Alpha7 integrin mediates cell adhesion and migration on specific laminin isoforms.

The laminin-binding α7β1 integrin receptor is expressed at high levels by skeletal and cardiac muscles and by certain melanocytic cells. We have assessed the potential role of the α7A/B integrin isoforms in mediating cell adhesion and motility and determined the laminin isoform specificity of this integrin. When MCF-7 breast carcinoma cells, normally nonadherent to laminin 1, were stably transfected with cDNA for mouse α7, they adhered with high efficiency and migrated on laminin 1 substrates. Function-perturbing monoclonal antibodies generated to mouse α7 subunit blocked both adhesion and migration of α7 transfectants on laminin 1 substrates. Additional studies with MCF-7 transfectants revealed that α7β1 binds well to laminin 1 and to a mixture of laminin 2 and 4 but not to laminin 5. Importantly, α7β1 was capable of promoting motility on both laminin 1 and laminin 2/4 substrates. However, MCF-7 cells transfected with cDNA for either α7A or α7B showed no significant differences in cell adhesion or motility on laminin 1 substrates. Although the role for the alternatively spliced cytoplasmic variants of α7 remains unknown, the results establish that α7β1 mediates cell adhesive activities on a restricted number of laminin isoforms.

The laminin-binding ␣7␤1 integrin receptor is expressed at high levels by skeletal and cardiac muscles and by certain melanocytic cells. We have assessed the potential role of the ␣7A/B integrin isoforms in mediating cell adhesion and motility and determined the laminin isoform specificity of this integrin. When MCF-7 breast carcinoma cells, normally nonadherent to laminin 1, were stably transfected with cDNA for mouse ␣7, they adhered with high efficiency and migrated on laminin 1 substrates. Function-perturbing monoclonal antibodies generated to mouse ␣7 subunit blocked both adhesion and migration of ␣7 transfectants on laminin 1 substrates. Additional studies with MCF-7 transfectants revealed that ␣7␤1 binds well to laminin 1 and to a mixture of laminin 2 and 4 but not to laminin 5. Importantly, ␣7␤1 was capable of promoting motility on both laminin 1 and laminin 2/4 substrates. However, MCF-7 cells transfected with cDNA for either ␣7A or ␣7B showed no significant differences in cell adhesion or motility on laminin 1 substrates. Although the role for the alternatively spliced cytoplasmic variants of ␣7 remains unknown, the results establish that ␣7␤1 mediates cell adhesive activities on a restricted number of laminin isoforms.
Laminins are adhesive glycoproteins found in basement membranes that promote diverse cellular responses. Cell adherence to laminin matrices plays an important role in maintaining normal tissue organization and in tissue renewal and repair. The interaction of cells with extracellular matrix macromolecules like laminins is mediated primarily by heterodimeric receptors from the integrin superfamily (reviewed in Ref. 1). Integrins provide linkage between the component elements of the extracellular matrix and the structural constituents inside the cell. Besides serving as adhesion receptors, integrins can transmit signals from the extracellular matrix to the cell interior that can activate several pathways, ultimately influencing an array of cellular properties including proliferation, differentiation, survival, and apoptosis (2).
The biological response to laminin appears to be cell typespecific, and this may be due in part to the specific integrin receptors expressed by individual cells. ␣7␤1, originally found in melanoma cells, is a muscle-specific integrin (6) that binds to laminin (7)(8)(9). Although there is only one ␣7 gene, complicated splicing mechanisms result in several ␣7 isoforms. Several studies have shown that alternative splicing generates two isoform subsets: (i) X1 and X2 and (ii) A, B, and C, which differ at extracellular and cytoplasmic regions, respectively (10 -13). Upon terminal differentiation of myoblasts, isoform switching and up-regulation of ␣7 expression are detected. The extracellular variants have altered sequence in the ligand binding domain and may have different laminin isoform specificity or affinity. The cytoplasmic isoforms, which share the common extracellular and transmembrane domains but differ at the cytoplasmic region, may trigger different biological functions when cells interact with laminin.
