Human neutrophil adherence to laminin in vitro. Evidence for a distinct neutrophil integrin receptor for laminin.

We used mAbs against polymorphonuclear leukocyte (PMN) surface proteins to investigate the mechanisms by which stimulated human neutrophils (PMNs) adhere in vitro to laminin, the major glycoprotein of mammalian basement membrane. mAb IB4, which is directed against the common beta 2 chain of the CD11/CD18, only partially inhibited the adherence of PMA-stimulated PMNs to both laminin and to subendothelial matrices. In contrast, IB4 completely inhibited PMA-stimulated PMN adherence to gelatin, fibronectin, collagen IV, and endothelial cell monolayers. PMA-stimulated PMNs from a patient with severe congenital CD11/CD18 deficiency also adhered to laminin, but not to gelatin or endothelial cell monolayers. Therefore, PMA-stimulated PMNs adhere to laminin by both CD11/CD18-dependent and CD11/CD18-independent mechanisms. Expression of CD11/CD18-independent adherence to laminin was agonist dependent, occurring after stimulation with the calcium ionophore A23187 and recombinant TNF-alpha, but not with the chemotactic factors FMLP, platelet activating factor, or recombinant human C5a. Expression of CD11/CD18-independent adherence was also divalent cation dependent, occurring in the presence of Mg2+ but not Ca2+ as the sole added divalent cation. The mAbs AIIB2 and 13, which are directed against the beta 1 subunit of the VLA integrins, significantly inhibited the CD11/CD18-independent adherence of normal PMNs to laminin, and completely abolished the adherence of CD11/CD18-deficient PMNs to laminin. Both anti-beta 1 mAbs bound to PMNs, as demonstrated by flow cytometry, and immunoprecipitated a membrane molecule of Mr 130,000 daltons from 125I-labeled, detergent-solubilized PMNs. These data suggest that human PMNs possess beta 1 and beta 2 classes of integrins, and that both mediate PMN adherence.

Activation of CD4 cells by fibronectin and anti-CD3 antibody. A synergistic effect mediated by the VLA-5 fibronectin receptor complex.

In this study, fibronectin synergized with anti-CD3 antibody to promote CD4 cell proliferation in a serum-free culture system. The cell-adhesive domain plus additional regions of the fibronectin molecule are involved in this synergy. Anti4B4(CDw29) antibody blocked the activation of CD4 cells in this system. Furthermore, it is the VLA-5 protein within the set of molecules recognized by anti-4B4 that serves as a fibronectin receptor on the CD4 lymphocytes. The VLA-5 fibronectin receptor was mainly expressed on CD4+ CD45R-CDw29+ cells and may in part contribute to the unique function of these cells.

Characterization of a 140-kD avian cell surface antigen as a fibronectin-binding molecule.

A 140,000-D protein cell surface antigen (140k) complex has been implicated in fibronectin-mediated cell-substratum attachment. We have used three different experimental systems to evaluate the hypothesis that this 140k complex can function as a fibronectin receptor. A monoclonal antibody that binds to the 140k complex specifically inhibits the direct binding of 3H-labeled 75,000-D fibronectin cell-binding fragment (f75k) to chicken embryo fibroblasts in suspension. The 140k complex is retarded in its passage through an affinity column consisting of immobilized f75k, and this interaction is specifically inhibited by a synthetic peptide that contains the fibronectin cell-recognition signal sequence. Finally, exogenous purified 140k complex inhibits the attachment and spreading of chicken embryo fibroblasts on fibronectin-coated substrates. Thus, our results indicate that the 140k complex can bind directly to fibronectin and is likely to be a fibronectin receptor for chicken cells.

Regulation of fibronectin receptor distribution.

