Substrate of MMP-9 NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46 plates were blocked with 10% goat serum in PBS. Bound ␤ig-h3 was detected using an antibody specific for ␤ig-h3, anti-␤ig-h3-c (Cell Signaling Technology), or anti-␤ig-h3-m (Proteintech), followed by horseradish peroxidase (HRP, Cell Signaling Technology)-conjugated anti-rabbit secondary antibody. The plates were washed, treated with substrate and then stop solution, and the absorbance was read at 450 nm. The "ratio" was obtained by dividing the ␤ig-h3 level detected by the anti-␤ig-h3 antibody in the ␤ig-h3 or ␤ig-h3 plus MMP-9 samples by the ␤ig-h3 level detected by anti-␤ig-h3 antibody in nonadded samples.
Invasion Assay-Invasion assays were performed using 24-Transwell inserts with 8-m microporous membranes with or without Matrigel (BD Biosciences) coating on the upper side (Costar). The outer membranes of the Transwell inserts were coated with gelatin (1 mg/ml). Cells (5 ϫ 10 4 ) preincubated in 100 l of DMEM or RPMI were placed in the upper chamber, and 600 l of DMEM containing 2% FBS was added to the lower chamber. After 24 h, the cells remaining on the inner membrane were removed with a cotton swab. The Transwells were stained with hematoxylin (Merck) and eosin Y (Sigma). The cell number was counted using a light microscope.
Spheroid Assay-The spheroid assay was modified as described (40). In brief, 5 ϫ 10 5 cells were seeded into nonadhesive culture plates and cultured with 293 serum-free media (293SFM; Invitrogen) supplemented with 1 mM glutamine (Invitrogen). Spheroids were allowed to form for 3 or 6 weeks at 37°C and 5% CO 2 . Spheroid formation was observed using a light microscope (4ϫ, 3 weeks of culture) and was visible to the naked eye. "Naked eye 1" was defined as the spheroids, which could be visible to the naked eye after 3 weeks of incubation. "Naked eye 2" was defined as the spheroids, which could be visible to the naked eye after 6 weeks of incubation.
Cell Surface Binding Assay by Flow Cytometric Analysis-PMA-differentiated human macrophage U937 cells (1 ϫ 10 5 ) were washed in cold RPMI containing 20 mM Tris-HCl (pH 7.4), 1 mM CaCl 2 , and 1 mM MgCl 2 . The cells (50-l aliquots) were incubated with a recombinant human ␤ig-h3 fragment (␤ig-h3-3rd, amino acid residues 502-683, 25 nM, ProSpec) for 2 h at 4°C with gentle shaking. After two washes, the cells were blocked with goat serum and incubated for 30 min on ice with anti-␤ig-h3 antibody or isotype IgG, followed by incubating for 30 min with FITC-conjugated secondary anti-rabbit antibody. After two washes, the cells were fixed in 1% paraformaldehyde in PBS and analyzed by flow cytometry. Flow cytometry was performed on a FACSCalibur (BD Biosciences), and data were analyzed using WinMDI software version 2.8 (The Scripps Research Institute).
Chemotaxis Assay-The chemotaxis assay was performed using 24-well chemotaxis chambers (Costar). The outer membrane of the Transwell insert was coated with gelatin (1 mg/ml). PMA-differentiated human macrophage U937 cells (5 ϫ 10 4 ) were preincubated in 100 l of RPMI and placed in the upper chamber. Then 600 l of RPMI-containing lipopolysaccharide (LPS) (Sigma, 10 ng/ml), the recombinant human ␤ig-h3 fragment (␤ig-h3-3rd, amino acid residues 502-683, 25 nM, ProSpec), and the recombinant human ␤ig-h3 (␤ig-h3 fulllength, 25 nM, Sino Biological) were added to the lower cham-ber. All recombinant human ␤ig-h3 were commercially available. After 16 h, the cells remaining on the inner membrane were removed with a cotton swab. Cells on the transwell plates were fixed in 4% formaldehyde and stained with hematoxylin. The cell number was counted using a light microscope. Endotoxin in the recombinant human ␤ig-h3 fragment was not detected.
Immunoprecipitation-Cells were lysed with Lysis Buffer (50 mM HEPES, 150 mM NaCl, 1% Nonidet P-40 (pH 8.0)) for 15 min on ice and then centrifuged at 12,000 ϫ g and at 4°C. Supernatants were pre-cleared for 1 h at 4°C with G protein beads (Invitrogen). The pre-cleared samples were immunoprecipitated at 4°C for 18 h using anti-␤ig-h3 or control rabbit IgG, which were coupled to G protein beads (Invitrogen). Samples were examined by immunoblotting with anti-mouse Myc antibody followed by HRP-conjugated anti-mouse IgG secondary antibody.
