Adhesion-Independent α6β4 Integrin Clustering Is Mediated by Phosphatidylinositol 3-Kinase

Integrins are protein heterodimers that function as the major receptors for cell adhesion to extracellular matrix in multicellular animals (1). Integrins also mediate signaling through a variety of signal transduction pathways. Monovalent ligand interactions with dispersed integrin heterodimers generally induce limited signaling, whereas clustering of cell-surface integrins greatly augments signal transduction (2). Breast carcinoma cell line MDA-MB-231 is known to have a high level of cell-surface α6β4 integrin expression (3). Ligation of the α6β4 integrin in adherent MDA-MB-231 cells results in activation of phosphatidylinositol 3-kinase (PI3K) and phosphorylation of Akt (4). The PI3K-Akt signal transduction pathway inhibits apoptosis, and activation of this pathway demonstrates the prosurvival function of the α6β4 integrin in adherent cells (5). Tumor cell motility has also been shown to be mediated by the α6β4 integrin, and although tumor cell motility requires PI3K, it appears to be adhesion independent and does not require activation of Akt (6).

Integrin-mediated signaling in epithelial cells is generally investigated by adhering cells to antibody- or ligand-coated substrates, but evaluation of some signaling mechanisms may be hindered by simultaneous activation of non-integrin-mediated events during cell adhesion (7). As part of a long-term goal to delineate the adhesion-independent integrin signaling pathways important for cell motility, we began by examining the mechanism of adhesion-independent α6β4 integrin clustering in MDA-MB-231 cells.

Breast carcinoma cell line MDA-MB-231 was cultured in Eagle’s MEM supplemented with 5% fetal bovine serum, l-glutamine, sodium pyruvate, and nonessential amino acids and vitamins (Life Technologies, Inc., Rockville, MD). The cells were maintained in monolayer culture in a humidified incubator at 37°C in an atmosphere of 5% CO₂ and 95% air.

MDA-MB-231 cells were analyzed on a flow cytometer (FACScan, Becton Dickinson, San Jose, CA) with standard techniques and the following commercially available antibodies: rat monoclonal anti-α6 (clone GoH3, BD Biosciences, San Diego, CA), mouse monoclonal anti-β4 (clone 3E1, Chemicon, Temecula, CA), fluorescein isothiocyanate (FITC)-labeled goat polyclonal antirat IgG (BD Biosciences), and phycoerythrin-labeled goat polyclonal antimouse IgG (Zymed, South San Francisco, CA).

Cells were serum-starved overnight, trypsinized from the culture dishes, and washed twice with PBS. The cells were then resuspended in MEM containing 0.1% bovine serum albumin at a concentration of 5 × 10⁶ cells/mL. For integrin cross-linking, cells in suspension were incubated with mouse monoclonal anti-β4 (clone 3E1, Chemicon) or mouse monoclonal anti-MHC I control (clone W6.32, Sigma, St. Louis, MO) on ice for 40 minutes, washed, and then incubated with polyclonal antimouse IgG (Sigma) at 37°C for 30 minutes. Alternatively, cells in suspension were incubated with mouse monoclonal anti-β4 or mouse monoclonal anti-MHC I control on ice for 40 minutes, washed, and then added to plates coated with polyclonal antimouse IgG and incubated at 37°C for 1 h. For some experiments, cells were preincubated with PI3K inhibitor LY294002 (50 μmol/L, Cell Signaling, Beverly, MA), wortmannin (100 nmol/L, Sigma), protein kinase C inhibitor GF109203X (5 μmol/L, Biosource, Camarillo, CA), rapamycin (50 nmol/L, Biosource), or heparin (0.5 mg/mL, MP Biomedicals, Irvine, CA) followed by integrin cross-linking in the presence of the inhibitors.

After cross-linking α6β4 on cells in suspension in the presence or absence of inhibitors and cytocentrifuging the cells onto a glass slide, immunofluorescence staining was performed on treated and control cells with mouse monoclonal anti-β4 (clone ELF1, Novocastra, Norwell, MA) as the primary antibody and FITC-labeled antimouse IgG (Zymed) as the secondary. We defined clustering as an aggregation of fluorescent β4 integrin subunit signals resulting in at least a 10-fold increase in the size of the individual signals compared with controls. The percentage of cells with clusters was determined by counting at least 100 cells in each experimental group. Any cell with one or more prominent clusters was regarded as a cell with clustering.

