14-3-3ζ Mediates Integrin-induced Activation of Cdc42 and Rac

Integrin-induced cytoskeletal reorganizations are initiated by Cdc42 and Rac1 but little is known about mechanisms by which integrins activate these Rho GTPases. 14-3-3 proteins are adaptors implicated in binding and regulating the function and subcellular location of numerous signaling molecules. In platelets, the 14-3-3ζ isoform interacts with the glycoprotein (GP) Ibα subunit of the adhesion receptor GP Ib-IX. In this study, we show that integrin-induced activation of Cdc42, activation of Rac, cytoskeletal reorganizations, and cell spreading were inhibited in Chinese hamster ovary cells expressing full-length GP Ibα compared with GP Ibα lacking the 14-3-3ζ binding site. Activation of Rho GTPases and cytoskeletal reorganizations were restored by expression of 14-3-3ζ. Spreading in cells expressing truncated GP Ibα was inhibited by co-expressing a chimeric receptor containing interleukin 2 receptor α and GP Ibα cytoplasmic domain. These results identify a previously unrecognized function of 14-3-3ζ, that of mediating integrin-induced signaling. They show that 14-3-3ζ mediates Cdc42 and Rac activation. They also reveal a novel function of platelet GP Ib-IX, that of regulating integrin-induced cytoskeletal reorganizations by sequestering 14-3-3ζ. Signaling across integrins initiates changes in cell behavior such as spreading, migration, differentiation, apoptosis, or cell division. Thus, introduction of the 14-3-3ζ binding domain of GP Ibα into target cells might provide a method for regulating integrin-induced pathways in a variety of pathological conditions.

The 14-3-3 family of proteins are expressed in all eukaryotic cells. There are at least seven highly conserved isoforms encoded by different gene products. These proteins have molecular weights of 29,000 -32,000 and bind numerous cytoplasmic and nuclear signaling molecules including signaling molecules such as Raf, protein kinase C, p130 Cas , BAD, and phosphatidylinositol 3-kinase (1)(2)(3)(4)(5), proteins such as Cdc25 and Wee1 that are involved in cell cycle control (6,7), and proteins such as FKHRLl and DAF-16 (8,9) that are involved in regulation of transcription.
Interaction of 14-3-3 proteins with signaling molecules can localize, activate, inhibit, or stabilize the target molecules. Because 14-3-3 proteins are dimers, they have been considered as adaptor proteins that recruit and regulate the function of signaling molecules. Whereas they have been implicated in regulation of cell proliferation, cell cycle progression, apoptosis, and differentiation (10 -13), little is known about the possibility that 14-3-3 proteins might serve as adaptors in recruiting and regulating signaling molecules involved in transmission of signals across transmembrane receptors. It is interesting in this regard that one of the first proteins shown to bind 14-3-3 in intact cells was the GP 1 Ib-IX complex (14), an adhesion receptor that is expressed almost exclusively in platelets. The GP Ib-IX complex consists of a disulfide-linked heterodimer GP Ib␣/GP Ib␤ non-covalently associated with GP IX (15) and the complex associates with the leucine-rich repeat protein GPV, which appears to link two GP Ib-IX trimers (16,17). The extracellular domain of GP Ib␣ binds a number of proteins, including the extracellular matrix protein, von Willebrand factor (vWf), and the plasma proteins, thrombin (16, 18 -20), factor XI (21), and factor XIIa (22). The cytoplasmic domain of GP Ib␣ interacts with 14-3-3 (14). Platelets contain high levels of the , ␤, and ␥ isoforms and lower levels of the ⑀ and isoforms of 14-3-3 (2). The only isoform that has been shown to interact with GP Ib␣ is 14-3-3. This interaction was first identified when a 29-kDa protein was found to copurify with GP Ib-IX from platelet membrane extracts (23). Protein sequencing of internal fragments of the protein revealed that it was 14-3-3. Subsequent studies have shown interaction between GP Ib␣ and 14-3-3 in a two-hybrid system (24), shown that 14-3-3 and GP Ib-IX interact in platelet lysate (14), and that it coimmunoprecipitates with GP Ib␣ expressed in Chinese hamster ovary (CHO) cells (25). Studies with CHO cells expressing GP Ib-IX complex with truncated forms of GP Ib␣ have shown that truncation of the C-terminal 19 amino acids of GP Ib␣ abolishes 14-3-3 binding to the GP Ib-IX complex (14,25) and the use of synthetic peptides has located the binding site to the C-terminal five amino acids of GP Ib␣ (14).
