INTRODUCTION

Cell migration is essential for many biological processes, including development, wound healing, and hemostasis. In addition, several pathologic processes such as cancer metastases, inflammation, thrombosis, and restenosis are dependent on cell migration. To generate the necessary traction forces required for movement, cells depend on adhesive interactions with the substratum, mediated at least in part by integrins. The ₅₁ and ₄₁ integrins mediate migration toward fibronectin (Fn)¹ (1, 2). The vitronectin receptors (VnR) _v3 and _v5 have been implicated in the migration of a variety of cell types including smooth muscle (3), keratinocytes (4), leukocytes (5, 6), endothelial cells (7), and neural crest cells (8), and were shown to play a role in melanoma metastases (9, 10) and angiogenesis (11). Migration requires the fine control of integrin association with and release from the extracellular matrix. These interactions generate signals that are subsequently transmitted to the cytoskeleton. It has been suggested that integrin activity is itself regulated by interaction with substrate or by inside-out signaling (12, 13) and that cell migration is regulated, in part, by the cycling of integrins between cytoplasmic compartments and the cell surface (14, 15).

Modulation of integrin affinity or avidity has been clearly observed in the platelet integrin _IIb3, and the ₁ and ₂ integrins of lymphocytes and leukocytes (12). In resting cells, these receptors exist in a low affinity ligand binding state and change to a high affinity binding state upon cellular activation. Platelets are activated by various agonists such as thrombin, collagen, and ADP (16) that increase intracellular pH and Ca²⁺. Cellular activation of ₂ integrins varies with cell type. Neutrophils and monocytes are activated by phorbol esters and by inflammatory mediators, such as tumor necrosis factor, platelet-activating factor, fMet-Leu-Phe, and lipids (12). The T-lymphocyte _L2 can be activated by phorbol esters or by cross-linking of the cell surface molecules CD2 or CD3 (17, 18). These are some examples of integrin activation by inside-out signaling in response to extracellular agonists.

An additional mechanism for modulating integrin activity is by another integrin heterodimer within the same cell. Ligation of the fibronectin receptor (FnR) by attachment of monocytes to Fn-coated surfaces promoted _M2-mediated phagocytosis of complement fragment C3b (19). Furthermore, ligation of the FnR expressed on the basal plasma membrane of these cells activated _M2 on the apical cell surface (20), suggesting a signal transduction pathway. Ligation of the leukocyte response integrin, a ₃ and unique  subunit-containing receptor (21), and subsequent formation of the leukocyte response integrin/integrin-associated protein (IAP) complex was reported to enhance _M2 binding activity (22, 23). In addition, binding of ₅₁ to its ligand stimulates, in a protein kinase C-dependent manner, ₂₁-mediated adherence to collagen type I (Col I), which subsequently results in secretion of interleukin-1 (24). An inflammatory response is thus induced by cell adhesion to extracellular matrix proteins. Furthermore, it was recently reported that _IIb3 regulates ₅₁- and ₂₁-mediated cell attachment to Fn and Col I by a conformation change induced by receptor occupancy (25).

The _v3 integrin has previously been shown to negatively regulate ₅₁ integrin function. Blystone and colleagues (26) have demonstrated that ligation of _v3 by antibodies or by Vn-coated surfaces inhibits ₅₁-mediated phagocytosis of Fn-coated beads. It was suggested that this phenomenon is mediated by a phosphoserine/phosphothreonine signaling cascade, since it is blocked by the inhibitor H7 (26). Furthermore, it was shown that _v3 inhibition of ₅₁ phagocytosis occurs as a result of _v3 interaction with IAP, and that this integrin cross-talk requires the cytoplasmic domain of the ₃ integrin subunit (27).

The present study examines the role of _v3 in the regulation of cell migration. We show that antibody ligation of _v3, addition of an _v3 peptide inhibitor, or a Glanzmann mutation in the ligand binding site of ₃ not only inhibit migration toward Vn but also toward Fn. The _v3 modulation of ₅₁-mediated migration toward Fn appears to be specific and unidirectional. In contrast to the cross-talk between integrins observed in phagocytosis (27), deletion and mutation studies indicate that the transmembrane domain of ₃ is important for generating the regulatory signal for ₅₁-dependent migration.

