Abstract
Previous findings indicated that the activated leukocyte cell adhesion molecule (ALCAM) is expressed by tumors and plays a role in tumor biology. In this study, we show that ALCAM is shed from epithelial ovarian cancer (EOC) cells in vitro, leading to the generation of a soluble ALCAM (sALCAM), consisting of most of the extracellular domain. A similar sALCAM molecule was also found in the ascitic fluids and sera from EOC patients, suggesting that this process also occurs in vivo. sALCAM is constitutively produced by EOC cells, and this process can be enhanced by cell treatment with pervanadate, phorbol 12-myristate 13-acetate (PMA), or epidermal growth factor (EGF), a known growth factor for EOC. Pharmacologic inhibitors of matrix metalloproteinases (MMP) and of a disintegrin and metalloproteases (ADAM), and the tissue inhibitor of metalloproteinase-3, significantly inhibited sALCAM release by EOC cells. The ADAM17/TACE molecule was expressed in EOC cell lines and ADAM17/TACE silencing by specific small interfering RNA–reduced ALCAM shedding. In addition, inhibitors of ADAM function blocked EOC cell motility in a wound-healing assay. Conversely, a recombinant antibody blocking ALCAM adhesive functions and inducing ALCAM internalization enhanced EOC cell motility. Altogether, our data suggest that the disruption of ALCAM-mediated adhesion is a relevant step in EOC motility, and ADAM17/TACE takes part in this process, which may be relevant to EOC invasive potential. (Mol Cancer Res 2007;5(12):1246–53)
Introduction
Cell adhesion molecules (CAM) are essential for homeostasis and cellular architecture in multicellular organisms, being involved in cell-cell and cell-matrix interactions. In neoplastic development, multiple adhesive interactions, in concert with the activation of proteolytic cascades, play critical roles in determining cell release from the primary tumor and invasiveness (reviewed in refs. 1, 2). Activated leukocyte cell adhesion molecule (ALCAM), or CD166, is a member of the immunoglobulin superfamily belonging to the subgroup with five extracellular immunoglobulin-like domains (VVC2C2C2). ALCAM mediates cell-cell clustering through homophilic (ALCAM-ALCAM) and heterophilic (ALCAM-CD6) interactions (3). Moreover, the transmembrane and cytoplasmic domains are involved in anchoring of ALCAM to the actin, thus connecting ALCAM clustering to cytoskeleton mobility (4).
ALCAM has been involved in several biological processes, such as hematopoiesis, immune response, migration of neurons during brain development (reviewed in ref. 3), and transendothelial monocyte migration (5). Several observations indicate that ALCAM expression and its regulation may play a relevant role in tumor biology and progression. The overexpression or de novo expression of ALCAM at the cell surface can promote cell adhesion and clustering (4, 6-8). These observations have a direct correlation in pathology because ALCAM has been involved in a pathway to melanoma metastasis. In melanoma, in fact, ALCAM membrane expression correlates with the vertical growth phase of tumor progression but not with metastatic diffusion (9). Indeed, transfection of a cDNA encoding for a dominant negative, amino-terminally truncated ALCAM, unable to support homotypic cell clustering, increased spontaneous lung metastasis in a transplant tumor model (ref. 10, reviewed in ref. 11).
We previously showed that ALCAM is expressed at the cell surface of epithelial ovarian cancer (EOC) cells by an ALCAM-specific recombinant antibody, and that ALCAM can be internalized through a clathrin-dependent pathway upon ligand binding (12). We recently found that cytoplasmic localization of ALCAM is a marker of poorer outcome in ovarian carcinoma patients, as compared with its membrane expression (13). A similar correlation has been reported in breast cancer patients (14).
Altogether, these observations suggested that delocalization of ALCAM from the cell membrane may be involved in reduced cell adhesiveness and possibly in metastatic behavior. The relevance of ALCAM interaction in cell clustering has been extensively explored by means of a soluble recombinant chimeric ALCAM-IgG-Fc molecule, which could mimic the function of a natural soluble ALCAM (sALCAM), inhibiting cell-cell adhesion (6) and promoting the migration of osteoprogenitor cells in vitro (15). The existence of natural sALCAM is known, but no data on the biochemical characteristics and on the origin of this form(s) are available. A soluble isoform of ALCAM, generated by alternative splicing and consisting only of the single amino-terminal immunoglobulin-like V1 domain, has been recently described as a promoter of endothelial cell migration and inhibitor of endothelial tube formation (8). However, the translation into sALCAM protein of this alternatively spliced mRNA by cells naturally expressing this transcript has not been shown.
sALCAM might also be generated from full-length ALCAM by proteolysis, as reported for other adhesion molecules (reviewed in ref. 16), and a recent report, based on a proteomic approach, showed that ALCAM is an ADAM17/TACE substrate (17), suggesting a possible role of this molecule in the regulation of ALCAM-dependent functions.