In the present study, we have stably transfected MCF-7 carcinoma cells, which normally do not adhere to laminin 1, with mouse ␣7 cDNA. We also generated function-perturbing monoclonal antibodies to mouse ␣7 integrin to inhibit ␣7-extracellular matrix interactions. Using these approaches, we demonstrated that both ␣7A and ␣7B mediate adhesion and migration of MCF-7 transfectants on laminin 1 and laminin 2/4 substrates; however, ␣7␤1 does not bind to laminin 5.

EXPERIMENTAL PROCEDURES
Materials-The human breast carcinoma cell line MCF-7 was from American Type Culture Collection and was grown in Dulbecco's modified Eagle's medium H-16 with 10% fetal bovine serum. Laminin 1 was purified from mouse Engelbreth-Holm-Swarm tumor as described previously (7). Human placental laminin was purchased from Life Technologies, Inc. and is known to be a mixture of laminin 2 and 4 (14,15). Purified human laminin 5 was kindly provided by Dr. Robert Burgeson (Cutaneous Biology Research Center, Boston, MA). Human plasma fibronectin was purchased from Collaborative Biomedical Products (Bedford, MA).
Antibodies against integrin subunits included the rat anti-human ␤1 mAb 1 A2B2 and rat anti-human ␣5 mAb B2G2, kindly provided by Dr.
Caroline Damsky (University of California, San Francisco, CA); mouse anti-human ␣2 mAb VM1, kindly provided by Dr. Vera Morhenn (SRI International, Menlo Park, CA). Anti-human ␣3 mAb P1B5 was purchased from Life Technologies, Inc.; rat anti-human ␣6 mAb GoH3 was purchased from AMAC (Westbrook, ME). The rabbit polyclonal antibody 22780 and 1211 were prepared in this laboratory against peptide sequences specific to ␣7A cytoplasmic region (NSPSSSFRTNYHR) and to ␣7B cytoplasmic region (GTIQRSNWGNSQWEGSDAH), respectively. The peptides coupled to keyhole limpet hemocyanin using carbodiimide were injected subcutaneously in New Zealand White rabbits. Serum titers were monitored and verified by immunoprecipitation and immunoblotting of ␣7A or ␣7B transfectants and of myoblasts and myotubes. Rat anti-mouse ␣7 mAbs CA5, CY4, and CY8 were also generated in our laboratory. Production and characterization of these monoclonal antibodies against ␣7 will be described in detail elsewhere. 2 CA5, CY4, and CY8 were used for immunoprecipitation and fluorescence-activated cell sorting; CY4 and CY8 were used for function-perturbing assays. Fluorescein-conjugated secondary antibodies were obtained from Jackson ImmunoResearch Laboratories (West Grove, PA). Streptavidin-horseradish peroxidase and ECL kit were purchased from Amersham Corp.
␣7 cDNA Constructs-The cDNA encoding the complete sequence of mouse ␣7 was amplified from reverse-transcribed cDNA and then subjected to polymerase chain reaction with primers AGAGCGTTGATCCC and CTGCTGTCCCAAG and ligated into Bluescript (Stratagene). For constructing ␣7A, the NsiI-XbaI fragment was amplified from a reverse transcription product of C2C12 mRNA with primers abtran (GCTGCT-CAGAGATGCATCC) and NheI-XbaI (AGTAAGTTGCTAGCATACG-TCTAGAGC) and ligated into pGEM 11f (Promega). The NsiI-EcoRI fragment from pGEM was cut and ligated into Bluescript containing the ␣7B sequence from which the NsiI-EcoRI cytoplasmic region had been deleted. Both ␣7A and ␣7B cDNA in Bluescript were further cloned into pRc/CMV (Invitrogen).
Transfection and Selection of ␣7 Transfectants-Transfection of MCF-7 cells was performed by the calcium phosphate precipitation method (Mammalian Transfection Kit, Stratagene). MCF-7 cells at 30% confluency in 10-cm plates were transfected with 25 g DNA/plate. Cells subsequently were selected in growth medium containing 500 g/ml G418. Individual clones were isolated after 2 weeks with cloning rings. At least 10 clones for each isoform were isolated and then tested for enhanced adhesion to immobilized laminin 1. Those showing positive results (ϳ30%) were verified to express ␣7 subunits by Western blot with polyclonal antibody 1211 for ␣7B, and polyclonal antibody 22780 for ␣7A. Clone G, expressing a high level of ␣7B, and clone 114, expressing a comparable amount of ␣7A, were chosen for further studies.