MATERIALS AND METHODS

Human vitronectin (Vn) and mouse laminin (Ln) were purchased from Life Technologies, Inc. Human fibronectin (Fn) was purchased from NY Blood Center (New York, NY). Collagen type IV (Col IV) was purchased from Collaborative Biomedical Products (Bedford, MA). Antibodies against: _v3 (mAb LM609), ₅ subunit (mAb CLB-705), _v5 (mAb P1F6), and ₅₁ (mAb JB55). Anti-₁ antibody (mAb 13) was purchased from Becton-Dickinson (San Jose, CA). Polyclonal rabbit anti-_v antiserum were purchased from Chemicon (Temecula, CA). Polyclonal rabbit anti-₃ antibodies was a generous gift from Dr. Daniel Bollag and Patricia McQueney (Merck Research Laboratories, West Point, PA). These antibodies were raised against human _IIb3 (28). These antibodies were also recognize by immunoprecipitation the ₃ subunit of the _v3 integrin expressed in HEK 293 cells (Fig. 1B); we therefore refer to these antibodies as anti-₃ antibodies.

[View Larger Version of this Image (24K GIF file)]

Full-length cDNA of the _v integrin subunit (selectable by hygromycin resistance) was generously provided by Dr. A. Schmidt (Merck Research Laboratories). The full-length cDNAs of the ₃ and ₅ integrin subunits were cloned from a human umbilical cord endothelial cell gt11 5 stretch cDNA library (CLONTECH, Palo Alto, CA). The  subunit constructs were cloned into pcDNA3 (InVitrogen, San Diego, CA) selectable by neomycin resistance. Deletion mutants of the ₃ subunit, ₃(717) and ₃(693), were generated by introducing a stop codon after Lys-716 and Asp-692, respectively. These constructs were made by polymerase chain reaction using a common 5 primer spanning the BamHI site and 3 primer containing the appropriate mutation and a XhoI restriction site at its 3 end. Amplified products were digested with the same restriction enzymes. The ₃(D119Y) construct was made by polymerase chain reaction using a 5 primer including the ₃ 5-sequence with a HindIII site, and a 3 primer spanning the KpnI site and a point mutation at the appropriate position. The amplified product was inserted directly into pcDNA3-₃ digested with HindIII-KpnI. The ₃(D119A/717) construct was made as a combination of both mutations. The ₃(S752P) construct was made as described previously (29). All constructs were characterized by sequence analysis and purified by CsCl centrifugation prior to transfection into cells. Restriction enzymes were purchased from Stratagene (La Jolla, CA) or New England Biolabs (Beverly, MA).

Human embryonic kidney 293 cells (ATCC, Rockland, MD) were cultured in minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS), 0.1 mg/ml kanamycin, and 2 mM L-glutamine (Life Technologies, Inc.), and maintained at 37 °C and 5% CO₂. Cells transfected with human _v and ₃ integrin subunits were maintained in complete media with added 400 µg/ml G418 (Life Technologies, Inc.) and 50 µg/ml hygromycin (Calbiochem, San Diego, CA). Human umbilical vein endothelial cells (HUVECs; Cell Systems, Kirkland, WA) were maintained in MCDB (Sigma) supplemented with: 15% FBS (heat-inactivated), 0.2 mM KCl, 3 mM KH₂PO₄, 0.3 mM glycine, 90 µg/ml heparin (Sigma), 25 µg/ml endothelial mitogen (Sigma), 50 µg/ml kanamycin (Life Technologies, Inc.) on tissue culture plates coated with 50 µg/ml Col I (Celtrix Pharmaceuticals, Santa Clara, CA). Cells were used before passage 8.

The constructs described above were transfected into HEK 293 cells by electroporation at 200 V, 960 microfarads using a GenePulser (Bio-Rad). Briefly, cells at 50% confluence were collected using trypsin-EDTA. After two washes in serum-free media, the cells (1 × 10⁶ cells/ml) were incubated with 5 µg of plasmid DNA on ice for 30 min prior to electroporation. Cells were subjected to differential selection after 48 h in complete media containing 800 µg/ml G418 (Life Technologies, Inc.) and 100 µg/ml hygromycin (Calbiochem, San Diego, CA). In this study, all cell lines represent pools of at least six single clones.

