Extracellular matrix metalloproteinase inducer (EMMPRIN; CD147) is a heavily glycosylated protein containing two immunoglobulin superfamily domains. It is enriched on the surface of tumor cells and stimulates the production of matrix metalloproteinases (MMPs) by adjacent stromal cells. Here we use CD147 transfectants and immobilized recombinant CD147-Fc fusion protein to show that CD147/EMMPRIN engages in a homophilic interaction, predominantly through the first immunoglobulin domain. Anti-CD147 antibody 8G6 and recombinant CD147-Fc fusion protein markedly inhibited not only homophilic interaction, but also the production of secreted MMP-2 by breast cancer cell line MDA-435 and the MMP-2-dependent invasion of MDA-435 cells through reconstituted basement-membrane Matrigel. Purified native CD147 induced the production of secreted MMP not only by dermal fibroblasts (MMP-1) but also by MDA-435 cells themselves (MMP-2), suggesting homophilic CD147-binding may occur in the context of both heterotypic and homotypic cell-cell interactions. Purified deglycosylated CD147 failed to induce MMP-1 or MMP-2, but instead antagonized the MMP-1-inducing activity of purified native CD147. Our results suggest that homophilic CD147 interactions may play a key role in MMP-2 production and tumor cell invasion, and that perturbation of this molecule may have potential therapeutic uses in the prevention of MMP-2 and MMP-1-dependent cancer metastasis.

Degradation of basement membrane by MMPs3 has been implicated in several aspects of tumor progression, including growth, invasion, metastasis, and angiogenesis (1, 2). Among the various MMPs, MMP-2 and MMP-9 are key enzymes for degrading type IV collagen, a major component of basement membrane in human cancer tissues. In many cases, MMPs are secreted from associated stromal fibroblasts rather than from the tumor cells themselves (3, 4, 5, 6, 7). A search for tumor cell-derived MMP-inducing factors led to the discovery and characterization of CD147/EMMPRIN (8, 9). This protein contains two extracellular immunoglobulin domains: a transmembrane domain, and a 39-amino acid cytoplasmic domain (10). Various independent laboratories have discovered the EMMPRIN protein, naming it basigin (11), OX-47 (12), Neurothelin (13), M6 antigen (14), and HT7 antigen (15). Mice lacking the gene for CD147/EMMPRIN/basigin showed defects in embryogenesis, spermatogenesis, and female fertilization (16, 17), as well as an altered mixed lymphocyte reaction and loss of an aversive response to a strong odor (18).

CD147 is more highly expressed on human carcinoma cells than on normal cells (8, 19, 20, 21). Normal oral mucosa showed localized keratinocyte membrane-staining of CD147, whereas oral carcinoma cells showed staining diffusely through the entire tumor specimen (22). Both native and recombinant purified CD147 may induce the production of interstitial collagenase (MMP-1), gelatinase A (MMP-2), and stromelysin (MMP-3; Refs. 9 and 23) by dermal fibroblasts. A role for CD147 during cell invasion and MMP-2 production was recently confirmed for squamous cell carcinoma cells cocultured with peritumor fibroblasts (22). In addition, the assembly of extracellular matrix by squamous cell carcinoma-fibroblast cocultures was inhibited by anti-CD147 mAb, which is consistent with CD147-dependent regulation of MMPs (22).

The precise molecular function of CD147/EMMPRIN is still largely unclear. Its ability to induce MMP production in fibroblasts implies that a fibroblast counterreceptor may exist, but, as yet, none has been identified. In this study, we address the possibility that CD147 could be a receptor for itself, and we identify the key extracellular domain required for CD147-dependent binding. Next, we establish that CD147 may induce MMP production not only in primary fibroblast cells, but also in tumor cells themselves, thus facilitating tumor cell invasion. Finally, we demonstrate a key role for glycosylation in determining whether soluble CD147 will be an agonist or antagonist with respect to MMP-1 and MMP-2 production.

Antibodies, Proteins, and Cell Culture.

Anti-CD147 mouse mAbs 8G6 (24) and AAA6 (14) were described previously. Rabbit polyclonal antibody B10 was prepared against purified placental CD147. Mouse monoclonal antibodies to MMP-1 and MMP-2 were from Calbiochem (La Jolla, CA). Recombinant soluble VCAM-1 (25) was provided by Dr. Roy Lobb, Biogen, Inc. Breast carcinoma cell line MDA-435, monkey kidney epithelial line COS-7, and HT1080 fibrosarcoma cells were cultured in DMEM supplemented with 10% fetal bovine serum and penicillin and streptomycin. Primary human dermal fibroblasts were purchased from Clonetics (San Diego, CA) and cultured in fibroblast basal medium Bulletkit fibroblast complete growth medium provided by Clonetics.

Generation of Chimeric Fc Fusion Protein.