Statistical Analysis-All experiments were repeated at least three times. Data represent the means Ϯ S.E. A value of p Յ 0.05 was considered to be statistically significant.
In IL-1␤ or TNF-␣-stimulated cells, the levels of ECM proteins, including fibronectin, laminin, collagen IV, and SPARC were unchanged, even though fibronectin and collagen IV are known as substrates of MMP-9 ( Fig. 1D) (17). Interestingly, an unknown protein band was detected in immunoblots with anti-SPARC antibody (Fig. 1, D and E, upper panel, *). The identity of the protein (*) is unknown, but it is distinct, and its disappearance coincides with MMP-9 induction (Fig. 1, A and D). These results suggest that the unknown protein may be involved in MMP-9-mediated cell migration. The unknown protein was identified as ␤ig-h3 by LC/MS/MS analysis (Fig. 1E, lower  panel). Results of immunoblotting analysis revealed that ␤ig-h3 undergoes degradation in MMP-9-induced cells (Fig. 1F). To confirm that IL-1␤ and TNF-␣ do not directly affect ␤ig-h3 mRNA, RT-PCR was conducted. In glioma CRT-MG cells, ␤ig-h3 mRNA expression levels were not changed by IL-1␤ or TNF-␣ treatment (Fig. 1G).
To confirm if full-length MMP-9 could cleave ␤ig-h3, the full-length commercially available recombinant human MMP-9 was tested. APMA is known for its ability to enhance the autocleavage of the proenzyme to the active and lower molecular mass form. Active and lower molecular mass form MMP-9 was confirmed by immunoblotting using anti-MMP-9 antibody (Fig. 2D). When ␤ig-h3 secreted from ␤ig-h3-transfected HEK293F cells was treated with the full-length MMP-9, the ␤ig-h3 protein band also disappeared, indicating cleavage by MMP-9 (Fig. 2E). In addition, pretreatment with the MMP-9 inhibitor, TIMP-1, prevented the loss of ␤ig-h3. However, when the ␤ig-h3 protein (P135E/P501E) with mutations in both MMP-9 cleavage sites was treated with APMA-activated fulllength MMP-9, there was no change (Fig. 2E).
A time course study of ␤ig-h3 degradation by MMP-9 was conducted and is shown in Fig. 2F. In the absence of MMP-9 treatment, there was no ␤ig-h3 degradation (Fig. 2F). After MMP-9 treatment for 1 h, the cleaved ␤ig-h3 fragment was detected using anti-Myc antibody (Fig. 2G, right panel). Because the cleaved ␤ig-h3 fragment was not detected by anti-␤ig-h3 antibody, which recognizes the C terminus of the protein (Fig. 2G, middle panel), antibody specific for different regions of the ␤ig-h3 protein were used for detecting both uncleaved and cleaved products (Fig. 2, H and I). ␤ig-h3 fragments cleaved by active full-length MMP-9 increased in a time-and concentration-dependent manner, whereas uncleaved ␤ig-h3 decreased (Fig. 2, H and I, left  panels). However, when the ␤ig-h3 protein (P135E/P501E) with mutations in both MMP-9 cleavage sites was treated with activated full-length MMP-9, there was no change (Fig.  2, H and I, right panels).
Full-length MMP-9 Purified from Different Sources Also Cleaved ␤ig-h3 Protein-To confirm that full-length MMP-9 secreted from different sources can cleave ␤ig-h3 protein, MMP-9 from MMP-9-transfected HEK293F cells, differentiated U937 macrophages, and human blood neutrophils were isolated and examined. MMP-9 transfected HEK293F cells were established by stable transfection with the MMP-9 expression vector (MMP-9-pcDNA3.1-myc) into HEK293F cells (Fig.  3A). MMP-9 expression was confirmed by immunoblotting of the cell culture media with antibodies specific for Myc or MMP-9. Unexpectedly, pro-MMP-9 was overexpressed and secreted from MMP-9-transfected HEK293F cells (Fig. 3,  A-D). The protein was detected by gelatin zymography that HEK293F cells expressed low levels of endogenous, active MMP-9. However, the level was very low when compared with overexpressed pro-MMP-9 (Fig. 3B). To induce the conversion of pro-MMP-9 to the lower molecular mass active form of MMP-9, the conditioned media of MMP-9-transfected HEK293F cells were incubated with APMA (Fig. 3D). The active form of MMP-9 observed after APMA treatment was confirmed by gelatin zymography (Fig. 3E, upper panel). The
MMP-9 production is primarily observed in neutrophils and macrophages in humans (44 -46). PMA-differentiated U937 macrophages or human blood neutrophils release large amounts of active MMP-9 (Fig. 3, G and H). MMP-9 from neutrophils without TIMP were more effective than from other cells (47). To obtain MMP-9 from macrophages, human monocytes (U937) were differentiated with PMA, which is known to stimulate macrophage differentiation from monocytes (48). Active MMP-9 from the culture media of PMA-differentiated U937 macrophages or from human blood neutrophils was purified by gelatin affinity-Sepharose chromatography and detected by gelatin zymography (Fig. 3, G and H, upper panel). Treatment with this purified MMP-9 caused ␤ig-h3 degradation (Fig. 3, G and H, lower panel). Overall, these results suggest that ␤ig-h3 contains a cleavage site, which is recognized by MMP-9.