When GP Ib-IX binds to vWf, signals are induced across the adhesion receptor. One consequence of this signaling is activation of intracellular signaling pathways leading to activation of ␣ IIb ␤ 3 , the ␤ 3 -containing integrin that mediates platelet adhesion and aggregation (25)(26)(27)(28). The finding that 14-3-3 interacted with the cytoplasmic domain of GP Ib␣ raised the possi-bility that 14-3-3 might serve to recruit signaling molecules involved in transmission of these signals from the cytoplasmic domain of GP Ib␣. However, studies in which the GP Ib-IX complex containing full-length or truncated GP Ib␣ were expressed in CHO cells suggested that the GP Ib/14-3-3 interaction is not required for transmission of the signals that convert CHO cell ␣ v ␤ 3 from an inactive form to a form that can bind immobilized vWf (28).
Integrins are another family of adhesion receptors that transmit signals upon ligand binding. Evidence from in vitro binding assays has suggested that 14-3-3 may also interact with the cytoplasmic domain of these adhesion receptors. In one study, ␤ 1 -and ␤ 3 -integrin subunits bound to 14-3-3␤ in a two-hybrid screen (29). In another study, 14-3-3␣␤ and 14-3-3␦ in leukocyte extracts bound to phosphopeptides with the sequence of the cytoplasmic domain of the ␤ 2 -integrin (30). Although there has been no clear evidence that any isoform of 14-3-3 interacts directly with integrin in intact cells, 14-3-3␤ has been reported to colocalize with ␤ 1 -integrin in human foreskin fibroblasts spreading on fibronectin (29).
The adhesion-induced signals transmitted across integrins induce cytoskeletal reorganizations that lead to cell spreading and migration and regulate pathways involved in development, inflammation, wound repair, cell division, differentiation, and apoptosis (31,32). Previous work has shown that the integrininduced cytoskeletal reorganizations are initiated by the Rho GTPases, Cdc42, and Rac (33,34). Whereas mechanisms by which these GTPases are recruited and activated following transmission of signals across other families of receptors have been well characterized, little is known about the way in which signals transmitted through integrin cytoplasmic domains activate these Rho GTPases.
In the present study, we considered the possibility that 14-3-3 might be involved in mediating integrin-induced Rho GTPase activation. To investigate this possibility we sequestered endogenous 14-3-3 in CHO cells, observed the functional consequences on integrin-induced signaling, and then expressed additional 14-3-3 to determine whether normal integrin-induced events were restored. The approach that we used to sequester 14-3-3 was to express GP Ib-IX in CHO cells. Normal CHO cells spreading on an integrin substrate extended filopodia and membrane ruffles and eventually formed well spread cells filled with stress fibers. However, if the GP Ib-IX complex was present, integrin-induced cytoskeletal reorganizations were inhibited. Cells expressing GP Ib-IX complex containing a truncated form of GP Ib␣, which lacked the 14-3-3 binding site, spread normally and normal cytoskeletal reorganizations occurred. Overexpression of 14-3-3 in cells in which endogenous 14-3-3 had been sequestered by GP Ib-IX restored cell spreading. Sequestration of 14-3-3 inhibited integrin-induced Cdc42 and Rac activation; overexpression of 14-3-3 restored GTPase activation.
These studies describe a previously unrecognized function of 14-3-3 proteins. They show that 14-3-3 has a critical role in inducing the cascades of signaling reactions that occur as integrins mediate cell adhesion. They show that 14-3-3 mediates step(s) upstream of Cdc42 and Rac activation. In addition, the studies show that the 14-3-3 binding domain on the cytoplasmic domain of GP Ib␣ can inhibit integrin-induced Cdc42 and Rac activation by a mechanism that involves sequestration of 14-3-3. Thus, introduction of the 14-3-3 binding domain of GP Ib␣ into target cells might provide a method for regulating integrin-induced events such as migration, cell division, differentiation, or apoptosis in a variety of pathological conditions.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-CHO cells were grown in Ham's F-12 medium (Invitrogen) containing 10% fetal bovine serum and a mixture of penicillin/streptomycin antibiotics (Invitrogen). Cells were stably transfected with the cDNA encoding GP Ib␤, GP IX, and full-length or truncated GP Ib␣ as previously described (35). The truncated form of GP Ib␣ was generated by introducing a stop codon at position 591 of the cytoplasmic domain of GP Ib␣, utilizing Stratagene's Double Take double-stranded mutagenesis system (35). The resulting protein lacked the C-terminal 19 amino acids. Expression of GP Ib-IX on the cell surface was analyzed by flow cytometric analysis using a monoclonal antibody against GP Ib␣ (Immunotech, Westbrook, ME).
In some experiments, cells expressing truncated GP Ib-IX were cotransfected with a chimeric receptor consisting of IL2R␣ extracellular and transmembrane domains linked to the GP Ib␣ cytoplasmic domain (kindly provided by Dr. Xiaoping Du, Department of Pharmacology, University of Illinois, College of Medicine, Chicago, IL) along with pSV2-hph. Cells co-transfected with IL2R␣ lacking a cytoplasmic domain were used as a control (kindly provided by Dr. Susan LaFlamme, Center for Cell Biology and Cancer Research, Albany Medical College, Albany, NY). Membrane expression of IL2R␣ was assessed using flow cytometry after staining of the cells with a monoclonal antibody against IL2R␣ (Immunotech). A FACScan flow cytometer was used for flow cytometry analysis (BD Biosciences, Advanced Cellular Biology, San Jose, CA) as described previously (36).
