INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Invasion of basement membranes and of stromal extracellular matrix (ECM) is a rate-limiting step for establishment of tumor metastases. Invasive behavior is a multistep process consisting of adhesion to, proteolysis of, and migration along ECM (Liotta et al., 1983). Although adhesion to ECM is an a priori requirement for ECM invasion, cell migration along the ECM involves the dynamic establishment and dissolution of cell-ECM contacts, i.e. adhesion and deadhesion. One critical parameter that determines the rate of cell migration is the degree of adhesiveness of cells to their ECM substrata. This has been corroborated by mathematical models (DiMilla et al., 1991), as well as by experimental evidence (Goodman et al., 1989; Halfter et al., 1989; Duband et al., 1991; DiMilla et al., 1993), showing that intermediate levels of cell-to-substratum adhesion are required for maximal cell migration. Thus, the composition of ECM and the repertoire of ECM receptors on the cell surface are intimately involved in the regulation of these processes (Damsky and Werb, 1992; Heino, 1996).

Integrins, the major and best characterized group of ECM receptors, have attracted considerable attention in the effort to unravel the mechanism of cell migration and/or invasion. In mammals, currently 16 different integrin  and 8 different integrin  subunits are known (Giancotti, 1997). In most cell types, integrins appear to be essential for cell adhesion to individual ECM constituents (Hynes, 1992). However, because cells in vivo interact with an ECM of complex composition as well as with other cells, the participation of individual integrins in adhesive events cannot be studied easily under such conditions. Furthermore, nonintegrin ECM receptors, such as cell surface proteoglycans or immunoglobulin (Ig) superfamily adhesion molecules, also play a role in cell adhesion to the ECM (Zisch et al., 1992; Mercurio, 1995; Noat et al., 1997; Powell and Kleinman, 1997).

Although some studies show expression of certain integrin subunits as positively correlated with invasion (Cannistra et al., 1995; Melchiori et al., 1995; Chao et al., 1996; Matsuura et al., 1996; Vihinen et al., 1996; Trikha et al., 1997), others show that integrins can diminish invasion (Damsky et al., 1994; Paulus and Tonn, 1994; Danen et al., 1996). One well characterized mechanism by which integrins affect tumor cell invasion is the establishment of short-lived adhesive contacts with the ECM (DiMilla et al., 1991; Duband et al., 1991; Palecek et al., 1997), a process that is accompanied by rapid changes in cytoskeletal microarchitecture, which are prerequisite for cell motility (Kassner et al., 1995; Lauffenburger and Horwitz, 1996). Because integrins can trigger diverse signaling events that lead to alterations in gene expression (Schwartz et al., 1995; Kheradmand et al., 1998), modulation of synthesis of other proteins that are important for invasion, including ECM-degrading enzymes (Heino, 1996), may be a critical event regulating tumor cell invasion. Integrins also facilitate invasion by recruiting ECM-degrading proteinases to sites where proteolysis is required (Brooks et al., 1996).

In this study we used function-blocking antibodies against 1, 2, 5, 6, v, and 1 integrins to dissect their specific functions in cell invasion, migration, adhesion, and matrix metalloproteinase production. We show that different integrins mediate these processes via distinct mechanisms in mammary epithelial cells.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Recombinant Stromelysin-1 and Antibodies

Human stromelysin-1 with the C-terminal hemopexin domain truncated was expressed and purified from the methyltrophic yeast Pichia pastoris (Smith, Sharkov, and Navre, unpublished results). In brief, a portion of the human stromelysin-1 cDNA (from the start of the prodomain to the hinge region: PLDGAA to PDSPET) (Whitham et al., 1986) was inserted behind the  factor signal sequence in the vector pPIC9 (Invitrogen, San Diego, CA). The resultant plasmid was transformed into Pichia pastoris, and eight methanol use-deficient clones were screened for production of recombinant stromelysin-1. One overproducing clone was selected and used for all expression and purification work. Induction of an AOX1 promoter was accomplished using the Pichia Expression Kit (Invitrogen), according to the manufacturer's instructions. The cell culture medium containing the recombinant stromelysin-1 was cleared of cells by centrifugation and dialyzed against two to three changes of 0.02% Brij-35 in 20 mM HEPES buffer, pH 7.0. The dialysate was loaded on a 100-ml column of reactive red agarose (Sigma, St. Louis, MO) that was equilibrated with HCB buffer (0.02% Brij-35 and 5 mM CaCl₂ in 20 mM HEPES, pH 7.0). After the column was washed with HCB buffer, the protein was eluted using a 160-ml 0-2 M NaCl gradient in HCB buffer. For further purification, the eluted protein was dialyzed against HCB buffer, bound to a 5-ml column of Q-Sepharose Fast Flow (Pharmacia, Piscataway, NJ), and eluted with 160 ml of a 0-1 M NaCl gradient in HCB buffer. Mass spectrometry and N-terminal sequencing of the final product were consistent with the protein being intact and with no apparent modifications, such as glycosylation. Casein zymography indicated that the recombinant stromelysin-1 was proteolytically active. Before use in cell culture experiments, stromelysin-1 was dialyzed against DMEM/F12 (Life Technologies, Gaithersburg, MD). Stromelysin-1 at a concentration of 1 mg/ml was activated by incubation with trypsin (Life Technologies) at 1 µg/ml for 1 h at 37°C. Trypsin was subsequently inhibited with soybean trypsin inhibitor (Sigma) added at a final concentration of 10 µg/ml.

