Nectin-4, a new serological breast cancer marker, is a substrate for tumor necrosis factor-alpha-converting enzyme (TACE)/ADAM-17.

Serum markers are extensively used in diagnostic and follow-up of cancer patients. We recently described Nectin-4, a 66-kDa adhesion molecule of the Nectin family, which is a valuable new histological and serological marker for breast carcinoma. In vivo, Nectin-4 is re-expressed in breast carcinoma, and a circulating form of Nectin-4 is detected in the sera of patients with metastatic breast cancer. In vitro, a soluble form of Nectin-4 is produced in the supernatant of breast tumor cell lines (S. Fabre-Lafay, C. Ginestier, S. Garrido-Urbani, C. Berruyer, R. Sauvan, N. Reymond, J. Adelaide, J. Geneix, P. Dubreuil, J. Jacquemier, D. Birnbaum, and M. Lopez, manuscript in preparation). We have investigated the mechanisms that regulate the production of this soluble form. It was found that the soluble form of Nectin-4 detected in the sera of patients and the supernatant of breast tumor cell lines share similar biochemical and immunological features. The soluble Nectin-4 form (43 kDa) is formed by the entire Nectin-4 ectodomain. Nectin-4 shedding is constitutive, strongly enhanced by 12-O-tetradecanoylphorbol-13-acetate activation, and reduced tumor necrosis factor-alpha protease inhibitor TAPI-1 or by the tissue inhibitor of metalloproteinase-3 (TIMP-3). TAPI-1 and TIMP-3 are inhibitors of the endoprotease tumor necrosis factor-alpha-converting enzyme (TACE)/ADAM-17. Overexpression or small interfering RNA-mediated silencing of TACE enhanced or reduced Nectin-4 shedding, respectively. Nectin-4 is not shed when expressed in TACE-deficient fibroblasts. Interestingly, the active form of TACE is overexpressed in breast tumors and may indicate that TACE is responsible for Nectin-4 shedding not only in vitro but also in vivo.

Cancer serum markers result from cell surface shedding from the tumor from which they originate. Ectodomain shedding is a major process regulating various biological events such as cell differentiation, proliferation, migration, and survival. Various proteins such as growth factors, growth factor receptors, or cell adhesion molecules are substrates for sheddases. Among the Zn 2ϩ -dependent protease superfamily, matrix metalloproteases (MMPs) 2 and ADAM (a disintegrin and metalloproteinase) proteins, both members of the metzincin subgroup, are largely implicated in the degradation and remodeling of the extracellular matrix (12,13). MMPs and ADAMs are largely implicated in most steps of tumorigenesis, such as invasion of the basement membrane and stroma, tumor growth and angiogenesis, extravasation, and migration (12,13). Among the 24 ADAMs described so far, tumor necrosis factor-␣-converting enzyme (TACE)/ADAM-17 is involved in various biological processes and cleaves numerous substrates including tumor necrosis factor (TNF)-␣, TNF receptor, epidermal growth factor receptor L, c-fms, c-kit, p75NTR, growth hormone receptor, interleukin-6 receptor, interleukin-1 receptor, vascular cell adhesion molecule-1, L-selectin, collagen VII, MUC1, Notch, CX3CL-1, CD40, ␤-amyloid precursor, and prion protein (14 -30). The mutations that inactivate TACE metalloprotease uncover its importance during development (30).
In the present study we show that the soluble form of Nectin-4 found in the supernatant and in patient sera have similar immunological and biochemical characteristics suggesting that they result from a common mechanism. Using breast tumor cell lines expressing endogenous Nectin-4 and Chinese hamster ovary cells (CHO) cells transfected with Nectin-4, we demonstrate that this mechanism is an active proteolytic process involving the TACE/ADAM-17 metalloprotease. We show that the active form of TACE is overexpressed in breast tumor samples, suggesting that TACE could also be involved in the shedding of Nectin-4 in vivo.
