Secreted and Membrane-Bound Isoforms of Protease ADAM9 Have Opposing Effects on Breast Cancer Cell Migration

Published OnlineFirst August 24, 2010; DOI: 10.1158/0008-5472.CAN-09-4231

Tumor and Stem Cell Biology Cancer Research Secreted and Membrane-Bound Isoforms of Protease ADAM9 Have Opposing Effects on Breast Cancer Cell Migration

Jessica L. Fry and Alex Toker

Abstract Tumor cell migration is mediated by cell-autonomous signaling mechanisms as well as paracrine and auto- crine factors secreted by activated stromal cells in the tumor microenvironment. Like other members of the ADAM (a disintegrin and metalloproteinase) family, the integrin-binding metalloproteinase ADAM9 modulates cell-cell and cell-matrix interactions as well as ectodomain shedding of cell surface receptors and ligands, thereby modifying intracellular and extracellular signaling. ADAM9 transcripts are alternatively spliced to express a transmembrane protein (ADAM9-L) and a secreted variant (ADAM9-S). In this study, we show that ADAM9-S promotes breast cancer cell migration in a manner requiring its metalloproteinase activity, whereas ADAM9-L suppresses cell migration independent of its metalloproteinase activity. Suppression of migration by ADAM9-L requires a functional disintegrin domain and integrin binding. Expression analysis revealed that both ADAM9 isoforms are expressed in breast cancer cell lines and tissues. Therefore, relative levels of membrane-tethered and secreted variants of ADAM9 are a key determinant in manifestation of aggressive migratory phenotypes associated with breast cancer progression. Cancer Res; 70(20); 8187–98. ©2010 AACR.

Introduction growth factor (5, 6). MMPs proteolytically cleave and release the ectodomains of multiple signaling molecules from the cell One of the hallmarks of advanced breast cancer is the abil- membrane in a process known as membrane shedding (7). Shed- ity of the tumor cells to lose their epithelial phenotype, disen- ding of substrates such as transforming growth factor-β (8) and gage from neighboring cells, degrade the basement membrane, tumor necrosis factor-α (9) activates signaling pathways that and invade surrounding tissues, ultimately metastasizing to dis- promote cell migration and survival. The ADAM (a disintegrin tant organs. The interactions between the tumor and its stromal and metalloproteinase) family of proteases has functional microenvironment are a key determinant in breast cancer pro- similarity to the MMPs in their zinc-binding metalloprotei- gression from carcinoma in situ to advanced invasive and nase domains, and is also referred to as the metalloproteinase, metastatic carcinoma. Tumor cells proteolyze the basement disintegrin, cysteine-rich (MDC) family. ADAMs control base- membrane and stroma to invade the vasculature, and also alter ment membrane proteolysis and shedding of proteins from stromal signals to stimulate angiogenesis and alternately re- the cell membrane. Comprising 40 isoforms, the ADAM family lease and tether to the extracellular matrix (ECM) to allow for of proteases also contains an integrin-binding disintegrin do- efficient migration. Tumor cells also release growth factors and main. ADAMs function in cell adhesion through their cysteine- chemokines that alter the stroma, inducing inflammation, rich domains that bind to syndecans (10) and fibronectin (11). angiogenesis, and mechanisms of tissue repair (1, 2). The Src homology-3 (SH3)–binding domain in the cytoplasmic Proteases play a major signaling role at the tumor-stromal regions of some ADAMs mediates signaling by activation of boundary. Enzymes comprising the matrix metalloproteinase Src and Grb proteins (12). Recent studies point to a functional (MMP) family proteolyze basement membrane and function importance for some ADAM family members in cancer. in promigratory signaling mechanisms. Proteolysis by MMPs ADAM12 is expressed in carcinoma and promotes breast can- exposes cryptic sites in ECM constituents such as laminin-5 cer progression by inducing the apoptosis of surrounding stro- and collagen IV that in turn promote tumor cell migration mal cells (13). Consequently, ADAM12 protein levels correlate (3, 4). Proteolysis of the basement membrane also releases with advanced breast cancer (14, 15). In contrast, the disintegrin ECM-boundgrowthfactorssuchasinsulinandfibroblast domain of ADAM15 inhibits angiogenesis and tumor growth (16). Moreover, ADAM15 expression decreases integrin-mediated ovarian cancer cell adhesion and motility (17). Authors' Affiliation: Department of Pathology, Beth Israel Deaconess ADAM9, also known as MDC9 (metalloproteinase/disintegrin/ Medical Center, Harvard Medical School, Boston, Massachusetts cysteine-rich protein 9), MCMP (myeloma cell metalloprotei- Corresponding Author: Alex Toker, Department of Pathology, Beth γ Israel Deaconess Medical Center, 330 Brookline Avenue, EC/CLS- nase), and meltrin- , comprises a subgroup of ADAMs that also 633A, Boston, MA 02215. Phone: 617-735-2482; Fax: 617-735-2480; contains ADAM12 and ADAM15 (18). The ADAM9 metallo- E-mail: [email protected]. proteinase activity cleaves heparin-binding epidermal growth doi: 10.1158/0008-5472.CAN-09-4231 factor (EGF; ref. 19), which binds activates EGF receptor to ©2010 American Association for Cancer Research. promote tumor growth and angiogenesis (20). ADAM9 also