To determine the role of each intracellular domain of the fibronectin receptor in receptor distribution, chimeric receptors were constructed containing the human interleukin-2 receptor (gp55 subunit) as the extracellular and transmembrane domains, in combination with either the alpha 5 or beta 1 intracellular domain of the fibronectin receptor as the cytoplasmic domain. These chimeric receptors were transiently expressed in normal fibroblasts, and their localization on the cell surface was determined by immunofluorescence using antibodies to the human interleukin-2 receptor. The alpha 5 chimera was expressed diffusely on the plasma membrane. The beta 1 chimera, however, colocalized with the endogenous fibronectin receptor at focal contacts of cells spread on fibronectin. On cells spread in the presence of serum, the beta 1 chimera colocalized both with the fibronectin receptor at sites of extracellular fibronectin fibrils and with the vitronectin receptor at focal contacts. The beta 1 intracellular domain alone, therefore, contains sufficient information to target the chimeric receptor to regions of the cell where ligand-occupied integrin receptors are concentrated. The finding that the beta 1 chimeric protein behaves like a ligand-occupied receptor, even though the beta 1 chimera cannot itself bind extracellular ligand, suggests an intracellular difference between occupied and unoccupied receptors, and predicts that the distribution of integrin receptors can be regulated by ligand occupancy. We tested this prediction by providing a soluble cell-binding fragment of fibronectin to cells spread on laminin. Under conditions preventing further ligand adsorption to the substrate, this treatment nevertheless resulted in the relocation of diffuse fibronectin receptors to focal contacts. Similarly, a redistribution of diffuse vitronectin receptors to focal contacts occurred on cells spread on laminin after the addition of the small soluble peptide GRGDS. We conclude that the propensity for receptor redistribution to focal contacts driven by the beta 1 cytoplasmic domain alone is suppressed in heterodimeric unoccupied fibronectin receptors, and that ligand occupancy can release this constraint. This redistribution of integrin receptors after the binding of a soluble substrate molecule may provide a direct means of assembling adhesion sites.

Neurite extension of chicken peripheral nervous system neurons on fibronectin: relative importance of specific adhesion sites in the central cell-binding domain and the alternatively spliced type III connecting segment.

Fibronectin contains at least two domains that support cell adhesion. One is the central cell-binding domain that is recognized by a variety of cell types, including fibroblasts. The second, originally identified by its ability to support melanoma cell adhesion, is located in the alternatively spliced type III connecting segment (IIICS). Using specific adhesive ligands and inhibitory probes, we have examined the role of each of these domains in fibronectin-mediated neurite extension of neurons from chick embryo dorsal root and sympathetic ganglia. In studies using explanted ganglia, both fl3, a 75-kD tryptic fragment of human plasma fibronectin containing the central cell-binding domain, and CS1-IgG, a synthetic peptide-IgG conjugate containing the principal cell adhesion site from the IIICS, supported neurite outgrowth after adsorption onto the substrate. The maximal activities of fl3 and CSl-IgG were 45-55% and 25-30% that of intact fibronectin, respectively. Co-coating of the substrate with f13 and CS1-IgG produced an additive stimulation of neurite outgrowth, the extent of which approached that obtained with fibronectin. Similar results were obtained with purified neuronal cell preparations isolated by tryptic dissociation of dorsal root ganglia. In complementary studies, blockage of the adhesive function of either the central cell-binding domain (with mAb 333, an antiadhesive monoclonal antibody) or the IIICS (with CS1 peptide), resulted in approximately 60 or 30% reduction in fibronectin-mediated neurite outgrowth, respectively. When tested in combination, the inhibitory activities of mAb 333 and CSl were additive. From these results, we conclude that neurons from the peripheral nervous system can extend neurites on both the central cell-binding domain and the IIICS region of fibronectin, and that these cells are therefore the first normal, embryonic cell type shown to adhere to the IIICS. These results suggest that spatiotemporal fluctuations in the alternative mRNA splicing of the IIICS region of fibronectin may be important in regulation of cell adhesive events during development of the peripheral nervous system.

Identification of an alternatively spliced site in human plasma fibronectin that mediates cell type-specific adhesion.

