A Membrane-distal Segment of the Integrin a IIb Cytoplasmic Domain Regulates Integrin Activation*

Previous evidence suggests that interactions between integrin cytoplasmic domains regulate integrin activation. We have constructed and validated recombinant structural mimics of the heterodimeric a IIb b 3 cytoplas- mic domain. The mimics elicited polyclonal antibodies that recognize a combinatorial epitope(s) formed in mix-tures of the a IIb and b 3 cytoplasmic domains but not present in either isolated tail. This epitope(s) is present within intact a IIb b 3 , indicating that interaction between the tails can occur in the native integrin. Furthermore, the combinatorial epitope(s) is also formed by introducing the activation-blocking b 3 (Y747A) mutation into the b 3 tail. A membrane-distal heptapeptide sequence in the a IIb tail ( 997 RPPLEED) is responsible for this effect on b 3 . Membrane-permeant palmitoylated peptides, con- taining this a IIb sequence, specifically blocked a IIb b 3 activation in platelets. Thus, this region of the a IIb tail causes the b 3 tail to resemble that of b 3 (Y747A) and suppresses activation of the integrin.

The integrin family of adhesion receptors is essential for the development and functioning of multicellular animals (1). Integrin-mediated adhesion is rapidly and precisely regulated, a process that is often central to integrin functions (2). One important regulatory mechanism is cellular modulation of integrin affinity for ligand (activation). These affinity changes arise from both changes in the conformation of the extracellular domain and lateral clustering of integrins in the plane of the membrane (3). Integrin activation has been widely documented among members of this receptor family and controls cell adhesion, migration, and the assembly of the extracellular matrix (4,5). Thus, integrin activation plays an important role in mediating integrin functions.
Integrins are noncovalent heterodimers of type I transmembrane protein subunits termed ␣ and ␤. Each subunit has a large (Ͼ700 residues) N-terminal extracellular domain, a single membrane-spanning domain, and a generally short  residues) cytoplasmic domain (5). In general, integrin ␣ and ␤ subunits contain a remarkably conserved 7-10-residue motif near the junction of the transmembrane and cytoplasmic domains (membrane-proximal segment) (6). The remainders of the cytoplasmic domain sequences are generally less well conserved. Integrin ␤ cytoplasmic domains are required for activation because mutations or truncations of specific membranedistal sequences in ␤ cytoplasmic tails can disrupt activation (7)(8)(9)(10). A particularly sensitive site is a highly conserved NPX(Y/F) motif in the ␤ cytoplasmic domain where substitution of the Tyr with Ala (e.g. ␤ 3 (Y747A)) blocks activation (11). This region of the ␤ tail is important for binding to cytoskeletal proteins, such as talin because a Tyr 3 Ala substitution in the NPX(Y/F) motif also disrupts talin binding to ␤ tails in vitro (12,13). Moreover, overexpression of an integrin-binding fragment of talin in cells activates integrin ␣ IIb ␤ 3 (14). Thus, interactions of membrane distal portions of the ␤ cytoplasmic domain with proteins such as talin appear to be important in integrin activation.
In addition to a role for the membrane-distal portion of the ␤ cytoplasmic domain, interactions between integrin ␣ and ␤ subunit cytoplasmic tails may regulate activation. Deletion of the membrane-proximal region of either ␣ or ␤ tail activates integrins (7,10,(15)(16)(17). In addition, specific point mutations in this segment of both the ␣ and ␤ subunits promote constitutive bidirectional signaling in integrin ␣ IIb ␤ 3 (18) and other integrins (19). Complementary mutations in the ␣ and ␤ subunits suggest that these activating mutations disrupt an interaction between the highly conserved membrane-proximal portions of the ␣ and ␤ cytoplasmic tails (18). In vitro integrin ␣ and ␤ cytoplasmic domain interactions have been reported by surface plasmon resonance analysis (20,21) and by alterations in circular dichroism and intrinsic fluorescence (22). Furthermore, replacement of the ␣ and ␤ cytoplasmic domains with acidic and basic peptides that form an ␣-helical coiled-coil caused inactivation of integrin ␣ L ␤ 2 (23). In contrast, replacement of these cytoplasmic domains with two basic peptides that do not form an ␣-helical coiled-coil activated ␣ L ␤ 2 . Thus, there is evidence to suggest that an interaction between integrin ␣ and ␤ tails regulates activation.
We previously used a synthetic strategy to produce a model of the cytoplasmic domain of integrin ␣ IIb ␤ 3 (24). The integrin cytoplasmic domains are tethered at their N termini to membrane spanning presumptive ␣-helices. Moreover, they are laterally constrained because the subunits interact with each other. More importantly, they have vertical constraints, because they are initially parallel to each other and are in a specific vertical stagger as they exit the membrane. The model protein was made by the covalent ligation of two synthetic polypeptides (called "minisubunits") consisting of an integrin cytoplasmic tail at the C terminus, connected to a 28-residue stretch comprised of four heptad repeats derived from tropomyosin. The parallel orientation and stagger of the two tails is defined by the formation of the covalent linkage and by a noncovalent helical coiled-coil structure between heptad repeats on each minisubunit. We now describe a completely recombinant model of the ␣ IIb ␤ 3 cytoplasmic domain. By using recombinant proteins, we avoided limitations of polypeptide length and modest yield encountered in the initial synthetic approaches. Moreover, we modified the design of the model protein to contain four heptad repeats derived from the GCN4 transcription factor in place of those derived from tropomyosin. We used two variant sequences that preferentially heterodimerize (25) to obviate an inherent 50% loss in yield of heterodimer.