In this study, we show that ALCAM is released from EOC cells by a metalloprotease-dependent mechanism leading to the generation of a natural sALCAM, which contains the great part of the extracellular domain. ALCAM shedding from EOC can be enhanced by stimuli such as pervanadate, phorbol 12-myristate 13-acetate (PMA), and epidermal growth factor (EGF) and can be blocked by inhibitors of ADAMs and by ADAM17/TACE silencing. A recombinant antibody blocking ALCAM molecular interactions significantly increased EOC cell motility, whereas inhibitors of ADAM17/TACE had an opposite effect. Altogether, these data suggest that the release of ALCAM adhesive functions may play a role in EOC cell motility and invasiveness.
Results
EOC Cell Lines Constitutively Release ALCAM Ectodomain in a Soluble Form
The presence of sALCAM in the conditioned media from EOC cell lines, expressing surface ALCAM, was first investigated by ELISA. Two experiments done with seven human EOC cell lines showed that sALCAM is constitutively released by all the cell lines tested, ranging from 1.1 to 5.8 ng/mL (mean ± SD, 3 ± 1 ng/mL). The biochemical characteristics of sALCAM were then analyzed by immunoprecipitation with the anti-ALCAM I/F8 scFv, which interacts with the ligand-binding site of ALCAM in the extracellular domain (12). The shed ALCAM molecule shows a 10-kDa smaller Mr in 11% acrylamide SDS-PAGE analysis, as compared with its membrane counterpart from cell lysates (Fig. 1A). Matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) analysis of peptides derived from the I/F8 immunoprecipitates confirmed the identity of sALCAM and identified peptides spanning from the NH2 terminus to amino acid position 419, corresponding roughly to four out of the five extracellular domains (Fig. 1B). These data are consistent with the immunoreactivity pattern obtained by Western blot. Indeed, both the anti-ALCAM monoclonal antibody (mAb) L50, directed against the membrane-proximal C domain, and the mAb MOG/07, raised against a 200-amino-acid-long peptide from the extracellular region, recognized the sALCAM spontaneously released by the EOC cell lines (Fig. 1C).
EOC cell lines constitutively release a soluble ALCAM ectodomain. A. Silver staining of the ALCAM molecules immunoprecipitated from serum-free conditioned media (right) shows a different Mr in SDS-PAGE analysis as compared with its membrane counterpart (left). B. Scheme of the ALCAM molecule showing the NH2 terminus (N), signal peptide (L), and transmembrane region (T; shaded boxes), and COOH terminus (C). Black boxes, peptides identified by MALDI-TOF analysis of the sALCAM excised from the above gel; numbers, their location in the protein. C. Western blot analysis of the I/F8-immunoprecipitated molecules by the anti-ALCAM murine mAbs MOG/07 and L50. Similar results were obtained by probing immunoprecipitated molecules from SKOV-3 with mAb L50 or from A2774 with mAb MOG/07 (data not shown). Arrowhead, sALCAM; arrow, membrane ALCAM.
EOC cell lines constitutively release a soluble ALCAM ectodomain. A. Silver staining of the ALCAM molecules immunoprecipitated from serum-free conditioned media (right) shows a different Mr in SDS-PAGE analysis as compared with its membrane counterpart (left). B. Scheme of the ALCAM molecule showing the NH2 terminus (N), signal peptide (L), and transmembrane region (T; shaded boxes), and COOH terminus (C). Black boxes, peptides identified by MALDI-TOF analysis of the sALCAM excised from the above gel; numbers, their location in the protein. C. Western blot analysis of the I/F8-immunoprecipitated molecules by the anti-ALCAM murine mAbs MOG/07 and L50. Similar results were obtained by probing immunoprecipitated molecules from SKOV-3 with mAb L50 or from A2774 with mAb MOG/07 (data not shown). Arrowhead, sALCAM; arrow, membrane ALCAM.
sALCAM Is Detected in Sera and Ascitic Fluids from EOC Patients
To gain information on the possible generation of sALCAM in vivo, we measured the sALCAM levels in ascites and sera from EOC patients. As shown in Fig. 2A, the average serum sALCAM levels were significantly higher in EOC patients (44 ± 15 ng/mL; M ± SE) than in healthy donors (29 ± 16 ng/mL), although an overlap was observed. In addition, even higher sALCAM levels (78 ± 31 ng/mL) were found in the patient ascites, thus suggesting local production of sALCAM in the peritoneal cavity of EOC patients. To explore the biochemical characteristics of sALCAM released in vivo, I/F8 immunoprecipitates from serum, ascites, and lysate of ascitic cells obtained from the same patients were compared by Western blot analysis using both the MOG/07 and the L50 mAbs. As shown in Fig. 2B, the sALCAM molecule detected in sera and ascitic fluids from patients displays the same Mr as the sALCAM shed in vitro, i.e., about 10 kDa less than the membrane form. Moreover, such difference in Mr is not due to different glycosylation because it is maintained also after PNGaseF treatment of the immunoprecipitates (Fig. 2C). Deglycosylated sALCAM resolved in multiple bands of 52 to 59 kDa, possibly resulting from incomplete deglycosylation or differential proteolysis. Similar data were obtained using cultured EOC cells and their conditioned media (Fig. 2C). The apparent Mr of the deglycosylated soluble forms is consistent with that of the ALCAM ectodomain amino acidic backbone (56 kDa from amino acid position 27 to 528). Altogether, these data indicate that the sALCAM detected in biological fluids or in culture media includes most of the ALCAM ectodomain and suggest that it could be generated by proteolytic cleavage in the extracellular region proximal to the cell membrane.