We have compared the molecular specificities of the adhesive interactions of melanoma and fibroblastic cells with fibronectin. Several striking differences were found in the sensitivity of the two cell types to inhibition by a series of synthetic peptides modeled on the Arg-Gly-Asp-Ser (RGDS) tetrapeptide adhesion signal. Further evidence for differences between the melanoma and fibroblastic cell adhesion systems was obtained by examining adhesion to proteolytic fragments of fibronectin. Fibroblastic BHK cells spread readily on fl3, a 75-kD fragment representing the RGDS-containing, "cell-binding" domain of fibronectin, but B16-F10 melanoma cells could not. The melanoma cells were able to spread instead on f9, a 113-kD fragment derived from the large subunit of fibronectin that contains at least part of the type III connecting segment difference region (or "V" region); f7, a fragment from the small fibronectin subunit that lacks this alternatively spliced polypeptide was inactive. Monoclonal antibody and fl3 inhibition experiments confirmed the inability of the melanoma cells to use the RGDS sequence; neither molecule affected melanoma cell spreading, but both completely abrogated fibroblast adhesion. By systematic analysis of a series of six overlapping synthetic peptides spanning the entire type III connecting segment, a novel attachment site was identified in a peptide near the COOH-terminus of this region. The tetrapeptide sequence Arg-Glu-Asp-Val (REDV), which is somewhat related to RGDS, was present in this peptide in a highly hydrophilic region of the type III connecting segment. REDV appeared to be functionally important, since this synthetic tetrapeptide was inhibitory for melanoma cell adhesion to fibronectin but was inactive for fibroblastic cell adhesion. REDV therefore represents a novel adhesive recognition signal in fibronectin that possesses cell type specificity. These results suggest that, for some cell types, regulation of the adhesion-promoting activity of fibronectin may occur by alternative mRNA splicing.

Inhibition of binding of fibronectin to matrix assembly sites by anti-integrin (alpha 5 beta 1) antibodies.

Fibroblasts have cell surface sites that mediate assembly of plasma and cellular fibronectin into the extracellular matrix. Cell adhesion to fibronectin can be mediated by the interaction of an integrin (alpha 5 beta 1) with the Arg-Gly-Asp-Ser (RGDS)-containing cell adhesion region of fibronectin. We have attempted to elucidate the role of the alpha 5 beta 1 fibronectin receptor in assembly of fibronectin in matrices. Rat monoclonal antibody mAb 13, which recognizes the integrin beta 1 subunit, completely blocked binding and matrix assembly of 125I-fibronectin as well as binding of the 125I-70-kD amino-terminal fragment of fibronectin (70 kD) to fibroblast cell layers. Fab fragments of the anti-beta 1 antibody were also inhibitory. Antibody mAb 16, which recognizes the integrin alpha 5 subunit, partially blocked binding of 125I-fibronectin and 125I-70-kD. When cell layers were coincubated with fluoresceinated fibronectin and either anti-beta 1 or anti-alpha 5, anti-beta 1 was a more effective inhibitor than anti-alpha 5 of binding of labeled fibronectin to the cell layer. Inhibition of 125I-fibronectin binding by anti-beta 1 IgG occurred within 20 min. Inhibition of 125I-fibronectin binding by anti-beta 1 Fab fragments or IgG could not be overcome with increasing concentrations of fibronectin, suggesting that anti-beta 1 and exogenous fibronectin may not compete for the same binding site. No beta 1-containing integrin bound to immobilized 70 kD. These data indicate that the beta 1 subunit plays an important role in binding and assembly of exogenous fibronectin, perhaps by participation in the organization, regeneration, or cycling of the assembly site rather than by a direct interaction with fibronectin.

Analysis of fibronectin receptor function with monoclonal antibodies: roles in cell adhesion, migration, matrix assembly, and cytoskeletal organization.

We have developed two rat mAbs that recognize different subunits of the human fibroblast fibronectin receptor complex and have used them to probe the function of this cell surface heterodimer. mAb 13 recognizes the integrin class 1 beta polypeptide and mAb 16 recognizes the fibronectin receptor alpha polypeptide. We tested these mAbs for their inhibitory activities in cell adhesion, spreading, migration, and matrix assembly assays using WI38 human lung fibroblasts. mAb 13 inhibited the initial attachment as well as the spreading of WI38 cells on fibronectin and laminin substrates but not on vitronectin. Laminin-mediated adhesion was particularly sensitive to mAb 13. In contrast, mAb 16 inhibited initial cell attachment to fibronectin substrates but had no effect on attachment to either laminin or vitronectin substrates. When coated on plastic, both mAbs promoted WI38 cell spreading. However, mAb 13 (but not mAb 16) inhibited the radial outgrowth of cells from an explant on fibronectin substrates. mAb 16 also did not inhibit the motility of individual fibroblasts on fibronectin in low density culture and, in fact, substantially accelerated migration rates. In assays of the assembly of an extracellular fibronectin matrix by WI38 fibroblasts, both mAbs produced substantial inhibition in a concentration-dependent manner. The inhibition of matrix assembly resulted from impaired retention of fibronectin on the cell surface. Treatment of cells with mAb 16 also resulted in a striking redistribution of cell surface fibronectin receptors from a streak-like pattern to a relatively diffuse distribution. Concomitant morphological changes included decreases in thick microfilament bundle formation and reduced adhesive contacts of the streak-like and focal contact type. Our results indicate that the fibroblast fibronectin receptor (a) functions in initial fibroblast attachment and in certain types of adhesive contact, but not in the later steps of cell spreading; (b) is not required for fibroblast motility but instead retards migration; and (c) is critically involved in fibronectin retention and matrix assembly. These findings suggest a central role for the fibronectin receptor in regulating cell adhesion and migration.