DNA Transfection-CHO or MDAMB-231 cells grown to 50 -80% confluency were transfected with the appropriate cDNA expression plasmids (FLAG-tagged Nectin-4 (p3XFLR4.C1) (31), FLAG-tagged TACE (32)) by using the Fu GENE TM 6 or the Polyfect TM reagent method. The cells were cultivated for 1 day, and the medium was replaced. When necessary, cells were selected in the presence of G418 to establish stable transfectants. The TACE Ϫ/Ϫ EC2 cell line was transfected with Nectin-4 cDNA and selected with 0.5 mg/ml hygromycin.
Reverse Transcription PCR-Total RNA was extracted from T47D cells using the TRIzol reagent according to the manufacturer's instructions (Invitrogen). Quantification and estimation of purity were performed by measuring the absorbance of each RNA sample at UV wavelengths of 260 and 280 nm. Reverse transcription was performed using the SuperscriptII RNase H reverse transcriptase (Invitrogen) according to manufacturer's instructions.
Construction, Production, and Purification of Soluble Forms of Nectin-4 -Human Nectin-4 ectodomains made up with either the V-C-C or the V Ig-like domains were fused to the Fc fragment of the human IgG1. Construction and production of these two soluble forms (named N4VCC-Fc and N4V-Fc, respectively) are already described (31).
Antibodies-The M2 anti-FLAG mAb was purchased from Sigma. Rabbit anti-TACE pAb was purchased from Chemicon International. The anti-Nectin-4 mAbs were obtained after successive intraperitoneal injections of BALB/c mice with 20 g of N4VCC-Fc (31). Two hundred clones were tested for differential reactivity to COS cells and COS cells transiently expressing the human Nectin-4. mAbs were then assayed for the ability to detect cell surface expression of Nectin-4 by fluorescence-activated cell sorter and to react with recombinant soluble N4VCC-Fc and N4V-Fc by ELISA. Two clones were obtained. The N4.61 clone reacts with both soluble Nectin-4 forms and is specific to the V domain. The N4.40 clone reacts only with the VCC form and is specific to one of the two C domains (31).
ELISA-A sandwich enzyme-linked immunosorbent assay was used to detect soluble Nectin-4 in cell culture supernatants and in patient sera. Ninety-six-well trays were coated with 10 g/ml of either anti-Nectin-4 mAb N4.40 or a purified polyclonal antibody directed against the C-terminal sequence of Nectin-4 (TGNGIYINGRGHLV, N4-pAb) (5). After saturation of wells with phosphate-buffered saline (PBS) containing 1% bovine serum albumin, 100 l of supernatant or serum was incubated for 12h at 4°C, then with 2.5 g/ml biotinylated mAb N4.61 for 2h at 37°C. Streptavidin-peroxidase in PBS-bovine serum albumin (2 g/ml) was incubated for 1 h at 37°C. One hundred l of peroxidase substrate was added (One Step ABTS, Pierce), and optical density was read at 405 nm.

3-5 Wash
Steps Were Performed between Each Incubation with PBS Containing 0.5% Tween 20 -We analyzed duplicates and reported the medium value. Sensitivity of the test is 30 pM, and measurement is linear up to 10 Ϫ9 M. Nectin-4 concentration was calculated using serial dilutions of N4VCC-Fc (concentrations ranging from 10 Ϫ9 to 10 Ϫ12 M). This ELISA does not detect soluble forms of Nectin-1, -2, -3 and PVR-Fc used at the same concentrations.
Shedding Assay-CHO or T47D cells were grown on CHO-Dulbecco's modified Eagle's medium supplemented with 10% SVF. CHO or T47D cells were seeded on 24-well tissue culture plates and grown to 80 -90% of confluence. At the time of treatment, cells were washed once in PBS and grown on serum-free CHO medium in the absence or in the presence of PMA (100 ng/ml) and GM6001 (10 or 25 M). Supernatants was collected after 1, 3, or 8 h of incubation and filtered before analysis.