Flow Cytometry-Subconfluent cells were briefly trypsinized. Singlecell suspensions of 10 6 /ml were incubated with optimal concentrations of primary antibodies in wash buffer (2% normal goat serum in PBS) for 1 h on ice, washed three times, and incubated with the secondary fluorescein-labeled antibodies for 30 min on ice. After washing again three times, the cells were stained with propidium iodide (1 g/ml) to identify nonviable cells. Flow cytometry was performed on a FACScan flow cytometer (Becton Dickinson). Control samples consisted of cells with or without secondary antibody binding. Any nonviable cells stained with propidium iodide were eliminated from the analysis.
Immunoprecipitation of Surface Biotin-labeled Cells-Confluent cultures of cells were washed twice with PBS and then labeled with NHS-LC-Biotin (Pierce), 1 mg/ml in cold PBS at 4°C for 90 min. Cells were washed twice with 50 mM glycine blocking buffer and incubated in this buffer for 10 min at 4°C. The cells were lysed in lysis buffer (PBS with 0.1 M Tris, pH 7.5, 2% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride, and 1 mM N-ethylmaleimide). After preclearing with protein A beads, the lysate was mixed by rotation for Ն3 h with primary antibody and protein A beads. The beads were washed with the wash buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM MgCl 2 , 0.5% Nonidet P-40, 0.1% BSA) three times and heated at 100°C in SDS sample buffer for 5 min. The supernatant was divided into two aliquots: one for nonreducing samples and one for reducing with 2-mercaptoethanol. Samples were separated by 7.5% SDS-polyacrylamide gel electrophoresis under reduced and nonreduced conditions. The biotinylated proteins were detected by streptavidin-horseradish peroxidase and then ECL.
Cell Adhesion Assay-Microtiter plates (96-well Immulon plates, Dynatech) were coated with matrix proteins at the indicated concentrations in PBS for 1 h at 37°C in a humidified atmosphere. Plates were washed with PBS and incubated with medium containing 0.1% BSA for 60 min in a CO 2 incubator to block nonspecific adhesion. Single-cell suspensions were prepared in Dulbecco's modified Eagle's medium with 0.1% BSA at 4 ϫ 10 5 cells/ml, added in triplicate to 96-well plates, and then incubated for 30 -90 min at 37°C. Nonadherent cells were removed by shaking on a titer plate shaker (Lab-line Instruments) and washing with PBS. Cells were fixed with 1% formaldehyde, stained with 1% crystal violet, solublized in 2% SDS, and then read at 562 nm. Cells bound to wheat germ agglutinin (10 g/ml) or collagen type I (100 g/ml) on a separate 96-well plate were used to indicate 100% attachment. Background cell adhesion to 1% BSA-coated wells was subtracted. The effect of specific antibody was tested by preincubating the cells with the hybridoma supernatants or dilutions of purified antibody on ice for 30 min prior to the assay.
Migration Assay-Cell migration was assayed in a modified Boyden chamber (Neuroprobe, Bethesda, MD) as described previously (16). Briefly, an 8-m porosity polyvinylpyrolidone-free polycarbonate filter (Nucleopore, Pleasanton, CA) was precoated with ligand at the indicated concentration. The lower well of the chamber was filled with serum-free medium containing 0.1% BSA. In some studies, the lower chamber contained medium with or without basic fibroblast growth factor as indicated. Cell suspensions were prepared from subconfluent cultures and resuspended to a final concentration of 4 ϫ 10 5 cells/ml in serum-free medium containing 0.1% BSA. A 50-l aliquot of cell suspension was added to the upper chamber and then incubated for the indicated time at 37°C. Cells on the top of the filter were removed by wiping, and the filter was then fixed in 1% formaldehyde in PBS. Migrating cells were stained with 1% crystal violet, and nine randomly chosen fields from triplicate wells were counted at 400ϫ magnification.