All function-blocking monoclonal antibodies against integrin subunits were obtained from PharMingen (San Diego, CA). Azide-free preparations of antibodies against 1 (clone Ha31/8), 5 (clone 1A29), v (clone H9.2B8), and 1 (clone Ha2/5) were purchased. Custom-made azide-free preparations of 6 antibodies (clone GoH3) were purchased. Azide-containing preparations of 2 antibodies (clone HM2) were purchased and used with similar results either before or after dialysis against phosphate-buffered saline, pH 7.4 (PBS).

Promoter Constructs

A genomic DNA clone containing the mouse stromelysin-1 gene identified in a SuperCos I (Stratagene, La Jolla, CA) library (gift from Dr. John S. Mudgett, Merck Research Laboratories, Rahway, NJ) was used to isolate an EcoRI-PstI genomic fragment encompassing ~1.3 kb of 5'-untranslated sequences and ~0.2 kb of the first exon of stromelysin-1. The EcoRI-PstI fragment was subcloned into Bluescript KS (Stratagene) and sequenced with the CircumVent sequencing kit (New England BioLabs, Beverly, MA) according to the manufacturer's instructions. The 5'-untranslated sequences were then amplified by PCR and subcloned into the KpnI-BglII cloning sites of the pGL2 vector (Promega, Madison, WI) upstream from a luciferase reporter gene. The control vector containing the Rous sarcoma virus (RSV) promoter attached to a -galactosidase reporter gene has been described (Li et al., 1992).

Cell Culture and Transfection

The mouse mammary carcinoma cell line SCg6 (Desprez et al., 1993; Lochter et al., 1997b) was maintained and passaged routinely in medium containing 5% fetal bovine serum, 5 µg/ml insulin (Sigma), and 50 µg/ml gentamycin (Life Technologies) as described (Desprez et al., 1993; Lochter et al., 1997b). All assays were performed in chemically defined medium consisting of DMEM/F12, 5 µg/ml insulin, 5 µg/ml transferrin, 5 ng/ml selenium (added as ITS medium supplement; Sigma), and 50 µg/ml gentamycin.

SCg6 cells were transfected with promoter constructs by the use of the Lipofectin reagent (Life Technologies) and pools of stable clones selected as described (Lochter et al., 1997a). In brief, 1.2 × 10⁶ cells were maintained in culture dishes 10 cm in diameter (Falcon; Becton Dickinson, Franklin Lakes, NJ) and incubated for 24 h with 5 ml of OptiMEM (Life Technologies) containing 40 µl of Lipofectin, 0.5 µg of SV40neo (Schmidhauser et al., 1992), 3.75 µg of stromelysin-1 promoter construct in pGL2, and 3.75 µg of RSV--galactosidase vector. Subsequently, the medium was replaced with medium containing serum. Two days after transfection, cells were selected by addition of 200 µg/ml Geneticin (Life Technologies) to the culture medium. Surviving cells, originating from ~200 surviving clones per dish, were pooled and expanded in medium containing serum.

Zymograms and Immunoprecipitations

For zymography of proteinases in conditioned medium, 10⁶ cells were plated in 1 ml of chemically defined medium into dishes 3.5 cm in diameter (Falcon). Medium was collected 2 d later, and gelatin and casein zymography was performed as described (Fisher and Werb, 1995; Lochter et al., 1997b). To detect gelatinases and caseinases, 2 and 20 µl, respectively, of conditioned medium was loaded per lane.

Immunoprecipitations on cell lysates were performed as described (Lochter et al., 1997a). In brief, cells were metabolically labeled for 16 h with 200 µCi of ³⁵S-methionine (Amersham, Arlington Heights, IL) per milliliter of culture medium. Radiolabeled cells were washed with chemically defined medium and lysed in Nonidet P-40 (NP-40) lysis buffer (150 mM NaCl, 50 mM Tris, pH 7.5, 1% NP-40). Antibodies to integrin subunits were added to cell lysates at a final concentration of 10 µg/ml and were incubated overnight at 4°C. Protein G agarose (Sigma) was then added, and incubation was performed for 1 h at 4°C. Protein G agarose beads were washed with NP-40 lysis buffer and subsequently with 50 mM Tris, pH 7.5. Immunoprecipitates were separated on 5% SDS-polyacrylamide gels under nonreducing conditions. Gels were then dried and processed for autoradiography.