In the present work, we utilized these model proteins as a tool to probe the potential role of integrin ␣␤ tail interaction in regulating activation. Antibodies raised against these model proteins revealed the presence of combinatorial epitopes formed by a mixture of the ␣ IIb and ␤ 3 tails but not present in either isolated tail. These combinatorial epitopes were present in intact integrin ␣ IIb ␤ 3 isolated from platelets, suggesting that the ␣ IIb and ␤ 3 tail can interact in the intact receptor. Moreover, the same set of epitopes was present in the ␤ 3 tail bearing a Tyr 747 3 Ala mutation, even in the absence of the ␣ IIb tail. This result indicates that ␣ IIb cytoplasmic domain causes the ␤ 3 tail to resemble the activation-defective ␤ 3 (Y747A) mutant. Furthermore, we mapped the ␣ IIb residues required for this interaction into a minimal heptapeptide sequence and found that palmitoylated peptides containing this sequence specifically blocked the agonist-induced activation of integrin ␣ IIb ␤ 3 in platelets. Thus, our data provide evidence that the ␣ IIb and ␤ 3 tails can interact in the native integrin and indicate that the interaction opposes integrin activation by altering the structure of the ␤ 3 cytoplasmic domain.

MATERIALS AND METHODS
Model Protein Synthesis and Peptide Production-Polymerase chain reaction as described (12) was used to create a cDNA encoding the modified GCN4 heptad repeat protein sequences reported by John et al. (25). The cDNAs were ligated into a NdeI-HindIII-restricted modified pET15b vector (12) (Novagen). ␣ IIb and ␤ 3 integrin cytoplasmic domain cDNAs were generated by polymerase chain reaction from appropriate cDNAs using forward oligonucleotides introducing a 5Ј-HindIII site and reverse oligonucleotide creating a 3Ј-BamHI site directly after the stop codon. Integrin cytoplasmic domains were joined to the helix as HindIII-BamHI fragments and verified by sequencing. Recombinant expression in BL21(DE3)pLysS cells (Novagen) and purification of the recombinant products was performed as described with an additional final purification step on a reverse phase HPLC 1 column (Vydac) (12). Typical yields were ϳ10 mg/liter of bacterial culture. To form heterodimers, a mixture of 45 M ␤ 3 and 90 M ␣ IIb cytoplasmic tail model proteins were denatured for 20 min at 100°C in the presence of 10 mM dithiothreitol. The dithiothreitol was removed by gel filtration through a PD 10 column (Bio-Rad), and the mixture was then air-oxidized by stirring overnight. The disulfide-bonded products were purified by reverse phase C18 HPLC. The products were analyzed by electrospray ionization mass spectrometry using an API-III quadrupole spectrometer (Sciex). Typically we recovered ϳ30 mg of heterodimer/preparation. The identity of each model protein is specified by the nature of the GCN4 helix (Jun-like ϭ J, Fos-like ϭ F), the number of Gly spacers (G0 -G4), and the identity of the integrin tail (␣ IIb and ␤ 3 ), e.g. JG3␣ IIb .
Synthetic peptides were prepared by the Scripps Microchemistry Core using a Gilson AMS 422 or ABI 430A Peptide Synthesizers. One ␤ 3 cytoplasmic domain peptide (␤ 3 (Ile 719 -Thr 762 )) was a generous gift of Drs. Ed Plow and Tom Haas (Cleveland Clinic). The peptides were routinely Ͼ90% homogeneous as judged by reverse phase HPLC using an analytical Vydac C-18 column. Their mass varied by less than 0.1% from that predicted from their sequence as judged by electrospray ionization mass spectroscopy. Peptides were desalted prior to use, and peptide concentrations were verified by quantitative amino acid analysis.
Antibodies-Antibodies to the ␣ IIb ␤ 3 cytoplasmic domain heterodimer model protein (see Fig. 1) (anti-␣ IIb ␤ 3 -Cyt) were raised in 2.5-kg New Zealand White rabbits. Antibodies were also produced against the ␤ 3 (Y747A) model protein (anti-␤ 3 (Y747A)). 100 l of a water solution containing 1 mg/ml of the model protein was emulsified in 1 ml of incomplete Freund's adjuvant and administered subcutaneously to the rabbits. Two additional injections at 2-week intervals were followed by bleeding (ϳ50mls) at monthly intervals until the rabbits were sacrificed. The blood was permitted to clot at room temperature, and the serum was recovered by centrifugation. Serum was heat-inactivated at 56°C for 30 min and stored at Ϫ20°C. A preimmune serum, obtained prior to immunization, was used as a control. Antibodies against a synthetic peptide containing the last 20 residues of the ␤ 3 cytoplasmic domain (anti-␤ 3 -Cyt, rabbit 8275; Ref. 15) and the extracellular domain of ␣ IIb ␤ 3 (D57; Ref. 26) have been described.