Generation of Stable Transfectants Expressing High Levels of
␣7A and ␣7B-To analyze ␣7 function, we generated ␣7-transfected expressers from cells that lack endogenous functional laminin 1-binding integrins. We transfected cDNA encoding the ␣7A or ␣7B subunits into MCF-7 cells. MCF-7 cells normally adhere poorly to laminin 1 (17). At least 10 clones of each transfectant were isolated and characterized, and several high expressing clones were obtained for both ␣7A and ␣7B. High expressers for ␣7A (clone 114) and ␣7B (clone G) were chosen for further analysis. Fluorescence-activated cell sorting analysis with mAbs against different laminin-binding integrins showed that one of the high ␣7A-expressing cell lines, clone 114, expresses moderate levels of ␣2, ␣3, and ␣6 and high amounts of ␣7 that correspond to means of fluorescence intensity of 84.8, 34.1, 35.9, and 645.3, respectively (Fig. 1A). For the parental MCF-7, ␣2 (104.1) and ␣6 (25.4) levels were similar, but ␣3 was expressed at a mean fluorescence intensity of 67.0. We also found that the ␤1 integrin level was increased in clone 114 in compensation to the increased level of ␣7 on the cell surface (data not shown).
The immunoprecipitation analysis of surface biotinylated parental MCF-7 cells and clone 114 cells verified specificity of the mAb and in addition showed that ␣7␤1 integrin was expressed in transfectants but was not detectable in the parental cells (Fig. 1B). Immunoprecipitation of cell lysates with CY8, a mAb against the extracellular domain of ␣7, yielded the ␣7 subunit in clone 114; as expected for this integrin under nonreducing conditions, the ␤1 subunit partner comigrated with the ␣7 subunit (Fig. 1B, lane 2) (7,18). Following reduction, the ␤1 subunit exhibited decreased mobility, whereas the ␣7 subunit was cleaved to yield a 100-kDa fragment and an ϳ30-kDa fragment containing the cytoplasmic tail (Fig. 1B, lane 3). The ␣7␤1 bands were also detected by using polyclonal antibody 22780 to ␣7A and polyclonal antibody 1211 to ␣7B in clone 114 and clone G, respectively (data not shown). In other studies, we have found that transfection of MCF-7 cells with cDNA of ␣6 integrin leads to significant expression of the ␣6␤4 complex. 3 In the case of MCF-7 cells transfected with ␣7, the integrin does not associate with ␤4 but preferentially pairs with the ␤1 subunit. This is interesting in view of the fact that ␣6 and ␣7 have high amino acid sequence homology (6,10).
Adhesion of ␣7 Transfectants to Laminin-We initially assessed the ligand specificity of ␣7A and ␣7B by testing the transfectants in standard adhesion assays with laminin 1 and its fragments as substrates. Both the MCF-7 ␣7A-transfectant clone 114 and ␣7B-transfectant clone G adhered effectively to laminin 1 and its E8 fragment. In contrast, the parental MCF-7 cells did not adhere to laminin 1 or the E8 fragment ( Fig. 2A). As expected, neither the parental cells nor transfectants attached to laminin 1 fragments E4, E1Ј, or P1, which lack the E8 region containing the ␣7 binding site. The adhesion of ␣7Aexpressing clone 114 cells to laminin 1 was further evaluated using function-perturbing mAbs to ␣7 and other potential laminin-binding integrins. Whereas mAbs to ␣2, ␣3, and ␣6 integrins failed to inhibit the strong laminin 1-binding activity of these cells, mAbs to ␣7 (CY4, CY8) completely blocked adhesion (Fig. 2B). A nonfunction-perturbing mAb to ␣7 (CA5) had no effect on adhesion. CY4 and CY8 had a similar blocking effect on ␣7B-transfectant clone G cells (data not shown). These results confirm that ␣7 expressed in MCF-7 cells effectively binds to laminin 1 and the E8 fragment and that both ␣7 isoforms show similar activities for these ligands.