MMP-9 Cleavage Alters the Adhesiveness of ␤ig-h3 to ECM and to Cell to Cell Interactions-We examined the ability of ␤ig-h3 to bind ECM proteins in a solid-phase assay using commercially available recombinant human ␤ig-h3 with the ECM proteins collagen IV and fibronectin. To determine whether the binding of ␤ig-h3 to ECM proteins is influenced by MMP-9, we used a solid-phase assay to quantitate the amount of ␤ig-h3 bound to collagen IV and fibronectin with or without MMP-9 pretreatment. Two antibodies specific for recombinant human ␤ig-h3 were used for detection of ␤ig-h3; one recognizes the C terminus of ␤ig-h3 (anti-␤ig-h3-c) and the other recognizes the middle region of ␤ig-h3 (anti-␤ig-h3-m). Untreated ␤ig-h3 strongly bound to collagen IV and fibronectin, whereas ␤ig-h3 pretreated with MMP-9 had decreased binding to collagen IV (Fig. 4A) and fibronectin (Fig. 4B).
In addition, when the anti-␤ig-h3-c antibody that recognizes the ␤ig-h3 C terminus was used, the effect of MMP9 on ␤ig-h3 binding to collagen IV was greater. This effect was less when the anti-␤ig-h3-m antibody recognizing the ␤ig-h3 middle region was used. We suggest that the ␤ig-h3 C-terminal region would be involved in binding to collagen to a greater extent.
These results further suggest that ␤ig-h3 adheres to collagen IV and fibronectin. MMP-9 cleaves ␤ig-h3, and this leads to ␤ig-h3 release from collagen IV and fibronectin, thereby weakening the binding between cells.
To determine whether MMP-9 and ␤ig-h3 interact on the cell surface, immunoblotting analysis of anti-␤ig-h3 antibody immunoprecipitates from lysates of MMP-9-transfected U87MG cells was performed. The association of ␤ig-h3 and MMP-9 in these immunoprecipitates was observed, although the quantity of the associated complex between ␤ig-h3 and MMP-9 was small (Fig. 5C). These results suggest that ␤ig-h3 may provide a docking site as well as a cleavage site for MMP-9 on the cell surface.
␤ig-h3 Fragment Generated by MMP-9 Cleavage Affects Macrophage Migration-〈s ␤ig-h3 has two cleavage sites recognized by MMP-9, three ␤ig-h3 fragments were expected to be generated by MMP-9 treatment. The decrease of ␤ig-h3 binding affinity to the ECM after MMP-9 cleavage was observed in a solid-phase assay (Fig. 4, A and B). The ␤ig-h3 cleavage by MMP-9 leads to cell migration and FAK/Src activation in glioma cells (Figs. 1 and 5). However, among these fragments, the C-terminal ␤ig-h3 fragment composed of amino acid residues 502-683 still has the RGD and YH motifs, which interact with cell surface integrins. To determine whether this ␤ig-h3 fragment can bind to neighboring cell surfaces after its release, the cell surface binding assay was performed by flow cytometry, using the commercially available recombinant human ␤ig-h3 fragment (␤ig-h3-3rd) that contains 182 amino acids (residues 502-683). This recombinant human ␤ig-h3 fragment (␤ig-h3-3rd) corresponds to the C-terminal fragment of ␤ig-h3, which is expected to be generated after MMP-9 cleavage. The recombinant human ␤ig-h3 fragment (␤ig-h3-3rd) induced macrophage migration (Fig. 6A), bound to the surface of macrophages, and increased phosphorylated FAK/ Src (Fig. 6, B and C). This result demonstrates that the small C-terminal ␤ig-h3 fragment functions as a peptide chemoattractant to neighboring cells, resulting in cell migration.