After cross-linking α6β4 on cells in suspension or under adherent conditions as described above, cells were lysed on ice for 30 minutes with lysis buffer containing 50 mmol/L HEPES at pH 7.4, 150 mmol/L NaCl, 1% Triton X-100, 1 mmol/L CaCl₂, 1 mmol/L MgCl₂, 10% glycerol, 100 mmol/L NaF, 1 mmol/L sodium orthovanadate, 10 mmol/L sodium PP_i, 1 mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL leupeptin, and 10 μg/mL aprotinin. Aliquots of lysates with equal amounts of total protein were separated on 7.5% SDS-PAGE gels under reducing conditions and transferred to nitrocellulose filters. Filters were probed with rabbit polyclonal antibodies to Akt and phospho-Akt (serine 473, Cell Signaling). After incubating the filters with horseradish peroxidase-linked streptavidin (Vector, Burlingame, CA), proteins were detected with the ECL Western blotting detection reagents (Amersham, Piscataway, NJ).

Anti-p85 immunoprecipitation products were prepared in quadruplicate from cell lysates containing the same amount of protein, and the immunoprecipitated PI3K enzyme was used in a competitive ELISA-based assay (Echelon Biosciences, Salt Lake City, UT) according to the manufacturer’s instructions. Briefly, the reaction products were mixed and incubated with a PI(3,4,5)P₃ detector protein and then added to a PI(3,4,5)P₃-coated microplate for competitive binding. A peroxidase-linked secondary detection reagent was used to detect PI(3,4,5)P₃ detector protein binding to the plate, and the amount of PI(3,4,5)P₃ produced was calculated from a standard curve prepared from known concentrations of lipid product.

Comparison of the means of paired data was performed with a t test, and a P value less than 0.05 was considered to be significant.

Expression of α6β4 on MDA-MB-231 breast carcinoma cells was confirmed by flow cytometry (Fig. 1). Cell-surface α6β4 integrins were cross-linked in suspension by treating cells with a monoclonal antibody to the β4 integrin subunit at 4°C, followed by anti-IgG at 37°C. Immunofluorescence microscopy revealed that adhesion-independent cross-linking of the surface integrins results in prominent integrin cluster formation (Fig. 2,A and B). Clustering of α6β4 was significantly blocked when cross-linking was performed in the presence of either LY294002 (ref. 8; P < 0.001) or wortmannin (ref. 9; P = 0.001), both of which are PI3K inhibitors (Fig. 2,C and F). In contrast, no significant inhibition of clustering was observed with protein kinase C inhibitor GF109203X (ref. 10; Fig. 2,D), rapamycin (ref. 11; Fig. 2,E), or heparin (ref. 12; Fig. 2 G). The inhibition of α6β4 integrin clustering by LY294002 and wortmannin was not a result of decreased cell viability. As assessed by trypan blue exclusion, viability of MDA-MB-231 cells after cross-linking α6β4 in suspension without inhibitors was 89%. Cell viability after cross-linking α6β4 in the presence of LY294002 and wortmannin was essentially unchanged at 92% and 92%, respectively.

Consistent with previously reported data (4), an increase in phospho-Akt was observed when α6β4 was cross-linked by treating cells with a monoclonal antibody to the β4 integrin subunit at 4°C, followed by adhesion to an anti-IgG-coated plate at 37° (Fig. 3, Lanes 1 and 2). However, although the amount of phospho-Akt increased compared with actin controls, the nonphosphorylated Akt also increased, indicating that increased Akt available for phosphorylation, rather than activation of PI3K lipid kinase activity alone, appears to account for some of the increased phospho-Akt. In contrast, no increase in either nonphosphorylated Akt or phospho-Akt was observed when α6β4 was cross-linked in suspension for the amount of time (30 minutes) required for adhesion-independent α6β4 clustering (Lanes 3 and 4), although the baseline level of phospho-Akt in suspended cells was higher than in cells adherent to an anti-IgG coated plate. There was actually a small decrease in phospho-Akt after cross-linking α6β4 in suspension.

After cross-linking α6β4 under adherent conditions, there was a suggestion of increased PI3K lipid kinase activity (Fig. 3,A and B), although the increase did not quite reach statistical significance (P = 0.06). In contrast, we observed a decrease in PI3K lipid kinase activity when α6β4 was cross-linked on cells in suspension (P = 0.02; Fig. 3 C and D).