www.aacrjournals.org 8187

Fry and Toker

cleaves Delta-like 1, a Notch ligand (21), as well as ADAM10 GAATGCGGC-3′ and 5′-GCCGCATTCACACGTCTCTC- (22, 23). The ADAM9 disintegrin domain is a ligand for specific CAGGGTCC-3′. Deletion mutations in mADAM9L.Δcyt.myc integrin heterodimers, including multiple β1 (24, 25), α6β4 and mADAM9L.Δdis.myc were derived as follows: To delete α β Kpn (26), and v 5 integrins (27, 28). the cytoplasmic domain, a 1 restriction site was inserted Recent studies have revealed that ADAM9 message is up- into pcDNA3-mADAM9-L 5′ of the cytoplasmic region with regulated in breast tumors as well as in breast cancer cell primers 5′-GGCTAACTAGAGAACCCACTGGTACCTG‐ lines (29, 30). The chromosomal region of the ADAM9 gene GCTTATCG-3′ and 5′-CGATAAGCCAGGTACCAGTG‐ (8p11-12) is amplified in several breast cancers and cell lines, GGTTCTCTAGTTAGCC-3′. BamHI from the vector backbone and ADAM9 is overexpressed in cell lines encompassing the and the inserted Kpn1 restriction sites were used to isolate the luminal, basal A, and basal B gene expression clusters (31). mADAM9L.Δcyt sequence, which was cloned into digested The ADAM9 transcript is alternatively spliced into membrane- pcDNA4A/TO/myc/his vector in frame with the myc tag. To de- bound and secreted isoforms. Alternative splicing of exon 12 lete the disintegrin domain, Kpn1 sites were inserted 5′ and 3′ of in the ADAM9 message leads to a shorter ADAM9-S (ADAM9- the disintegrin domain in pcDNA3-mADAM9L.myc using site- secreted) transcript that contains eight unique amino acids directed mutagenesis using the following sequences: 5′ Kpn1, not present in ADAM9-L (ADAM9-long). ADAM9-S also lacks 5′-GTGGAGCAAAGAGCGGTACCATGAATTCAGGAGC-3′ and the transmembrane and cytoplasmic domains and is a secreted 5′-GCTCCTGAATTCATGGTACCGCTCTTTGCTCCAC-3′; protein (26). ADAM9 is implicated in cancer cell phenotypes, as 3′ Kpn1, 5′-GAAGGAGTGTGAGGGTACCCCATGCTGTGAAG- secretion of ADAM9-S promotes invasion of breast cancer cells GAAG-3′ and 5′-CTTCCTTCACAGCATGGGGTACCCTCA- in Matrigel assays (26). Furthermore, adhesion of keratinocyte CACTCCTTC-3′. Kpn1 digestion was used to isolate and cells to recombinant ADAM9 disintegrin/cysteine-rich domain discard the disintegrin sequence, and the vector backbone increases cell migration, and overexpression of ADAM9 also Kpn1 sites were ligated. For RNAi, ADAM9 sequences from increases pro-MMP9 expression (25). However, the functional the Mission small interfering RNA (siRNA) project (Sigma- role of ADAM9-L and ADAM9-S in modulating cell migration Aldrich) were cloned into the pLKO.1 vector directing expression and invasion in breast cancer cells has not been evaluated. of short hairpin RNA (shRNA; ref. 32). The following sequences Here, we use specific ADAM9-L and ADAM9-S antibodies were used: ADAM9shRNA.1, 5′-CCGGCCCAGAGAAGTT- to detect expression of both variants in breast cancer cell CCTATATATCTCGAGATATATAGGAACTTCT‐ lines and breast cancer tissues. Using RNA interference CTGGGTTTTTG-3′ (sense) and 5′-AATTCAAAAACCCAGA- (RNAi) and gene replacement, we show that ADAM9-S and GAAGTTCCTATATATCTCGAGATATATAGGAACTTCTCTGG- ADAM9-L do not function in a redundant manner in the reg- 3′ (antisense); ADAM9shRNA.2, 5′-CCGGGCCAGAATAA- ulation of cell migration, whereby ADAM9-S promotes and CAAAGCCTATTCTCGAGAATAGGCTTTGTTATTCTGGCTTT- ADAM9-L attenuates migration. These studies identify for 3′ (sense) and 5′-AATTCAAAAAGCCAGAATAACAAAGCC- the first time an ADAM as a migration suppressor, and have TATTCTCGAGAATAGGCTTTGTTATTCTGGC-3′ (antisense). implications for the functional roles of ADAM9 proteins in Second-generation lentiviral packaging plasmids pCMV-dR8.2 breast cancer progression. dvpr and pCMV-VSVG were obtained from Addgene for lentiviral packaging. Materials and Methods Cell lines Antibodies and reagents All cell lines were obtained from the American Type Culture Anti-myc antibody purified from the 9B11 hybridoma was Collection, with the exception of the estrogen-independent purchased from Cell Signaling Technology. Anti–β-actin was human breast cancer SUM-159-PT, which has been described from Sigma-Aldrich. Anti–ADAM9-L was from BioMol Re- (33). BT549, HCC38, and ZR75-1 cells were maintained in search Laboratories, Inc. Anti–ADAM9-S was designed and RPMI 1640 supplemented with 10% fetal bovine serum. produced in collaboration with Pro-Sci, Inc. as a rabbit poly- MDA-MB-231, HEK293T, MDA-MB-468, and NIH3T3 cells clonal antibody based on the immunization peptide sequence were maintained in DMEM supplemented with 10% fetal bo- CATGLSLKFHAPF. Transient transfections were performed vine serum. CAMA-1 cells were maintained in MEM supplemen- with Lipofectamine 2000 (Invitrogen) or polyethyleimine ted with 10% fetal bovine serum. SKBR3 cells were maintained (Polysciences, Inc). pcDNA3-mA9L, pcDNA3-mA9L.EA, in McCoy's modified medium 5A supplemented with 10% fetal pcDNA3-mADAM9L.myc, and pcDNA4-hADAM9S.myc have bovine serum. MCF10A cells were maintained in DMEM mod- been described previously (12, 26). Point mutants mADAM9L. ified with 5% equine serum, 20 ng/mL EGF, 10 μg/mL insulin, EA.myc, ADAM9S.EA.myc, and ADAM9L.TCE.myc were derived 100 ng/mL cholera toxin, 500 ng/mL hydrocortisone, and 1% from these vectors via Stratagene site-directed mutagenesis penicillin and streptomycin. Cell lines were verified by multiple according to the manufacturer's instructions and the follow- methods, including DNA bar coding, gene expression, and tran- ing primers: ADAM9S.EA, 5′-CCAAGATTATGACCCAATG- scriptome analysis, and kept in culture for <6 months after CATGAGCAACAATGG-3′ and 5′-CCATTGTTGCT‐ receipt. CATGCATTGGGTCATAATCTTGG-3′; mADAM9L.EA, 5′- CCATTGTTGCTCATGCATTGGGGCATAACCTTGG-3′ and Cell lysis 5′-CCAAGGTTATGCCCCAATGCATGAGCAACAATGG-3′; Cells were lysed in NP40 lysis buffer [20 mmol/L Tris-HCl mADAM9L.TCE.myc, 5′-GGACCCTGGAGAGCGTGT- (pH 7.0), 10% glycerol, 1% NP40, 10 mmol/L EDTA, 150 mmol/L