Histochemistry

To analyze the morphology of cells maintained on reconstituted basement membrane (rBM; Matrigel; Collaborative Research, Bedford, MA) gels in the presence of 10 µg/ml anti-integrin antibodies, we fixed cells with 5% glutaraldehyde in PBS and stained the cells with toluidine blue O (Sigma) as described (Lochter et al., 1991). rBM gels were prepared according to published procedures (Lochter et al., 1997b).

Dimethylthiazolyl-diphenyltetrazolium Bromide (MTT) Assay

To determine the number of viable cells, we added 20 µl of 5 mg/ml MTT (Sigma) in PBS to 100 µl of the culture medium of cells maintained for 2 d on rBM gels in 96-well tissue culture plates (Falcon) in the presence of 10 µg/ml antibodies against integrins. Cells were incubated for 1 h at 37°C, whereupon the culture medium was removed and the metabolized MTT was solubilized with 50 µl of dimethylsulfoxide. Absorbance was measured at 570 nm.

Assays for Luciferase and -Galactosidase Activity

To measure the activity of the luciferase reporter gene, we plated cells at a density of 10,000 cells per well in 96-well multiwell plates. Cells were maintained on untreated tissue culture plastic, on substrata coated with rBM gels, or in suspension culture in which cell adhesion was blocked by coating wells with 2 mg/ml poly(2-hydroxyethyl methacrylate) (poly-HEMA; Sigma) as described (Roskelley et al., 1994). Antibodies against integrins were added at the concentrations indicated (see Figure 8). As determined by the MTT assay, the number of viable cells was unaffected by the highest concentration of each antibody used (our unpublished results). After 2 d in culture, cells were lysed with 1 mM dithiothreitol and 1% (wt/vol) Triton X-100 in 100 mM sodium phosphate buffer, pH 7.5. Luciferase activity was assayed in a luminometer (Wallac, Gaithersburg, MD) after addition of an equal volume of 150 µg/ml beetle luciferin (Promega), 9 mM ATP, 20 mM MgCl₂, 10 mM potassium phosphate, pH 7.5, and 50 mM HEPES, pH 7.4. Activity of -galactosidase was analyzed with a luminometer with the Galacto-Light Plus kit (Tropix, Bedford, MA), according to the manufacturer's instructions.

Cell Adhesion Assays

For adhesion assays, 96-well tissue culture plates were coated for 2 h at 37°C with laminin (purified from Engelbreth-Holm-Swarm sarcomas; Sigma), type I collagen (Collagen, Fremont, CA), type IV collagen (Collaborative Research), or fibronectin (Collaborative Research), all at 20 µg/ml in PBS. Wells were then washed twice with PBS, and nonspecific binding sites were blocked by incubation with 2 mg/ml heat-inactivated, fatty acid-free bovine serum albumin (BSA; Sigma) in PBS for 1 h at 37°C. Heat inactivation was achieved by incubation of BSA in PBS for 6 min at 80°C. After blocking, wells were washed twice with PBS before addition of 50 µl of chemically defined medium containing antibodies at twice the final concentrations (see Figure 4). Subsequently, 20,000 cells in 50 µl of chemically defined medium were added per well and allowed to adhere for 1 h at 37°C. Adhesion assays were terminated by the addition of 100 µl per well of 5% glutaraldehyde in PBS to the cell culture medium. Cells were fixed for 15 min at ambient temperature, washed three times with water, and stained for 1 h with 1 mg/ml crystal violet (Sigma) in water (Gillies et al., 1986). Finally, cells were washed five times with water and incubated for 1 h with 0.3% (wt/vol) Triton X-100 in water to solubilize crystal violet. Absorbance was measured at 600 nm.

Cell Migration and Invasion Assays

Cell migration assays were performed in modified Boyden chambers with polyterephtalate filter inserts for 24-well plates containing 8-µm pores (Collaborative Research). Both the bottom and the top of the filters were coated with 20 µg/ml laminin, 20 µg/ml fibronectin, or 100 µg/ml rBM in PBS by incubation for 2 h at 37°C. Subsequently, chambers were washed twice with PBS and incubated for 1 h with heat-inactivated fatty acid-free BSA. After two washes with PBS, the lower chamber was filled with 300 µl of medium, and 20,000 cells were plated in each upper chamber in 200 µl of medium. Immediately after cell plating, antibodies (10 µg/ml final concentration), the matrix metalloproteinase inhibitor GM6001 (10 µM), its inactive structural homologue GM1210 (10 µM) (gifts from Dr. R. Galardy, Glycomed Corporation, Alameda, CA) (Grobelny et al., 1992), or activated recombinant stromelysin-1 (1 µg/ml) was added. Migration assays were terminated 6 h after cell plating by fixation with 5% glutaraldehyde in PBS. Cells that had not migrated to the bottom of the filters were then removed, and the remaining cells were stained with toluidine blue and counted as described (Lochter et al., 1997b).