For transient transfection of an HA-tagged 14-3-3 construct encoded in the vector pCMV5 (kindly provided by Dr. Charles Abrams, Hematology-Oncology Division, University of Pennsylvania, Philadelphia, PA), cells were transfected for 3 h with 3 g of the construct and LipofectAMINE Plus reagent according to the manufacturer's instructions (Invitrogen).
Immunofluorescence Microscopy-Cells were incubated with 5 g/ml botrocetin (Pentapharm, Basel, Switzerland) in serum-free media and plated on Lab-Tek glass chamber slides (Nunc, Inc., Naperville, IL) that had been previously coated with 5 g/ml vWf (American Diagnostics, Greenwich, CT) and saturated with 3% bovine serum albumin (35). In some experiments, as described in the text, suspensions of cells were incubated for 30 min at room temperature with 4 mM Arg-Gly-Asp-Ser (RGDS) peptide (Sigma) or 2 mM EDTA, prior to plating. The cells were fixed at various time points, permeabilized with 0.5% Triton X-100, and stained as previously described (35,36). Actin filaments were stained with 1 g/ml tetramethylrhodamine isothiocyanate (TRITC)-labeled phalloidin (Sigma).
For analysis of cells transiently expressing HA-tagged 14-3-3, transfected cells were allowed to recover for 48 h in serum-containing medium. Cells were then replated onto vWf-coated dishes for 2 h and subsequently examined by immunofluorescence. Actin filaments were detected with TRITC-labeled phalloidin and HA-tagged 14-3-3 detected with monoclonal antibody (clone 12CA5, Roche Diagnostics) against HA epitope followed by secondary antibodies conjugated to Alexa 488 (Molecular Probes, Eugene, OR). Fluorescence microscopy was performed using a Leica TCS-NT laser scanning confocal microscope as described previously (37). Cell area measurement was performed using Image-Pro Plus software (Media Cybernetics, Silver Spring, MD). Data were presented as the mean Ϯ S.D. and were compared using Student's t test.
Biochemical Assay of Activated GTPases in CHO Cell Lysates-GST-p21 binding domain of PAK (PBD) was produced in DH5␣ cells transformed with pGEX plasmid containing PBD cDNA (kindly provided by Dr. Martin Schwartz, Department of Microbiology, Cardiovascular Research Center, Mellon Prostate Cancer Research Center, University of Virginia, Charlottesville, VA) and subsequently bound to glutathionecoupled agarose beads (Sigma) as previously described (38). CHO cells were plated on vWf in the presence of botrocetin for 45 min, scraped off the dishes, and lysed in an ice-cold buffer containing 50 mM Tris, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, pH 7.2, and a mixture of protease inhibitors (Complete, Roche Diagnostics). Cell lysates were cleared by centrifugation at 16,000 ϫ g for 8 min and supernatants were incubated with GST-PBD beads for 45 min. Beads were washed four times in ice-cold buffer containing 50 mM Tris, protease inhibitors, 1% Triton X-100, and 150 mM NaCl, pH 7.2. Proteins bound to the beads were solubilized by addition of an electrophoresis buffer containing 3% SDS and 2% 2-mercaptoethanol and Western blots were prepared.
Western Blot Analysis-For analysis of 14-3-3 isoforms normally expressed in CHO cells, cells were solubilized in an electrophoresis buffer containing 3% SDS and 2% 2-mercaptoethanol, boiled for 10 min, electrophoresed through SDS-polyacrylamide gels containing 10% polyacrylamide, and transferred to polyvinylidene difluoride membrane (Millipore, Bedford, MA) by standard Western blotting techniques (39). Western blots were probed with polyclonal antibodies against 14-3-3 ␣, ␤, ␥, and (Santa Cruz Biotechnology, Santa Cruz, CA) followed by anti-rabbit horseradish peroxidase-coupled antibody (Amersham Biosciences). To quantitate the amount of HA-tagged 14-3-3 in CHO cells transiently expressing this protein, Western blots of whole cell extracts were probed with a monoclonal antibody against HA epitope (Roche Diagnostics) and anti-mouse horseradish peroxidase-coupled antibody (Amersham Biosciences). For analysis of Cdc42 or Rac in whole cell extracts or activated Cdc42 or Rac bound to GST-coupled beads, Western blots were probed with monoclonal Cdc42 or Rac antibodies (Transduction Laboratories, San Diego, CA) and anti-mouse horseradish peroxidase-coupled antibody. Antigen-antibody complexes on Western blots were detected by incubating the membranes with ECL reagent (Amersham Biosciences) and exposing them to Kodak X-Omat films (Eastman Kodak, Rochester, NY). Quantitation of antigen-antibody complexes was performed by NIH Image 1.62 software.