Next, we tested the adherence of parental MCF-7-and ␣7expressing clone 114 cells on preparations of human placental merosin (laminin 2 and 4) and purified human laminin 5.
In adhesion assays with laminin 5, we used as a positive control a human squamous carcinoma cell line (HSC-3) that binds strongly to laminin 5 via the ␣3 integrin, which is highly expressed in these cells (16). At laminin 5 concentrations from 0.3 to 3 g/ml, HSC-3 cells showed a dose-dependent increase in adhesion, whereas MCF-7 parental cells showed only moderate binding efficiency (Fig. 3A). However, the ␣7-transfected clone 114 cells bound poorly to the laminin 5 substrate. Under the same assay conditions, HSC-3 and clone 114 cells adhered efficiently to laminin 1 in a dose-dependent fashion, whereas the MCF-7 parental cells did not (Fig. 3B). In other studies with MCF-7 parental cells and ␣7 transfectants, adhesion to laminin 5 could be totally blocked with mAb against ␣3 (data not shown). As mentioned above, analysis of integrin profile indicates that in the ␣7-transfected clone 114 cells, ␣3␤1 levels are decreased ϳ50% compared with that of the parental cell population. Thus, there is a correlation between adhesion to laminin 5 and expression of ␣3. This clearly demonstrates that ␣7 integrin in MCF-7 cells binds to laminin 1 and laminin 2/4 but cannot efficiently mediate binding to laminin 5.
Laminin Induces Motility in ␣7 Transfectants-We next examined the locomotive response of parental MCF-7 cells and ␣7 transfectants to laminin 1 (Fig. 4, A and B) and laminin 2/4 ( Fig. 4C) substrates in a modified Boyden chamber assay. The parental MCF-7 cells are known to be poorly migratory on laminin 1 (20), and several growth factors have been shown to stimulate their motility on other ligands (21,22). However, both of the ␣7-transfected clones (114 and G) showed an enhanced motile response on laminin 1 compared with the parental cells; furthermore, in the presence of basic fibroblast growth factor (1 ng/ml) as a stimulant the motile response was enhanced. These results established that transfection of ␣7A or ␣7B is sufficient to convert MCF-7 cells into migratory cells on laminin 1. In addition, motility of clone 114 cells on this substrate was completely blocked by CY8 monoclonal antibody to ␣7 (Fig. 4B).

DISCUSSION
In this study, we have demonstrated that the ␣7␤1 integrin can mediate adhesion and migration on a restricted number of laminin isoforms. We used two approaches successfully: (i) a gain of function approach by transfecting ␣7 cDNA into MCF-7 cells and (ii) a loss of function approach by using functionperturbing antibodies against ␣7 on cells expressing this integrin. Evidence obtained from these approaches clearly shows that exogenously expressed ␣7 confers on MCF-7 cells both the ability to bind and the ability to migrate on laminins. While this work was in preparation, Echtermeyer et al. (9) reported that transfection of mouse ␣7 cDNA into the human 293 embryonic kidney cells conferred a motile phenotype on laminin 1. Interestingly, this enhancement in motility occurred even Parental MCF-7 cells served as a control. Cells (2 ϫ 10 4 ) were resuspended in culture medium and added to laminin-coated plates as described under "Experimental Procedures." Laminin 1 and its fragments were coated at 10 g/ml. Adherence of cells in 1% BSA-coated wells was treated as background binding and subtracted. In A-C, data are presented as percentages of the total cells added to each well. Values are the means of triplicate wells; bars show standard deviation. B, inhibition of clone 114 attachment to 10 g/ml laminin 1 by mAbs CY4 and CY8 to the ␣7 integrin subunit. Assays were done as in A except that the indicated samples were preincubated with 10 g/ml mAbs to integrin subunits. CY4 and CY8 mAbs inhibited adherence of the transfectants to laminin 1 to the same extent as A2B2 (anti-␤1). No inhibition by VM1 (anti-␣2), P1B5 (anti-␣3), or GoH3 (anti-␣6) was detected. Nonfunction-perturbing mAb CA5 (anti-␣7) is a control for CY4 and CY8 (anti-␣7) function-perturbing mAbs. C, inhibition of clone 114 attachment to human placental merosin (a mixture of laminin 2 and 4) by function-perturbing mAbs to integrins. Optimal concentrations of the blocking antibodies were predetermined by adhesion assay of though the parental cells expressed additional laminin 1-binding receptors.