Biochemical characteristics of the sALCAM detected in sera and ascitic fluids from EOC patients. A. ELISA assay of sALCAM in sera and ascitic fluids from EOC patients and in sera from age-matched normal female controls. P was calculated by the Mann-Whitney test. B. Comparison by Western blot analysis of the ALCAM molecules immunoprecipitated from sera, ascites, and lysates of ascitic cells obtained from the same patients with serous EOC (numbers). C. Western blot analysis of the ALCAM molecules immunoprecipitated by I/F8 from SKOV-3 cells lysate and conditioned media (left) and from ascitic fluid and lysate of ascitic cells (right), before and after deglycosylation by enzymatic digestion with PNGaseF. Arrowheads, sALCAM; arrows, membrane ALCAM.
Biochemical characteristics of the sALCAM detected in sera and ascitic fluids from EOC patients. A. ELISA assay of sALCAM in sera and ascitic fluids from EOC patients and in sera from age-matched normal female controls. P was calculated by the Mann-Whitney test. B. Comparison by Western blot analysis of the ALCAM molecules immunoprecipitated from sera, ascites, and lysates of ascitic cells obtained from the same patients with serous EOC (numbers). C. Western blot analysis of the ALCAM molecules immunoprecipitated by I/F8 from SKOV-3 cells lysate and conditioned media (left) and from ascitic fluid and lysate of ascitic cells (right), before and after deglycosylation by enzymatic digestion with PNGaseF. Arrowheads, sALCAM; arrows, membrane ALCAM.
Release of sALCAM Is Inducible by Various Stimuli
To address the mechanism responsible for the release of sALCAM, we first searched for inducer stimuli. 35S-Met-Cys–metabolically labeled SKOV-3 cells were treated with different stimuli for various time intervals during the chase. Conditioned media were then immunoprecipitated with the anti-ALCAM I/F8-sepharose. As shown in Fig. 3A, the release of sALCAM can be induced by pervanadate or by PMA. sALCAM accumulation by pervanadate treatment increased with time (Fig. 3B), concomitant with the cell rounding and detachment from substrate (Fig. 3C). Densitometric analysis of the autoradiographies (mean of three experiments) suggested that about 10 ± 3% of total ALCAM was released after 50 min of treatment. Immunohistochemistry indicated that control orthovanadate-treated SKOV-3 cell monolayer shows a marked ALCAM membrane staining, whereas pervanadate-treated cells progressively detach from the culture dish and show membrane staining limited to the cell-to-cell contact region (Fig. 3C). It is therefore conceivable that other mechanisms, for instance, internalization, participate in the relocalization of ALCAM, which was observed during pervanadate-induced cell detachment. Because pervanadate mimics hyperactivity of tyrosine kinases (TK) by inhibiting tyrosine phosphatase activity, we studied whether a more physiologic TK stimulus would also enhance ALCAM shedding. Indeed, EGF, a known EOC growth factor acting through a TK receptor (reviewed in ref. 18), increased the ALCAM shedding more than 2-fold above spontaneous release as assessed both by ELISA (Fig. 3D) and immunoprecipitation from 24 h culture conditioned media (Fig. 3E).
Release of sALCAM is inducible by pervanadate, PMA, and EGF. A. sALCAM is released in culture media following treatment with PMA (2 h) or with pervanadate (PV, 50 min) but not with sodium orthovanadate (OV) or H2O2, as determined by SDS-PAGE analysis of anti-ALCAM I/F8 scFv-immunoprecipitated molecules of metabolically labeled SKOV-3 cell extract and conditioned media. B. Kinetics (in minutes) of pervanadate-induced sALCAM release by metabolically labeled SKOV-3 cells. C. ALCAM localization in pervanadate- or orthovanadate-treated SKOV-3 cells visualized by immunohistochemistry with the I/F8 scFv. Original magnification, ×400. Insets, negative controls. D. sALCAM is released following treatment with EGF by EOC cells. Supernatants from SKOV-3 cells, cultured for 24 and 48 h with (black columns) or without (white columns) EGF, were assessed by ELISA for the presence of sALCAM. Columns, mean of two experiments; bars, SD. *, P < 0.05, versus no EGF 48-h sample; **, P < 0.005 versus no EGF 24-h sample. E. Biochemical characterization and kinetics of release of sALCAM in metabolically labeled, EGF-treated SKOV-3 cells. Arrowhead, sALCAM; arrow, membrane ALCAM.