Monoclonal antibody characterization of two distant sites required for function of the central cell-binding domain of fibronectin in cell adhesion, cell migration, and matrix assembly.

Site-directed mutagenesis studies have suggested that additional peptide information in the central cell-binding domain of fibronectin besides the minimal Arg-Gly-Asp (RGD) sequence is required for its full adhesive activity. The nature of this second, synergistic site was analyzed further by protein chemical and immunological approaches using biological assays for adhesion, migration, and matrix assembly. Fragments derived from the cell-binding domain were coupled covalently to plates, and their specific molar activities in mediating BHK cell spreading were compared with that of intact fibronectin. A 37-kD fragment purified from chymotryptic digests of human plasma fibronectin had essentially the same specific molar activity as intact fibronectin. In contrast, other fragments such as an 11.5-kD fragment lacking NH2-terminal sequences of the 37-kD fragment had only poor spreading activity on a molar basis. Furthermore, in competitive inhibition assays of fibronectin-mediated cell spreading, the 37-kD fragment was approximately 325-fold more active than the GRGDS synthetic peptide on a molar basis. mAbs were produced using the 37-kD protein as an immunogen and their epitopes were characterized. Two separate mAbs, one binding close to the RGD site and the other to a site approximately 15 kD distant from the RGD site, individually inhibited BHK cell spreading on fibronectin by greater than 90%. In contrast, an antibody that bound between these two sites had minimal inhibitory activity. The antibodies found to be inhibitory in cell spreading assays for BHK cells also inhibited both fibronectin-mediated cell spreading and migration of human HT-1080 cells, functions which were also dependent on function of the alpha 5 beta 1 integrin (fibronectin receptor). Assembly of endogenously synthesized fibronectin into an extracellular matrix was not significantly inhibited by most of the anti-37-kD mAbs, but was strongly inhibited only by the antibodies binding close to the RGD site or the putative synergy site. These results indicate that a second site distant from the RGD site on fibronectin is crucial for its full biological activity in diverse functions dependent on the alpha 5 beta 1 fibronectin receptor. This site is mapped by mAbs closer to the RGD site than previously expected.

Novel function for beta 1 integrins in keratinocyte cell-cell interactions.

We have examined the expression, localization, and function of beta 1 integrins on cultured human epidermal keratinocytes using polyclonal and monoclonal antibodies against the beta 1, alpha 2, alpha 3, and alpha 5 integrin subunits. The beta 1 polypeptide, common to all class 1 integrins, was localized primarily in areas of cell-cell contacts of cultured keratinocytes, as were alpha 2 and alpha 3 polypeptides, suggesting a possible role in cell-cell adhesion for these integrin polypeptides. In contrast, the fibronectin receptor alpha 5 subunit showed no such accumulations in regions of cell-cell contact but was more diffusely distributed in the keratinocyte plasma membrane, consistent with the absence of fibronectin at cell-cell contact sites. Colonies of cultured keratinocytes could be dissociated by treatment with monoclonal antibody specific to the beta 1 polypeptide. Such dissociation of cell-cell contacts also occurred under conditions where the monoclonal antibody had no effect on cell-substrate adhesion. Therefore, beta 1 integrin-dependent cell-cell adhesion can be inhibited without affecting other cell-adhesive interactions. Antibody treatment of keratinocytes maintained in either low (0.15 mM) or high (1.2 mM) CaCl2 also resulted in the loss of organization of intracellular F-actin filaments and beta 1 integrins, even when the anti-beta 1 monoclonal antibody had no dissociating effect on keratinocyte colonies at the higher calcium concentration. Our results indicate that beta 1 integrins play roles in the maintenance of cell-cell contacts between keratinocytes and in the organization of intracellular microfilaments. They suggest that in epithelial cells integrins can function in cell-cell interactions as well as in cell-substrate adhesion.