Immunoprecipitation and Immunoblot Analysis-Determination of the molecular weight of truncated Nectin-4 on supernatant was determined by immunoblot assay. Cells in 100-mm dishes were washed 3 times with ice-cold PBS and then resuspended for 30 min in 750 l of ice cold lysis buffer containing 50 mM Hepes, pH 7.5, 150 mM NaCl, 1.5 mM MgCl 2 , 1 mM EGTA, 1% Triton X-100, and 10% glycerol. A protease inhibitor mixture was added as recommended by the manufacturer (Roche Diagnostics). Breast tumor sample lysis was performed for 30 min at ϩ4°C in a 5 mM Tris, pH 8.0, lysis buffer I containing 0.3% SDS and 0.2 M dithiothreitol. Lysates were then treated for 15 min at ϩ4°C in a 50 mM Tris, pH 8.0, lysis buffer II containing 50 mM MgCl 2 , 1 mg/ml DNase, and 0.25 mg/ml RNase. Solubilized material was clarified by centrifugation at 13,000 rpm for 10 min at 4°C. When mentioned, conditioned supernatant or cell lysate were incubated for 12 h at 4°C with 10 g/ml concentrations of the appropriate antibody and then for 1 h at 4°C with 100 l of protein G-Sepharose (Amersham Biosciences). Immune complexes were washed 3 times with cold lysis buffer, heated in SDS sample buffer (60 mM Tris-HCl, pH 6.7, 3% SDS, 2% (v/v) 2-mercaptoethanol, and 5% glycerol), separated by 7.5% SDS-PAGE, semidry-transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore, Boston, MA), probed with the indicated antibody, and visualized with ECL (Amersham Biosciences).
Mass Spectrometry Analysis-Protein chip arrays (PS-20) from Ciphergen Biosystems (Fremont, CA) were coated with IgG1 irrelevant antibody or anti-Nectin-4 N4.61 antibody (1 g/ml) as recommended by the manufacturer. Serum or supernatant was adsorbed three times on the chip. After binding and washing, 0.8 l of a saturated matrix solution (sinapinic acid in 50% acetonitrile and 0.5% trifluoroacetic) was applied to each spot twice. Mass spectrometry analysis was performed using the surface-enhanced laser desorption/ionization-time of flight system PBS-IIc (Ciphergen Biosystems). Spectra were collected by the accumulation of 260 laser shots with a laser intensity of 205 and a detector sensitivity of 10. The protein masses were calibrated externally using purified protein standards.
Cell Surface Biotinylation and Time-chase Experiments-For timechase analysis of the release of Nectin-4 in parallel to cell expression, MDAMB-231 Nectin-4-transfected cells were washed extensively with pH 8 PBS solution and resuspended at 25 ϫ 10 6 /ml, and 500 g of biotin (Pierce) were added. After a 30-min incubation at room temperature, cells were washed with cold pH 8 PBS solution and plated in 60-mm dishes in fresh medium supplemented with PMA (100 ng/ml). Supernatant and cell lysates were prepared over 72 h. Supernatant and cell lysates were immunoprecipitated with N4.61 antibody. Precipitates were run on SDS-PAGE and immunoblotted with a streptavidin-coupled peroxidase to detect biotinylated Nectin-4. Supernatant and cell lysates were also analyzed by ELISA.

Soluble Forms of Nectin-4 Detected in Supernatants and in Patient Sera Present Similar Characteristics-Soluble
Nectin-4 can be detected both in the supernatant of breast tumor cell lines and in 51% of the sera of patients with metastatic breast carcinoma. Soluble Nectin-4 is never detected in normal sera. 1 Different mechanisms may account for Nectin-4 release in supernatant such as cell degradation, proteolytic process, or alternative splicing. Cell degradation seems unlikely as cellular viability was greater than 98% during the experiments for all Nectin-4 Is a Substrate for TACE/ADAM-17 the cell lines tested. Nectin-4 has three Ig-like domains of V, C, and C type in its extracellular region. We previously described two splice isoforms of Nectin-4 differing by the presence or the absence of a sequence of 25 amino acids at position Gln-411 in the cytoplasmic region (5). We failed to detect other splice variant by reverse transcription-PCR or in the GenBank TM and EST databases, suggesting that a soluble Nectin-4 sequence resulting from alternative splicing probably does not exist.