Effect of the Cytoplasmic Domain of GP Ib-IX on Integrininduced Signaling-GP Ib␣ interacts with 14-3-3 in platelets;
it also interacts with CHO cell 14-3-3 when transfected into these cells (23,25). A form that lacks 19 C-terminal amino acids (m591) (35) retains its interaction with its other known binding partner, actin-binding protein (35), but fails to interact with 14-3-3 (25). To gain insight into the function of 14-3-3 in CHO cells, we transfected cells with cDNA for full-length GP Ib␣ or m591. Because GP Ib␣ is not efficiently expressed in the membrane of CHO cells unless all three subunits of the GP Ib-IX complex are present (40), we also transfected the cells with cDNA for GP Ib␤ and GP IX. FACS analysis revealed that comparable amounts of full-length and truncated protein were expressed (Fig. 1).
Many integrins cannot bind extracellular ligands until intracellular signals "activate" the integrins by inducing clustering or conformational changes (41). Activation can be induced by signals transmitted across a variety of cytokine or adhesion receptors (41). In platelets, activation is induced by signals transmitted as GP Ib-IX binds vWf (42). Previously, we have shown that binding of vWf to GP Ib-IX expressed in CHO cells also induces signals that activate endogenous CHO cell integrins, converting them from a form that cannot interact with vWf to one that can (28). These signals are transmitted across GP Ib-IX whether the GP Ib␣ cytoplasmic domain contains the 14-3-3 binding site or not (35). Interaction of activated integrin with vWf activates pathways that induce cytoskeletal reorganizations, changes in cell shape, and cell spreading (28). In the present study, we investigated the possibility that 14-3-3 is involved in integrin-induced signal transmission by determining whether integrin-induced cytoskeletal reorganizations were different in cells expressing full-length GP Ib␣ (which sequesters CHO cell 14-3-3) compared with cells expressing GP Ib␣ that lacked the 14-3-3 binding site.
Cells were plated on vWf and allowed to spread for up to 2 h. As reported previously (28), the initial adhesion of the cells was induced by GP Ib-IX as shown by inhibition of adhesion by antibodies against GP Ib␣, whereas the spreading resulted from activation of integrins and signaling across ligand-occupied integrin, because RGDS, EDTA, and antibodies against ␣ v ␤ 3 inhibited cell spreading.
Spreading was markedly inhibited in cells expressing fulllength GP Ib␣ as compared with truncated GP Ib␣ (Fig. 2). The area of cells expressing full-length GP Ib␣ was 59% of the area of the cells expressing truncated GP Ib␣ after 20 min of spreading, 62% after 40 min, and 45.7% after 120 min of spreading (Fig. 3). The area of 100 cells was measured for each cell line and each time point. Typically, spreading of cells on integrin substrates leads to actin polymerization and formation of various actin filament organizations. The earliest changes are induced by Cdc42, which causes localized polymerization of filaments that push membranes outwards in fine filopodia (43). Activation of Rac subsequently causes the formation of networks of submembranous filaments and the extension of lamellipodia (43,44). At later times, activation of RhoA leads to the formation of stress fibers that mediate the stable attachment of spread cells (43,45).
To gain insight into the step at which the presence of GP Ib␣ affected integrin-induced spreading, the actin filament organization was examined by immunofluorescence (Fig. 2). Spreading and filament organizations in cells expressing truncated GP Ib␣ were similar to those typically seen as CHO cells spread on an integrin substrate. Thus, as shown in Fig. 2, cells had begun to spread by 20 min; many cells contained numerous fine filopodia (e.g. cell labeled with small arrowhead in panel A), others had extended membrane ruffles (e.g. cell labeled with large arrowhead in panel A). After 40 min the cells contained actin filaments in submembranous locations and in numerous filopodia (panel B). After 120 min they were well spread and typically filled with stress fibers (panel C). The actin organization in cells expressing full-length GP Ib␣ was very different. From the earliest time point examined, spreading was delayed and the actin filament organization was abnormal. Thus, after 20 min, the surface of the adherent cells remained quite smooth (panel D). After 40 min the cells had started to spread but the surface was still smooth; the filopodia present in the cells expressing truncated GP Ib␣ (panel B) were markedly absent (panel E). At 120 min, when cells expressing truncated GP Ib␣ were well spread and filled with stress fibers, cells expressing full-length receptor still contained few filopodia; similarly they contained few stress fibers (panel F). The almost complete absence of filopodia after 40 min or stress fibers after 120 min can be seen more clearly in the higher magnification images shown in Fig. 4. Quantitative analysis revealed that 92% Ϯ 3.3 of the cells expressing truncated GP Ib␣ contained stress fibers (600 cells were counted) as compared with 8% Ϯ 2.9 of the cells expressing full-length GP Ib␣ (600 cells were counted). Five different clones for each cell type were examined and showed identical results in over 10 independently performed experiments.