Invasion assays were performed in the same type of modified Boyden chambers in which the migration assays were performed and according to published procedures (Lochter et al., 1997b); 100,000 cells per well were plated on top of filters coated with 10 µl of rBM gels. After an incubation period of 24 h, cells were fixed and analyzed as described for the migration assay. All antibodies used in invasion assays were applied at a final concentration of 10 µg/ml immediately after cell plating. Recombinant stromelysin-1 was used at a final concentration of 1 µg/ml.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Integrin Subunits 1, 2, 6, and 1 Are Involved in rBM Invasion by Mammary Carcinoma Cells

To study the participation of integrins in tumor cell invasion, we used the highly aggressive mouse mammary carcinoma cell line SCg6 (Lochter et al., 1997b). In culture, SCg6 cells readily invade a rBM (Lochter et al., 1997b). SCg6 cells expressed integrin subunits 1, 2, 5, 6, and 1 but not v (Figure 1). Antibodies against integrin subunits 1, 2, 5, and 6 coimmunoprecipitated a band corresponding to the molecular weight of the 1 integrin subunit (Figure 1). The 4 integrin subunit was not found in immunoprecipitations with antibodies against the 6 integrin subunit (Figure 1), indicating that 61, not 64, was the predominant 6-containing integrin expressed by SCg6 cells. This is in contrast to nonmalignant mouse mammary epithelial cells that express the 64, but not the 61, integrin (Delcommenne and Streuli, 1995; Lochter et al., 1997a).

View larger version (93K): [in this window] [in a new window]   Figure 1.   Integrin expression by SCg6 cells. Autoradiogram of nonreducing SDS-polyacrylamide gels run on immunoprecipitations performed with antibodies against 1, 2, 5, 6, v, and 1 integrin subunits on radiolabeled cell lysates of SCg6 cells is shown. Note that, except for v, all integrin subunits examined could be detected in SCg6 cells and that all antibodies coimmunoprecipitated a band corresponding to the expected molecular mass (130 kDa) of the 1 integrin subunit (). Positions () of integrin subunits 1 at 180 kDa, 2 at 150 kDa, 5 at 160 kDa, and 6 at 150 kDa are indicated.

Antibodies against 1, 2, and 1, but not 5 or v or control IgGs, at 50 µg/ml (our unpublished results) or 10 µg/ml (Figure 2) abolished the invasion of SCg6 cells in modified Boyden chambers that were coated with rBM, whereas antibodies against 6 integrins reduced tumor cell invasion by one-half. There was no difference in plating efficiency or survival between the inhibitory and noninhibitory treatments (Figure 3A). However, antibodies against 1 integrins, but not any of the other antibodies tested, altered the morphology of cells plated on rBM. Whereas SCg6 cells on rBM formed a network of interconnected and elongated cell aggregates from which invasion occurred at the periphery of cell clusters, cells treated with anti-1 antibodies appeared to be primarily single and barely spread (Figure 3, B-F). The latter observation indicates that cell-rBM interactions are affected by antibodies against 1.

View larger version (24K): [in this window] [in a new window]   Figure 2.   Effects of antibodies against 1, 2, 6, and 1 integrin subunits on the invasion of rBM by SCg6 cells. Invasion assays were performed for 24 h in modified Boyden chambers coated with rBM. SCg6 cells were then fixed and stained, and the number of cells that had migrated through rBM was counted by microscopic inspection. Invasion assays were performed in the absence of antibodies (control [C]), in the presence of antibodies against 1, 2, 5, 6, v, or 1 integrin subunits, or in the presence of rat (r) or hamster (h) IgGs. Results are normalized with the number of cells migrating through rBM in the absence of antibodies set to 100. Means ± SD from three independent experiments are shown. In the absence of IgGs, 169.5 ± 8.89 cells/mm2 migrated through rBM.

View larger version (84K): [in this window] [in a new window]   Figure 3.   Cell number and morphologies of SCg6 cells maintained on rBM gels in the presence of antibodies against integrins. (A) SCg6 cells were maintained for 2 d on rBM, and the number of viable cells was determined by the MTT method. Means ± SD from three independent experiments are expressed as the absorbance (ABS) measured at 570 nm. Note that only minor differences in the number of viable cells were observed for the different antibody treatments. w/o, without antibody. (B-F) Micrographs of toluidine blue-stained SCg6 cells maintained for 2 d on rBM in the absence of antibodies (B) or in the presence of antibodies against 1 (C), 2 (D), 6 (E), and 1 (F) integrins are shown. Bar, 100 µm.