Integrin function: molecular hierarchies of cytoskeletal and signaling molecules.

Integrin receptors play important roles in organizing the actin-containing cytoskeleton and in signal transduction from the extracellular matrix. The initial steps in integrin function can be analyzed experimentally using beads coated with ligands or anti-integrin antibodies to trigger rapid focal transmembrane responses. A hierarchy of transmembrane actions was identified in this study. Simple integrin aggregation triggered localized transmembrane accumulation of 20 signal transduction molecules, including RhoA, Rac1, Ras, Raf, MEK, ERK, and JNK. In contrast, out of eight cytoskeletal molecules tested, only tensin coaccumulated. Integrin aggregation alone was also sufficient to induce rapid activation of the JNK pathway, with kinetics of activation different from those of ERK. The tyrosine kinase inhibitors herbimycin A or genistein blocked both the accumulation of 19 out of 20 signal transduction molecules and JNK- and ERK-mediated signaling. Cytochalasin D had identical effects, whereas three other tyrosine kinase inhibitors did not. The sole exception among signaling molecules was the kinase pp125FAK which continued to coaggregate with alpha 5 beta 1 integrins even in the presence of these inhibitors. Tyrosine kinase inhibition also failed to block the ability of ligand occupancy plus integrin aggregation to trigger transmembrane accumulation of the three cytoskeletal molecules talin, alpha-actinin, and vinculin; these molecules accumulated even in the presence of cytochalasin D. However, it was necessary to fulfill all four conditions, i.e., integrin aggregation, integrin occupancy, tyrosine kinase activity, and actin cytoskeletal integrity, to achieve integrin-mediated focal accumulation of other cytoskeletal molecules including F-actin and paxillin. Integrins therefore mediate a transmembrane hierarchy of molecular responses.

Vinexin: a novel vinculin-binding protein with multiple SH3 domains enhances actin cytoskeletal organization.

Using the yeast two-hybrid system and an in vitro binding assay, we have identified a novel protein termed vinexin as a vinculin-binding protein. By Northern blotting, we identified two types of vinexin mRNA that were 3 and 2 kb in length. Screening for full-length cDNA clones and sequencing indicated that the two mRNA encode 82- and 37-kD polypeptides termed vinexin alpha and beta, respectively. Both forms of vinexin share a common carboxyl-terminal sequence containing three SH3 domains. The larger vinexin alpha contains an additional amino-terminal sequence. The interaction between vinexin and vinculin was mediated by two SH3 domains of vinexin and the proline-rich region of vinculin. When expressed, vinexin alpha and beta localized to focal adhesions in NIH 3T3 fibroblasts, and to cell-cell junctions in epithelial LLC-PK1 cells. Furthermore, expression of vinexin increased focal adhesion size. Vinexin alpha also promoted upregulation of actin stress fiber formation. In addition, cell lines stably expressing vinexin beta showed enhanced cell spreading on fibronectin. These data identify vinexin as a novel focal adhesion and cell- cell adhesion protein that binds via SH3 domains to the hinge region of vinculin, which can enhance actin cytoskeletal organization and cell spreading.

Synergistic roles for receptor occupancy and aggregation in integrin transmembrane function.

Integrin receptors mediate cell adhesion, signal transduction, and cytoskeletal organization. How a single transmembrane receptor can fulfill multiple functions was clarified by comparing roles of receptor occupancy and aggregation. Integrin occupancy by monovalent ligand induced receptor redistribution, but minimal tyrosine phosphorylation signaling or cytoskeletal protein redistribution. Aggregation of integrins by noninhibitory monoclonal antibodies on beads induced intracellular accumulations of pp125FAK and tensin, as well as phosphorylation, but no accumulation of other cytoskeletal proteins such as talin. Combining antibody-mediated clustering with monovalent ligand occupancy induced accumulation of seven cytoskeletal proteins, including alpha-actinin, talin, and F-actin, thereby mimicking multivalent interactions with fibronectin or polyvalent peptides. Integrins therefore mediate a complex repertoire of functions through the distinct effects of receptor aggregation, receptor occupancy, or both together.