Inhibition of Integrin-dependent Signaling Results from the Presence of the GP Ib␣ Cytoplasmic Domain-Control experi-ments were performed to ensure that the inhibitory effects on integrin-induced cytoskeletal reorganizations observed in cells expressing full-length GP Ib␣ resulted from the presence of the 14-3-3 binding site on GP Ib␣ rather than from differential signaling across full-length and truncated GP Ib␣. In these experiments, we used only CHO cells expressing the GP Ib-IX

FIG. 2. Effect of GP Ib␣ cytoplasmic domain on integrin-induced reorganization of the cytoskeleton.
Cells transfected with the cDNAs for GP Ib␤ and GP IX together with either truncated GP Ib␣ (m591) (panels A-C) or full-length GP Ib␣ (panels D-F) were allowed to spread on vWf-coated slides in the presence of botrocetin for the indicated times. Cells were fixed, permeabilized, and actin filaments were stained with TRITC-labeled phalloidin. Cells were examined with a confocal microscope. In panel A, the arrow with a small arrowhead indicates a cell extending filopodia; the arrow with a large arrowhead indicates a cell extending membrane ruffles. Bar, 20 m.

FIG. 3. Comparison of the cell area during the spreading of CHO cells expressing GP Ib-IX containing full-length or truncated GP Ib␣ (m591).
Cells transfected with the cDNAs for GP Ib␤ and GP IX together with either full-length GP Ib␣ (black bars) or truncated GP Ib␣ (m591) (dotted bars) were allowed to spread on vWf-coated slides in the presence of botrocetin for the indicated times. Cells were fixed, permeabilized, and actin filaments were stained with TRITC-labeled phalloidin. The cell area was measured using Image-Pro software. *, cells expressing full-length GP Ib␣ showed a marked reduction in spreading (p Ͻ 0.0001).

FIG. 4. Actin filament organization in cells expressing GP
Ib-IX containing full-length or truncated GP Ib␣ (m591) in CHO cells. Cells transfected with cDNAs for GP Ib␤ and GP IX together with either truncated GP Ib␣ (m591) (panel A) or full-length GP Ib␣ (panels B and C) were allowed to spread on vWf-coated slides in the presence of botrocetin for the indicated times. Cells were fixed, permeabilized, and actin filaments were stained with TRITC-labeled phalloidin. Cells were examined with a confocal microscope. Cells expressing truncated GP Ib␣ contained numerous actin filaments and filopodia, as soon as 40 min after plating. In contrast, cells expressing full-length GP Ib␣ cytoplasmic domain showed few filopodia or actin filament bundles, even 2 h after plating. Bar, 5 m. complex containing truncated GP Ib␣. Thus, all of the signaling that induced integrin activation was induced by the GP Ib-IX complex containing truncated GP Ib␣. However, some of the cells were co-transfected with a chimeric receptor consisting of the extracellular and transmembrane domains of the IL2 receptor ␣ chain linked to the GP Ib␣ cytoplasmic domain (IL2R␣-Ib␣). As a control, some cells were transfected with a construct consisting of IL2R␣ extracellular and transmembrane domains only (IL2R␣-I). FACS analysis showed similar amounts of IL2R␣-I and chimeric IL2R␣ in the two cell lines (Fig. 5, compare the dotted lines in panels B and C). As shown in Fig. 5, the presence of the construct containing only extracellular and transmembrane domains of IL2R␣ had no apparent effect on cell spreading (compare panels D and E). In contrast, the presence of the chimeric molecule containing the GP Ib␣ cytoplasmic domain almost completely inhibited spreading. These cells also failed to form detectable stress fibers even after 120 min (panel F). They had an actin filament organization and shape resembling that of cells expressing only GP Ib-IX complex containing full-length GP Ib␣ (Fig. 6, compare panels C and D). Quantitation of 600 cells for each cell line showed that 83% Ϯ 3.7 of the cells expressing truncated GP Ib␣ alone contained stress fibers. In contrast, only 18% Ϯ 2.8 of the cells expressing the chimeric IL2 receptor/GP Ib␣ cytoplasmic domain in addition to truncated GP Ib␣ contained stress fibers. The results are representative of five independent experiments. Cells from six different IL2R␣ clones and from six different IL2R␣-GP Ib␣ showed similar results.