8188 Cancer Res; 70(20) October 15, 2010 Cancer Research

ADAM9 Isoforms in Breast Cancer Cell Migration

NaCl, 20 mmol/L NaF, 5 mmol/L sodium pyrophosphate] Migration assays supplemented with protease inhibitor cocktail (Sigma-Aldrich). Migration assays were carried out using Transwell cham- Protein concentration was calculated using Bio-Rad Protein bers (Corning) with 8-μm pore membranes. Cells were tryp- Assay solution. sinized and resuspended in serum-free medium containing 0.1% bovine serum albumin (BSA). Cells (1 × 105) were added Immunoprecipitation to the top of each Transwell chamber and allowed to migrate Equal amounts of protein were incubated at 4°C in a 1-mL toward cell-conditioned medium for 4 to 16 hours at 37°C volume with 2 μg antibody for 4 hours before addition of pro- depending on the cell type. Cells that had migrated to the tein A–Sepharose beads. Beads were incubated for 2 hours lower surface of the membrane were fixed and stained. Lenti- and then washed with 1% NP40 in PBS three times before virally infected cells were stained with Hema-3 Protocol Stain elution with 2× Laemmli lysis buffer [125 mmol/L Tris (pH (Fisher Scientific). Cells cotransfected with expression plas- 6.8), 23% glycerol, 4% SDS, 10% β-mercaptoethanol, brom- mids and β-galactosidase were stained with X-galactosidase. phenol blue] and immunoblotting. Experiments were conducted in triplicate, and equal fields of stained cells were counted in each well and normalized to Immunoblotting the plating efficiency or the transfection efficiency. Transfec- Cell lysates were equilibrated using the Bio-Rad Protein tion efficiency was calculated by counting the number of Assay solution, and protein (50–100 μg) was resolved by transfected cells per 100,000. Plating efficiency was calculated SDS-PAGE, transferred to nitrocellulose (Hoefer semidry by counting two wells of 100,000 cells plated in parallel to the transfer system, 160 mA, 80–120 minutes), and blocked in migration assays to control for minor discrepancies in plat- 5% nonfat milk in 1% TBST (1% Tween in TBS). Membranes ing. Statistical analysis was performed using ANOVA and the were incubated with the appropriate antibodies (primary InStat statistical analysis software. Cell-conditioned medium antibodies used at the concentration recommended by the was harvested from NIH3T3 fibroblast cells grown in full manufacturer, ADAM9-S used at 2 μg/mL) for 14 hours. serum for 48 hours to confluency, or HEK293T cells trans- Cell membranes were washed in 1% TBST and incubated fected with hADAM9-S expression plasmids, and incubated with peroxidase-conjugated anti-mouse or anti-rabbit anti- in serum-free medium for 48 hours. bodies (Jackson ImmunoResearch) and exposed using high- sensitivity enhanced chemiluminescence (Millipore). Integrin-blocking assays Cells were infected and transfected with plasmid DNA as Lentiviral infection described above. Before migration assay, cells were trypsi- pLKO.1 plasmid constructs were cotransfected into nized and incubated with 5 μg/mL of GOH3 antibody for HEK293T cells with packaging vectors pCMV-dR8.2 dvpr 10 minutes at room temperature. and pCMV-VSVG using polyethyleimine. Forty-eight hours after transfection, lentiviral particles in supernatant were Results harvested, passed through a 0.45-μm filter, and either stored at −80°C or used to infect target cells. Target cells at ∼80% ADAM9-S is expressed in breast cancer cell lines and confluency were infected with virus alone for a period of breast tumors 4 hours, at which point they were supplemented with full se- We first determined the expression pattern of ADAM9-S rum and allowed to recover for 24 hours before selection in and ADAM9-L in breast cancer cells and breast cancer tis- puromycin in full culture medium for 18 hours (2–5 μg/mL). sues. To detect specific endogenous isoforms of ADAM9, a polyclonal antibody directed toward the COOH terminus of Transfection of siRNA into BT549 cells ADAM9-L and an ADAM9-S–specific polyclonal antibody pCS2-(n)-β-galactosidase reporter plasmid (1 μg) was designed against the unique COOH terminus of human transfected alone or with 200 pmol/L of nontargeting or ADAM9-S were used (Fig. 1A). To confirm isoform specificity, ADAM9 siRNA pools according to the Lipofectamine 2000 COOH-terminal myc-tagged murine ADAM9-L and human protocol [5 μL Lipofectamine 2000, 500 μL Opti-MEM (Life ADAM9-S were transiently expressed in HEK-293T cells. Technologies)] into 5 × 105 BT549 cells per 6-cm dish (Corn- Equal amounts of protein were immunoblotted with myc ing). Medium was replaced with full-serum medium 4 hours antibody to validate equivalent expression and ADAM9 anti- after transfection. body specificity. The results show that, as expected, ADAM9-L is detected by myc and ADAM9-L antibodies as a pro–ADAM9-L Transfection of plasmid DNA variant that comprises the unprocessed form migrating at Plasmids of interest and the pCS2-(n)-β-galactosidase re- 114 kDa and the fully processed mature membrane-bound porter plasmid were cotransfected at a ratio of 6.5:1 into non- form at 84 kDa (Fig. 1B). Expression of ADAM9-S is detected infected or lentivirally infected cells 24 hours after infection by myc and ADAM9-S antibodies at 67 kDa, as previously with Lipofectamine 2000 according to the manufacturers' in- shown (26). Importantly, the ADAM9-L antibody does not structions. Medium was replaced with full culture medium cross-react with ADAM9-S, and the ADAM9-S antibody does for noninfected cells or puromycin selection medium 4 hours not detect ADAM9-L (Fig. 1B). Therefore, specific antibodies after transfection. Cells were assayed for migration and pro- against ADAM9-L and ADAM9-S are specific to the secreted tein expression 24 hours after transfection. and membrane-bound variants of ADAM9. It is important to

www.aacrjournals.org Cancer Res; 70(20) October 15, 2010 8189

Fry and Toker

Figure 1. Expression of ADAM9 isoforms in breast cancer cell lines and tissues. A, a schematic representation of the membrane-bound (ADAM9-L) and secreted (ADAM9-S) isoforms of ADAM9. B, HEK293T cells transiently transfected with a vector control (pcDNA4), murine ADAM9L.myc, and human ADAM9S.myc were lysed and proteins were resolved on an 8% SDS-PAGE gel. Equivalent lanes were blotted with anti-myc to validate protein expression. Equivalent lanes were blotted with ADAM9-L– and ADAM9-S–specific antibodies. C, a panel of cell lines representing breast cancer subtypes was used for immunoblotting with ADAM9-L COOH-terminal–specific antibody, ADAM9-S COOH-terminal–specific antibody, and β-actin. “#” denotes overexpression of ADAM9 and amplified 8p11-12; “*” denotes amplified 8p11.12. D, SDS-extracted lysates from frozen tumor blocks were immunoblotted with anti–ADAM9-S antibody. BT549 cell lysate was used as control. All results are representative of at least three independent experiments.