Surface expression of transfected integrins was characterized using flow cytometry analysis and immunoprecipitation, followed by Western blots. For flow cytometry analysis, cells were lifted by trypsin-EDTA and washed once with five volumes of MEM containing 10% FBS and twice in Dulbecco's phosphate-buffered saline. HEK 293 cells expressing _v3 and ₃ mutants (2 × 10⁵ cells/ml) were incubated with polyclonal anti-₃ antibodies (15 µg/ml), for 30 min at room temperature, followed by washing and incubation with FITC-conjugated donkey-anti-rabbit IgG antibodies (Jackson Laboratories, West Grove, PA) for 30 min at room temperature. Cells were then washed and resuspended in 250 µl of Flow buffer (100 mM HEPES buffer, pH 7.5, 150 mM NaCl, 3 mM KCl, and 1 mM CaCl₂) and analyzed by flow cytometry using a FACScalibur instrument (Becton-Dickinson). Similarly, HEK 293 cells expressing _v5 were incubated with mAb P1F6 (20 µg/ml), followed by incubation with FITC-conjugated goat anti-mouse IgG antibodies (Jackson). Endogenous expression of ₅₁ in all cell lines were detected using mAb JB55 (20 µg/ml) and followed by FITC-conjugated anti-mouse IgG antibodies as described above.

Additionally, transfectants (1 × 10⁶ cells) were surface-labeled with 2 mM Immunopure Sulfo-NHS-LC-Biotin (Pierce) and then solubilized in RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM CaCl₂, 1% Nonidet P-40, 0.5% deoxycholate) containing 1 mM phenylmethylsulfonyl fluoride, 10 µg/ml aprotinin, and 100 µg/ml leupeptin. Cell extracts were immunoprecipitated using the polyclonal anti-₃ antibodies, followed by protein G-Sepharose (30). Precipitated proteins were separated on 8% SDS-polyacrylamide gel (Novex, San Diego, CA), followed by Western blotting and developed with horseradish peroxidase-conjugated streptavidin (Amersham). Similarly, expression of _v1 in these cells was detected by immunoprecipitation with mAb 13 (anti-₁) and subsequent blotting with anti-_v cytoplasmic domain polyclonal antibodies, then detected using horseradish peroxidase-conjugated anti-rabbit IgG antibodies (Amersham), followed by enhanced chemiluminescence (ECL) system (NEN Life Science Products).

Cell migration was assayed using a Boyden chamber type apparatus (Neuroprobe, Cabin John, MD). Prior to assay, cells were loaded with a fluorescent marker, 5-chloromethyl fluorescein diacetate (Molecular Probes, Eugene, OR). Blocking antibodies or peptides were added to cells just prior to assay. Extracellular matrix proteins were diluted into serum-free media and placed in the bottom chamber. Labeled cells were washed with serum-free media and added to the upper chamber at a density of 20,000 cells/well. Normally, 3,000-10,000 cells (~15-50% of total cells added) migrate in these assays. Cells were allowed to migrate through a polycarbonate filter (pore size 8 µm) for 15 h in a humidified incubator at 37 °C. The cells migrating to the bottom of the filter were detected using the Cytofluor fluorescence plate reader (Millipore, Bedford, MA). No migrated cells were detected when ligand was added to the upper well of the migration chamber. The number of migrated cells was calculated based upon standard curves for each cell line used in the experiment. Results are expressed as a mean value of triplicate or quadruplicate samples.

Cells were lifted with trypsin-EDTA and washed four times with serum-free MEM. Cells (10,000 cells/well) were added to microtiter wells coated with Vn or Fn and allowed to attach at 37 °C in a humidified incubator for 30 min or 2 h. Non-attached cells were gently washed away, and attached cells were quantified by colorimetric detection of hexosaminidase enzymatic activity (31) in a Vmax plate reader (Molecular Devices, Menlo Park, CA). The number of attached cells was quantitated using a standard curve for each cell line assayed and expressed as a mean value of triplicate samples.

RESULTS

Stable transfectants of HEK 293 cells expressing _v3 (₃-WT) and _v5 (₅-WT) and ₃ mutants were used in this study. The following mutants of ₃ were constructed and co-expressed with wild type _v: truncations of the ₃ subunit cytoplasmic domain (₃-717) and of the transmembrane domain (₃-693), insertion of the Glanzmann thrombasthenia mutations in the ligand binding site (₃-D119Y) or in the cytoplasmic domain (₃-S752P), and the combined ₃-D119Y mutation with truncation of the cytoplasmic domain (₃-D119Y/717).