Experiments to Identify Mechanisms by Which 14-3-3 Regulates Integrin-induced Cytoskeletal Reorganizations-Because
full-length, but not truncated, GP Ib␣ is known to bind endogenous 14-3-3 when expressed in CHO cells (25), the inhibitory effect of full-length GP Ib␣ is likely to result from sequestration of endogenous 14-3-3. The finding that co-expression of the chimeric IL2R/GP Ib␣ cytoplasmic domain molecule in cells expressing truncated GP Ib␣ inhibited integrin-induced signaling is consistent with this interpretation. As a further test of this, we transiently transfected cells expressing full-length GP Ib␣ with an HA-tagged 14-3-3 construct. As shown in Fig. 7, the cells expressing only GP Ib-IX containing full-length GP Ib␣ showed little spreading; as seen in other experiments described above, there were few filopodia or membrane ruffles, even after 120 min of spreading. In contrast, the cells that also expressed HA-tagged 14-3-3 showed numerous long filopodia (e.g. cell labeled with a small arrowhead in Fig. 7) or extensive membrane ruffling (e.g. cell labeled with a large arrowhead in Fig. 7). Quantitation of 300 HA-tagged 14-3-3 expressing cells revealed that 75% showed increased spreading as compared with cells not expressing this construct. 25% showed numerous filopodia, indicative of Cdc42 activation, whereas the rest showed increased lamellipodia and membrane ruffles, indicative of Rac activation.
Experiments to Identify Mechanisms by Which 14-3-3 Regulates Integrin-induced Cytoskeletal Reorganizations-The re-

FIG. 5. Effect of a chimeric molecule containing the cytoplasmic domain of GP Ib␣ on spreading of CHO cells expressing GP Ib-IX containing truncated GP Ib␣.
Cells expressing GP Ib-IX complex containing truncated GP Ib␣ (m591) (panels A and D) were cotransfected with IL2R␣ extracellular and transmembrane domains alone (panels B and E) or a chimeric receptor comprising IL2R␣ extracellular and transmembrane domains linked to full-length GP Ib␣ cytoplasmic domain (panels C and F). Membrane expression of GP Ib-IX and IL2R␣ was assessed by FACS analysis (panels A-C); dashed lines represent the negative control with cells stained only with FITC-labeled secondary antibody, dotted lines represent expression of IL2R␣, and solid lines represent expression of GP Ib-IX, as detected with an antibody that binds to the extracellular domain of GP Ib␣. Cells were allowed to spread for 2 h on vWf, fixed, permeabilized, stained for actin, and examined with a confocal microscope (panels D-F). Cells expressing truncated GP Ib␣, whether in the presence or absence of IL2R␣, spread and contained numerous actin filaments and stress fibers. In contrast, cells that also expressed the cytoplasmic domain of GP Ib␣ (as a chimeric IL2R␣ molecule) showed little spreading or stress fiber formation. Bar, 20 m.
sults described so far suggest that 14-3-3 is involved in integrin-induced cytoskeletal reorganizations following signaling across ␣ v ␤ 3 in CHO cells. Signaling molecules known to be involved in integrin-induced cytoskeletal reorganizations are the Rho GTPases Cdc42, Rac, and RhoA (33,34,38). The inhibitory effects observed in our experiments suggest that when 14-3-3 is sequestered, integrin signaling is inhibited at a very early stage. The formation of numerous filopodia and membrane ruffles in cells in which HA-tagged 14-3-3 was coexpressed with full-length GP Ib␣ is consistent with a role for 14-3-3 upstream of Cdc42 and Rac activation. To directly test this, a biochemical assay was used to measure the activity of Cdc42 and Rac. Cells expressing truncated or full-length GP Ib␣ were allowed to spread on vWf for 45 min, lysed, and active (GTP-bound) Cdc42 and Rac isolated by binding to GST-PBD beads. Total and GTP-bound Cdc42 and Rac were visualized on Western blots. As shown in Fig. 8, active Cdc42 and Rac were present in cells expressing truncated GP Ib␣ (which could not sequester endogenous 14-3-3). However, active Cdc42 and Rac were markedly decreased in cells with full-length GP Ib␣. The Western blot shown is representative of three experiments. Quantitation of the three experiments showed that Cdc42 ac-tivation was reduced 14.3 Ϯ 8-fold and Rac activation 2.8 Ϯ 1.2-fold in cells expressing full-length compared with truncated GP Ib␣.
Immunofluorescence images of cells in which HA-tagged 14-3-3 was coexpressed with full-length GP Ib␣ (Fig. 7) suggested that the ability of ligand-occupied integrin to transmit signals leading to Cdc42 and Rac activation could be restored by expression of 14-3-3. To directly test this, cells expressing truncated or full-length GP Ib␣ were transiently transfected with HA-tagged 14-3-3 and allowed to spread on vWf for 45 min. The amount of 14-3-3 expressed was readily detected on Western blots with an HA antibody and was comparable in the two cell lines (Fig. 9, panel A). As in the experiments shown in Fig.  8, activation of Rac was markedly reduced in cells expressing full-length GP Ib␣ compared with those expressing truncated GP Ib␣ (Fig. 9, panel B). Expression of additional 14-3-3 in cells expressing truncated GP Ib␣ did not produce any detectable increase in Rac activation (presumably because endogenous 14-3-3, which is not sequestered by truncated GP Ib␣, is not a limiting factor in regulating integrin-induced Rac activation). In contrast, transient overexpression of 14-3-3 markedly increased activation of Rac in cells expressing full-length GP Ib␣ (i.e. cells in which endogenous 14-3-3 was sequestered). The results shown are representative of three independent experiments. Quantitation of GTP-bound Rac in these experiments showed that overexpression of 14-3-3 increased Rac activation 2.2 Ϯ 1.1-fold in cells expressing full-length GP Ib␣. DISCUSSION 14-3-3 proteins are involved in pathways regulating proliferation, cell cycle progression, apoptosis, and differentiation (46). This study identifies another function of 14-3-3 proteins. It shows that 14-3-3 mediates signaling across activated integrin as the integrin attaches to ligand in the extracellular matrix. It shows that 14-3-3 acts upstream of Cdc42 and Rac, two GTPases that induce the cytoskeletal reorganizations responsible for spreading and migration. These findings suggest that 14-3-3 may be a key player in the initial steps of the signaling pathways that regulate physiological and pathological events such as development, homeostasis, lymphocyte trafficking, metastasis, thrombosis, and the progression of arteriosclerosis that are dependent upon integrin-induced cell motility, spreading, and migration.