note that ADAM9-L and ADAM9-S migrate slower on SDS- breast cancer subtypes, whereas ADAM9-S is predominantly PAGE than their predicted molecular mass due to the pres- observed in basal-type cells. Notably, CAMA-1, a noninvasive ence of cysteine-rich regions as well as glycosylation. cell type, expresses ADAM9-L, but not ADAM9-S. To evaluate the expression of ADAM9-S and ADAM9-L in To evaluate expression of ADAM9 isoforms in breast can- breast cancer cell lines, lysates were collected from BT549, cer tissues, ADAM9-S levels were evaluated in stage II and III HCC38, MDA-MB-231, SUM159PT, MCF10A, MDA-MB-468, invasive ductal carcinoma tumor and normal breast lysate CAMA-1, ZR75-1, MCF-7, and SKBR3 cells at approximately samples kindly provided by Dr. Gerburg Wulf (Beth Israel 80% to 90% confluence, and ADAM9 expression was deter- Deaconess Medical Center; ref. 34). The results show that mined by immunoblotting. ADAM9-L expression is detected ADAM9-S is expressed in breast cancers of both grades as in all cell lines tested (Fig. 1C, dark exposure). Importantly, well as in normal tissue (Fig. 1D). cells previously characterized to overexpress ADAM9 (BT549, HCC38, and CAMA-1; ref. 31) reveal a substantially higher level Overexpression of ADAM9-S promotes breast cancer of expression of ADAM9-L compared with other lines (Fig. 1C, cell migration light exposure). In addition, the variance in ADAM9-L bands Previous studies have shown that cancer cells exhibit in- detected by immunoblotting reflects differences in processing creased translocation through Matrigel in the presence of ex- as well as posttranslational modifications that include pro- ogenous recombinant ADAM9-S (26). To evaluate the role of teolytic cleavage and glycosylation. ADAM9-S endogenous ADAM9 isoforms in cell migration independent of matrix expression is detected in BT549, HCC38, SUM159PT, MCF10A, proteolysis, motility was assayed using uncoated Transwells. and MDA-MB-468 cells (Fig. 1C). Thus, endogenous ADAM9-L Plasmid constructs directing the expression of wild-type and expression is detected in cells of both the luminal and basal mutant murine ADAM9-L and human ADAM9-S (Fig. 2A)

8190 Cancer Res; 70(20) October 15, 2010 Cancer Research

ADAM9 Isoforms in Breast Cancer Cell Migration

were transfected into BT549 breast cancer cells. Over- activity–dependent manner. However, this phenotype is not expression of wild-type ADAM9-S significantly increases recapitulated by ADAM9-L. BT549 migration. In contrast, overexpression of a metallo- To confirm that extracellular ADAM9-S stimulates cell mi- proteinase-deficient ADAM9-S.EA mutant with a Glu to Ala gration, ADAM9-S, ADAM9S.EA, or vector control was ex- substitution in the metalloproteinase active site (35) does pressed in HEK293T cells and serum-free conditioned not increase migration and is comparable with control- medium was collected after transfection. These supernatants transfectedcells(Fig.2B).Similarly,overexpressionofthe were added to the bottom chamber of uncoated Transwells, ADAM9-L variant did not promote BT549 cell migration, and BT549 breast cancer cells were added to the top chamber. nor did overexpression of the equivalent metalloproteinase- Under these conditions, BT549 cells require an extended 16- deficient ADAM9L.EA mutant. Therefore, as previously re- hour incubation period to migrate. However, BT549 cells exhibit ported, ADAM9-S promotes cell migration in a protease increased migration when exposed to ADAM9-S–containing

Figure 2. ADAM9-S promotes breast cancer cell migration. A, schematic of human and murine ADAM9 constructs used in migration experiments. B, migration of BT549 cells overexpressing ADAM9-L and ADAM9-S. Cells were migrated for 4 h in Transwells. Expression of transfected proteins was confirmed with anti–ADAM9-L COOH-terminal–specific antibody or anti-myc. ADAM9-L is denoted by arrows. Overexpressed murine ADAM9-L migrates slower than endogenous human ADAM9-L. *, P < 0.001, ANOVA. C, migration of BT549 cells stimulated with exogenous ADAM9-S. Cells were migrated for 16 h in Transwells in which the bottom chamber contained medium collected from HEK293T cells expressing pcDNA control, ADAM9-S, or ADAM9S.EA. Expression of ADAM9-S was confirmed by anti-myc immunoblotting. *, P < 0.05, ANOVA. Columns, mean of triplicate measurements of at least three independent experiments; bars, SD.

www.aacrjournals.org Cancer Res; 70(20) October 15, 2010 8191

Fry and Toker

Figure 3. ADAM9 silencing increases cell migration. A, BT549 cells were transfected with β-galactosidase alone, or in conjunction with nontargeting siRNA oligonucleotide, or ADAM9 siRNA oligonucleotide pool. Cells were migrated for 16 h. *, P < 0.01, ANOVA. Silencing was confirmed by immunoblotting using ADAM9-L COOH-terminal–specific antibody and β-actin. B, BT549 cells infected with ADAM9shRNA.1 and ADAM9shRNA.2 were lysed and immunoblotted against ADAM9-L and ADAM9-S. ADAM9-S was immunoprecipitated from equal cell lysates before immunoblotting. C, BT549 cells were infected with control or ADAM9shRNA.1 lentiviruses before transfection of pcDNA control, ADAM9-L, ADAM9-S, and ADAM9S.EA. Cells were migrated for 4 h. *, P < 0.05; **, P < 0.001, ANOVA. ADAM9-L immunoblot confirms silencing. ADAM9-S expression is detected with anti-myc.

conditioned medium in comparison with control-transfected ingly, ADAM9 siRNA increases migration in BT549 cells as or ADAM9S.EA-transfected medium, indicating that exoge- measured by Transwell migration assays compared with con- nous ADAM9-S stimulates haptotactic migration through its trol or nontarget siRNA (Fig. 3A). This is concomitant with metalloproteinase activity (Fig. 2C). quantitative silencing of ADAM9-L protein expression. The phenotype of increased migration by ADAM9 siRNA recapi- ADAM9 silencing increases breast cancer cell migration tulates that observed on expression of ADAM9-S (Fig. 2A). To determine the functional role of endogenous ADAM9 in To evaluate that this is not due to an off-target effect of cell migration, we performed loss-of-function experiments the siRNA pool, two distinct lentiviral shRNA constructs using a siRNA oligonucleotide pool targeting five distinct se- were generated to target both isoforms in the 3′ untranslated quences spanning both ADAM9 isoforms. We chose to tran- region of ADAM9 mRNA (ADAM9shRNA.1) and the cysteine- siently transfect these siRNAs into BT549 cells, which express rich domain (ADAM9shRNA.2). After lentiviral infection high levels of ADAM9 and are also highly migratory. Surpris- and selection, robust silencing of ADAM9-L and ADAM9-S