To analyze the mechanism that lead to the generation of soluble Nectin-4, we used in vitro models of cells expressing Nectin-4. To control that these models could faithfully reproduce in vivo processes, we tried to compare the soluble form of Nectin-4 found in the sera and in the supernatants. Two approaches were selected; one is based on their immunological characteristics, and the other is based on their biochemical properties.
Epitope mapping of Nectin-4 was performed by ELISA (31) using antibodies specific of the IgV (N4.61), the IgC (N4.40), or the cytoplasmic region of Nectin-4 (anti-Cter). Two different ELISA were developed; an anti-IgC/anti-IgV sandwich (expected to identify both soluble and transmembrane forms of Nectin-4 (ECϩTM)) and an anti-Cter/anti-IgV sandwich (expected to detect specifically the transmembrane form of Nectin-4 (TM)). The CHO cell line transfected with the cDNA encoding a Nter tagged full-length transmembrane form of Nectin-4 (CHO-N4) was used as control. Both ELISA readily detected Nectin-4 from CHO-N4 lysates, pointing out that the form detected corresponds to the transmembrane form of Nectin-4 (Fig. 1). The slight difference between the levels of detection may suggest that anti-IgC/anti-IgV ELISA detected a Nectin-4 form undetected by the anti-Cter/anti-IgV ELISA or that the ELISA had a different sensitivity. The Nectin-4 soluble form in the supernatant was detected by the anti-IgC/anti-IgV ELISA and not by the anti-Cter/anti-IgV ELISA. The analysis of the circulating form of Nectin-4 in the sera of two patients with metastatic breast carcinoma indicates that this form is only detected by the anti-IgC/anti-IgV ELISA and not by the anti-Cter/anti-IgV ELISA (Fig. 1). Together these results show that the circulating form of Nectin-4 found in the sera of pa-tients with metastatic breast carcinoma is similar to that detected in vitro in the supernatant. These results suggest that the soluble form produced in the supernatant of CHO-N4 cells may result from cell surface cleavage.
Because anti-Nectin-4 mAbs failed to detect serum Nectin-4 by Western blot, serum Nectin-4 molecular mass was analyzed by mass spectrometry and compared with soluble Nectin-4 produced in culture. To this end the anti-Nectin-4 N4.61 mAb was covalently bound to a preactivated surface protein chip array (see "Experimental Procedures"). After incubating the protein chip with the different biological fluids, mass of retained proteins was determined via time of flight mass spectrometry. As shown in Fig. 2 (black arrow), we detected a protein at 38.5 kDa in the supernatant of CHO-N4. Peak height decreased when supernatant was partially depleted from soluble Nectin-4 before analysis (data not shown). This peak was absent in the control CHO medium. Interestingly, a peak at 35.1 kDa was reproducibly found in the serum of a patient with a breast tumor and not in normal serum (white arrow). This peak was not detected when the protein chip was coated with control IgG1. The difference in molecular mass can be attributable to the FLAG epitope (2.8 kDa) present at the N terminus of the soluble form of Nectin-4 produced in the supernatant. ELISA and mass spectrometry experiments do not demonstrate that the in vitro and in vivo soluble forms of Nectin-4 are strictly identical but at least suggest that they are closely related.