The approach that we used to determine whether 14-3-3 regulates integrin signaling was to express the GP Ib-IX complex, a 14-3-3-binding protein normally expressed only in platelets, in CHO cells. It has been shown previously that this complex binds endogenous 14-3-3 when expressed in CHO cells and that the interaction is ablated by deletion of the C terminus of the GP Ib␣ subunit (14,25). We reasoned that comparison of signaling in cells expressing full-length and truncated GP Ib␣ would provide insights into potential functions of 14-3-3. Our results show that cytoskeletal changes induced by integrin-induced signaling in CHO cells were markedly inhibited when GP Ib-IX containing full-length GP Ib␣ was expressed. In contrast, inhibition did not occur in cells in which GP Ib␣ lacked the 14-3-3 binding site, a finding that suggests that inhibition resulted from sequestration of a protein by the C-terminal 19 amino acids of GP Ib␣. Experiments in which cells were transfected with a chimeric molecule containing the transmembrane and extracellular domains of the IL2 receptor and the GP Ib␣ cytoplasmic domain supported this conclusion.
It is conceivable that molecules other than 14-3-3 can associate with full-length GP Ib␣ but not with the truncated form of GP Ib␣. The 14-3-3 isoforms show a high level of functional conservation (46) so it is possible that other isoforms of 14-3-3 interact with GP Ib␣. We found that CHO cells express proteins cross-reacting with antibodies against ␤, ␥, and isoforms on Western blots. However, 14-3-3 is the only isoform that has been shown to interact with GP Ib␣. Based on the well char- acterized interaction between 14-3-3 and GP Ib␣ in the GP Ib␣ expressing CHO cells, we can assume that this isoform was sequestered in the present study. Together with the observation that integrin-induced cytoskeletal reorganizations and Rho GTPase activation were restored by overexpression of 14-3-3, the present studies show that the isoform can mediate integrin-induced Cdc42 and Rac activation. Our previous studies have shown that the cytoskeletal reorganizations induced in the CHO cell system utilized in the present study are induced as a consequence of signaling across ␣ v ␤ 3 . Future studies will be needed to determine whether signaling across other integrins is also dependent upon 14-3-3.
Although these studies show that 14-3-3 mediates the activation of ␣ v ␤ 3 , they do not comment on the possibility that other isoforms of CHO cell 14-3-3 may also be able to interact with GP Ib␣ and are involved in integrin-induced signaling. Future studies will be needed to determine whether GP Ib␣ can interact with and sequester any of the other CHO cell 14-3-3 isoforms in addition to 14-3-3. If so, it is conceivable that they might also be able to mediate integrin-induced Rho GTPase activation. It is of interest in this regard that 14-3-3␣␤ and 14-3-3␦ were among the proteins in a leukocyte extract that bound to a ␤ 2 -integrin cytoplasmic domain peptide (30). In a study testing the interaction between 14-3-3 and ␤ 1 -and ␤ 3integrin subunits, it was the ␤ isoform of 14-3-3 that was used (29). Future studies will be needed to determine whether these isoforms can also mediate integrin-induced Cdc42 and Rac activation.