Release of sALCAM is inducible by pervanadate, PMA, and EGF. A. sALCAM is released in culture media following treatment with PMA (2 h) or with pervanadate (PV, 50 min) but not with sodium orthovanadate (OV) or H2O2, as determined by SDS-PAGE analysis of anti-ALCAM I/F8 scFv-immunoprecipitated molecules of metabolically labeled SKOV-3 cell extract and conditioned media. B. Kinetics (in minutes) of pervanadate-induced sALCAM release by metabolically labeled SKOV-3 cells. C. ALCAM localization in pervanadate- or orthovanadate-treated SKOV-3 cells visualized by immunohistochemistry with the I/F8 scFv. Original magnification, ×400. Insets, negative controls. D. sALCAM is released following treatment with EGF by EOC cells. Supernatants from SKOV-3 cells, cultured for 24 and 48 h with (black columns) or without (white columns) EGF, were assessed by ELISA for the presence of sALCAM. Columns, mean of two experiments; bars, SD. *, P < 0.05, versus no EGF 48-h sample; **, P < 0.005 versus no EGF 24-h sample. E. Biochemical characterization and kinetics of release of sALCAM in metabolically labeled, EGF-treated SKOV-3 cells. Arrowhead, sALCAM; arrow, membrane ALCAM.
sALCAM Is Shed by ADAM17/TACE-Dependent Proteolysis
Both PMA and pervanadate are known activators of metalloproteases (19). To evaluate if the sALCAM form was generated by metalloproteases in EOC cells, we treated cells with pervanadate in the presence of a variety of protease inhibitors. As shown in Table 1, the release of [35S]-labeled sALCAM in pervanadate-treated SKOV-3 cell supernatants was greatly reduced in the presence of nonselective zinc endopeptidase synthetic inhibitors, such as o-phenanthroline and GM6001, whereas inhibitors of serine or cysteine proteases had no effect. In addition, a strong inhibition of pervanadate-inducible ALCAM shedding was observed by the use of the nonselective and potent matrix metalloproteinases (MMP) and ADAMs inhibitor CGS27023A (CGS; IC50 = 159 ± 9 nmol/L on recombinant ADAM17 in vitro)5
Our unpublished data.
Effects of Protease Inhibitors on the Pervanadate-Induced ALCAM Ectodomain Shedding by the SKOV3 Human Ovarian Carcinoma Cell Line
Agent . | Specificity . | Concentration . | % Inhibition* . |
---|---|---|---|
Leupeptin | Serine proteases | 2 μg/mL | 0 |
Aprotinin | Serine proteases | 2 μg/mL | 0 |
E-64 | Cysteine proteases | 10 μmol/L | 0 |
o-Phenanthroline | Metalloproteases | 0.1 mmol/L | 0 |
1 mmol/L | 60 | ||
GM6001 | Metalloproteases | 10 μmol/L | 19 |
50 μmol/L | 55 | ||
EDTA | Metalloproteases | 5 mmol/L | 65 |
EGCG | Metalloproteases | 10 μmol/L | 0 |
CGS | Metalloproteases | 1 μmol/L | 65 |
10 μmol/L | 85 | ||
50 μmol/L | 92 | ||
TIMP-1 | Metalloproteases | 5 μg/mL | 0 |
TIMP-2 | Metalloproteases | 5 μg/mL | 15 |
TIMP-3 | Metalloproteases | 0.5 μg/mL | 25 |
5 μg/mL | 95 |
Agent . | Specificity . | Concentration . | % Inhibition* . |
---|---|---|---|
Leupeptin | Serine proteases | 2 μg/mL | 0 |
Aprotinin | Serine proteases | 2 μg/mL | 0 |
E-64 | Cysteine proteases | 10 μmol/L | 0 |
o-Phenanthroline | Metalloproteases | 0.1 mmol/L | 0 |
1 mmol/L | 60 | ||
GM6001 | Metalloproteases | 10 μmol/L | 19 |
50 μmol/L | 55 | ||
EDTA | Metalloproteases | 5 mmol/L | 65 |
EGCG | Metalloproteases | 10 μmol/L | 0 |
CGS | Metalloproteases | 1 μmol/L | 65 |
10 μmol/L | 85 | ||
50 μmol/L | 92 | ||
TIMP-1 | Metalloproteases | 5 μg/mL | 0 |
TIMP-2 | Metalloproteases | 5 μg/mL | 15 |
TIMP-3 | Metalloproteases | 0.5 μg/mL | 25 |
5 μg/mL | 95 |
Inhibition was evaluated comparing band intensities, quantitated by densitometric scanning, of I/F8-immunoprecipitated sALCAM from metabolically labeled SKOV-3 cells.