Biochemical Characterization of Soluble Nectin-4 -We then analyzed the biochemical properties of the soluble Nectin-4 form produced in the supernatant of CHO-N4 cells using three different mAbs directed against the Nectin-4 ectodomain. Two of three mAbs were already used in ELISA experiments (see Fig. 1), and the third mAb was a M2 mAb directed against the Nter FLAG epitope. A form at ϳ43 kDa was immunoprecipitated by these mAbs, which is lower than the 66 kDa of the transmembrane Nectin-4 form (Fig. 3A). This result strongly suggested that this form corresponds to the ectodomain of Nectin-4. To further measure this phenomenon, a kinetic of cell surface release of Nectin-4 was followed over time by monitor- ing soluble and membrane forms of Nectin-4. A breast tumor cell line overexpressing Nectin-4 (MDA-MB231-N4) was used in this experiment. Cells were surface biotinylated then analyzed at various times. A similar form corresponding to the biotinylated Nectin-4 ectodomain could be precipitated with the anti-Nectin-4 mAb 2 h after the addition of fresh medium (Fig. 3B, bottom). This form migrated slightly slower probably because of covalently bound biotin molecules. Levels increased to reach a plateau at 24 h and slightly decreased at 72 h. The 66-kDa full-length transmembrane Nectin-4 was progressively cleaved to reach undetectable levels after 24 h (Fig. 3B, top). Analysis of supernatants and cell lysates by ELISA confirmed Western blot experiments (Fig. 3C). Interestingly, the Nectin-4 ectodomain was still recovered in the supernatant after 72 h, pointing out the marked stability of this soluble protein. These results showed that a 43-kDa soluble Nectin-4 is produced in the supernatant and result from cell surface cleavage of fulllength Nectin-4. The difference between molecular mass measured by surface enhanced laser desorption/ionization-time of flight (38.5 kDa) and SDS-PAGE (43 kDa) could be attributable to the technique used. Surface-enhanced laser desorption/ionization-time of flight molecular mass determination is strictly correlated to the mass of the protein, whereas SDS-PAGE migration defines an apparent molecular mass that could be influenced by protein charges partially neutralized by SDS leading to an altered migration (35).
Together, these results highlight (i) that soluble forms of Nectin-4 found in serum and supernatants are similar, (ii) that the generation of this soluble form stems from the shedding of the cell surface expressed transmembrane Nectin-4 form, (iii) that this process is not restricted to the CHO model as it can also occur in breast tumor cells.
Nectin-4 Cleavage Is Regulated by a Metalloproteasic Activity-The proteolytic cleavage of many cell surface receptors is the result of metalloproteic sheddases that belong to the MMPs or ADAMs families. Phorbol esters, such as PMA, activate ectodomain shedding through activation of protein kinase C. Conversely, GM6001, a broad-spectrum synthetic inhibitor of metalloproteases, blocks cell surface shedding. Both drugs were used to assess the involvement of metalloproteases in Nectin-4 shedding. PMA enhanced the level of Nectin-4 in the supernatant of both CHO-Nectin-4 (5-fold) and T47D cell line (1.7-fold) (Fig. 4A). When used at 10 and 25 M, GM6001 inhibited the levels of Nectin-4 in the supernatants. Similar results were found with the 1,10-phenanthrolin inhibitor (data not shown).
As found by ELISA, the level of the 43-kDa form increased upon PMA activation and was lowered upon GM6001 treatment (Fig. 4B, top). This form was not detected by mAb directed against the cytoplasmic tail of Nectin-4 (Fig. 4B, bottom).
Together, our results show that Nectin-4 produced in the supernatant as a 43-kDa form is the result of a marked cell

Nectin-4 Is a Substrate for TACE/ADAM-17
surface proteolysis of its entire ectodomain by cellular proteases that belong to the MMPs or ADAMs families. Using reverse transcription-PCR, we found that TACE and ADAM-10 were predominantly expressed in the T47D cell line and that various MMPs (1,2,3,7,9,10,13) were not detected as already reported (data not shown) (36). To characterize the protease activity involved in Nectin-4 shedding, we used the TAPI-1 extensively used to block TACE-mediated tumor necrosis factor ␣ shedding. TAPI-1 markedly inhibits spontaneous and PMAinduced shedding of Nectin-4 in CHO cells, suggesting that the protease could be TACE (Fig. 4C, left). We analyzed the effects of the TIMPs, which are physiological protein inhibitors of MMPs and ADAMs. Four members have been described (TIMP-1 to TIMP-4) that are inhibitors of MMPs and ADAMs. Whereas most MMPs are inhibited by TIMP-1, TIMP-2, and TIMP-3, ADAM-10 is inhibited by TIMP-1 and faintly by TIMP-3, and TACE is only inhibited by TIMP-3 (37,38). CHO-N4 cells were incubated with TIMP-1, TIMP-2, or TIMP-3 in the presence of PMA. Inhibition of Nectin-4 release was noted only in the presence of TIMP-3 and was dose-dependent (Fig. 4C, right). This indicates that MMPs are probably not involved in Nectin-4 cleavage and that TACE may regulate this process.