1 Integrins Are Used for Adhesion of SCg6 Cells to Basement Membrane Constituents

To analyze whether the integrins involved in rBM invasion by SCg6 cells are used to mediate cell adhesion to major basement membrane constituents, we subjected cells to adhesion assays in the presence of the integrin-blocking antibodies that interfered with rBM invasion. Within 1 h after plating, >90% of cells adhered to substrata coated with laminin, type IV collagen, fibronectin, and type I collagen (our unpublished results). Antibodies against 1 integrins prevented cell adhesion to laminin and type I collagen and reduced adhesion to type IV collagen and fibronectin (Figure 4). Antibodies against 1 integrins reduced cell adhesion to type I and IV collagens at 10 µg/ml and to laminin slightly at 50 and 100 µg/ml (Figure 4). In contrast, antibodies against 2 and 6 integrins alone (Figure 4) and in combination (our unpublished results) had little effect on SCg6 cell adhesion. Therefore, none of the  subunits tested is a major laminin receptor, whereas 1 integrins expressed by SCg6 cells appear to be the major adhesion receptor for type I collagen. These results raise the question of how 1, 2, and 6 integrins inhibit rBM invasion.

View larger version (37K): [in this window] [in a new window]   Figure 4.   Effect of integrin antibodies on adhesion of SCg6 cells to ECM constituents. Adhesion assays on substrata coated with laminin (A), type IV collagen (B), fibronectin (C), or type I collagen (D) were performed for 1 h in the presence of increasing concentrations of antibodies against 1 (), 2 (), 6 (), or 1 () integrin subunits. Results are normalized with the number of cells adhering to ECM substrata in the absence of antibodies set to 100. Means ± SD from three independent experiments are shown.

Integrin 6 and 1 Subunits Regulate Cell Motility

One possible mechanism by which integrins 1, 2, and 6 could regulate tumor cell invasion may be by interfering with cell motility. In cell migration assays performed using modified Boyden chambers coated with soluble rBM material to yield two-dimensional substrata (2D rBM), as opposed to the three-dimensional rBM gels used for the invasion assays, cell migration was almost completely inhibited by antibodies against 1 integrins (Figure 5A), and antibodies against 6 integrins inhibited SCg6 cell migration to an extent similar to that seen in rBM invasion (~50%; Figure 5A). Anti-1 and -6 antibodies gave similar results when laminin, instead of rBM, was used as a substratum (Figure 5B). Antibodies against 1, 2, 5, or v integrins or control IgGs had no significant effect on SCg6 cell migration on 2D rBM or laminin (Figure 5, A and B). Migration of SCg6 cells on fibronectin was not affected by anti-6 antibodies (Figure 5C). Thus, 6 integrins appear to contribute to rBM invasion by affecting the ability of SCg6 cells to migrate.

View larger version (25K): [in this window] [in a new window]   Figure 5.   Effects of antibodies against 6 and 1 integrins on SCg6 cell migration along 2D rBM and laminin substrata. Cell migration assays were performed for 6 h in modified Boyden chambers coated with rBM (A), laminin (B), or fibronectin (C). SCg6 cells were then fixed and stained, and the number of cells that had migrated along the ECM was counted by microscopic inspection. Cell migration assays were performed in the absence of antibodies (control [C]), in the presence of antibodies against 1, 2, 5, 6, v, or 1 integrin subunits, or in the presence of rat (r) or hamster (h) IgGs. Results are normalized with the number of cells that had migrated along the ECM in the absence of antibodies set to 100. Means ± SD from three independent experiments are shown. In the absence of IgGs, the number of cells migrating per square millimeter was 60.1 ± 7.55 on 2D rBM, 90.3 ± 12.25 on laminin, and 102.9 ± 13.27 on fibronectin.

Integrin Subunits 1 and 2 Mediate Invasion by Regulating Proteinase Expression

The results obtained thus far suggest that the mechanism by which 1 and 2 integrins contribute to SCg6 cell invasion is different from that of 6 and other 1 integrins, because neither cell attachment or adhesion to rBM nor cell migration was markedly affected by antibodies against these integrins. ECM-degrading proteinases are required for basement membrane invasion by tumor cells (MacDougall and Matrisian, 1995) and are regulated by integrins in some cell types (Heino, 1996). We therefore analyzed whether the secretion of proteinases by SCg6 cells was affected by antibodies against 1 and 2 integrins. SCg6 cells secreted the proenzyme form of gelatinase B, latent and active isoforms of gelatinase A, stromelysin-1, and an unidentified matrix metalloproteinase migrating at 80 kDa in casein substrate gels (Lochter et al., 1997b) (Figure 6). Antibodies against 2 and 6 integrins, but not against 1 or 1 integrins, reduced expression of both gelatinases slightly (Figure 6). Expression of stromelysin-1 activity was unaffected by antibodies against 6 and 1 integrins but was strongly inhibited by antibodies against 1 and 2 integrins (Figure 6). Expression of the 80-kDa caseinase was increased by antibodies against integrin subunits 1, 2, and 1 (Figure 6). These data are further evidence that the antibody effects were specific rather than attributable to toxicity of the antibody preparations.