Shedding of sALCAM is metalloprotease dependent. A. Immunoprecipitation with I/F8 scFv of metabolically labeled conditioned media from SKOV-3 cells untreated or treated for 50 min with pervanadate (PV) in the presence of the tissue inhibitors of metalloprotease TIMP-1, TIMP-2, and TIMP-3. B. Supernatants from SKOV-3 cells cultured with pervanadate (50 min), PMA (2 h), EGF (24 h), or medium alone (ctr, 24 h), in the presence (black columns) or in the absence (white columns) of the metalloproteinase inhibitor CGS27023A (CGS), were assessed by ELISA for sALCAM. Columns, mean of two experiments; bars, SD. *, P < 0.05, and **, P < 0.005 versus same stimulus no CGS sample. C. Supernatants from A2774 cells cultured for 24 h with EGF or medium alone in the presence (black columns) or in the absence (white columns) of the EGFR inhibitor AG1478 (0.5 μmol/L), were assessed by ELISA for sALCAM. Columns, mean; bars, SD. *, P < 0.05, and **, P < 0.005. D. Expression of ADAM17/TACE mRNA and protein by EOC cell lines, as assessed by RT-PCR and Western blot analysis. The 130-kDa glycosylated ADAM17/TACE zymogen and the 80-kDa active form are indicated. E. Expression of ADAM17/TACE protein by A2774 cells transfected with an ADAM17/TACE-specific siRNA (OTP17), or with non-targeting siRNA (NT) as detected by Western blot. The amount of protein was calculated by comparative densitometric scanning with actin. F. ELISA detection of sALCAM release by A2774 EOC cells after transfection with siRNA specifically inhibiting ADAM-17/TACE (OTP17, black columns) or with non-targeting siRNA (NT, white columns), in the presence or in the absence of EGF. **, P < 0.005 versus NT siRNA samples. A2774 cells were used as they showed a better transfection efficiency than SKOV-3 cells.
Shedding of sALCAM is metalloprotease dependent. A. Immunoprecipitation with I/F8 scFv of metabolically labeled conditioned media from SKOV-3 cells untreated or treated for 50 min with pervanadate (PV) in the presence of the tissue inhibitors of metalloprotease TIMP-1, TIMP-2, and TIMP-3. B. Supernatants from SKOV-3 cells cultured with pervanadate (50 min), PMA (2 h), EGF (24 h), or medium alone (ctr, 24 h), in the presence (black columns) or in the absence (white columns) of the metalloproteinase inhibitor CGS27023A (CGS), were assessed by ELISA for sALCAM. Columns, mean of two experiments; bars, SD. *, P < 0.05, and **, P < 0.005 versus same stimulus no CGS sample. C. Supernatants from A2774 cells cultured for 24 h with EGF or medium alone in the presence (black columns) or in the absence (white columns) of the EGFR inhibitor AG1478 (0.5 μmol/L), were assessed by ELISA for sALCAM. Columns, mean; bars, SD. *, P < 0.05, and **, P < 0.005. D. Expression of ADAM17/TACE mRNA and protein by EOC cell lines, as assessed by RT-PCR and Western blot analysis. The 130-kDa glycosylated ADAM17/TACE zymogen and the 80-kDa active form are indicated. E. Expression of ADAM17/TACE protein by A2774 cells transfected with an ADAM17/TACE-specific siRNA (OTP17), or with non-targeting siRNA (NT) as detected by Western blot. The amount of protein was calculated by comparative densitometric scanning with actin. F. ELISA detection of sALCAM release by A2774 EOC cells after transfection with siRNA specifically inhibiting ADAM-17/TACE (OTP17, black columns) or with non-targeting siRNA (NT, white columns), in the presence or in the absence of EGF. **, P < 0.005 versus NT siRNA samples. A2774 cells were used as they showed a better transfection efficiency than SKOV-3 cells.
To visualize the involvement of ALCAM proteolytic cleavage in EOC cell motility, we used a wound-healing assay in the presence of EGF. The results were recorded at the 18-h time point because under the conditions used, the doubling time of A2774 was 36 h. As shown in Fig. 5A, 10 μmol/L CGS produced 40 ± 10% (mean ± SD of three experiments) inhibition of cell migration with respect to the solvent (DMSO) control in A2774 cells. These data suggest that ALCAM-mediated adhesive interactions must be resolved to permit cell motility and invasion. To confirm this hypothesis, we did the wound-healing assay in the presence of the scFv I/F8, which is known to block both ALCAM-ALCAM and ALCAM-CD6 interactions and can also induce ALCAM internalization (12). As shown in Fig. 5B and C, cell migration was significantly increased in the presence of I/F8 scFV, whereas an irrelevant scFv (anti-NIP) had a nonsignificant effect.