Localization of integrin receptors for fibronectin, collagen, and laminin in human skin. Variable expression in basal and squamous cell carcinomas.

VLA integrins in human skin were examined by indirect immunofluorescence utilizing antibodies recognizing the beta 1, alpha 2, alpha 3, or alpha 5 subunits. Staining of fetal, newborn, or adult skin with antibodies to beta 1, alpha 2, or alpha 3 subunits gave essentially similar staining patterns: intense staining was associated with the basal layer of the epidermis, hair follicles, and blood vessel walls. The alpha 5 subunit could be detected only in epidermis and the inner root sheath of hair follicles in fetal skin. In epidermis, the staining reaction for the beta 1 subunit was not only found in sites interfacing with the basement membrane zone, but also around the entire periphery of these cells. We speculate that these receptors might have previously unrecognized functions in cell-cell interactions or that these findings may suggest the presence of previously unrecognized ligands in the intercellular spaces of keratinocytes. Examination of nine nodular basal cell carcinomas revealed a prominent staining reaction with anti-beta 1 and anti-alpha 3 antibodies at the periphery of the tumor islands. In contrast, staining of five squamous cell carcinomas revealed either the absence of integrins or altered and variable expression. Thus, matrix components and their receptors may participate in modulation of growth, development, and organization of human skin.

Differences in the biosynthesis and localization of the fibronectin receptor in normal and transformed cultured human cells.

We examined the biosynthesis and localization of the fibronectin receptor integrin from normal and transformed cultured human cells. Normal cells required a minimum of 20 h for the biosynthesis of completely mature fibronectin-receptor beta-subunit, while transformed cells required only 6-8 h. There was a correspondingly major decrease in the amount of the intracellular beta-chain precursor in the transformants. Immunostaining of normal fibroblastic cells with monoclonal antibodies indicated that both alpha- and beta-polypeptides of the fibronectin receptor are localized in cell surface streaks and focal contact areas. In contrast, both subunits lacked this clustering and had a more diffuse distribution on the surfaces of transformed cells, even though quantitative immunofluorescence experiments indicated that similar or larger amounts of each subunit were present on a per cell basis. Both immunostaining and biochemical analyses also indicated the presence of a relatively large intracellular pool of beta-polypeptides in normal fibroblasts that is not present in transformed cells. There was no major transformation-dependent change in total quantities of mature fibronectin receptor subunit expressed and inserted into the plasma membrane, when normalized to total protein synthesis. Our results indicate that malignant transformation of cultured human cells results in altered localization and processing of the fibronectin receptor. Such changes involving pathways of crucial cell surface molecules may contribute to alterations in their interactions with extracellular macromolecules, including during the process of cellular invasion.

Monoclonal antibody and synthetic peptide inhibitors of human tumor cell migration.

The processes of migration and invasion by human tumor cells are likely to involve specific cell surface receptors, such as receptors for the extracellular matrix molecules fibronectin, laminin, and collagen. We have examined the roles of several of these receptors using a set of monoclonal antibodies directed against the beta 1 integrin family, as well as a series of synthetic peptides reported to inhibit various interactions of each of these proteins with the cell surface. The most general inhibitor of tumor cell migration was found to be the anti-beta 1 monoclonal antibody 13, which inhibited the migration of human HT-1080 fibrosarcoma cells, 5637 bladder carcinoma cells, VA13 viral transformants, and HCT 116 colon carcinoma cells when fibronectin was the migration substrate. Moreover, this antibody was particularly effective in blocking cell migration on laminin, as well as migration within 3-dimensional collagen gels. It also inhibited in vitro invasiveness in a reconstituted basement membrane invasion assay (Matrigel assay) at concentrations as low as 1 microgram/ml. Integrins of the beta 1 class thus appear to play a central role in several types of migration by a variety of human tumor cell lines. Anti-alpha 5 fibronectin receptor monoclonal antibody 16 also significantly inhibited migration on fibronectin, but not on other substrates, in 3 of the 4 cell lines. Conversely, anti-alpha 2 monoclonal antibody F17 strikingly inhibited migration in 3-dimensional collagen gels, but not on other substrates, implicating the alpha 2 beta 1 integrin system in migration of tumor cells within collagenous matrices. A series of synthetic peptides previously reported to inhibit interactions of normal cells with fibronectin, laminin, and collagen were also tested as inhibitors of tumor cell migration. Peptides containing the Arg-Gly-Asp adhesive recognition signal were partially inhibitory, but with occasional exceptions, most other peptides had no effects on migration. Our results indicate the central importance of several specific beta 1 integrins in human tumor cell migration and show the effectiveness of monoclonal antibody treatment in blocking this process in vitro.