View larger version (42K): [in this window] [in a new window]   Figure 6.   Effect of antibodies against integrins on the production of ECM-degrading proteinases. Medium conditioned by SCg6 cells maintained for 2 d on rBM in the absence of antibodies (w/o) or in the presence of antibodies against 1, 2, 6, or 1 integrin subunits was analyzed by gelatin and casein substrate gel zymography. Negative images of zymograms are shown. Molecular masses of major gelatinolytic and caseinolytic bands are indicated. Left, gelatinases of 72, 66, and 62 kDa represent gelatinase A, and the 102-kDa gelatinase represents gelatinase B. Right, the 57-kDa caseinase represents stromelysin-1, and the 80-kDa caseinase represents an unidentified metalloproteinase secreted by SCg6 cells (Lochter et al., 1997b).

Stromelysin-1 is required for the invasion of rBM by SCg6 cells (Lochter et al., 1997b). Therefore we hypothesized that antibodies against 1 and 2 integrins interfere with rBM invasion by SCg6 cells by impairing stromelysin-1 expression. To test this hypothesis, we reconstituted the system by adding back recombinant stromelysin-1. Addition of stromelysin-1 to the culture medium overcame the block in invasion induced by anti-1 and -2 antibodies (Figure 7A) but did not affect SCg6 cell invasion in the presence of antibodies against 6 or 1 integrins (Figure 7B). These data support the concept that 6 and other 1-containing integrins inhibit tumor cell invasion by interfering with the ability of cells to interact with and/or migrate along rBM. When recombinant stromelysin-1 was added to cultures not treated with antibodies, invasion was increased approximately threefold (288 ± 53.6% of control). This observation supports the notion that increased stromelysin-1 activity correlates with increased invasive activity of SCg6 cells, as long as 6 and 1 integrin functions are not compromised.

View larger version (19K): [in this window] [in a new window]   Figure 7.   Effect of recombinant stromelysin-1 (SL-1) on invasion and migration of SCg6 cells. (A and B) Invasion assays were performed for 24 h in modified Boyden chambers coated with rBM gels. SCg6 cells were then fixed and stained, and the number of cells that had migrated through rBM was counted by microscopic inspection. Invasion assays were performed in the absence of antibodies (control [C]) or in the presence of antibodies against 1, 2, 6, or 1 integrin subunits with (+ SL-1, black bars) or without ( SL-1, white bars) the addition of recombinant stromelysin-1 to the culture medium. Results are normalized with the num-ber of cells migrating through rBM in the absence of antibodies and stromelysin-1 set to 100. Means ± SD from three independent experiments are shown. In the absence of antibodies, 183.2 ± 16.21 cells/mm2 migrated through rBM. (C) Two-dimensional cell migration assays were performed for 6 h in modified Boyden chambers coated with rBM. SCg6 cells were then fixed and stained, and the number of cells that had migrated along rBM was counted by microscopic inspection. Cell migration assays were performed in the absence of additives (control) or in the presence of recombinant stromelysin-1, the matrix metalloproteinase inhibitor GM6001, or its inactive homologue GM1210. Results are normalized with the number of cells that had migrated along rBM in the absence of additives set to 100. Means ± SD from three independent experiments are shown. Under control conditions, 72.8 ± 9.16 cells/mm2 migrated along rBM.

SCg6 cells did not use stromelysin-1 to migrate on rBM. Addition of recombinant stromelysin-1 to cells plated on 2D rBM substrata did not affect cell migration (Figure 7C). Moreover, cell migration on 2D rBM was unaffected by the inhibitor of matrix metalloproteinases GM6001 (Figure 7C), which abrogates rBM invasion by SCg6 cells (Lochter et al., 1997b). Thus matrix metalloproteinases in general and stromelysin-1 in particular are not required for migration of SCg6 cells on rBM but, instead, modulate the invasive behavior of these cells by regulating proteolysis.

When we transfected SCg6 cells with stromelysin-1 promoter-luciferase reporter constructs, antibodies against 1 and 2, but not 6 or 1, integrins inhibited activity of the stromelysin-1 promoter in cells maintained on rBM, on tissue culture plastic, or in suspension culture (Figure 8, A, C, and D). Activity of the control RSV promoter construct, which was cotransfected with the stromelysin-1 promoter construct into the same cell population, was not inhibited by any of the antibodies (Figure 8B). Thus, the antibodies directed against 1 and 2 integrins interfered in the signaling pathway that regulates transcription of the stromelysin-1 gene.