Involvement of ALCAM in EOC cell motility. A. The pharmacologic blockade of MMP/ADAMs by CGS reduces cell migration with respect to the experimental control consisting of only solvent (DMSO) in a wound-healing assay on EGF-treated A2774 EOC cells. Columns, mean percent migration representative of three different experiments each done in quadruplicate; bars, SD. *, P < 0.05 versus DMSO control. B. Stimulatory effect of IF/8 scFv on EOC cell motility detected as above. *, P < 0.05 versus control scFv. C. Representative image of the induction of EOC cell migration by the I/F8 scFv in a wound-healing assay on A2774 cells. Original magnification, ×400. Images are taken immediately after scratching the cultures (t0) and 18 h later (t18).
Involvement of ALCAM in EOC cell motility. A. The pharmacologic blockade of MMP/ADAMs by CGS reduces cell migration with respect to the experimental control consisting of only solvent (DMSO) in a wound-healing assay on EGF-treated A2774 EOC cells. Columns, mean percent migration representative of three different experiments each done in quadruplicate; bars, SD. *, P < 0.05 versus DMSO control. B. Stimulatory effect of IF/8 scFv on EOC cell motility detected as above. *, P < 0.05 versus control scFv. C. Representative image of the induction of EOC cell migration by the I/F8 scFv in a wound-healing assay on A2774 cells. Original magnification, ×400. Images are taken immediately after scratching the cultures (t0) and 18 h later (t18).
Discussion
The present study shows that EOC cells spontaneously release sALCAM molecules, and that shedding can be enhanced by stimuli, such as pervanadate, EGF, or phorbol esters, known to activate metalloproteases (19, 23). Our present data also show that ADAM17/TACE was indeed expressed by all the EOC cell lines tested, and that either TIMP-3, pharmacologic inhibitors of MMPs and ADAMs, or the specific silencing of ADAM17/TACE inhibited ALCAM shedding. By biochemical analyses and mass spectrometry, we showed that sALCAM consists of the large part of the ectodomain, comprising the ligand-binding domain and the oligomerization domain identified by the L50 mAb (4, 7). These analyses were made by immunoprecipitation with the I/F8 scFv, which reacts to an epitope within the external V1 domain of ALCAM (12). Our data seem to exclude a major contribution of the previously described ALCAM alternative transcript (8), encoding for a short sALCAM consisting of the external V1 domain, on the generation of sALCAM by EOC cell lines, because we were unable to detect a low-molecular-weight form by metabolic labeling and immunoprecipitation experiments.
The loss of surface ALCAM expression, associated to cytoplasmic expression, has been related to poor prognosis in both breast (13) and ovarian cancer. In addition, transgenic expression of a truncated dominant negative ALCAM in melanoma cells increased the metastatic behavior of these cells, indicating that suppression of surface ALCAM adhesive functions is required to mobilize cells from primary tumors (10).
It thus seems that the dynamic control of ALCAM at the cell surface is relevant to tumor progression. Our present data provide the first demonstration that surface ALCAM can be actively cleaved by ADAM17-mediated proteolysis in EOC cells, and that this process can be enhanced by tumor growth factors acting through TK stimulation, such as EGF. In this context, EGF is a known activator of other MMPs, which are specifically involved in the degradation of the extracellular matrix (23), thus resulting in increased tumor invasiveness. In addition to proteolytic cleavage, our previous data showed that ALCAM can be modulated from the cell surface by internalization through a clathrin-dependent pathway, upon binding of the scFv I/F8 or of a soluble ALCAM/Fc chimera (12). It is thus possible that the proteolytic generation of sALCAM in the tumor microenvironment may induce membrane ALCAM endocytosis upon homophilic interaction. Moreover, previous reports indicated that the use of recombinant sALCAM chimeric molecules might inhibit ALCAM adhesive functions through competitive binding and act as negative regulators of cell aggregation. Therefore, ALCAM proteolysis may act in different ways to down-regulate ALCAM-mediated intercellular adhesion, thus controlling the transition between cell clustering and cell movement (reviewed in ref. 3). Here, we also show that a recombinant antibody blocking ALCAM adhesive functions and inducing ALCAM internalization (12) can increase EOC cell motility in a wound-healing assay. Conversely, inhibitors of ADAM17/TACE, which block ALCAM proteolytic cleavage, inhibited cell motility. Altogether, these data indicate that the disruption of ALCAM-mediated adhesiveness is a relevant step in cell motility in EOC, thus providing a new insight in the molecular mechanisms underlying EOC cell invasiveness.
On the other hand, a recent report indicated that ALCAM may directly act as a sensor of cell density and activate certain metalloproteases expressed by melanoma cells, such as MT1-MMP and MMP2, suggesting a role of ALCAM in an outside-in signaling (24). Although our present data indicate an involvement of ALCAM proteolysis through a different member of the metalloprotease family, namely, ADAM17/TACE, and regulation through TK receptors, a possible regulatory role of cell density signals on ADAM17 cannot be excluded.