Surface expression of _v3 integrin and its mutants was determined by flow cytometry and immunoprecipitation followed by Western blots. In Fig. 1A, the surface expression of _v3 and its mutants was analyzed by flow cytometry using polyclonal anti-₃ antibodies. The level of integrin expression is compared with that in parental HEK 293 cells, which lack endogenous _v3 expression. Surface expression of the _v3 mutants is comparable to that present in cells expressing the wild-type _v3 integrin (₃-WT), with the exception of the ₃-693 cells, which express approximately 10-fold lower levels of mutant integrin. Therefore, we chose for comparison another HEK 293 cell line (₃-L) that expresses wild type _v3 at levels comparable to those in ₃-693 cells. The ₃-S752P cells appear to be a mixed population as indicated by the broad histogram indicating varied levels of receptor expression (Fig. 1A). In addition, heterodimer formation and surface expression of _v3 and all ₃ mutants were also confirmed by surface biotinylation followed by immunoprecipitation with the anti-₃ antibodies (Fig. 1B). Both _v (130 kDa) and ₃ (110 kDa) subunits were immunoprecipitated from ₃-WT and the ₃ mutants (₃-S752P and ₃-D119Y). Deletion of the cytoplasmic domain (₃-717, ₃-D119Y/717) and transmembrane domain (₃-693) of the ₃ subunit leads to a shift in the mobility of the ₃ subunit bands (97 kDa) on the gels. Therefore, the subunit mutations do not appear to disrupt normal subunit association or cell surface expression. The _v5 expression in transfected cells was also relatively high, and the ₅₁ levels were similar to those in parental HEK 293 cells as shown in Fig. 1A. Therefore, the level of expression of the endogenous ₅₁ integrin was not affected by overexpression of exogenous VnRs.

HEK 293 cells express _v1 integrins, which function as Vn and Fn receptors in these cells (32). We examined the relative levels of _v1 in ₃-WT and ₃ mutants, using immunoprecipitation from cell lysates with anti-₁ antibody (mAb 13), followed by immunoblotting with anti _v-cytoplasmic domain antibodies. A small reduction in _v1 was observed in ₃-WT cells in the experiment presented in Fig. 1C; however, we detected no significant difference in _v1 between cells expressing _v3 or its mutants in repeated experiments.

HEK 293 cells expressing _v3 attached (Fig. 3A), and spread on Vn (Fig. 2A). Parental cells attach loosely but fail to spread on Vn. In contrast, the ₃-WT and ₅-WT cells exhibit a well spread morphology. Cells with the Glanzmann mutations (₃-S752P and ₃-D119Y) or the transmembrane truncation (₃-693) show diminished spreading on and attachment to Vn by comparison to ₃-WT cells (Figs. 2A and 3A). Cells with the cytoplasmic domain truncation (₃-717) plated on Vn have at the periphery projections of thin ruffled lamellipodia (Fig. 2A), which appear lucid and free of organelles. Cells overexpressing _v3 or its mutants spread (Fig. 2B) and attach (Fig. 3B) to Fn similarly to parental cells. Interestingly, the ₅-WT cells are very well spread on Fn and contain many vacuoles (Fig. 2B). Cells transfected with _v3 acquire, as expected, the ability to migrate toward Vn (Fig. 4). Migration toward Fn via the endogenous ₅₁ integrin is not altered by the presence of _v3 integrin (Fig. 4).

Attachment of HEK 293 cell lines expressing _v with ₃-WT or the ₃ mutants: -S752P, -D119Y, -717, -693, and -D119Y/717 to either Vn or Fn.

[View Larger Version of this Image (38K GIF file)]

Morphology of cell lines expressing the indicated integrins: ₃-WT, the ₃ mutants, and ₅-WT seeded on Vn or Fn.

[View Larger Version of this Image (98K GIF file)]

Surface expression of _v3 enables migration toward vitronectin.

[View Larger Version of this Image (27K GIF file)]

Antibody Ligation of _v3 Inhibits Cell Migration toward Fibronectin

Treatment of ₃-WT expressing cells with the anti-_v3 blocking monoclonal antibody LM609 inhibits migration toward Vn by ~98% (Fig. 5A). Surprisingly, _v3 ligation also inhibits migration toward Fn by ~81%. This cross-regulation phenomenon appears to be specific to _v3. Overexpression of the VnR _v5 also enables HEK 293 cells to migrate toward Vn. However, while antibodies against _v5 (mAb P1F6) block migration toward Vn by ~89%, they do not affect migration toward Fn. The parental HEK 293 cells, which do not express _v3 or _v5, were used as controls. As expected, they do not migrate toward Vn and the addition of LM609 (Fig. 5A) or P1F6 (data not shown) does not alter migration toward Fn.

The integrin _v3 cross-regulates ₅₁-mediated migration toward in ₃-WT cells.

[View Larger Version of this Image (44K GIF file)]

To determine if _v3 cross-regulation of ₅₁ integrin function is a consequence of the overexpression of the exogenous VnR, a cell line expressing lower levels of _v3 was also examined (₃-L) (Figs. 1A and 5B). Treatment of these cells with mAb LM609 resulted in similar inhibition of cell migration toward both Vn and Fn. As shown in Fig. 5B, the level of inhibition of migration toward Fn is similar to that in highly expressing ₃-WT. Therefore, in this range, the level of _v3 expression does not seem to alter the cross-regulation of ₅₁ activity.