View larger version (35K): [in this window] [in a new window]   Figure 8.   Effect of antibodies against 1 and 2 integrins on stromelysin-1 promoter activity in SCg6 cells. Luciferase (A, C, and D) and -galactosidase (B) activity of SCg6 cells cotransfected with SL-1 promoter-luciferase and RSV promoter--galactosidase constructs was analyzed 2 d after maintenance of cells on rBM gels (A and B), on tissue culture plastic (C), or in suspension culture (poly-HEMA, D). Cells were cultured in the presence of increasing concentrations of antibodies against 1 (), 2 (), 6 (), or 1 () integrin subunits (A and C) or in the presence of 10 µg/ml antibodies (B and D). Results are normalized with values obtained for cells maintained in the absence of antibodies (control [C] in B and D) set to 100. Means ± SD from three independent experiments are shown. In the absence of antibodies, the average number of relative light units obtained per well was 22371 on rBM, 6580 on plastic, and 20689 on poly-HEMA for the stromelysin-1 promoter and 9618 on rBM for the RSV promoter.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

Most cells have a large repertoire of integrins, yet individual integrins may play distinct roles in regulating proliferation, differentiation, and apoptosis (Hynes, 1992; Giancotti, 1997). Integrins play a central role in tumor cell invasion (Dedhar, 1995; Juliano, 1996). Because integrins could modulate invasion by affecting cell adhesion, cell motility, and gene expression, the molecular mechanisms underlying integrin function can only be understood by dissecting the function of individual integrins. In this study, we showed that different 1 integrins have distinct functions in cell adhesion, motility, and MMP expression in an aggressive tumor cell model system and that they act in concert to modulate invasive activity.

Antibodies against the 1 integrin subunit that block the interaction of all 1-containing integrins with ECM ligands completely inhibited the ability of SCg6 mammary carcinoma cells to invade rBM (Matrigel). These antibodies also abrogated cell migration along two-dimensional rBM and laminin substrata, impaired cell spreading on rBM, and reduced cell adhesion to laminin and type IV collagen without affecting plating efficiency or viability on rBM. The rBM contains a heterogenous mixture of ECM molecules (Kleinman et al., 1986) that could serve as ligands for integrin receptors that do not contain the integrin 1 subunit and for nonintegrin ECM receptors such as cell surface proteoglycans (Timpl, 1993; Mercurio, 1995). These receptors appear to be sufficient for maintaining adhesion and survival in the absence of functional 1 integrins.

Although antibodies against 1 integrins interfered with tumor cell invasion by inhibiting various aspects of cell-ECM interactions, antibodies against 6 integrin, a major laminin receptor (Hynes, 1992), affected cell migration and invasion but not adhesion or morphology of cells maintained on rBM substrata. Thus, 6 integrins appear to facilitate SCg6 tumor cell invasion selectively by promoting tumor cell migration. The predominant 6 integrin expressed by SCg6 cells was 61 and not 64, the major 6-containing integrin of nonmalignant mammary epithelial cells in culture (Delcommenne and Streuli, 1995; Lochter et al., 1997a; Stahl and Mueller, 1997; Weaver et al., 1997). Whereas loss of 64 and increased expression of 61 are positively correlated with tumor cell aggressiveness in prostate cancer and in mammary carcinoma cells (Cress et al., 1995), 64 and not 61 promotes tumor cell invasion in colorectal carcinoma (Chao et al., 1996; Rabinovitz and Mercurio, 1996). And in several glioma cell lines, antibodies against 6 increase tumor cell invasion (Paulus and Tonn, 1994). It was also shown that integrin 61 is required for the interaction of murine melanoma cells with laminin (Hangan et al., 1997). In contrast to our study, both tumor cell adhesion to and cell migration on laminin could be inhibited with the anti-6 antibody GoH3. However, another antibody binding in close proximity to the GoH3 epitope on 6 only interfered with cell migration but not with cell adhesion. Thus, the 61 integrin appears to have distinct but overlapping functions in different tumor cell types.