Cis-polyunsaturated fatty acids stimulate beta1 integrin-mediated adhesion of human breast carcinoma cells to type IV collagen by activating protein kinases C-epsilon and -mu.

We have investigated the effects of various fatty acids (FAs) on integrin-mediated MDA-MB-435 breast carcinoma cell adhesion to type IV collagen (collagen IV) in vitro. Arachidonic acid (AA) and linoleic acid both induced a dose-dependent increase in cell adhesion to collagen IV with no significant increase in nonspecific adhesion to polylysine and BSA. Oleic acid (a monounsaturated FA), AA methyl ester, and linoelaidic acid (a trans-isomer of linoleic acid) failed to stimulate adhesion to collagen IV, suggesting that these effects required cis-polyunsaturation and a free carboxylic moiety and that they were not due to membrane perturbations. Calphostin C, a protein kinase C (PKC) inhibitor, blocked cis-polyunsaturated FA (cis-PUFA)-induced cell adhesion in a dose-dependent manner, suggesting a role for a calcium-dependent PKC in this signal transduction pathway. Immunoblotting revealed that cis-PUFAs induced the translocation of PKCepsilon and PKCmu, two of the novel PKC isozymes, from the cytosol to the membrane. In contrast, a conventional PKC isozyme, PKCalpha, as well as the atypical isozymes, PKCzeta and PKCiota, did not translocate after cis-PUFA treatment. Function-blocking antibodies specific for alpha1, alpha2, and beta1, integrin subunits inhibited cell adhesion to collagen IV, whereas antibodies to alpha3 and alpha5 did not. No increase in the expression of these integrins on the cell surface was detected after the incubation of cells with cis-PUFAs, suggesting that there is an increase in the activity, but not in the amount, of these beta1, integrins. Altogether, these data suggest that cis-PUFAs enhance human breast cancer cell adhesion to collagen IV by selectively activating specific PKC isozymes, which leads to the activation of beta1 integrins.

A potential marker protease of invasiveness, seprase, is localized on invadopodia of human malignant melanoma cells.

Seprase, a large, gelatin-degrading membrane-protease complex, is expressed at the invasive front of malignant melanoma cells on invadopodia, and its surface expression contributes to the invasive phenotype. An in vitro assay was used to determine the matrix-degrading activity of four malignant human melanoma cell lines. The lines differ in matrix-degrading activity with LOX > RPM17951 > A375 > SKMEL28. The seprase and Gelatinase A activities of these cell lines were also investigated. Seprase and active gelatinase A are found in cell membranes of LOX and RPM17951 cells but not those of SKMEL28 cells. Experiments using anti-seprase monoclonal antibodies in conjunction with a cell fractionation technique indicate that seprase consists of M(r) 97,000 polypeptides and is enriched on the ventral membrane of LOX in contact with planar extracellular matrix substratum. Confocal microscopy further substantiates our biochemical findings that seprase, as well as Gelatinase A, is localized on invadopodia membranes with a 6-fold increase of seprase and 4-fold increase of Gelatinase A intensity over the level expressed on dorsal membranes. In addition, LOX cells expressing higher levels of seprase at the cell surface, as selected by fluorescence-activated cell sorting, are significantly more degradative than LOX cells with lower seprase expression. Taken together, our data show a concordance between seprase and Gelatinase A expression on the cell surface at invadopodia and the matrix-degrading activity of human malignant melanoma cells. Seprase and major secreted proteases may act in concert to degrade components of the extracellular matrix during invasion.