Other reports indicated that also L1, another member of the CAM family, plays an important role in EOC invasiveness (25), and that this molecule can be released in a soluble form by the combined action of ADAM17/TACE and ADAM10 (26). Our data seem to exclude a role of ADAM10 in ALCAM proteolytic cleavage because sALCAM generation seemed completely blocked by the ADAM17/TACE inhibitor TIMP-3, whereas the ADAM10 inhibitor TIMP-1 was ineffective. The finding that two adhesion molecules of the CAM family can be cleaved and regulated by a similar enzymatic process suggests that these molecules may have similar and redundant adhesive functions in EOC.
In conclusion, our data indicate that the expression of surface ALCAM is regulated through an ADAM17/TACE-mediated proteolytic process in EOC cells, and that this process might play a role in EOC cell biology and invasiveness. The finding that average sALCAM levels are higher in EOC patients than in healthy controls and are even higher in EOC ascites supports the concept that ALCAM shedding occurs at the site of tumor growth. This observation may deserve further studies on case materials with known clinical history to gain information on the possible role of sALCAM levels as a progression marker of EOC.
Materials and Methods
Cells, Reagents, and Antibodies
EOC cell lines SKOV-3, OVCAR-3, OVCAR-5, A2780 (American Type Culture Collection), IGROV-I and A2774 (from J. Bénard, Institut Gustave Roussy, Paris, France), OC315, OC136, and OC314 (from A. Alama, Istituto Nazionale per la Ricera sul Cancro, Genoa, Italy) were grown in RPMI 1640 supplemented with 10% FCS (Bio Whittaker Cambrex) or in serum-free CD CHO medium (Life Technologies/Invitrogen).
Chemicals were from Sigma Chemical Co. Recombinant human EGF was purchased from PeproTech EC. E64, o-phenanthroline, recombinant human (rh) TIMP-1, and rhTIMP-2 were from Calbiochem (Merck Biosciences), rhTIMP-3 was from R&D Systems, GM6001 was from Chemicon (Millipore), and AG1478 was from BioSource Int. CGS27023A was obtained as described (27).
Anti-ALCAM scFv I/F8 and anti-NIP scFv have been described (12). Monoclonal anti-ALCAM antibodies MOG/07 and L50 were purchased from NovoCastra Laboratories and Monosan, respectively. Anti-ADAM17/TACE rabbit polyclonal antibody was from Abcam. Biotin- or horseradish peroxidase (HRP)–conjugated secondary antibodies and streptavidin were from Caltag/Invitrogen.
Biological Samples
Ascites and/or sera were obtained upon Institutional Review Board approval from 42 EOC patients and from 23 age-matched female healthy donors who gave informed consent. Ascitic fluids (n = 16) were collected during surgical procedures from EOC patients. Tumor cells were isolated as previously described (28).
Cell Treatments and ELISA
To assess the shedding of the ALCAM molecule, subconfluent cells were cultured in 24-well plates in medium 0.1% FCS plus the indicated concentration of protease inhibitors or the equivalent amount of their solvent. After 30 min at 37°C, 200 μmol/L pervanadate (29) or 100 ng/mL PMA, or 100 ng/mL EGF was added, and incubation was prolonged for the indicated times. Untreated and treated conditioned media were then collected, centrifuged at 1000 × g, and used undiluted for sALCAM detection by an ELISA assay (DuoSet ELISA Development kit, R&D System), based on the anti-human ALCAM mAb clone 105902 as capturer and on a biotinylated polyclonal goat anti-human ALCAM serum as detection antibody. Sera and ascites were diluted 1:50 in PBS [137 mmol/L NaCl, 10 mmol/L Na2HPO4, 2.7 mmol/L KCl (pH 7.4)] for ELISA. Assays were done in triplicates, and background values were subtracted. Data were expressed as the mean ± SD and were analyzed using a two-tailed Student's t test or the Mann-Whitney test, where indicated.
Metabolic Labeling and Cell Treatments
Subconfluent cells cultured in six-well plates were washed twice and starved 1 h with Met-Cys–free, serum-free RPMI. [35S]-Met-Cys (5.5 MBq per sample; GE Healthcare) was then added to each culture dish in 2 mL of Met-Cys–free medium supplemented with 0.1% FCS and incubated overnight. Cultures were then chased with fresh medium. Variability of labeling was evaluated by counting [35S]-Met-Cys incorporation in trichloroacetic acid precipitates from supernatants. Labeled cultures were treated with the indicated protease inhibitors or their solvent for 30 min at 37°C, and stimuli were subsequently added. Conditioned media were then collected, centrifuged at 1000 × g, supplemented with Brij97 (0.5%) and protease inhibitors (Complete C Mini, Roche Diagnostics), and immunoprecipitated. Cell monolayers were lysed and immunoprecipitated.