Our data show that 1 and 2 integrins contribute to rBM invasion. A role for 1 integrins in tumor cell invasion has not been reported previously, but 2 integrins have been shown to be both stimulatory and inhibitory for invasion of different cell types (Zutter et al., 1995; Vihinen et al., 1996). The mechanism by which 1 and 2 integrins affect tumor cell invasion in our study appears to be different from that of 6 and other 1-containing integrins. Antibodies against 2 integrins had little effect on plating efficiency or cellular morphology on rBM, on the ability of cells to migrate on two-dimensional rBM or laminin substrata, or on cell attachment to laminin or type IV collagen. Antibodies against 1 integrins inhibited SCg6 cell adhesion to type IV collagen by ~50% only, without affecting cell adhesion to laminin. Instead, 1 and 2 integrins selectively regulated expression of matrix metalloproteinase stromelysin-1 at the transcriptional level. The block in invasion by antibodies against 1 and 2 integrins could be overcome by addition of recombinant stromelysin-1, indicating that loss of stromelysin-1 was responsible for the loss-of-function. The amount of stromelysin-1 used to recover invasive activity was similar to the amount that the cells secreted in the absence of anti-1 or anti-2 antibodies and similar to the amount secreted by functionally normal mouse mammary epithelial cells transfected with a stromelysin-1 cDNA (Lochter et al., 1997a,b; our unpublished results), a manipulation that resulted in acquisition of premalignant properties in these cells (Lochter et al., 1997a). Whereas the recombinant stromelysin-1 was activated before its use in cell culture, most of the stromelysin-1 produced by mouse mammary carcinoma cells is in its latent form. We have argued previously that stromelysin-1 is locally activated by cells at sites where basement membrane degradation is required for invasion (Lochter et al., 1997b). This is supported by the fact that, when a latent form of recombinant stromelysin-1 was used in the invasion assay, inhibition of invasion by antibodies against 1 and 2 integrins could also be overcome, although less efficiently than with activated stromelysin-1 (our unpublished results). Our data additionally support our previous observations that stromelysin-1 plays a critical role in mouse mammary tumor cell progression (Lochter et al., 1997a) and is indispensable for rBM invasion by mouse mammary carcinoma cells (Lochter et al., 1997b).

The antibodies against 1 and 2 integrins used in this study are function blocking (Miyake et al., 1994; Mendrick et al., 1995; Noto et al., 1995) rather than function stimulating. Thus, it is likely that these antibodies inhibit stromelysin-1 expression by interfering with integrin binding to a ligand produced endogenously. Laminin, which is expressed by SCg6 cells (Galosy, Werb, Bissell, unpublished results), is the most likely candidate to mediate regulation of stromelysin-1 expression by signaling via 11 and 21 integrins. Laminin positively modulates stromelysin-1 expression (Lochter et al., 1997b), whereas type I collagen, type IV collagen, and tenascin-C, which are other potential ligands for 11 and 21 integrins, do not affect stromelysin-1 expression (Lochter et al., 1997b; our unpublished results). Therefore, we propose that the basal level of stromelysin-1 expression in SCg6 cells is maintained by the binding of endogenously produced laminin to 11 and 21 integrins and that stromelysin-1 expression increases as a result of further receptor occupancy by exogenous laminin (or rBM) (Lochter et al., 1997b). Both of these processes would then be blocked by antibodies against 1 and 2 integrins. Antibodies against 1 integrins, which should also block integrins 11 and 21, were without effect on stromelysin-1 protein expression or promoter activity. It is at present not clear why this is so; however, one possible explanation is that antibodies against 1 will block other integrins that affect processes that negatively impact on stromelysin-1 expression, leading to simultaneous inhibition of integrins that promote stromelysin-1 production (11 and 21) and of others that reduce it and thus neutralizing the effect of individual integrins on stromelysin-1 expression. One such integrin may be 31 that is expressed by SCg6 cells (our unpublished results) and mediates cell adhesion to laminin in other cell types (Gehlsen et al., 1989; Mercurio, 1995). Interestingly, in two glioblastoma cell lines, antibodies against 31 integrin stimulated expression of matrix metalloproteinase gelatinase A (Chintala et al., 1996), and increased gelatinase A production was associated with increased invasiveness of these cells (Chintala et al., 1996). However, because of the lack of blocking antibodies that react with the murine 31 integrin, we were unable to investigate the role of 31 in invasion and metalloproteinase production of SCg6 cells. Alternatively, the antibodies may have different effects on integrins, blocking in one case but maintaining the integrin in an active conformation, whereas in the other case, the integrin may be locked in an inactive conformation that would signal differently.

The involvement of integrins in regulation of matrix metalloproteinase expression is well documented (Werb et al., 1989; Seftor et al., 1992; Larjava et al., 1993; Huhtala et al., 1995; Riikonen et al., 1995; Kheradmand et al., 1998). In a recent study, it was shown that antibodies against 2 integrins block collagenase-1 production that is required for keratinocyte migration on type I collagen substrata (Pilcher et al., 1997). Our study provides the first direct demonstration that integrins modulate tumor cell invasion by altering expression of matrix metalloproteinase genes. These results provide a rationale for further investigation of how ECM molecules, ECM receptors, and ECM-degrading enzymes work together to bring about tumor progression.