The DNA fragments encoding human CD147/EMMPRIN extracellular domain (10) and mouse IgG2a Fc domain (26) were amplified by PCR. Primers used for amplifying CD147 extracellular domains were: (a) domain1-domain2: sense, 5′ GAAAGCTTGCTGCTGGCACAGTCTTC; and antisense, 5′ TCTGGGATCCGCACGCGGAGCGTGATG; (b) domain1: sense, 5′ GAAAGCTTGCTGCTGGCACAGTCTTC; and antisense, TCTGGGATCCGCATCTGGGAGGCCCGTGGAGC; and (c) domain 2: sense, 5′ GAAAGCTTGTGAAGGCCGTGAAGTCGTC; and antisense, 5′ TCTGGGATCCGCACGCGGAGCGTGATG. Primers used for amplifying mouse IgG2a Fc region were: sense, 5′ TCTGGGATCCCAGAGGGCCCACAATC; and antisense, 5′ AGACTCGAGTCATTTACCCGGAGTCCGG. PCR products for CD147 extracellular domains and mouse IgG2a Fc domain were digested with HindIII/BamHI and BamHI/XhoI, respectively. The digested two fragments were then ligated with Vector pSecTag2B (Invitrogen, Carlsbad, CA) predigested with HindIII/XhoI. The CD147 Fc plasmids were transiently transfected into COS-7 cells. After 48- to 72-h transfection, the supernatants were harvested. The fusion proteins were affinity-isolated on protein A-Sepharose according to standard methods.

Purification of Native CD147 and Deglycosylated CD147 from HT1080 Cells.

CD147 was purified from lysates of HT1080 cells by immunoaffinity chromatography using mAb 8G6 conjugated to Sepharose beads. After the column was washed several times with PBS, CD147 was eluted with 0.1n acetic acid and concentrated in Centriprep and Centricon 10 columns (Amicon; Beverly, MA). To obtain deglycosylated EMMPRIN, HT1080 cells were treated with tunicamycin at 5 μg/ml for 24 h before making cell lysates.

Adhesion Assay.

Ninety-six-well adhesion plates were prepared by precoating wells with 1 μg/well of purified goat antimouse Fc antibody (Pierce) overnight at 4°C. Plates were washed three times in PBS, and remaining unbound sites were blocked with PBS/0.4% BSA (fraction V, Sigma) for at least 2 h at room temperature. 50 μl recombinant proteins at 10 μg/ml were then bound, and the plates were washed three times in PBS. COS cells were transfected with full-length CD147 by using Fugene 6 transfection reagent (Roche Molecular Biochemicals, Indianapolis, IN). Transfection efficiency was typically 40–60% as determined by flow cytometry. After transfection (48 h) cells harvested in PBS with 2 mm EDTA were labeled using BCECF-AM [2′,6′-bis(2-carboxyethyl)-5 (6) -carboxyfluorescein acetoxymethyl ester; Molecular Probes, Eugene, OR] for 30 min, pelleted, and suspended in assay buffer (RMPI-1640 and 20 mm Hepes/0.2% BSA) at 3–4 × 104 cells/well. After a 60-min incubation at 37°C, unbound cells were gently washed with prewarmed assay buffer. The bound cells were quantitated using a Cytofluor 2300 fluorescence measurement system (Millipore Corp., Bedford, MA).

MMP Assays.

The activity of MMP-2 and MMP-9 was assayed by Gelatin Zymography. Conditioned cell supernatants were electrophoresed under nonreducing conditions (Laemmli) into 10% gelatin gels. After electrophoresis, gels were washed twice with 2.5% Triton X-100 (37°C, 15 min) and incubated for 16 h with 40 mm Tris-Cl (pH 7.5), 10 mm CaCl2, and 1 μm ZnCl2(27). Gels were stained with 0.5% (w/v) Coomassie Blue in 50% methanol and 10% acetic acid for 1 h, then destained. MMPs were detected as clear bands against a blue background. The level of MMP-2 was also determined using an MMP-2 ELISA kit (Amersham Pharmacia Biotech, Piscataway, NJ). Briefly, MMP-2 is captured by antibody and fully activated upon treatment with p-aminophenylmercuric acetate (APMA). Active MMP-2 then cleaves modified pro-urokinase, which then cleaves its chromogenic peptide substrate, yielding an increase in A405 nm.

Homophilic Adhesion Mediated by CD147.

Because other members of the immunoglobulin superfamily are well established to mediate homophilic adhesion [e.g., neural cell adhesion molecule (28) and carcinoembryonic antigen (29)], we considered that CD147 also might interact in a homophilic manner. To test this idea, we first generated a series of truncated CD147 fusion proteins. Each contained the Fc region of mouse IgG2a fused to the two extracellular immunoglobulin domains of CD147 (D1–2) or to D1 or D2 alone. The fusion proteins were expressed in COS-7 cells, and secreted proteins were purified to homogeneity using protein A beads. Each of the CD147-Fc fusion proteins formed the expected disulfide-linked dimers as determined by SDS-PAGE under reduced and nonreduced conditions (Fig. 1). Sizes of the Fc fusion proteins are consistent with incomplete glycosylation, as seen previously for recombinant CD147 expressed on COS cells (23). For example, the size of reduced CD147(D1-D2)-Fc (Mr ∼55,000) is more than the predicted size of the core protein (Mr ∼46,000), but less than the size expected if fully glycosylated (Mr ∼70,000–75,000).