8192 Cancer Res; 70(20) October 15, 2010 Cancer Research

ADAM9 Isoforms in Breast Cancer Cell Migration

Figure 3. Continued. D, BT549, CAMA-1, and ZR75-1 cells were infected with control or ADAM9shRNA.1 or ADAM9shRNA.2 lentiviruses. Selected cells were migrated in uncoated Transwells for 4 h (BT549) or 16 h. *, P < 0.03, ANOVA. Silencing is confirmed by immunoblot with ADAM9-L COOH- terminal antibody. In all experiments, cell counts were normalized. Columns, mean of triplicate measurements of at least three independent experiments; bars, SD.

protein is observed in cells transduced with both sequences shRNA. The murine ADAM9-L (mADAM9-L) would be pre- as revealed by immunoblotting (Fig. 3B). Importantly, trans- dicted to be refractory to silencing using human ADAM9. duction of both shRNA sequences also results in an increase Consistent with this, expression of mADAM9-L protein in BT549 cell migration (Fig. 3D). Moreover, increased cell was resistant to silencing in cells expressing ADAM9 shRNA migration resulting from transduction of ADAM9 shRNA is compared with cells with ADAM9 shRNA alone that show a also reproduced in distinct breast cancer cell lines, includ- robust reduction of ADAM9-L protein (Fig. 3C). In Transwell ing CAMA-1 and ZR75-1 (Fig. 3D). Therefore, in different lum- assays, expression of mADAM9-L rescues the increase in inal and basal breast cancer type cells, silencing ADAM9 migration observed with shRNA alone. Therefore, the expression results in enhanced cell migration observed with increase in migration observed on silencing of ADAM9 is ADAM9shRNA.1 and ADAM9shRNA.2 sequences. specific to ADAM9-L expression and is not due to off-target To further confirm that ADAM9 silencing is responsible effects. The implication is that endogenous ADAM9-L is a for the increase in migration observed with shRNA, a recon- suppressor of migration of breast cancer cells. stitution experiment was performed by introducing murine We also evaluated the contribution of ADAM9-S in the ADAM9-L and human ADAM9-S in cells harboring ADAM9 enhanced migration phenotype observed with ADAM9

www.aacrjournals.org Cancer Res; 70(20) October 15, 2010 8193

Fry and Toker

shRNA. As observed in Fig. 2A, expression of wild-type, (compare lane 5 with lane 1). However, treatment with GOH3 but not protease-deficient, ADAM9-S increases cell migra- significantly increased migration (compare lane 6 with lane 5), tion in control BT549 cells (Fig. 3C). In contrast to the effect an effect that was quantitatively rescued by expression of the of expression of mADAM9-L, the increase in migration in- shRNA-resistant mADAM9-L allele (compare lane 8 with duced by ADAM9 silencing is not rescued by expression of lane 6). This shows that ADAM9-L suppresses cell migration ADAM9-S that is refractory to silencing, as the shRNA used by altering integrin-mediated signals, either at the level of (ADAM9shRNA.1) targets the 3′ untranslated region. In- integrin binding or by interacting with a parallel signaling stead, expression of ADAM9-S leads to a further increase in pathway that engages in cross talk downstream of integrin the migratory phenotype that is additive to ADAM9 shRNA receptors. and ADAM9-S. As expected, this additive effect is not recapitulated by expression of the ADAM9-S protease-inactive EA Discussion mutant (Fig. 3C). Therefore, ADAM9-S does not phenocopy the migration suppression function of ADAM9-L, and more- The ADAM family of metalloproteinases controls multiple over, the implication from these results is that suppression phenotypes associated with tumor progression, although the of migration is the dominant role of ADAM9 in breast specific role of individual isoforms in tumorigenesis is poorly cancer cells. understood. Several ADAM isoforms, including ADAM9, have been shown to promote cell migration. In this study, we iden- Functional role of ADAM9-L domains in suppression tify nonredundant roles for the secreted and membrane-bound of cell migration ADAM9 variants in breast cancer cell migration. Consistent The data thus far show that ADAM9-S is an enhancer of cell with previous observations, we show that ADAM9-S can pro- migration, whereas ADAM9-L is a suppressor of migration. We mote breast cancer cell migration in a cell-autonomous man- next determined the contribution of the ADAM9-L functional ner. Surprisingly, this phenotype is not recapitulated by domains in the suppression of migration. We generated muta- ADAM9-L, which functions as a migration suppressor. tions or deletions in the metalloproteinase (ADAM9L.EA), dis- The ADAM9 gene (8p11.23) is located in the 8p11-12 clus- integrin (ADAM9L.Δdis), and cytoplasmic (ADAM9L.Δcyt) ter that is frequently amplified or deleted in breast cancer, domains (Fig. 4A). Again the murine ADAM9-L allele was used with instances of abnormal copy number that correlate with to generate these mutations, as it is refractory to silencing in reduced patient survival (37). Although ADAM9-null mice have human breast cancer cells. Wild-type and mutant mADAM9-L no overt phenotypes (38), in mouse models of breast, colon, and alleles were introduced into BT549 cells after silencing with prostate carcinoma such as MMTV-PyMT, APC/Min/+, and ADAM9 shRNA. As already observed, expression of the wild- W10, overexpression of ADAM9 in neoplastic regions is detected − − type ADAM9-L allele rescued the phenotype of increased mi- (35). ADAM9 / W10 mice have well-differentiated tumors com- gration on ADAM9 silencing (Fig. 4B). This was also observed pared with tumors in ADAM9+/+ W10 (35). Immunohistochem- in cells expressing the metalloproteinase-deficient ADAM9L.EA ical analysis of ADAM9 expression in breast tumors and cell mutant, indicating that the proteolytic activity of ADAM9-L is lines supports our finding that secreted ADAM9-S is expressed dispensable for the suppression of migration. In contrast, there in breast cancer tissues (29, 30). In our studies, the ADAM9- was no rescue of increased migration with the cytoplasmic do- specific antibodies were unable to provide a clear detection main deletion mutant (ADAM9L.Δcyt), showing that localization of protein expression by immunohistochemistry or immuno- and downstream signaling mediated by the ADAM9-L cytoplas- cytochemistry, but regardless, this is the first clear identifica- mic domain is required for suppression of cell migration (Fig. 4C). tion of ADAM9-S expression in tumors and cell lines. To address the role of the ADAM9-L disintegrin domain Breast cancer cell lines are separated into different clusters and integrin-binding activity, point mutations in the ECD based on gene expression profiles (39). Luminal cell lines ex- motif as well as whole deletions in the integrin-binding do- press genes associated with a more differentiated, noninvasive main were generated. However, expression of these mutant phenotype, whereas basal cell lines are less differentiated and proteins followed by immunoblotting revealed that both more invasive. In a survey of gene expression profiles of 51 breast mutants are not properly processed from the prodomain- cancer cell lines, the basal gene cluster was separated into two containing form (114 kDa) to the active and mature 84-kDa groups, basal A and B, of which B has a stem cell–like expres- form compared with wild-type ADAM9-L (Fig. 5A). Therefore, sion profile closest to the clinical “triple-negative” tumor profile we assessed the contribution of the disintegrin domain in the (31). Cell lines with ADAM9 amplification and with or without regulation of cell migration using integrin function-blocking ADAM9 mRNA overexpression were identified in this analysis. antibodies. BT549 cells were transduced with ADAM9 shRNA In our present study, we find that ADAM9-L is expressed in cells and reconstituted with wild-type murine ADAM9-L, followed of each gene cluster, whereas ADAM9-S is expressed primarily by treatment with GOH3, an antibody that blocks α6 integrin in the more aggressive basal genotype (Fig. 1C), correlating with binding to its ligand laminin (Fig. 5B), or control IgG. GOH3 the finding that ADAM9-S promotes cell migration. is also routinely used as direct binding ligand to activate in- Our previous studies had identified ADAM9-S as a promi- tegrin clustering (36). Interestingly, treatment with GOH3 gration factor that is secreted by activated liver myofibro- modestly increased the migration of BT549 cells with control blasts in the wound-healing response (26). Tumors are sites pLKO. As predicted by the data thus far, silencing ADAM9 of chronic wounding, and cancer cells adapt to the wound by with shRNA increased cell migration compared with control increasing cell migration and angiogenesis. The success of