Our data show that adhesion-independent α6β4 integrin clustering can be induced in breast carcinoma cells in suspension by cross-linking the surface α6β4 integrins with antibodies. Clustering does not occur passively but is driven by a signaling mechanism that can be blocked with the PI3K inhibitors LY294002 and wortmannin. LY294002, although highly specific for PI3K, is also known to inhibit casein kinase 2 (13). Casein kinase 2 is known to phosphorylate nonmuscle myosin II heavy chain (14), which appears to play a role in the “capping” of some cell surface antigens (15, 16). Integrin clustering may share some features of antigen capping, hence, it is important to exclude involvement of casein kinase 2. Casein kinase 2 is extremely sensitive to inhibition by heparin (17). Lack of inhibition of integrin clustering by GF109203X or heparin indicates that neither protein kinase C nor casein kinase 2 are required for integrin clustering. Inhibition of integrin clustering by both LY294002 and wortmannin and lack of inhibition by GF109203X or heparin effectively demonstrates that adhesion-independent clustering of α6β4 requires PI3K activity.

Ligation of the α6β4 integrin on adherent epithelial cells has previously been shown to activate PI3K (6). There are three classes of mammalian PI3K enzymes (18). Class I enzymes are generally heterodimers composed of a catalytic subunit and a regulatory subunit. The regulatory p85 subunit of PI3K is thought to stabilize the p110 catalytic subunit and to inhibit PI3K activity. Activation of PI3K occurs after tyrosine phosphorylation of p85, which relieves the inhibition of p110 (19). The NH₂-terminal src homology 2 (SH2) domain of p85 docks with a phosphotyrosine-containing binding motif, usually in activated membrane receptor complexes, allowing translocation of p85–p110 to the cell membrane. The α6β4-mediated activation of PI3K in adherent epithelial cells produces active lipid products which, in turn, lead to phosphorylation and activation of Akt. The PI3K-Akt signal transduction pathway inhibits apoptosis, and activation of this pathway demonstrates the prosurvival function of the α6β4 integrin in adherent epithelial cells (5).

Although activation of the PI3K-Akt pathway has been reported for MDA-MB-231 cells after cross-linking α6β4 under adherent conditions (4), the increase in PI3K lipid kinase activity that we observed when α6β4 was cross-linked under adherent conditions did not appear substantial. Moreover, at least part of the increase in phospho-Akt under these conditions resulted from increased synthesis of Akt rather than increased PI3K lipid kinase activity alone. Nevertheless, there was clearly no increase in either phospho-Akt or PI3K lipid kinase activity when α6β4 was cross-linked in suspension. In fact, both appeared to decrease; nevertheless, the same adhesion-independent conditions induced α6β4 clustering, and the clustering could be blocked by PI3K inhibitors.

The p110α subunit of PI3K has both lipid and protein kinase activity (18). The protein kinase is a serine kinase with little known biological significance. Although the principal target of the PI3K protein kinase appears to be p85, additional protein substrates have been identified (20, 21). The lipid kinase phosphorylates phosphatidylinositol and activates a number of downstream effectors, thereby accounting for most of the known biological functions of PI3K, including stimulation of the PI3K-Akt pathway (18). LY294002 and wortmannin are known to inhibit both lipid and protein kinase activities of PI3K by blocking the catalytic pocket that contains both kinases (20). Because cross-linking surface α6β4 on suspended cells produces no increase in either PI3K lipid kinase activity or phosphorylated Akt, PI3K-dependent clustering of α6β4 in nonadherent cells may be mediated by the protein kinase of PI3K rather than by the lipid kinase. Because protein kinase activity of PI3K occurs at the expense of lipid kinase activity, this may explain the decrease we observed in PI3K lipid kinase activity. If this is the case, it will be an important challenge to identify the direct substrate of the PI3K protein kinase in this pathway.

A recent report indicates that chemokine-induced clustering of the αLβ2 integrin in murine T lymphocytes requires PI3K lipid kinase activity (22). Chemokine-induced clustering, however, may involve multiple pathways that lead to integrin clustering, and it is unclear whether lipid kinase activity is absolutely necessary for αLβ2 integrin clustering in lymphocytes. Moreover, αLβ2 is expressed in leukocytes, and integrin signaling in leukocytes may be substantially different from signaling in epithelial cells (22). Minimum requirements for integrin clustering may be more clearly identified from studies with ligand- or antibody-induced integrin clustering rather than with activation of other cell surface receptors such as chemokine receptors.

Our data show that adhesion-independent integrin clustering can be induced in breast carcinoma cells in suspension by cross-linking the surface integrins with antibodies. Clustering does not occur passively but is driven by a PI3K-dependent signaling mechanism. Our findings suggest that a novel PI3K pathway may be involved (Fig. 4). If PI3K plays different roles in adhesion-independent integrin clustering and adhesion-dependent cell survival pathways, it may serve as an important regulatory switch in integrin-mediated signal transduction.