COS cells transiently expressing full-length CD147 adhered moderately well to immobilized CD147(D1-D2)-Fc fusion protein (Fig. 2,A). In control experiments, CD147-COS cells did not adhere to immobilized VCAM-Fc, and COS cells transfected with empty vector did not adhere to CD147(D1-D2)-Fc fusion protein. These results are consistent with CD147 mediating a homophilic interaction. To map the domain(s) responsible for mediating homotypic adhesion, adhesion of CD147-COS transfectants to a series of CD147-Fc fusion proteins was investigated. As shown (Fig. 2,B), immobilized CD147(D1)-Fc protein again supported adhesion by CD147-COS cells. In addition, CD147(D1)-Fc also supported substantial adhesion, whereas CD147-COS cells did not adhere well to either immobilized CD147(D2)-Fc or VCAM-Fc proteins. These results indicate that CD147 extracellular immunoglobulin domain 1 is most critical for CD147 homophilic interaction. This conclusion was confirmed in adhesion blockade experiments. As indicated in Fig. 3, the adhesion of CD147 COS transfectants to CD147(D1-D2)-Fc fusion protein was substantially inhibited by soluble CD147(D1-D2)-Fc protein and was inhibited almost as much by soluble CD147(D1)-Fc protein. In contrast, VCAM-Fc gave no inhibition, and CD147(D2)-Fc showed only slight inhibition. Thus again, CD147(D1) plays a major role, whereas CD147(D2) plays only a minor role during homophilic adhesion. The anti-CD147 mAb 8G6 and polyclonal rabbit antibody B10 also inhibited CD147-dependent COS-cell adhesion. The 8G6 epitope and most of the B10 recognition epitopes reside on D1 of CD147, as determined by flow cytometry analysis of cell surface-expressed CD147 containing or lacking D1 (not shown).

A Role for CD147 during MMP-2 Production.

In contrast with previous studies of CD147-dependent MMP production that have relied on tumor cell/fibroblast coculture assays (9, 22, 23), here we demonstrate CD147-dependent MMP production by a tumor cell line in the absence of cocultured fibroblasts. Production of MMP-2 by MDA-435 breast carcinoma cells was substantially inhibited by two different anti-CD147 mAbs (8G6 and AAA6) as well as by the polyclonal antibody B10 (Fig. 4, A and B). Importantly, MMP-2 production in Fig. 4,A was essentially abolished not only by mAb 8G6, which engages cell-bound CD147, but also by recombinant soluble CD147(D1-D2)-Fc itself, which presumably also engages cell-bound CD147 in a homophilic fashion. In both cases, inhibition was highly specific, inasmuch as MMP-9 production was not altered (Fig. 4 A).

A Role for CD147 during MDA-435 Cell Invasion.

To assess the possible consequences of CD147-dependent MMP-2 production, an in vitro MDA-435 cell invasion experiment was conducted using Matrigel. Previously, production of MMP-2 has been shown to correlate strongly with invasion of Matrigel (30, 31). MDA-435 cells were allowed to invade for 24 h through a filter precoated with Matrigel, and then cells appearing on the lower surface of the filter were counted under a microscope. As indicated in Fig. 5, both anti-CD147 mAbs 8G6 and CD147(D1-D2)-Fc protein each inhibited MDA-435 cell invasion by ∼80% compared with invasion seen in the presence of control antibody or control fusion protein (VCAM-Fc). Importantly, an anti-MMP2 antibody inhibited cell invasion to a similar extent (∼80%), thus suggesting that the invasion inhibited by mAb 8G6 and CD147(D1-D2)-Fc is MMP-2-dependent.

Can Purified CD147 Directly Promote Tumor Cell MMP-2 Production?

Results shown in Fig. 4 and 5 suggest that MMP-2 production and MMP-2-dependent invasion may be dependent on a homophilic CD147 interaction between CD147 on adjacent tumor cells. To determine whether CD147 may indeed directly stimulate tumor cell MMP production, first we purified mature, glycosylated CD147 (Mr ∼54,000) from the HT1080 fibrosarcoma tumor cell line (Fig. 6, right Lane). Next, we confirmed that this material was active on dermal fibroblasts, as expected from previous reports (23). Indeed, our purified CD147 (at 1 or 3 μg/ml) stimulated MMP-1 production to nearly the same extent as the positive control agent, interleukin 1 (Fig. 7, A and B; Lanes 3 and 4). Finally, we demonstrated that purified CD147 (at 3 μg/ml) markedly stimulated MMP-2 production by MDA-435 cells (Fig. 7 C, Lane 3). In the same experiment, MMP-9 production was relatively unaffected.

What Distinguishes a CD147 Agonist from an Antagonist?

Why did recombinant soluble CD147 fusion protein antagonize MMP production and invasion in Figs. 4 and 5, whereas purified CD147 was an agonist in Fig. 7? Incomplete glycosylation of CD147-Fc proteins (Fig. 1) suggested that glycosylation may be required for agonist activity. To test this idea, CD147 was purified in a largely deglycosylated form (Mr ∼28,000) from tunicamycin-treated HT1080 cells (Fig. 6, left Lane). When added to dermal fibroblasts, the deglycosylated protein (either at 3 μg/ml or even 30 μg/ml) failed to induce MMP-1 production by dermal fibroblasts (Fig. 7, A and B; Lanes 5 and 6). Furthermore, the induction of MMP-1 by 3 μg/ml of glycosylated CD147 was abolished either partially or fully upon coincubation with either 3 μg/ml or 30 μg/ml of deglycosylated CD147, respectively (Fig. 7, A and B; Lanes 7 and 8). Deglycosylated CD147 also failed to induce MMP-2 production in MDA-453 cells (Fig. 7,C, middle Lane). However, inhibition of MMP-2 production was not observed (Fig. 7,C), in contrast with the strong inhibitory effect of unglycosylated CD147-Fc on MMP-2 production seen in Fig. 4,A. The difference in inhibition may be explained by the 20 μg/ml of bivalent CD147-Fc reagent used in Fig. 4,A, compared with the 3 μg/ml of monovalent purified material used in Fig. 7 C.