The integrin cross-regulation produced by ligation of _v3 appears to be unidirectional, with signals from _v3 modulating ₅₁ migratory function. Ligation of ₅₁ with the anti-₅ blocking antibody (mAb CLB-705) inhibits migration toward Fn, but does not affect migration toward Vn in _v3 expressing cells (Fig. 5C). In addition, the _v3-mediated cross-regulation appears to be specific for Fn, since _v3 ligation did not affect cell migration on Ln or Col IV (data not shown).

It has been reported that _v3 can mediate attachment to Fn (32). Therefore, ligation of _v3 with mAb LM609 may directly inhibit its interaction with and, subsequently, migration toward Fn. To address this possibility, blocking antibodies to either the ₅ subunit (mAb CLB-705) or the ₁ subunit (mAb 13) were used. Ligation of ₅₁ with these antibodies in ₃-WT cells resulted in 85-88% inhibition of migration toward Fn (Fig. 5D), indicating that in the cell system used here, cell migration toward Fn requires accessible ₅₁ integrins.

The Role of ₅₁, _v3, and _v5 in Cell Attachment to Fn and Vn

Attachment of ₃-WT cells to Fn was reduced by about 30% by the presence of the anti-_v3 blocking antibody LM609 (Fig. 6A), and separately by 60% by anti-₁ integrin antibodies. Combining both antibodies caused additive effects on attachment to Fn. The _v3 integrin thus participates in the attachment of ₃-WT cells to Fn (33). Although the VnR _v5 was also shown to act as an FnR (24, 34), in ₅-WT cells only ₁ blocking antibodies, not anti-_v5, inhibit attachment to Fn (Fig. 6A).

Attachment to Fn is mediated by both ₅₁ and _v3 integrins in ₃-WT cells.

[View Larger Version of this Image (25K GIF file)]

Attachment of ₃-WT to Vn is reduced by 70% by LM609 (Fig. 6B) and was not affected by ₁ antibodies. Therefore, _v3 functions as the predominant VnR in attachment of ₃-WT cells to Vn. Similarly, in ₅-WT cells, _v5 is the primary receptor for Vn, although anti-₁ antibodies seem to have an additive effect in the presence of anti-₅ antibodies. Antibodies to _v5 (mAb P1F6) inhibit attachment to Vn by 72%, whereas anti-₁ antibodies alone had little or no effect (Fig. 6B). These data suggest that there may be a component of Vn attachment that is mediated by ₁ integrins and that this activity is enhanced by _v5 ligation, or conversely that ₅₁ ligation modulates _v5 activity. The endogenous integrin _v1 in HEK 293 cells may be responsible for additional binding to Vn and Fn, and for the lack of complete inhibition of attachment to Vn by anti-_v3 or anti-_v5 antibodies.

An _v3-binding RGD Peptide Inhibits Cell Migration toward Fibronectin

The data presented above show that antibodies that ligate _v3 and block ligand binding to this receptor cross-regulate ₅₁-mediated migration. We therefore examined whether _v3 ligands can inhibit ₅₁ function. As shown in Fig. 7, the preferential peptide inhibitor of _v3, the cyclic RGD peptide G-Pen-GRGDSPC-A (35), is a potent inhibitor of ₃-WT cell migration toward Vn (IC₅₀ ~ 1 nM). In addition, it also strongly inhibits migration toward Fn (IC₅₀ ~ 2.5-5 nM). The cyclic peptide also weakly inhibits parental HEK 293 cell migration toward Fn (~25% at 10 nM); however, this inhibition was much lower than for ₃-WT cells migrating toward either Vn (~92%) or Fn (~80%). The effect on parental HEK 293 cells may be due to cross-reactivity of the peptide with endogenous integrins such as _v1. The effects of the _v3 ligand further support a role for _v3 in cross-regulation of migration toward Fn, and the low concentration (<5 nM) suggests that partial ligand occupancy of _v3 receptors may suffice to produce this effect.

A cyclic RGD peptide induces _v3 cross-regulation of ₅₁.