8194 Cancer Res; 70(20) October 15, 2010 Cancer Research

ADAM9 Isoforms in Breast Cancer Cell Migration

Figure 4. ADAM9-L suppresses cell migration in a metalloproteinase- independent manner. A, schematic representation of murine wild-type ADAM9 isoforms and functional domain mutations used to assess the role of each domain in cell migration. B, BT549 cells were infected with control or A9shRNA.1 lentiviruses before transfection with pcDNA control, mADAM9-L wild-type, or mADAM9L.EA after selection. Cells were migrated in uncoated Transwells for 4 h. *, P < 0.01, ANOVA. Cell lysates were immunoblotted against actin and ADAM9-L to detect both murine (top arrows) and human (bottom arrows) variants. C, BT549 cells were infected with control or A9shRNA.1 lentiviruses before transfection with pcDNA control, mADAM9-L wild-type, or mADAM9L.Δcyt after selection. Cells were migrated in uncoated Transwells for 4 h. *, P < 0.05; *** P < 0.001, ANOVA. ADAM9-L expression is detected with anti-myc, ADAM9 silencing is detected with anti–ADAM9-L, and actin is control. ns, nonspecific immunoreactive protein. In all experiments, cell counts were normalized to transfection efficiency. Columns, mean of triplicate measurements of at least three independent experiments; bars, SD.

tamoxifen treatment in recurrent breast cancer with a high ADAM9-S promotes cell migration via its metalloproteinase stromal content is correlated with elevated levels of ADAM9 activity (Fig. 2A), ADAM9-L functions as a migration sup- (40). Similarly, ADAM9 is upregulated in the stroma surround- pressor in a manner that is independent of proteolytic activity ing invasive melanoma (41) and colon cancer (26), as well as (Fig. 4B). ADAM9-S may proteolyze specific substrates on the prostate cancer cells (42, 43). These studies emphasize the im- cell surface or ECM that are not targeted by ADAM9-L, as its portance of ADAM9 function at the tumor-stromal boundary, localization is restricted by membrane tethering. Consistent suggesting that ADAM9 may function downstream of para- with this, cells expressing ADAM9-S have a higher rate of mi- crine factors involved in wound healing. We find that whereas gration than cells exposed to a more diffuse exogenously

www.aacrjournals.org Cancer Res; 70(20) October 15, 2010 8195

Fry and Toker

Figure 5. The ADAM9-L integrin- binding activity modulates cell migration. A, HEK293T cells were transfected with wild-type mADAM9-L, point mutant mADAM9L.TCE.myc, and mADAM9.Δdis.myc. Lysates were immunoblotted with ADAM9-L COOH-terminal antibody. B, BT549 cells infected with control pLKO or ADAM9shRNA.1 virus were transfected with pcDNA control or mADAM9-L wild-type constructs. Cells were resuspended in 0.01% BSA in PBS with 5 μg/mL GOH3 antibody and incubated for 10 min before a 4-h migration assay. Lysates were immunoblotted with ADAM9-L COOH-terminal antibody to confirm silencing and expression. Columns, mean of triplicate measurements of two independent experiments; bars, SD. P values were calculated using ANOVA. *, P < 0.05; ***, P < 0.001.

added ADAM9-S (Fig. 2A and B), indicating that ADAM9-S vation of p38MAPK, cPLA2, and MMP-9 synthesis (25, 27). In might function in a cancer cell–autonomous manner rather breast carcinoma cells, antibody binding to α6 integrin en- than acting on the tumor stroma. Regardless, the opposing hances migration, a phenotype that is significantly enhanced functions of ADAM9-L and ADAM9-S in cell migration suggest in the absence of ADAM9 expression (Fig. 5B). ADAM9-L may that the two splice variants control tissue homeostasis, with control the localization of integrins or syndecans to specific isoform-specific functions in response to signals from both lipid rafts (47), or alternatively modulate endocytosis of cell neighboring tumor and stromal cells. A recent shRNA screen surface receptors. ADAM9 has also been shown to interact for cell migration enhancers and suppressors identified with and promote the recycling of E-cadherin in colon cancer ADAM9 as a migration suppressor in nontumorigenic MCF10A cells, and this may represent a metalloproteinase-independent epithelial cells, consistent with our findings (44). Surprisingly, mechanism for ADAM9-L (48). we find that the ADAM9-L metalloproteinase domain is The opposing roles for ADAM9-S and ADAM9-L as migration dispensable for this phenotype, whereas the cytoplasmic and enhancers and suppressors in breast cancer cells raise the question disintegrin domains are required, indicative of a role for mem- as to how the expression profile of each splice variant is regulated. brane localization and cytoplasmic signaling. This is predicted to have a major effect in tumor progression, as The disintegrin domain of ADAM9 influences cell migra- cells overexpressing ADAM9-L would presumably have a reduced tion in a cell type–dependent manner. Keratinocytes adhering migration phenotype. Conversely, any tumors in which ADAM9-S to recombinant ADAM9-L disintegrin domain exhibit in- is the predominant isoform would be predicted to have aggres- creased migration (25), a result also observed with fibroblasts sive invasive and metastatic characteristics. Although pres- adhering to recombinant ectodomain (45). In contrast, recom- ently unknown, it is highly likely that one important mechanism binant disintegrin domain inhibits the adhesion of platelets to is the control of alternative splicing of the ADAM9 message. collagen I (46) and interacts with multiple integrins, resulting In summary, our results show that the secreted and in extracellular signal-regulated kinase phosphorylation, acti- membrane-tethered isoforms of ADAM9 have opposing