Previous studies have demonstrated that the CD147/EMMPRIN molecule on tumor cells may induce production of various MMPs on peritumor stromal fibroblasts or on cocultured dermal fibroblasts (9, 22, 23). These results implied that a CD147 counterreceptor might exist on the fibroblast cell surface, but such a counterreceptor had not been identified. Here, we show that immobilized CD147-Fc fusion protein can support specific adhesion by CD147-transfected COS cells, thus providing evidence that CD147 may be a counterreceptor for itself. Of the two extracellular immunoglobulin domains of CD147, D1 was clearly the one most important for homophilic adhesion. Thus, CD147 resembles many other immunoglobulin-superfamily proteins that use D1 to support homophilic or heterophilic adhesion.

The same agents [CD147(D1-D2)-Fc fusion protein and mAb 8G6] that strongly inhibited homophilic adhesion also strongly inhibited MMP-2 production by MDA-435 cells. Thus, CD147-dependent MMP-2 production is likely a consequence of homophilic CD147 interactions. However, at present we cannot exclude a possible role for additional counterreceptor interactions between CD147 and unidentified cell surface molecules. The availability of CD147 null cells (16) should allow for future resolution of this issue.

The anti-CD147 mAb AAA6 was shown previously to bind selectively to multimerized CD147 on cell surfaces (32). Our result showing inhibition of MMP-2 production by mAb AAA6 thus indicates that multimerized CD147 may induce MMP-2 production. If the affinity of CD147 for itself (or for other counterreceptors) is weak, as is seen typically for immunoglobulin superfamily proteins, multimerization could provide a mechanism for increasing the overall avidity of CD147 interactions. The mechanism by which CD147 may regulate MMP-2 (and MMP-1) production is largely unclear, although a recent study has suggested that the mitogen-activated protein kinase p38 pathway may be involved during MMP induction in dermal fibroblasts (33).

Previous studies of MMP production during coculture assays were based on the assumption that tumor cell CD147 induces MMP production by surrounding fibroblasts. We now expand the role of CD147 to include induction of MMP-2 by tumor cells themselves in the absence of cocultured fibroblasts. Thus homophilic CD147 adhesion may occur in both heterotypic (tumor cell-fibroblast) and homotypic (tumor cell-tumor cell) circumstances. The relatively high levels of CD147 on nearly all tumor cell lines suggests a potentially widespread role as an inducer of MMPs by tumor cells. CD147 is known to promote MMP-1, MMP-2, and MMP-3 production by fibroblasts (9, 22, 23). It remains to be determined whether CD147 may induce tumor cell production of other MMPs besides MMP-2.

Our tunicamycin treatment results confirm previous endoglycosidase F experiments indicating that N-linked oligosaccharide chains may contribute ∼50% of the size of mature CD147 (14). Also, earlier studies showing that CD147 maturation is critical for MMP induction capability (23) are consistent with our results demonstrating that N-linked glycosylation of CD147 is essential for MMP-1 and MMP-2 induction. However, glycosylation was not needed for homophilic adhesion or for the inhibition of MMP-2 induction. Thus, the ability to participate in homophilic adhesion alone is not sufficient for MMP induction. It remains to be determined how glycosylation may convert CD147 from an antagonist to an agonist. The CD147 protein may be shed by tumor cells and has been found in the urine of bladder cancer patients (8, 19). Thus, depending on the glycosylation state of the shed CD147, potentially it could either promote or inhibit MMP production in vivo. Furthermore, our results suggest that CD147 shed from tumor cells could promote MMP production in either paracrine or autocrine fashion.

In our studies, the apparent consequence of CD147-dependent MMP-2 induction was enhanced MMP-2-dependent tumor cell invasion. In vivo, MMP-2/gelatinase A is a key enzyme that degrades the extracellular matrix and facilitates tumor invasion and metastasis (1). Consistent with the critical role of MMP-2 and other MMPs during cancer invasion and metastasis, inhibitors of MMPs have exhibited potent anti-angiogenic and antitumor activity (34, 35, 36). Because of its pivotal role in the regulation of MMP production, we suggest that CD147 could be a novel target for anticancer therapies. In this regard, CD147 is more highly expressed on human carcinoma cells than on normal cells, and its expression correlates with MMP expression level and cancer invasive potential (8, 19, 20, 21). Perturbation of CD147 function by agents that mimic the effects of anti-CD147 antibodies, CD147(D1-D2) Fc fusion protein or deglycosylated native CD147 may have potential therapeutic value in the prevention of cancer invasion and metastasis.

Fig. 1.