[View Larger Version of this Image (32K GIF file)]

A Point Mutation in the ₃ Ligand Binding Domain Mimics the Effect of Antibody Ligation in the _v3 Cross-regulation of ₅₁ Activity

To examine the structural requirements for the _v3 cross-regulation effects on migration, a Glanzmann thrombasthenia mutation was introduced into the ₃ ligand binding domain. The substitution of tyrosine for the aspartic acid at residue 119 (₃-D119Y) in the ₃ subunit abolishes _IIb3 binding to fibrinogen (36). Surface expression and characterization of the ₃-D119Y mutant is described above. Cells that express _v3-D119Y (₃-D119Y) do not attach to (Fig. 3A) or migrate toward Vn (Fig. 8). Thus, this Glanzmann mutation inhibits _v3 ligand binding activity, similar to its effects on _IIb3.

The ligand binding site of _v3 is required for cross-regulation of ₅₁ integrin.

[View Larger Version of this Image (31K GIF file)]

Interestingly, the migratory activity of the ₃-D119Y mutant cells toward Fn is constitutively suppressed (Fig. 8), even in the absence of the _v3 antibody LM609. The level of inhibition is similar to that produced by ligation of ₃-WT with mAb LM609 or by 10 nM cyclic RGD peptide. Thus, the negative regulation of ₅₁ by _v3 seems to be induced by the changes produced by this point mutation. Importantly, this inhibition occurs in the presence of normal levels of ₅₁ on the cell surface (Fig. 1), and normal attachment to Fn comparable to that of ₃-WT cells (Fig. 3B). It should be noted that migration of the ₃-D119Y cells toward Ln (Fig. 9) and Col IV (data not shown) was similar to that of parental and ₃-WT cells. This migration is most likely mediated by other endogenous ₁ integrins expressed in HEK 293 cells. These findings support the notion that _v3 cross-regulation is selective for ₅₁-mediated migration toward Fn. In addition, these data suggest that the ₃-D119Y mutation mimics the effects of mAb LM609 ligation, and binding of the RGD peptide.

The cells expressing _v with wild-type ₃ or the ₃ mutants: -S752P, -D119Y, -717, -693, and -D119Y/717 migrate on Ln at similar levels to the parental HEK 293 cells.

[View Larger Version of this Image (40K GIF file)]

A Point Mutation in the ₃ Cytoplasmic Domain Has No Effect on the _v3 Cross-regulation of ₅₁-mediated Migration

The data presented so far suggest that a functional _v3 ligand binding domain is required for cross-regulation of ₅₁, implicating integrin activation by outside-in signaling. To evaluate a possible role for inside-out signaling in this cross-regulation, another ₃ Glanzmann mutation was generated, substitution of the serine residue 752 by proline (₃-S752P) in the ₃ cytoplasmic domain. In platelets, this mutation abolishes activation of _IIb3 by inside-out signals, and maintains the integrin in a low affinity ligand binding state (37). Expression of _v3-S752P abolished attachment to (Fig. 3A) and migration toward Vn (Fig. 8). Thus, similar to its effect on _IIb3, the Glanzmann ₃-S752P mutation affects the ligand binding ability of _v3, suggesting the possibility for regulation of _v3 activity by the cytoplasmic domain of the ₃ subunit, although good evidence for inside-out signaling in _v3 is still lacking.

In contrast to the mutation in the ligand binding site (₃-D119Y), which constitutively inactivates migration toward Fn, the cytoplasmic domain ₃-S752P mutants migrate on Fn at a level comparable to ₃-WT cells. Furthermore, like in the ₃-WT cells, ligation of the mutant receptor by mAb LM609 results in 60% inhibition of cell migration toward Fn (Fig. 8). Thus inactivation of _v3-mediated attachment to Vn by the S752P mutation has no effect on the cross-regulation activity of _v3, suggesting that the cytoplasmic domain mutation and the D119Y ligand binding domain mutation produce different changes in the receptor.

The Transmembrane Domain of ₃ Is Required for Cross-regulation Activity

Thus far, this study, indicates that (i) outside-in signaling of _v3 either by antibody ligation, ligand binding, or point mutation (₃-D119Y) results in cross-inhibition of ₅₁-mediated migration, and (ii) inside-out signaling, presumably mediated by the ₃ cytoplasmic domain does not appear to be involved in this signaling pathway. The signals induced by ligand binding are communicated into the cell by the transmembrane and cytoplasmic domains. To further examine the structure/function relationship of the FnR cross-regulation, two ₃ subunit deletion mutants were made: truncation of the cytoplasmic domain (₃-717) or of the cytoplasmic and transmembrane domains (₃-693).