8196 Cancer Res; 70(20) October 15, 2010 Cancer Research

ADAM9 Isoforms in Breast Cancer Cell Migration

functions in breast cancer cell migration. The expression of Acknowledgments both isoforms in breast cancer tumors and stroma suggests a model in which the secreted ADAM9-S acts as a migratory We thank Dr. Gerburg Wulf for kindly providing tumor tissue samples; Sara Courtneidge for providing ADAM9-L cDNA; Dr. Carl Blobel for providing stimulant, whereas ADAM9-L tethered to the plasma mem- murine ADAM9-L cDNA; Dr. Isaac Rabinovitz for assistance with the brane binds cell surface proteins and mediates localization integrin-binding studies; Drs. Marsha Moses, Bruce Zetter, and Jeffrey Settleman for guidance; and all members of the Toker laboratory for and signaling. One important implication from these findings discussions and advice. is that isoform-specific functions of secreted and membrane- tethered proteases exist, which has obvious consequences for Grant Support the development of small-molecule inhibitors for therapeutic NIH grant CA096710 (A. Toker) and Department of Defense Breast Cancer intervention. Research Program Pre-Doctoral Traineeship Award W81XWH 06-1-046 (J.L. Fry). 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. Disclosure of Potential Conflicts of Interest Received 11/25/2009; revised 07/21/2010; accepted 08/19/2010; published No potential conflicts of interest were disclosed. OnlineFirst 08/24/2010.

References 1. De Wever O, Mareel M. Role of tissue stroma in cancer cell invasion. 17. Beck V, Herold H, Benge A, et al. ADAM15 decreases integrin αvβ3/ J Pathol 2003;200:429–47. vitronectin-mediated ovarian cancer cell adhesion and motility in an 2. Zigrino P, Loffek S, Mauch C. Tumor-stroma interactions: their role in RGD-dependent fashion. Int J Biochem Cell Biol 2005;37:590–603. the control of tumor cell invasion. Biochimie 2005;87:321–8. 18. Kleino I, Ortiz RM, Huovila AP. ADAM15 gene structure and differen- 3. Giannelli G, Falk-Marzillier J, Schiraldi O, Stetler-Stevenson WG, tial alternative exon use in human tissues. BMC Mol Biol 2007;8:90. Quaranta V. Induction of cell migration by matrix metalloprotease-2 19. Izumi Y, Hirata M, Hasuwa H, et al. A metalloprotease-disintegrin, cleavage of laminin-5. Science 1997;277:225–8. MDC9/meltrin-γ/ADAM9 and PKCδ are involved in TPA-induced ec- 4. Xu J, Rodriguez D, Petitclerc E, et al. Proteolytic exposure of a cryp- todomain shedding of membrane-anchored heparin-binding EGF- tic site within collagen type IV is required for angiogenesis and tumor like growth factor. EMBO J 1998;17:7260–72. growth in vivo. J Cell Biol 2001;154:1069–79. 20. Ongusaha PP, Kwak JC, Zwible AJ, et al. HB-EGF is a potent inducer 5. Fowlkes JL, Thrailkill KM, Serra DM, Suzuki K, Nagase H. Matrix of tumor growth and angiogenesis. Cancer Res 2004;64:5283–90. metalloproteinases as insulin-like growth factor binding protein- 21. Dyczynska E, Sun D, Yi H, Sehara-Fujisawa A, Blobel CP, Zolkiewska A. degrading proteinases. Prog Growth Factor Res 1995;6:255–63. Proteolytic processing of delta-like 1 by ADAM proteases. J Biol 6. Whitelock JM, Murdoch AD, Iozzo RV, Underwood PA. The degrada- Chem 2007;282:436–44. tion of human endothelial cell-derived perlecan and release of bound 22. Parkin E, Harris B. A disintegrin and metalloproteinase (ADAM)- basic fibroblast growth factor by stromelysin, collagenase, plasmin, mediated ectodomain shedding of ADAM10. J Neurochem 2009; and heparanases. J Biol Chem 1996;271:10079–86. 108:1464–79. 7. Murphy G. The ADAMs: signalling scissors in the tumour micro- 23. Tousseyn T, Thathiah A, Jorissen E, et al. ADAM10, the rate-limiting environment. Nat Rev Cancer 2008;8:929–41. protease of regulated intramembrane proteolysis of Notch and other 8. Mizgerd JP, Spieker MR, Doerschuk CM. Early response cytokines proteins, is processed by ADAMS-9, ADAMS-15, and the γ-secretase. and innate immunity: essential roles for TNF receptor 1 and type I J Biol Chem 2009;284:11738–47. IL-1 receptor during Escherichia coli pneumonia in mice. J Immunol 24. Mahimkar RM, Visaya O, Pollock AS, Lovett DH. The disintegrin 2001;166:4042–8. domain of ADAM9: a ligand for multiple β1 renal integrins. Biochem 9. Yu Q, Stamenkovic I. Cell surface-localized matrix metalloproteinase-9 J 2005;385:461–8. proteolytically activates TGF-β and promotes tumor invasion and an- 25. Zigrino P, Steiger J, Fox JW, et al. Role of ADAM-9 disintegrin- giogenesis. Genes Dev 2000;14:163–76. cysteine-rich domains in human keratinocyte migration. J Biol Chem 10. Iba K, Albrechtsen R, Gilpin B, et al. The cysteine-rich domain of 2007;282:30785–93. human ADAM 12 supports cell adhesion through syndecans and 26. Mazzocca A, Coppari R, De Franco R, et al. A secreted form of triggers signaling events that lead to β1 integrin-dependent cell ADAM9 promotes carcinoma invasion through tumor-stromal inter- spreading. J Cell Biol 2000;149:1143–56. actions. Cancer Res 2005;65:4728–38. 11. Gaultier A, Cousin H, Darribere T, Alfandari D. ADAM13 disintegrin 27. Karadag A, Zhou M, Croucher PI. ADAM-9 (MDC-9/meltrin-γ), a and cysteine-rich domains bind to the second heparin-binding do- member of the a disintegrin and metalloproteinase family, regulates main of fibronectin. J Biol Chem 2002;277:23336–44. myeloma-cell-induced interleukin-6 production in osteoblasts by 12. Seals DF, Courtneidge SA. The ADAMs family of metalloproteases: direct interaction with the α(v)β5 integrin. Blood 2006;107:3271–8. multidomain proteins with multiple functions. Genes Dev 2003;17: 28. Maschler S, Wirl G, Spring H, et al. Tumor cell invasiveness corre- 7–30. lates with changes in integrin expression and localization. Oncogene 13. Kveiborg M, Frohlich C, Albrechtsen R, et al. A role for ADAM12 in 2005;24:2032–41. breast tumor progression and stromal cell apoptosis. Cancer Res 29. Lendeckel U, Kohl J, Arndt M, Carl-McGrath S, Donat H, Rocken C. 2005;65:4754–61. Increased expression of ADAM family members in human breast cancer 14. Pories SE, Zurakowski D, Roy R, et al. Urinary metalloproteinases: and breast cancer cell lines. J Cancer Res Clin Oncol 2005;131:41–8. noninvasive biomarkers for breast cancer risk assessment. Cancer 30. O'Shea C, McKie N, Buggy Y, et al. Expression of ADAM-9 mRNA Epidemiol Biomarkers Prev 2008;17:1034–42. and protein in human breast cancer. Int J Cancer 2003;105:754–61. 15. Roy R, Wewer UM, Zurakowski D, Pories SE, Moses MA. ADAM 12 31. Neve RM, Chin K, Fridlyand J, et al. A collection of breast cancer cell cleaves extracellular matrix proteins and correlates with cancer sta- lines for the study of functionally distinct cancer subtypes. Cancer tus and stage. J Biol Chem 2004;279:51323–30. Cell 2006;10:515–27. 16. Trochon-Joseph V, Martel-Renoir D, Mir LM, et al. Evidence of anti- 32. Moffat J, Grueneberg DA, Yang X, et al. A lentiviral RNAi library for angiogenic and antimetastatic activities of the recombinant disinte- human and mouse genes applied to an arrayed viral high-content grin domain of metargidin. Cancer Res 2004;64:2062–9. screen. Cell 2006;124:1283–98.