Transmembrane Nectin-4 Is Processed by the TACE-To confirm the implication of TACE in the shedding of Nectin-4, 0.5 g/ml Nectin-4 cDNA was co-transfected with an increasing amount of TACE cDNA in CHO cells. Expression of TACE was monitored by Western blot, and the Nectin-4 cleavage level was measured by ELISA (N4.40/N4.61) and Western blot (Fig. 5A). Nectin-4 cleavage was significantly increased by TACE in a dose-dependent manner. The molecular mass of the soluble form generated by TACE was identical to the form produced before and after PMA activation ( Fig. 5A and Fig. 4B), suggesting that TACE is a protease involved in both Nectin-4 constitutive and TPA-inducible shedding. Implication of TACE in Nectin-4 shedding was verified by blocking endogenous expression of TACE by RNA interference. Suppression of TACE was monitored by Western blot analysis after siRNA transfection (Fig. 5B, left). siRNA-directed silencing of TACE leads to the reduction of soluble Nectin-4 levels in the supernatants (Fig.  5B, right). This partial inhibition is probably due to the persistence of residual TACE activity. Last, mouse fibroblasts derived from TACE-deficient mice (14) were stably transfected with Nectin-4 cDNA (EC2 Ϫ/Ϫ N4) (Fig. 5C). Cell surface expression of Nectin-4 was assessed by fluorescence-activated cell sorter analysis (Fig. 5C, left). This level of expression is similar to CHO-N4 cells (data not shown). Nectin-4 cleavage was barely detected even after TPA activation, indicating that TACE-deficient cells are unable to process Nectin-4 (Fig. 5C,  right). Together, these results described a novel effector function for TACE in the shedding of the breast carcinoma marker Nectin-4.
Expression of TACE in Breast Tumor Samples-We have shown that TACE is actually involved in Nectin-4 shedding in vitro. To strengthen the implication of TACE in the breast tumor biology, we analyzed the expression of TACE in six FIG. 3. Biochemical analysis of the soluble Nectin-4 form. A, Western blot (WB) analysis of soluble Nectin-4 produced by CHO-N4 cells. One ml of supernatant was immunoprecipitated (IP) with 10 g/ml mAbs directed against the tagged Nectin-4 ectodomain (N4.40, N4.61, and M2 mAbs). An immunoblot was performed with 1 g/ml anti-FLAG M2 mAb. Ten g of CHO-N4 cell lysate was loaded to detect the Nectin-4 transmembrane form (66 kDa) (right lane). The 43-kDa band immunoprecipitated in the supernatant corresponds to the soluble Nectin-4 form. Note that this band is absent from the cell lysate. A slight difference in the migration of the 43-kDa form was reproducibly observed when immunoprecipitation was carried out with the N4.61 mAb. The reason is unclear, but we found that the N4.61 heavy chain migrates slightly faster than the N4.40 and M2 heavy chains and may modify the migration of the soluble Nectin-4 form within the gel. B, time-chase experiments with MDAMB-231 N4 cells. Cells were surface biotinylated as described under "Experimental Procedure." Cells were then incubated in biotinylated-free fresh medium in the presence of 100 ng/ml PMA for the indicated times. Cell lysate and supernatant were immunoprecipitated with 10 g/ml N4.61 mAb and analyzed by western-blot using a 1/5000 dilution of horseradish peroxidase (HRP)labeled streptavidin. C, time-chase experiments were performed as in B, and Nectin-4 levels in both supernatant and cell lysate were analyzed by means of a modified ELISA, made with a coated N4.40 mAb at 10 g/ml and a 1/5000 dilution of horseradish peroxidase-labeled streptavidin in the final step.