As expected, removal of the cytoplasmic domain inhibits cell migration toward Vn, but not toward Fn (Fig. 10). Again, these data indicate that _v3 is not a major receptor for migration toward Fn in _v3 transfectants. However, ligation of ₃-717 cells with mAb LM609 inhibits cell migration toward Fn, as it does in ₃-WT (Fig. 10). On the other hand, truncation of the ₃ subunit at the start of the transmembrane domain (₃-693), abolishes the ability of the _v3 blocking antibody LM609 to inhibit migration toward Fn. The failure to inhibit cross-regulation in ₃-693 cells was not due to the lack of antibody recognition, since the mAb LM609 binds to this mutant in flow cytometry and immunoprecipitation. As shown previously for the two Glanzmann mutants, migration toward Ln (Fig. 9) and Col IV (data not shown) was not altered by expression of the truncated ₃ subunits.

The transmembrane domain of ₃ is required for cross-regulation of ₅₁ integrin.

[View Larger Version of this Image (30K GIF file)]

The findings from truncation mutants, and the ₃-S752P mutation reinforce the idea that the cytoplasmic domain of the ₃ subunit is not required for _v3 regulation of migration toward Fn. To further test this finding, a double mutant was constructed: ₃-D119Y lacking the cytoplasmic domain (₃-D119Y/717). As seen in Fig. 10, this double mutant exhibits the combined phenotype of D119Y and 717, migration toward both Vn and Fn being greatly inhibited, by comparison to ₃-WT cells. In contrast, ₃-D119Y/717 cells migrate toward Ln (Fig. 9) and Col IV (data not shown) similar to parental HEK 293 and ₃-WT cells. Thus, like the ₃-D119Y cells, the double mutant constitutively exerts negative cross-regulation on ₅₁-mediated migration. It should be noted that, as described, these cells do not attach to Vn; however, attachment to Fn is similar to that of parental HEK 293 cells (Fig. 3, A and B). Therefore, the cross-regulation between _v3 and ₅₁ appears to be dependent on the transmembrane domain of the ₃ subunit. We also attempted to construct the ₃-D119Y/693 but were unable to produce viable transfectants in HEK 293 cells. The reason for this is presently unclear.

Ligation of _v3 in HUVECs Inhibits Migration toward Fibronectin

The experiments presented above have used HEK 293 cells overexpressing exogenous _v3. It was therefore important to determine if cells endogenously expressing _v3 and ₅₁ exhibit the same integrin cross-regulation. HUVECs express both _v3 and ₅₁ (38). As shown in Fig. 11, ligation of HUVEC-_v3 using mAb LM609 resulted in ~95% inhibition of migration toward Fn substrates. Although HUVECs showed lower levels of migration toward Vn by comparison to HEK 293 cells expressing ₃-WT, their motility was also inhibited by the anti-_v3 antibody. Similar results were obtained for MG-63 cells, a human osteosarcoma cell line expressing _v3 and ₅₁ (data not shown). Therefore, cross-regulation between _v3 and ₅₁-mediated migration may occur in a physiological setting. These results again indicate that overexpression levels of _v3 are not essential for cross-regulation of ₅₁ activity, since the level of endogenous _v3 in HUVEC is lower than that in the ₃-WT.

The _v3 cross-regulation of ₅₁-mediated migration occurs in cells naturally expressing _v3 and ₅₁.

[View Larger Version of this Image (30K GIF file)]

DISCUSSION

This study examines the role of _v3 integrin in the cross-regulation of cell migration, using human embryonic kidney (HEK) 293 cells expressing exogenous _v3 and ₃ mutants. Its two main findings are: (i) _v3 ligand binding or changes that may mimic this binding inhibit ₅₁ migration toward Fn, and (ii) these changes appear to be independent of mutations in the cytoplasmic domain, but require the transmembrane domain.

The cross-regulation of _v3 and ₅₁ affecting migration in HEK 293 cells differs in that respect from the previously reported effect of _v3 on ₅₁-mediated phagocytosis in K562 cells, which required the ₃ cytoplasmic domain (27). The findings are not strictly comparable since in that study the truncation ₃-728 left 11 amino acids of the cytoplasmic domain (LLITIHDRKEF), absent in this study. This domain is highly conserved among the majority of  subunits (39) and has been implicated in modulating the integrin activation state (29, 40), and the localization of ₁ and ₃ integrins to focal adhesions (41-44). Additionally, this region, particularly the HDRK residues, has been hypothesized to associate with a highly conserved region of the  subunit cytoplasmic domain (KVGFFK), via hydrophobic interactions (45). These 11 residues could affect the activation state of _v3, as suggested by the recent report that the truncated _IIb3 mutant _IIb3-724, transfected into CHO cells is in a low affinity ligand binding state, while the _IIb3-717 mutant is expressed in a high affinity ligand binding state (46). Taken together, these observations suggest either that expression of the ₃-717 and ₃-728 mutations have different effects on ₃ function or that migration and phagocytosis are governed by different signaling events.