Immunoprecipitation, PNGaseF Treatment, and Western Blot Analysis
Immunoprecipitation was done as described (12), and precipitated molecules were resolved by 11% acrylamide SDS-PAGE. Unlabeled samples were revealed by Western blotting, according to standard procedures, and peroxidase activity was visualized by enhanced chemiluminescence (ECL, GE Healthcare). For deglycosylation experiments, I/F8 scFv immunoprecipitates were treated with recombinant N-glycosidase F (PNGaseF, Roche Diagnostics), followed by electrophoresis and Western blot. Protein identification by MALDI-TOF peptide mass fingerprinting was done on the silver-stained protein excised from a preparative SDS-PAGE (12).
Pervanadate-Induced Cell Detachment from Substrate and Immunohistochemistry
SKOV-3 cells grown on cover glasses in 24-well tissue culture plates were treated for different time intervals with pervanadate or sodium orthovanadate or H2O2 as controls. Acetone-fixed monolayers were stained with scFv I/F8 and anti-FLAG M2 mAb (Sigma), followed by a biotin-conjugated goat anti-mouse K chain antibody and by HRP-conjugated streptavidin. Antibody binding was detected after reaction with 3-amino-9-ethylcarbazole (AEC, Calbiochem)/hydrogen peroxide. Slides were counterstained with Mayer's hematoxylin.
RT-PCR Analysis of ADAM17/TACE mRNA Expression
Total RNA was isolated by the use of NucleoSpin RNA II kit (Macherey-Nagel) and was reverse-transcribed using the SuperScript II Reverse Transcriptase (RT; Invitrogen). Two-microliter aliquots of cDNA were separately amplified in a final volume of 25 μL, with 2.5 IU Taq polymerase (Genecraft) in the presence of 1 μmol/L of the primers specific for ADAM17/TACE (upper primer GCACAGGTAATAGCAGTGAGT and lower primer CTCAGCATTTCGACGTTACTG) and for the housekeeping gene β-actin in a PCR Sprint thermal cycler (Hybaid). PCR products were analyzed on 1% agarose gel stained with ethidium bromide.
For quantitative RT-PCR analysis of ADAM17/TACE, the following primers were used: upper primer ATGTTTCACGTTTGCAGTCTCCA, lower primer CATGTATCTGTAGAAGCGATGATCTG; RNA polymerase IIA subunit upper primer GACAATGCAGAGAAGCTGG, lower primer GCAGGAAGACATCATCATCC; and glyceraldehyde-3-phosphate dehydrogenase upper primer GAAGGTGAAGGTCGGAGT, lower primer CATGGGTGGAATCATATTGGAA. Amplification was carried out in an iCycler instrument (Bio-Rad) using the SYBRgreen super mix system (Bio-Rad) according to the manufacturer's protocol. Relative quantification of ADAM17/TACE mRNA was calculated by the ΔΔCt method.
siRNA Transfection
ON-TARGET plus SMART pool for human ADAM17/TACE or siCONTROL Non-Targeting siRNA pool (Dharmacon) were transfected in A2774 cells using Lipofectamine RNAiMAX (Invitrogen) with the Forward protocol provided. Transfected cultures were assayed for constitutive or induced ALCAM shedding 48 h after siRNA delivery. Cells were harvested by PBS-EDTA treatment and processed for RNA and protein extraction as described above to detect ADAM17/TACE mRNA and protein.
Migration Assay
In vitro wound-healing assay was used to assess cell motility in two dimensions. A2774 cells were plated overnight to achieve a subconfluent cell layer in 24-well plates. A scratch was made on the cell layer with a micropipette tip, and cultures were washed twice with serum-free medium to remove floating cells. Cells were then incubated in culture medium containing 0.1% FCS, with 10 μmol/L CGS27023A (27) or with solvent only, or 10 μg/mL of the anti-ALCAM I/F8 scFv, or of the control (anti-NIP) scFv. Thirty minutes later, 100 ng/mL EGF were added, and cultures continued for 18 h. Wound healing was visualized by comparing photographs taken at the time of addition of EGF and 18 h later, by a Nikon DS-5M Camera System mounted on a phase-contrast Leitz microscope. The distance traveled by the cells was determined by measuring the wound width at time 18 h and subtracting it from the wound width at time 0. The values obtained were then expressed as % migration, setting the gap width at t0 as 100%. Three experiments were done in quadruplicates.
Grant support: Associazione Italiana per la Ricerca sul Cancro; Comitato Interministeriale Programmazione Economica (02/07/04, Centro Biotecnologie Avanzate project); Regione Liguria; Cariplo Foundation (2003-1740); and Italian Ministry of Health. T. Piazza was supported by Fondazione Italiana per la Ricerca sul Cancro.
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.