Characterization of CD147-Fc fusion proteins. CD147 Fc plasmids were transfected into COS cells using lipofectin, and after 48 h, cell supernatants were harvested and loaded onto protein A beads. CD147 Fc fusion proteins were then eluted with 0.1 m glycine (pH 3.0) and resolved on 12% SDS-PAGE under either reducing or nonreducing conditions. Proteins were stained with Coomassie Blue.

Fig. 1.

Characterization of CD147-Fc fusion proteins. CD147 Fc plasmids were transfected into COS cells using lipofectin, and after 48 h, cell supernatants were harvested and loaded onto protein A beads. CD147 Fc fusion proteins were then eluted with 0.1 m glycine (pH 3.0) and resolved on 12% SDS-PAGE under either reducing or nonreducing conditions. Proteins were stained with Coomassie Blue.

Close modal
Fig. 2.

CD147-dependent cell adhesion assay. A, COS cells transfected with full-length CD147 or vector alone were allowed to adhere for 1 h to immobilized CD147(D1-D2) Fc or to VCAM Fc fusion protein. Results are expressed as the mean percentage of total input cells adhering ± SE (n = 6). B, COS cells transfected with full-length CD147 were allowed to adhere for 1 h to immobilized CD147(D1-D2) Fc, CD147(D1) Fc, CD147(D2) Fc, and VCAM Fc fusion proteins. Results are expressed as the mean percentage of CD147-positive cells adhering ± SE (n = 6), with ∼50% of COS cells expressing CD147.

Fig. 2.

CD147-dependent cell adhesion assay. A, COS cells transfected with full-length CD147 or vector alone were allowed to adhere for 1 h to immobilized CD147(D1-D2) Fc or to VCAM Fc fusion protein. Results are expressed as the mean percentage of total input cells adhering ± SE (n = 6). B, COS cells transfected with full-length CD147 were allowed to adhere for 1 h to immobilized CD147(D1-D2) Fc, CD147(D1) Fc, CD147(D2) Fc, and VCAM Fc fusion proteins. Results are expressed as the mean percentage of CD147-positive cells adhering ± SE (n = 6), with ∼50% of COS cells expressing CD147.

Close modal
Fig. 3.

Mapping the CD147 homophilic binding site by adhesion blockade. COS cells transfected with full-length CD147 were allowed to adhere to immobilized CD147(D1-D2) Fc fusion protein for 1 h. Cells were preincubated at room temperature for 10 min with either the indicated antibodies (at 20 μg/ml) or soluble fusion proteins (at 200 μg/ml), and these agents were also included during the adhesion assay.

Fig. 3.

Mapping the CD147 homophilic binding site by adhesion blockade. COS cells transfected with full-length CD147 were allowed to adhere to immobilized CD147(D1-D2) Fc fusion protein for 1 h. Cells were preincubated at room temperature for 10 min with either the indicated antibodies (at 20 μg/ml) or soluble fusion proteins (at 200 μg/ml), and these agents were also included during the adhesion assay.

Close modal
Fig. 4.

Effects of CD147 antibodies and CD147(D1-D2) Fc fusion protein on MMP-2 and MMP-9 production by MDA-435 carcinoma cells. A, MDA-435 cells (80% confluent, in a 24-well dish) were incubated with anti-CD147 mAb 8G6, CD147(D1-D2) Fc fusion protein, or control mouse IgG2a for 48 h (each at 20 μg/ml), and then cell supernatants were harvested for MMP2 and MMP9 assay by Zymography. B, MDA-435 cells were incubated with antibodies (each at 20 μg/ml) including control mouse IgG2a, anti-CD147 mAb 8G6, anti-CD147 mAb AAA6, or anti-CD147 polyclonal antibody B10 for 48 h, and then cell supernatants were harvested and assayed for MMP2 levels by ELISA. Note: 100% of MMP-2 corresponds to A405 nm = 0.57.

Fig. 4.

Effects of CD147 antibodies and CD147(D1-D2) Fc fusion protein on MMP-2 and MMP-9 production by MDA-435 carcinoma cells. A, MDA-435 cells (80% confluent, in a 24-well dish) were incubated with anti-CD147 mAb 8G6, CD147(D1-D2) Fc fusion protein, or control mouse IgG2a for 48 h (each at 20 μg/ml), and then cell supernatants were harvested for MMP2 and MMP9 assay by Zymography. B, MDA-435 cells were incubated with antibodies (each at 20 μg/ml) including control mouse IgG2a, anti-CD147 mAb 8G6, anti-CD147 mAb AAA6, or anti-CD147 polyclonal antibody B10 for 48 h, and then cell supernatants were harvested and assayed for MMP2 levels by ELISA. Note: 100% of MMP-2 corresponds to A405 nm = 0.57.

Close modal
Fig. 5.

Effects of antibodies and fusion proteins on MDA-435 cell invasion through Matrigel. MDA-435 cells (1 × 104/well) were placed onto the upper surfaces of 24-well Matrigel invasion chambers (Becton Dickinson Labware, Bedford, MA) in the presence of the indicated fusion proteins or antibodies (each at 20 μg/ml). Then cells were allowed to invade (at 37°C for 24 h) through the filters precoated with Matrigel. Upper surfaces of filters were then wiped clean, filters were fixed and stained, and the cells that migrated to the underside were counted using a microscope to examine the entire filter. Results are expressed as the mean ± SE of the percentage relative to the control (n = 3). One hundred percent invasion represents ∼1250 migrated cells.