We examined the ligand specificity of the ␣7 receptor using available laminin isoforms. In MCF-7 ␣7 transfectants, ␣7␤1 mediated binding to preparations of laminin 1 and to human placental laminins (a mixture of laminin 2 and 4). In contrast, laminin 5 was a poor substrate for ␣7-expressing cells. Eventually, when pure preparations are available, the functions of ␣7 should be tested on additional members of the laminin superfamily, especially laminin 2 (merosin), which is present in the basement membrane surrounding adult skeletal myofibers where ␣7 normally is detected. It is interesting that another laminin-binding integrin, ␣6␤1, sharing high amino acid sequence homology with ␣7, not only binds to laminin 1 and human placental merosins (laminin 2 and 4) but also binds efficiently to laminin 5 (23). In addition, ␣6 can pair with ␤1 or ␤4 subunit, whereas as we show here in the MCF-7 transfectants, ␣7 pairs only with ␤1 subunit. Thus, even though ␣6 and ␣7 share strong amino acid sequence identity, there must be distinct domains that define both ligand specificity and pairing preferences. In the parental MCF-7 or transfectants, moderate levels of ␣2, ␣3, and ␣6 are expressed, yet these integrins are not capable of mediating adhesion or migration on laminin 1. It appears that this set of integrins, in contrast to ␣7, is not constitutively active for laminin 1. However, on laminin 2/4, this same set of integrins in the parental MCF-7 cells can mediate both adhesion and motility, yet in the transfectants is not fully competent for these adhesive interactions.
It is interesting that ␣7 transfection appeared to decrease the ability of existing integrins on the MCF-7 cells to interact with laminin 2/4 and with laminin 5. One possibility for the decrease in the activity of integrins ␣2 and ␣3 is due to a decrease in their expression. Fluorescence-activated cell sorting analysis confirms that the high expression of ␣7 caused a modest decrease in the expression level of ␣3. Another contributing factor may be that the high ␣7 expression produces a dominant negative effect that down-regulates the activity of the other integrins. Studies have suggested that certain integrins can produce a modulating effect on the function of other integrins. For example, in lymphocytes activation of LFA-1 can down-regulate ␣4 activity (24). A different phenomenon is observed in the ␣5-deficient CHO cells where full activity of ␣v␤3 receptor requires the presence of transfected ␣5 integrin (25). Thus an integrin may induce down-regulation of another integrin's function, or alternatively two integrins may cooperate with each other to modulate function.
It is conceivable that the cytoplasmic variants of ␣7 function differently. RNA alternative splicing events in the ␣ cytoplasmic region have been detected in several integrin molecules, including ␣3, ␣6, and ␣7 (10 -12, 26 -29). However, functional significance of the ␣ chain-cytoplasmic isoforms has not been well established. Results from recent studies searching for functional differences between ␣6A and ␣6B are still controversial (23, 30 -33). Our results indicate that ␣7A and ␣7B receptors are equally active in their adhesive or migratory activities. It is possible that ␣7A/B, a muscle-specific integrin, will show differential activities only in the context of a musclespecific environment.
In summary, the results presented here show that ␣7␤1 can mediate both cell adhesion and migration on laminin 1 and laminin 2/4. Importantly, we have demonstrated that ␣7 can interact with these different laminin substrates but not with epithelial cell-specific laminin 5. These results strongly support the role of the ␣7 receptor in mediating interactions with specific laminin isoforms. The tissue-specific expression of different family members of integrins and laminins (e.g. ␣7␤1 and merosins in skeletal muscle (13,34)) suggests that there is a selective interaction that may be important in both embryonic development and tissue homeostasis.