Nectin-4 Is a Substrate for TACE/ADAM-17
breast tumor samples and three normal breast tissues randomly selected using an antibody directed against the cytoplasmic tail of TACE. No, or faint expression of TACE was detected in the normal breast tissues (Fig. 6, left). Interestingly, we found that 100% of tumors overexpressed the metalloprotease at various levels ( Fig. 6, right). The processed active form of TACE was predominant in these samples, whereas the unprocessed form was slightly detected. These results provide support for an implication of TACE in Nectin-4 shedding in breast tumors.

DISCUSSION
Nectin-4 is the last cloned member of the Nectin family (5,31). We have recently shown that Nectin-4 is a new tumor marker in breast carcinoma. A circulating form of Nectin-4 is indeed found in the sera of patients with metastatic breast carcinoma and may improve their follow-up. 1 As described for other serum markers, Nectin-4 is probably released from tumor cells into body fluids. We aimed to characterize the biochemical properties of soluble Nectin-4 and the protease involved in its shedding.
We show here that the cDNA of the Nectin-4 long isoform (form exclusively found in breast tumors) encodes for a transmembrane protein of 66 kDa and that a soluble form of 43 kDa is produced in the supernatant of Nectin-4-expressing cells. Using mass spectrometry and ELISA, we showed that serum Nectin-4 shares common structural characteristics with the in vitro soluble form. We demonstrate that the cleavage of Nectin-4 is an active process mediated by the endoprotease TACE.
Overexpression of TACE increases Nectin-4 cleavage, and siRNA-mediated TACE silencing decreases cleavage. In vitro, Nectin-4 cleavage by recombinant TACE results in the generation of a soluble form with a slightly higher molecular mass (data not shown). Differences in TACE cleavage position have been already reported in vitro using recombinant TACE and may explain this difference (29). The physiological inhibitor of TACE, TIMP-3, inhibits Nectin-4 cleavage. TIMP-1 and TIMP-2 did not affect Nectin-4 cleavage, suggesting that MMPs do not play a central role in the Nectin-4 shedding process. However, our results do not rule out that other ADAMs may contribute to Nectin-4 cleavage. Interestingly, a marked overexpression of the active form of TACE was found in breast tumor samples (Fig. 6). Recently overexpression of TACE in breast carcinoma has been shown to promote tumor growth through pro-transforming growth factor-␣ shedding (39). TACE is involved in MUC1 cleavage, which may account for the presence of soluble form (CA15.3) in patients with carcinoma (23). These data suggest a major role of TACE is breast carcinogenesis. TIMP-3 is the only known physiological inhibitor of TACE. TIMP-3 transcripts have been detected by in situ hybridization in ductal epithelial and fibroblastic cells in the mammary gland (40,41). Interestingly, TIMP-3 is silenced by gene methylation in many malignancies, including breast cancer (42). This deregulated balance between TACE and TIMP-3 expression could foster the progression of the tumor through the shedding of cell surface molecules that control epithelial homeostasis. To date there is no data available about the expression of TIMP-3 protein in the mammary gland. Using a mAb directed against the C-terminal end of TIMP-3 (Chemicon International) we readily detected the 24-and the 30-kDa forms of TIMP-3 in kidney, heart, and placenta. However TIMP-3 protein expression was weak in normal and tumor breast tissues. Two of five tumors were found completely negative for TIMP-3 (data not shown). These results suggest that TACE activity is not or is at least marginally inhibited by TIMP-3 in breast tumors.
As for CEA and CA15.3, serum Nectin-4 is rarely detected in patient with non-metastatic tumor. The reason is unclear, but detection of these markers could be associated with tumor size and/or the presence of circulating tumor cells during metastasis progression.