Fig. 5.

Effects of antibodies and fusion proteins on MDA-435 cell invasion through Matrigel. MDA-435 cells (1 × 104/well) were placed onto the upper surfaces of 24-well Matrigel invasion chambers (Becton Dickinson Labware, Bedford, MA) in the presence of the indicated fusion proteins or antibodies (each at 20 μg/ml). Then cells were allowed to invade (at 37°C for 24 h) through the filters precoated with Matrigel. Upper surfaces of filters were then wiped clean, filters were fixed and stained, and the cells that migrated to the underside were counted using a microscope to examine the entire filter. Results are expressed as the mean ± SE of the percentage relative to the control (n = 3). One hundred percent invasion represents ∼1250 migrated cells.

Close modal
Fig. 6.

Effect of tunicamycin on CD147 glycosylation. HT1080 cells were treated with tunicamycin at 5 μg/ml for 24 h. Cell surface proteins were then biotinylated, and CD147 was immunoprecipitated and then visualized by Western blot with horseradish peroxidase-conjugated avidin.

Fig. 6.

Effect of tunicamycin on CD147 glycosylation. HT1080 cells were treated with tunicamycin at 5 μg/ml for 24 h. Cell surface proteins were then biotinylated, and CD147 was immunoprecipitated and then visualized by Western blot with horseradish peroxidase-conjugated avidin.

Close modal
Fig. 7.

Effect of CD147 glycosylation of CD147-dependent MMP induction. A, dermal fibroblasts (90% confluent in a 24-well dish) were cultured overnight in fibroblast basal medium containing 0.5% fetal bovine serum and then incubated for 48 h with interleukin 1β (10 ng/ml, Lane 1), VCAM-Fc (30 μg/ml, Lane 2), purified native CD147 (1 μg/ml, Lane 3; 3 μg/ml, Lane 4) or deglycosylated CD147 (3 μg/ml, Lane 5; 30 μg/ml, Lane 6), or native CD147 (3 μg/ml) plus deglycosylated CD147 (3 μg/ml, Lane 7; or 30 μg/ml, Lane 8). Supernatants were then harvested, resolved by 12% SDS-PAGE, electroblotted to a nitrocellulose membrane, and incubated with anti-MMP-1 mAb. The immunoreactive bands were detected with horseradish peroxidase-conjugated antimouse IgG and chemiluminescence reagent (New England Nuclear Life Science, Boston, MA). B, protein levels for MMP-1 were quantitated by densitometry (MultiImage device; Alpha Innotech Co.) of autoradiograms such as shown in part A. Results represent the mean ± SE of three separate Western blot assays. C, MDA-435 cells (80% confluent, in a 24-well dish) were incubated with control VCAM Fc protein, deglycosylated CD147, or native purified CD147 (each at 3μg /ml) for 48 h. Then supernatants were harvested and MMP2 and MMP9 activities were assayed by Zymography in the same experiment.

Fig. 7.

Effect of CD147 glycosylation of CD147-dependent MMP induction. A, dermal fibroblasts (90% confluent in a 24-well dish) were cultured overnight in fibroblast basal medium containing 0.5% fetal bovine serum and then incubated for 48 h with interleukin 1β (10 ng/ml, Lane 1), VCAM-Fc (30 μg/ml, Lane 2), purified native CD147 (1 μg/ml, Lane 3; 3 μg/ml, Lane 4) or deglycosylated CD147 (3 μg/ml, Lane 5; 30 μg/ml, Lane 6), or native CD147 (3 μg/ml) plus deglycosylated CD147 (3 μg/ml, Lane 7; or 30 μg/ml, Lane 8). Supernatants were then harvested, resolved by 12% SDS-PAGE, electroblotted to a nitrocellulose membrane, and incubated with anti-MMP-1 mAb. The immunoreactive bands were detected with horseradish peroxidase-conjugated antimouse IgG and chemiluminescence reagent (New England Nuclear Life Science, Boston, MA). B, protein levels for MMP-1 were quantitated by densitometry (MultiImage device; Alpha Innotech Co.) of autoradiograms such as shown in part A. Results represent the mean ± SE of three separate Western blot assays. C, MDA-435 cells (80% confluent, in a 24-well dish) were incubated with control VCAM Fc protein, deglycosylated CD147, or native purified CD147 (each at 3μg /ml) for 48 h. Then supernatants were harvested and MMP2 and MMP9 activities were assayed by Zymography in the same experiment.

Close modal

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

1

Supported by a grant from the Dana Farber Cancer Institute/Novartis Drug Discovery Program and by NIH Grant CA42368.

3

The abbreviations used are: MMP, matrix metalloproteinase; EMMPRIN, extracellular matrix metalloproteinase inducer; MMP-1, interstitial collagenase; MMP-2, gelatinase A; mAb, monoclonal antibody; D1, domain 1; D2, domain 2.