DISCUSSION
Cell migration plays a role in inflammation, cancer, and development. MMP-9 proteolytically acts on the ECM and activates signals involved in tumor cell migration (49). In glioma, tumor invasion is inhibited by suppressing MMP-9 secretion and dephosphorylation of FAK (50). It is believed that ␤ig-h3 may function as either an activator or an inhibitor of carcinogenesis, depending on the cell and tumor type. However, the mechanism of action of ␤ig-h3 remains unclear.
In our study, we demonstrated that ␤ig-h3 is a substrate of MMP-9 and identified its cleavage sites by site-directed mutagenesis. ␤ig-h3 adheres to collagen IV and fibronectin in the ECM. Cleavage of ␤ig-h3 by MMP-9 results in reduced binding of ␤ig-h3 and in the release of a ␤ig-h3 fragment from the ECM. ␤ig-h3 has the RGD and YH motifs that interact with cell surface integrins (28,31). Spheroids formed by increased cell-cell interactions were observed in cells overexpressing wild-type ␤ig-h3 or in the cells with the double mutant ␤ig-h3 (P135E/P501E). It is proposed that ␤ig-h3 functions as a bridge between cells or between cells and the ECM. These functions of ␤ig-h3 are directly influenced by MMP-9 cleavage.
Cell migration is involved in altering the cell-matrix interface on the cell surface. It is known that MMP-9 is involved in cell migration, but the detailed mechanism is not clear. In our study, ␤ig-h3 was expressed in glioma cells. It was also demonstrated that ␤ig-h3 directly binds with MMP-9 on cells and that MMP-9 overexpression leads to ␤ig-h3 cleavage. Thus, we suggest that ␤ig-h3 may provide both a docking site and a cleavage addition. Recombinant human ␤ig-h3 (rh␤ig-h3, 1 M) was pretreated or untreated with recombinant human MMP-9 (rhMMP-9, 0.25 M), and a solidphase binding assay was used to determine the adhesiveness of ␤ig-h3 to collagen IV and fibronectin (*, p Ͻ 0.05; **, p Ͻ 0.01, and n ϭ 3). The ratio refers to the ␤ig-h3 level detected by anti-␤ig-h3 antibody, in ␤ig-h3 or ␤ig-h3 plus MMP-9 samples, divided by the ␤ig-h3 level detected by anti-␤ig-h3 antibody, without additions. C, cell-cell interactions in the presence of ␤ig-h3, and the endogenous MMP-9 effects. Spheroids were formed in vehicles (vehicle/293F), in ␤ig-h3 (␤ig-h3/293F)-transfected, or the double mutant ␤ig-h3 (P139/501E/293F)-transfected HEK293F cells within 3 weeks, and spheroid formation was visible under light microscopy (4ϫ, upper panel) or with the naked eye (Naked eye 1). Spheroids were more visible with the naked eye at 6 weeks (Naked eye 2). Arrows marking the ␤ig-h3-transfected cells (␤ig-h3/ 293F) indicate small aggregates (C) (n ϭ 3).
␤ig-h3 Is a Substrate of MMP-9 NOVEMBER 9, 2012 • VOLUME 287 • NUMBER 46 site for MMP-9 on the cell surface. Glioma cells expressing intact ␤ig-h3 did not exhibit cell migration. In contrast, cells overexpressing MMP-9 in the presence of ␤ig-h3 exhibited cleaved ␤ig-h3 and increased cell migration. ␤ig-h3 knockdown also led to increased cell migration. Thus, we conclude that intact ␤ig-h3 inhibits cell migration, but degradation of ␤ig-h3 by MMP-9 leads to increased cell migration.