Formation of amyloid-like fibrils by self-association of a partially unfolded fibronectin type III module.

The ninth type III module of murine fibronectin was expressed in E. coli and folded into a compact homogeneous monomer whose unfolding and refolding were then investigated by fluorescence, circular dichroism, calorimetry and electron microscopy. The isolated module is unusually labile under physiological conditions. When heated at 1 deg. C/minute it exhibits an irreversible endothermic transition between 35 and 42 degrees C depending on the protein concentration. The transition is accompanied by changes in secondary and tertiary structure with partial exposure of the single tryptophan and increased binding of the hydrophobic probe, 1,8-anilinonaphthalene-sulfonate. The partially unfolded intermediate undergoes rapid self-association leading to the formation of large stable multimers that, like the original monomer, contain substantial amounts of beta sheet structure. The multimers melt and dissociate reversibly in a second endothermic transition between 60 and 90 degrees C also depending on the protein concentration. This second transition destroys the remaining secondary structure and further exposes the tryptophan. Visualization of negatively stained specimens in the electron microscope reveals that partially unfolded rmIII-9 slowly forms amyloid-like fibrils of approximately 10 nm width and indeterminate length. A subdomain swapping mechanism is proposed in which beta strands from one partially unfolded molecule interact with complementary regions of another to form oligomers and polymers. The possibility that similar interactions could play a role in the formation of fibrils by fibronectin in vivo is discussed.

Solution structure and dynamics of linked cell attachment modules of mouse fibronectin containing the RGD and synergy regions: comparison with the human fibronectin crystal structure.

We report the three-dimensional solution structure of the mouse fibronectin cell attachment domain consisting of the linked ninth and tenth type III modules, mFnFn3(9,10). Because the tenth module contains the RGD cell attachment sequence while the ninth contains the synergy region, mFnFn3(9,10) has the cell attachment activity of intact fibronectin. Essentially complete signal assignments and approximately 1800 distance and angle restraints were derived from multidimensional heteronuclear NMR spectra. These restraints were used with a hybrid distance geometry/simulated annealing protocol to generate an ensemble of 20 NMR structures having no distance or angle violations greater than 0.3 A or 3 degrees. Although the beta-sheet core domains of the individual modules are well-ordered structures, having backbone atom rmsd values from the mean structure of 0.51(+/-0.12) and 0.40(+/-0.07) A, respectively, the rmsd of the core atom coordinates increases to 3.63(+/-1.41) A when the core domains of both modules are used to align the coordinates. The latter result is a consequence of the fact that the relative orientation of the two modules is not highly constrained by the NMR restraints. Hence, while structures of the beta-sheet core domains of the NMR structures are very similar to the core domains of the crystal structure of hFnFn3(9,10), the ensemble of NMR structures suggests that the two modules form a less extended and more flexible structure than the fully extended rod-like crystal structure. The radius of gyration, Rg, of mFnFn3(9,10) derived from small-angle neutron scattering measurements, 20.5(+/-0.5) A, agrees with the average Rg calculated for the NMR structures, 20.4 A, and is ca 1 A less than the value of Rg calculated for the X-ray structure. The values of the rotational anisotropy, D ||/D perpendicular, derived from an analysis of 15N relaxation data, range from 1.7 to 2.1, and are significantly less than the anisotropy of 2.67 predicted by hydrodynamic modeling of the crystal coordinates. In contrast, hydrodynamic modeling of the NMR coordinates yields anisotropies in the range of 1.9 to 2.7 (average 2.4(+/-0.2)), with NMR structures bent by more than 20 degrees relative the crystal structure having calculated anisotropies in best agreement with experiment. In addition, the relaxation parameters indicate that several loops in mFnFn3(9,10), including the RGD loop, are flexible on the nanosecond to picosecond time-scale. Taken together, our results suggest that, in solution, the limited set of interactions between the mFnFn3(9,10) modules position the RGD and synergy regions to interact specifically with cell surface integrins, and at the same time permit sufficient flexibility that allows mFnFn3(9,10) to adjust for some variation in integrin structure or environment.