1
Stetler-Stevenson W. G., Aznavoorian S., Liotta L. A. Tumor cell interactions with the extracellular matrix during invasion and metastasis.
Ann. Rev. Cell Biol.
,
9
:
541
-573,  
1993
.
2
Werb Z. ECM and cell surface proteolysis: regulating cellular ecology.
Cell
,
91
:
439
-442,  
1997
.
3
Ralfkiaer P. C., Huhtala P., Hurskainen T., Dano K., Tryggvason K. Localization of messenger RNA for Mr 72,000 and 92,000 type IV collagenases in human skin cancers by in situ hybridization.
Cancer Res.
,
52
:
1336
-1341,  
1992
.
4
Gray S. T., Wilkins R. J., Yun K. Interstitial collagenase gene expression in oral squamous cell carcinoma.
Am. J. Pathol.
,
141
:
301
-306,  
1992
.
5
Poulsom R., Pignatelli M., Stetler-Stevenson W. G., Liotta L. A., Wright P. A., Jeffery R. E., Longcroft J. M., Rogers L., Stamp G. W. Stromal expression of 72 kDa type IV collagenase (MMP-2) and TIMP-2 mRNAs in colorectal neoplasia.
Am. J. Pathol.
,
141
:
389
-396,  
1992
.
6
Sasaguri Y., Komiya S., Sugama K., Suzuki K., Inoue A., Morimatsu M., Nagase H. Production of matrix metalloproteinases 2 and 3 (stromelysin) by stromal cells of giant cell tumor of bone.
Am. J. Pathol.
,
141
:
611
-621,  
1992
.
7
MacDougall J. R., Matrisian L. M. Contributions of tumor and stromal matrix metalloproteinases to tumor progression, invasion and metastasis.
Cancer Metastasis Rev.
,
14
:
351
-362,  
1995
.
8
Ellis S. M., Nabeshima K., Biswas C. Monoclonal antibody preparation and purification of a tumor cell collagenase-stimulatory factor.
Cancer Res.
,
49
:
3385
-3391,  
1989
.
9
Kataoka H., DeCastro R., Zucker S., Biswas C. Tumor cell-derived collagenase-stimulatory factor increases expression of interstitial collagenase, stromelysin, and 72-kDa gelatinase.
Cancer Res.
,
53
:
3154
-3158,  
1993
.
10
Biswas C., Zhang Y., DeCastro R., Guo H., Nakamura T., Kataoka H., Nabeshima K. The human tumor cell-derived collagenase stimulatory factor (renamed EMMPRIN) is a member of the immunoglobulin superfamily.
Cancer Res.
,
55
:
434
-439,  
1995
.
11
Miyauchi T., Kanekura T., Yamaoka A., Ozawa M., Miyazawa S., Muramatsu T. Basigin, a new, broadly distributed member of the immunoglobulin superfamily, has strong homology with both the immunoglobulin V domain and the β-chain of major histocompatibility complex class II antigen.
J. Biochem.
,
107
:
316
-323,  
1990
.
12
Fossum S., Mallett S., Barclay A. N. The MRC OX-47 antigen is a member of the immunoglobulin superfamily with an unusual transmembrane sequence.
Eur. J. Immunol.
,
21
:
671
-679,  
1991
.
13
Schlosshauer B., Bauch H., Frank R. Neurothelin: amino acid sequence, cell surface dynamics and actin colocalization.
Eur. J. Cell Biol.
,
68
:
159
-166,  
1995
.
14
Kasinrerk W., Fiebiger E., Stefanova I., Baumruker T., Knapp W., Stockinger H. Human leukocyte activation antigen M6, a member of the Ig superfamily, is the species homologue of rat OX-47, mouse basigin, and chicken HT7 molecule.
J. Immunol.
,
149
:
847
-854,  
1992
.
15
Seulberger H., Lottspeich F., Risau W. The inducible blood-brain barrier specific molecule HT7 is a novel immunoglobulin-like cell surface glycoprotein.
EMBO J.
,
9
:
2151
-2158,  
1990
.
16
Igakura T., Kadomatsu K., Kaname T., Muramatsu H., Fan Q. W., Miyauchi T., Toyama Y., Kuno N., Yuasa S., Takahashi M., Senda T., Taguchi O., Yamamura K., Arimura K., Muramatsu T. A null mutation in basigin, an immunoglobulin superfamily member, indicates its important roles in peri-implantation development and spermatogenesis.
Dev. Biol.
,
194
:
152
-165,  
1998
.
17
Kuno N., Kadomatsu K., Fan Q. W., Hagihara M., Senda T., Mizutani S., Muramatsu T. Female sterility in mice lacking the basigin gene, which encodes a transmembrane glycoprotein belonging to the immunoglobulin superfamily.
FEBS Lett.
,
425
:
191
-194,  
1998
.
18
Igakura T., Kadomatsu K., Taguchi O., Muramatsu H., Kaname T., Miyauchi T., Yamamura K., Arimura K., Muramatsu T. Roles of basigin, a member of the immunoglobulin superfamily, in behavior as to an irritating odor, lymphocyte response, and blood-brain barrier.
Biochem. Biophys. Res. Commun.
,
224
:
33
-36,  
1996
.
19
Muraoka K., Nabeshima K., Murayama T., Biswas C., Koono M. Enhanced expression of a tumor-cell-derived collagenase-stimulatory factor in urothelial carcinoma: its usefulness as a tumor marker for bladder cancers.