www.aacrjournals.org Cancer Res; 70(20) October 15, 2010 8197

Fry and Toker

33. Flanagan L, Van Weelden K, Ammerman C, Ethier SP, Welsh J. ulation in human skin melanoma and melanoma cell lines. Int J Can- SUM-159PT cells: a novel estrogen independent human breast cancer 2005;116:853–9. cer model system. Breast Cancer Res Treat 1999;58:193–204. 42. Shigemura K, Sung SY, Kubo H, et al. Reactive oxygen species me- 34. Wulf GM, Ryo A, Wulf GG, et al. Pin1 is overexpressed in breast diate androgen receptor- and serum starvation-elicited downstream cancer and cooperates with Ras signaling in increasing the tran- signaling of ADAM9 expression in human prostate cancer cells. scriptional activity of c-Jun towards cyclin D1. EMBO J 2001;20: Prostate 2007;67:722–31. 3459–72. 43. Sung SY, Kubo H, Shigemura K, et al. Oxidative stress induces 35. Peduto L, Reuter VE, Shaffer DR, Scher HI, Blobel CP. Critical function ADAM9 protein expression in human prostate cancer cells. Cancer for ADAM9 in mouse prostate cancer. Cancer Res 2005;65:9312–9. Res 2006;66:9519–26. 36. Lee EC, Lotz MM, Steele GD, Jr., Mercurio AM. The integrin α6β4isa 44. Simpson KJ, Selfors LM, Bui J, et al. Identification of genes that reg- laminin receptor. J Cell Biol 1992;117:671–8. ulate epithelial cell migration using an siRNA screening approach. 37. Chin K, DeVries S, Fridlyand J, et al. Genomic and transcriptional Nat Cell Biol 2008;10:1027–38. aberrations linked to breast cancer pathophysiologies. Cancer Cell 45. Nath D, Slocombe PM, Webster A, Stephens PE, Docherty AJ, 2006;10:529–41. Murphy G. Meltrin γ (ADAM-9) mediates cellular adhesion through 38. Weskamp G, Cai H, Brodie TA, et al. Mice lacking the metalloprotease- α(6)β(1) integrin, leading to a marked induction of fibroblast cell disintegrin MDC9 (ADAM9) have no evident major abnormalities motility. J Cell Sci 2000;113:2319–28. during development or adult life. Mol Cell Biol 2002;22:1537–44. 46. Cominetti MR, Martin AC, Ribeiro JU, et al. Inhibition of platelets and 39. Perou CM, Jeffrey SS, van de Rijn M, et al. Distinctive gene expres- tumor cell adhesion by the disintegrin domain of human ADAM9 to col- sion patterns in human mammary epithelial cells and breast cancers. lagen I under dynamic flow conditions. Biochimie 2009;91:1045–52. Proc Natl Acad Sci U S A 1999;96:9212–7. 47. Thodeti CK, Albrechtsen R, Grauslund M, et al. ADAM12/syndecan-4 40. Sieuwerts AM, Meijer-van Gelder ME, Timmermans M, et al. How signaling promotes β1 integrin-dependent cell spreading through ADAM-9 and ADAM-11 differentially from estrogen receptor predict protein kinase Cα and RhoA. J Biol Chem 2003;278:9576–84. response to tamoxifen treatment in patients with recurrent breast 48. Hirao T, Nanba D, Tanaka M, et al. Overexpression of ADAM9 cancer: a retrospective study. Clin Cancer Res 2005;11:7311–21. enhances growth factor-mediated recycling of E-cadherin in human 41. Zigrino P, Mauch C, Fox JW, Nischt R. Adam-9 expression and reg- colon cancer cell line HT29 cells. Exp Cell Res 2006;312:331–9.

8198 Cancer Res; 70(20) October 15, 2010 Cancer Research

Secreted and Membrane-Bound Isoforms of Protease ADAM9 Have Opposing Effects on Breast Cancer Cell Migration

Jessica L. Fry and Alex Toker

Cancer Res 2010;70:8187-8198. Published OnlineFirst August 24, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-09-4231

Cited articles This article cites 48 articles, 28 of which you can access for free at: http://cancerres.aacrjournals.org/content/70/20/8187.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/70/20/8187.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/70/20/8187. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.