Little is known about the way in which integrins activate Rho GTPases. Previously, we have shown that one of the earliest events in integrin-induced signaling is activation of calpain (47,48) and the calpain-induced formation of clusters containing calpain-cleaved integrin, skelemin, and active calpain (49,50). Dominant-negative Rac accumulates in these clusters and prevents Rac activation and cell spreading, suggesting that the clusters are the site of integrin-induced Rac activation (49). Colocalization of 14-3-3␤ and clusters of ␤ 1integrin in spreading cells has been reported (29) and is consistent with a model in which 14-3-3 is involved in integrininduced signaling. Based on known functions of 14-3-3, it appears likely that it serves as adaptor, recruiting signaling molecules or regulating their function at sites of clustered integrin. The evidence that 14-3-3␤ can bind integrin cytoplasmic domains in a two-hybrid screen (29) and 14-3-3␣␤ and -␦ in leukocyte extracts can bind to a phosphorylated ␤ 2 -integrin peptide (30) is consistent with the possibility that 14-3-3 provides a direct link between the cytoplasmic domain of the integrin and signaling molecules involved in Rho GTPase activation. However, clear evidence that any form of 14-3-3 interacts directly with integrin in intact cells has not yet been presented and future studies will be needed to directly test this possibility. One signaling molecule that has been implicated in integrin-induced signaling and shown to interact with 14-3-3 in a two-hybrid system is p130 Cas (3). This molecule colocalizes with clustered integrins (3) and is tyrosine phosphorylated during integrin-induced signaling (51,52). Phosphorylation induces its association with the adaptor, Crk (52,53). The resulting p130 Cas/crk complex can interact with DOCK180, an exchange factor for Rac (54,55), and is involved in Rac-induced membrane ruffling (56). Thus, a potential model is one in which 14-3-3 localizes p130 Cas/Crk complexes such that they can induce Rac activation at the site of clustered integrins.
In addition to describing a new function for 14-3-3 and providing further insights into the mechanisms involved in the initial steps of integrin-induced signaling, the present studies raise the possibility that the GP Ib-IX/14-3-3 interaction might provide a previously unrecognized mechanism for regulating integrin-induced signaling pathways in platelets. In the present studies, expression of GP Ib-IX in CHO cells was sufficient to sequester endogenous 14-3-3 to a level below that needed for integrin-induced Cdc42 and Rac activation. Whereas no information is available on the relative ratio of GP Ib/14-3-3 in platelets, it seems possible that, as in CHO cells, GP Ib␣/14-3-3 interactions could serve to inhibit integrininduced Rho GTPase activation and cytoskeletal reorganizations. As seen in the present study, signaling across integrins in cultured cells typically induces activation of Rho GTPases and cytoskeletal reorganizations that culminate in the formation of stress fibers and focal adhesions (34). In platelets, however, only a few filopodia and stress fibers form as during spreading on an integrin substrate; even in a fully spread platelet, only 3-4 stress fibers are present (57). Thus, integrininduced cytoskeletal reorganizations in spreading platelets more closely resemble those in CHO cells expressing full-length GP Ib␣ in the present study than those in cells expressing truncated GP Ib␣. Previous studies (25)(26)(27)(28), using both platelets and CHO cells have shown that GP Ib-IX induces signals that activate ␣ IIb ␤ 3 in platelets. The present study suggests that GP Ib-IX might also have a role in regulating outside-in signaling, limiting the activation of signaling molecules and subsequent cytoskeletal reorganizations that occur following integrin-ligand interactions in platelets.
In such a model, the potential would exist for differential regulation of integrin-induced activation of signaling pathways depending on the amount of 14-3-3 bound to GP Ib-IX. The model that we propose, in which sequestration of 14-3-3 by GP Ib-IX is responsible for the limited spreading and stress fiber formation in adherent platelets compared with other cell types, is based on the assumption that 14-3-3 remains bound to GP Ib-IX during spreading of platelets. It is generally considered that 14-3-3 remains associated with GP Ib-IX when platelets are activated. There is a report, however, that 14-3-3 is released from GP Ib-IX during shear-induced activation of platelets (58). Shear-induced activation is of particular interest because of its potential involvement in pathological events (26). If 14-3-3 does indeed remain associated with GP Ib-IX in platelets activated by soluble agonists but dissociates when platelets are activated by shear, the results in the present study suggest that platelets activated in the circulation at sites of shear might undergo enhanced integrininduced cytoskeletal changes as compared with those activated at normal sites of injury.
In summary, the present findings: 1) identify a previously unrecognized function of 14-3-3, showing that 14-3-3 acts as a key mediator of integrin-induced cytoskeletal changes; 2) provide information on the way in which signals from the cytoplasmic domain of ␣ v b 3 can activate Cdc42 and Rac, showing that 14-3-3 is a critical mediator in the transmission of these signals; 3) suggest a previously unrecognized role of GP Ib-IX in platelets, pointing to a mechanism by which sequestration of 14-3-3 by the complex limits integrin-induced platelet spreading while release of 14-3-3 from GP Ib-IX could increase integrin-induced cytoskeletal reorganizations; 4) show that when present in cultured CHO cells, the cytoplasmic domain of GP Ib␣ can sequester sufficient endogenous 14-3-3 to almost totally inhibit integrin-induced Cdc42 and Rac activation. Signaling across the integrin family of adhesion receptors initiates pathways leading to cytoskeletal reorganizations and a variety of changes in cell behavior such as cell spreading, migration, differentiation, apoptosis, or cell division. Thus the present studies also raise the possibility that introduction of the 14-3-3 binding domain of GP Ib␣ into target cells might provide a method for regulating integrin-induced pathways in a variety of pathological conditions.