Int. J. Cancer
,
55
:
19
-26,  
1993
.
20
Polette M., Gilles C., Marchand V., Lorenzato M., Toole B., Tournier J. M., Zucker S., Birembaut P. Tumor collagenase stimulatory factor (TCSF) expression and localization in human lung and breast cancers.
J. Histochem. Cytochem.
,
45
:
703
-709,  
1997
.
21
van den Oord J. J., Paemen L., Opdenakker G., de Wolf-Peeters C. Expression of gelatinase B and the extracellular matrix metalloproteinase inducer EMMPRIN in benign and malignant pigment cell lesions of the skin.
Am. J. Pathol.
,
151
:
665
-670,  
1997
.
22
Bordador L. C., Li X., Toole B., Chen B., Regezi J., Zardi L., Hu Y., Ramos D. M. Expression of emmprin by oral squamous cell carcinoma.
Int. J. Cancer
,
85
:
347
-352,  
2000
.
23
Guo H., Zucker S., Gordon M. K., Toole B. P., Biswas C. Stimulation of matrix metalloproteinase production by recombinant extracellular matrix metalloproteinase inducer from transfected Chinese hamster ovary cells.
J. Biol. Chem.
,
272
:
24
-27,  
1997
.
24
Berditchevski F., Chang S., Bodorova J., Hemler M. E. Generation of monoclonal antibodies to integrin-associated proteins: evidence that α3β1 complexes with EMMPRIN/basigin/OX47/M6.
J. Biol. Chem.
,
272
:
29174
-29180,  
1997
.
25
Lobb R. R., Chi-Rosso G., Leone D. R., Rosa M. D., Newman B. M., Luhowskyj S., Osborn L., Schiffer S. G., Benjamin C. D., Dougas I. G., Hession C., Chow E. P. Expression and functional characterization of a soluble form of vascular cell adhesion molecule 1 (VCAM1).
Biochem. Biophys. Res. Commun.
,
178
:
1498
-1504,  
1991
.
26
Ollo R., Auffray C., Morchamps C., Rougeon F. Comparison of mouse immunoglobulin γ 2a and γ 2b chain genes suggests that exons can be exchanged between genes in a multigenic family.
Proc. Natl. Acad. Sci. USA
,
78
:
2442
-2446,  
1981
.
27
Crawford H. C., Matrisian L. M. Tumor and stromal expression of matrix metalloproteinases and their role in tumor progression.
Invasion Metastasis
,
14
:
234
-245,  
1994
.
28
Rao Y., Wu X. F., Gariepy J., Rutishauser U., Siu C. H. Identification of a peptide sequence involved in homophilic binding in the neural cell adhesion molecule NCAM.
J. Cell Biol.
,
118
:
937
-949,  
1992
.
29
Zhou H., Fuks A., Alcaraz G., Bolling T. J., Stanners C. P. Homophilic adhesion between Ig superfamily carcinoembryonic antigen molecules involves double reciprocal bonds.
J. Cell Biol.
,
122
:
951
-960,  
1993
.
30
Kubota S., Ito H., Ishibashi Y., Seyama Y. Anti-α3 integrin antibody induces the activated form of matrix metalloprotease-2 (MMP-2) with concomitant stimulation of invasion through Matrigel by human rhabdomyosarcoma cells.
Int. J. Cancer
,
70
:
106
-111,  
1997
.
31
Miyake H., Hara I., Yoshimura K., Eto H., Arakawa S., Wada S., Chihara K., Kamidono S. Introduction of basic fibroblast growth factor gene into mouse renal cell carcinoma cell line enhances its metastatic potential.
Cancer Res.
,
56
:
2440
-2445,  
1996
.
32
Koch C., Staffler G., Huttinger R., Hilgert I., Prager E., Cerny J., Steinlein P., Majdic O., Horejsi V., Stockinger H. T cell activation-associated epitopes of CD147 in regulation of the T cell response, and their definition by antibody affinity and antigen density.
Int. Immunol.
,
11
:
777
-786,  
1999
.
33
Lim M., Martinez T., Jablons D., Cameron R., Guo H., Toole B., Li J. D., Basbaum C. Tumor-derived EMMPRIN (extracellular matrix metalloproteinase inducer) stimulates collagenase transcription through MAPK p38.
FEBS Lett.
,
441
:
88
-92,  
1998
.
34
Koivunen E., Arap W., Valtanen H., Rainisalo A., Medina O. P., Heikkila P., Kantor C., Gahmberg C. G., Salo T., Konttinen Y. T., Sorsa T., Ruoslahti E., Pasqualini R. Tumor targeting with a selective gelatinase inhibitor.
Nat. Biotechnol.
,
17
:
768
-774,  
1999
.
35
Khokha R. Suppression of the tumorigenic and metastatic abilities of murine B16–F10 melanoma cells in vivo by the overexpression of the tissue inhibitor of the metalloproteinases-1.
J. Natl. Cancer Inst.
,
86
:
299
-304,  
1994
.
36
Eccles S. A., Box G. M., Court W. J., Bone E. A., Thomas W., Brown P. D. Control of lymphatic and hematogenous metastasis of a rat mammary carcinoma by the matrix metalloproteinase inhibitor batimastat (BB-94).
Cancer Res.
,
56
:
2815
-2822,  
1996
.