The role of Nectin-4 re-expression in breast carcinomas is not FIG. 6. TACE expression in normal and tumor breast tissue samples. 20 g of tissue lysates were loaded and run on a 8% SDS-PAGE. Top, immunoblotting with anti-TACE pAb (1/1000) revealed two bands corresponding to the proform (120 kDa) and the active form (90 kDa) of TACE. Bottom, loading charge was controlled with an antibody directed against the p85 subunit of phosphatidylinositol 3-kinase (Upstate) used at 1/1000. N, normal breast tissues from three different donors; T, ductal breast carcinoma from six different patients. Note that tumor specimens used in this study were randomly selected. *, undefined band observed in some tumor samples with the anti-p85 pAb.

FIG. 5. Nectin-4 cleavage is regulated by TACE/ADAM-17.
A, top, effect of TACE transfection on Nectin-4 shedding. CHO cells were co-transfected with fixed concentration of Nectin-4 cDNA (0.5 g/ml) and increasing amounts of TACE cDNA or control plasmid (pEFBOS). After 12 h, cell supernatant was replaced with fresh medium for 8 h. Nectin-4 release was measured by ELISA (N4.40/N4.61) and subtracted from the release measured with the pEFBOS plasmid. Bottom, to detect TACE expression in transfected CHO cells, 20 g of cell lysate was run on 8% SDS-PAGE and blotted with the horseradish peroxidase-labeled M2 mAb. Note that a faint signal was detected with 0.25 g/ml TACE cDNA, probably due to the limit of the sensitivity of the assay. To analyze soluble Nectin-4, 1 ml of supernatant was immunoprecipitated as in Fig. 4 and blotted with 1 g/ml anti-FLAG M2 mAb. B, TACE siRNA treatment inhibits spontaneous and PMA-inducible Nectin-4 shedding in CHO-N4 cells. CHO-N4 cells were transfected with 0.5 g of siRNA couples as described under "Experimental procedures." Left, immunoblot analysis with anti-TACE pAb (1/1000) shows that a 3-day treatment of cells with TACE siRNA specifically blocks endogenous TACE expression compared with controls without siRNA or with irrelevant Nectin-2 and Lano siRNA. Loading charge was controlled with an antibody directed against the p85 subunit of phosphatidylinositol 3-kinase (Upstate) used at 1/1000. Right, Nectin-4 shedding was measured by ELISA (N4.40/N4.61) after treatment of cells with TACE siRNA for 3 days. As indicated, 100 ng/ml of PMA was added during the last 24 h. C, TACE-deficient mouse fibroblasts were stably transfected with Nectin-4 cDNA. Fluorescence-activated cell sorter analysis with the N4.61 mAb shows that 60% of cells express Nectin-4 (shaded histogram, control IgG1; clear histogram, Nectin-4 expression). Nectin-4 release in the supernatant was measured by ELISA before and after PMA activation and compared with the CHO-N4 cell line. elucidated. Overexpression of Nectin-4 does not alter the epithelial architecture of Madin-Darby canine kidney cell cysts embedded in a Matrigel matrix (data not shown). Recently, it has been shown that the E-cadherin-shed form was able to increase protease expression and invasive properties of lung tumor cells in vitro (43). Because Nectins are involved in leukocyte transmigration and diapedesis through endothelial cells, we can speculate that soluble Nectin-4 could take part in the process of tumor cell migration and invasion (7). This may be achieved by the interaction with cell surface-expressed Nectin-4 (homophilic interaction) and/or Nectin-1 (heterophilic interaction) or still unidentified ligands (5,31). Nectin-1 is expressed at the cell surface of all breast tumor cell lines tested (data not shown). Soluble forms of Nectin-1 resulting from alternative splicing or from cell surface shedding have been reported (44,45). However, using an ELISA with equivalent sensitivity, we failed to detect a soluble form of Nectin-1 in the supernatant of breast carcinoma cell lines (data not shown). We found that soluble Nectin-4 binds specifically to cell surface Nectin-1 expressed by the MCF-7 breast carcinoma cell line (data not shown). The role of this interaction in the biology of the tumor must now be addressed.