MMP-9 induction in IL-1␤-or TNF-␣-stimulated human glioma CRT-MG cells resulted in ␤ig-h3 cleavage, adhesionrelated FAK/Src activation, and increased cell migration. However, fibronectin, collagen IV, laminin, and SPARC in these cells did not undergo degradation even though fibronectin and collagen IV are known substrates of MMP-9. We cannot exclude the possibility that the degradation of other ECM proteins besides fibronectin, collagen IV, and laminin is involved in MMP-9 induction, cell migration, and FAK/Src signal activation. However, upon MMP-9 release from the cell surface, MMP-9 activity is inhibited by TIMP (13). The proteolytic activity of MMP-9 is most effective when MMP-9 is localized on the cell surface where its activity can be modulated by eliminating its docking site (13). Thus, it is also likely that ␤ig-h3 degradation by MMP-9 modulates cell migration and MMP-9 activity by eliminating the binding of ␤ig-h3 as its docking site on the cell surface. ␤ig-h3 degradation by MMP-9 leads to ␤ig-h3 release from the cell surface, resulting in the loss of MMP-9-docking site and in MMP-9 loss from the cell surface. MMP-9, once removed from the cell surface, may have less interaction with other ECM proteins due to docking site loss. Thus, we propose that ␤ig-h3 cleavage may be involved in decreasing the interaction of MMP-9 with ECM proteins. FIGURE 5. Endogenous ␤ig-h3 cleavage by MMP-9 in glioma cells. A and B, ␤ig-h3 and MMP-9 in MMP-9-transfected U87MG cells. A, ␤ig-h3 and MMP-9 expression in the vehicle (vehicle/U87MG)-transfected or MMP-9 (MMP-9/U87MG)-transfected U87MG cells were examined by immunoblotting. B, MMP-9 activity was detected by gelatin zymography. M is a marker for active MMP-9 and pro-MMP-2 (n ϭ 3). C, co-immunoprecipitation of MMP-9 with ␤ig-h3 from the vehicle (vehicle/U87MG)-transfected or MMP-9 (MMP-9/U87MG)-transfected U87MG cells. After the lysates from the vehicle (vehicle/U87MG)-transfected or MMP-9 (MMP-9/U87MG)-transfected U87MG cells were immunoprecipitated (IP) with anti-␤ig-h3 (ϩ) or control rabbit IgG (Ϫ), they were subjected to immunoblotting (IB) using mouse antibody for Myc (MMP-9) (n ϭ 4). D, MMP-9 overexpressed in cells alters cell invasion. Cell invasion in vehicle-transfected, MMP-9-transfected, or IL-1␤ (10 ng/ml)-treated U87MG cells was examined on Matrigels. ␤ig-h3 and MMP-9 proteins from IL-1␤-treated human glioma U87MG cells were examined by immunoblotting and zymography. The control represents untreated samples (n ϭ 3). E, ␤ig-h3 knockdown alters cell invasion. Cell invasion in control siRNA or ␤ig-h3 siRNA-transfected U87MG cells was examined on Matrigels. The ␤ig-h3 and MMP-9 from these cells were examined by immunoblotting and zymography. Control represents control siRNA. (*, p Ͻ 0.05; **, p Ͻ 0.01, n ϭ 3).

␤ig-h3 Is a Substrate of MMP-9
MMP-9 is mainly observed in neutrophils and macrophages in humans (45)(46)(47). Macrophages enhance tumor progression and metastasis (51). Tumor-associated macrophages, following their migration into the primary tumor sites, support metastatic cancer cell extravasation (51,52). Macrophage migration is often initiated in response to a chemotactic stimulus. In our study, intact ␤ig-h3 did not affect cell migration, but the degradation of ␤ig-h3 by MMP-9 leads to increased cell migration. The ␤ig-h3 fragments cleaved by MMP-9 were easily detached from the ECM. The fragment containing the C-terminal domain could bind to the surface of macrophages and increase their migration through the activation of FAK/Src-mediated signals. These results suggest that the fragment cleaved by MMP-9 stimulates migration of the surrounding cells by binding to them. Thus, the MMP-9-dependent, processed, and released ␤ig-h3 acts as a chemoattractant.
In this study, we found out that IL-1␤ and TNF-␣ treatment led to MMP-9 induction, FAK activation, and cell migration in glioma cells. With this correlative finding, we demonstrated that MMP-9 could cleave ␤ig-h3 in a molecular level. Through the use of cell-based assays, we demonstrated that ␤ig-h3 is an adhesion-related protein and that MMP-9 could cleave ␤ig-h3 in a cell culture model. The cleavage was not degraded because the resulting fragments played a role in cell migration. Thus, in conclusion, this study provides an explanation of how MMP-9 could induce cell migration.
Furthermore, in some tumor cells it was reported that IL-1␤ and TNF-␣ secreted from macrophages could enhance tumor cell invasion (53). In our study, IL-1␤ and TNF-␣ led to MMP-9 induction in glioma cells. MMP-9 increased cell migration through ␤ig-h3 cleavage. We hypothesize that the increased cell migration, by MMP-9-mediated ␤ig-h3 cleavage, plays an important role in enhancing tumor cell invasion.
In conclusion, this study suggests for the first time that ␤ig-h3 is a substrate of MMP-9 and that ␤ig-h3 plays a regulatory role in MMP-9-mediated cell migration of tumor cells and macrophages.