The molecular basis for the cross-regulation between _v3 and ₅₁ has not been elucidated. A direct interaction between the two integrins during cell migration toward Fn cannot be ruled out, although co-clustering of the two integrins was not observed by histochemistry when ₃-D119Y cells were plated on Vn or Fn (data not shown). Alternatively, cross-regulation could be mediated by signaling events. The cyclic RGD peptide, which binds preferentially to _v3, inhibited cell migration toward Fn in the nM range (IC₅₀ ~ 2.5-5 nM), suggesting that partial occupancy of _v3 receptors may be sufficient for cross-regulation, potentially via an amplifying signal transduction cascade. It was previously reported that the _v3-dependent inhibition of Fn phagocytosis is mediated by an H7-sensitive pathway (26). In our study, cell migration regardless of cell type or substrate was inhibited by low concentrations of H7 (data not shown). The 50-kDa transmembrane protein, IAP, has been implicated in regulating _v3 cross-regulation of ₅₁-mediated phagocytosis (27). The interaction between IAP and the ₃ integrins appears to require the IAP IgV-like extracellular domain (47, 48). Our findings are consistent with these observations, and since HEK 293 cells express high levels of IAP (49), its role should be further investigated.

There are several other examples of integrin cross-regulation. A conformation change induced by ligand binding to _IIb3 blocks ₅₁ and ₂₁-mediated cell attachment to Fn and Col I, respectively (25). Furthermore, the urokinase-type plasminogen activator receptor (uPAR) was also shown to regulate the ₁ integrin function (50), and the expression of uPAR has been associated with the presence of _v3 integrin (51). It would therefore be of interest to determine if uPAR plays a role in the observation reported in this study.

The _v3 integrin does participate (~30%) in the attachment of ₃-WT cells to Fn. Antibody ligation of _v3 could thus directly inhibit interaction with Fn and subsequently cell migration. However, this does not seem to be the case since in ₃-WT cells antibodies against the FnR inhibit migration by ~85%. Furthermore, expression of the Glanzmann ₃-S752P mutation, and ₃ truncation (₃-717, -693) abolish migration toward Vn, without affecting migration toward Fn. Interestingly, the ₃ Glanzmann thrombasthenia mutations (₃-S752P, ₃-D119Y) inhibit _v3 ligand binding as they do in _IIb3. In addition, it was recently shown that _v3-mediated ligand binding and migration can be activated by the AP5 antibody (52), raising the possibility of _v3 active and inactive states, which requires further studies. It was suggested that in non-platelet cells isolated from Glanzmann thrombasthenic patients _v3 function was compromised (53). It would be of interest to examine whether fibroblasts from these patients migrate normally toward Fn.

Integrin cross-regulation may play a role in vivo. In cells expressing both _v3 and ₅₁, ligand binding to _v3 could inhibit migration toward Fn. This mechanism could regulate, at the single cell level, the initiation of migration, selection of the matrix to follow, and arrest of migration, during development, inflammatory responses, wound healing, and pathologic conditions such as atherosclerosis or metastases. Cell types implicated in these processes include highly migratory cells, such as endothelial and smooth muscle cells, that express both _v3 and ₅₁ on their surface (38, 54). Interestingly, in endothelial cells (including HUVECs), both integrins were detected on the luminal (apical) and substrate-bound (basal) aspects of the plasma membrane (55).

The _v3 integrin was shown to be involved in several clinically relevant processes, including neovascularization (11, 56), and formation of atherosclerotic plaques (57). Additionally, migration of smooth muscle cells is suggested to be important during atherosclerotic plaque formation (57) and restenosis after balloon angioplasty (58). The implication of _v3 in these disease states has made it a target for drug development. The capacity of _v3 inhibitors to modulate migration toward Fn requires further examination.

In conclusion, this study shows that _v3 integrin cross-regulates ₅₁ activity in HEK 293 cells expressing exogenous _v3, as well as in cells that endogenously co-express _v3 and ₅₁. Interference with _v3 ligand binding, by antibodies, peptide inhibitor, or point mutation, inhibits migration but not attachment to Fn substrates. The cross-regulation is uni-directional, is specific for _v3, and appears to be mediated by the ₃ subunit transmembrane domain. The findings suggest that attachment and migration mediated by ₅₁ require different signaling pathways and that integrin cross-regulation may be a mechanism for local control of migration and adherence.