Enzyme-linked Immunosorbent Assays-␣ IIb ␤ 3 was purified by gel filtration as described previously (26) with omission of the heparin and Con A affinity chromatography steps. The final product was greater than 95% homogenous as judged by SDS-PAGE. In the enzyme-linked immunosorbent assay (ELISA) the ␣ IIb ␤ 3 was used at a concentration of 5 g/ml in a coating buffer containing 0.1 M NaHCO 3 and 0.05% NaN 3 . 50 l/well was used to coat Immulon II microtiter wells at 4°C overnight. After removal of the coating solution, 150 l of blocking buffer (coating buffer containing 5% bovine serum albumin) was added. After an additional 1-h incubation at 4°C, the blocking buffer was removed, and the plates were washed three times with wash buffer (0.01 M Tris, 0.15 M NaCl, 0.01% thimerisol, 0.05% Tween 20, pH 8.0). Preliminary titration experiments established a dilution of anti-model protein antibody resulting in 75% maximal binding. 25 l of the competitor was added to each well followed by 25 l of this dilution of the anti-model protein antibody. Following mixing, the plate was covered for 1 h at 37°C and washed four times with wash buffer. To quantify bound antibody, 50 l of horseradish peroxidase-conjugated goat anti-rabbit IgG (Bio-Rad) was diluted to a concentration of 1:1000 in wash buffer containing 1 mg/ml bovine serum albumin. These plates were then incubated at 37°C for 1.5 h. After four washes, bound antibody was assayed by measuring peroxidase activity with O-phenylenediamine as a substrate and quantifying reaction product by its optical density at 490 nm. The data were expressed as B/B 0 where B ϭ A 490 in the presence of competitor and B 0 ϭ A 490 in its absence. In some experiments, varying concentrations of ␣ IIb peptides were added to a fixed, saturating quantity of the ␤ 3 model protein (10 -50 nM). Competition was again expressed as B/B 0 ; however, B 0 was A 490 in the presence of the ␤ 3 protein and no added ␣ IIb peptide. EC 50 was defined as the dose of ␣ IIb resulting in B/B 0 ϭ 0.5.
Immunoprecipitations-The generation of Chinese hamster ovary cells expressing recombinant integrins ␣ IIb ␤ 3 and ␣ IIb ⌬996␤ 3 ⌬717 has been described (10). Chinese hamster ovary cell lines were cell surfacelabeled with sulfo-NHS-Biotin (Pierce) following the manufacturer's instructions. Cells were lysed on ice for 30 min in an immunoprecipitation buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 mM EDTA, 10 mM benzamidine HCl, 0.02% sodium azide, 1% Triton X-100, 0.05% Tween 20, 2 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, and 5 g/ml leupeptin). After clarification by centrifuging at 12,000 rpm for 20 min at 4°C, cell lysate was then incubated with protein G-Sepharose (Amersham Pharmacia Biotech) coated with antibody overnight at 4°C. The beads were washed with the immunoprecipitation buffer four times, and the precipitated polypeptides were extracted with SDS sample buffer. Precipitated cell surface biotin-labeled polypeptides were separated by SDS-PAGE under nonreducing conditions and detected with streptavidin-peroxidase followed by ECL (Amersham Pharmacia Biotech).
NMR Spectroscopy-The ␤ 3 minisubunit of the ␣ IIb ␤ 3 heterodimer model protein was biosynthetically labeled with 15 N as described. 2 NMR signal intensities were taken from a two-dimensional 15 N-1 H heteronuclear single-quantum coherence (28, 29) spectrum of 1.2 mM heterodimer model protein in 10 mM acetic acid-d 3 , pH 4.5, 37°C on a home-built spectrometer operating at a 1 H frequency of 750 MHz. Backbone assignment has been described previously, and 3 J HN␣ coupling constants for the tails have been reported previously 2 and were determined now for the coiled-coil region from the same data set as described previously.
Fibrinogen Binding Assays-Fibrinogen binding to platelets was performed as described previously (30). Briefly, 100 nM 125 I-labeled fibrinogen and the indicated concentration of peptide inhibitor were preincubated for 30 min at 22°C with a suspension of washed platelets (2 ϫ 10 8 /ml) in Tyrode's solution containing added 10 mM Hepes, 2 mg/ml bovine serum albumin, 2 mg/ml glucose, and 2 mM CaCl 2 . 20 M ADP and 20 M epinephrine were added to this suspension, and the resulting mixture was incubated for an additional 30 min. Bound 125 I-labeled fibrinogen was separated from free by centrifugation through a 20% sucrose cushion and was quantified by ␥ scintillation spectrometry.

Construction and Characterization of Recombinant Heterodimer Integrin Cytoplasmic Domain Model Proteins-To
test potential interactions between the ␣ IIb and ␤ 3 cytoplasmic tails, we first produced model proteins designed to mimic their normal arrangement in integrin ␣ IIb ␤ 3 . Recombinant ␣-helical coiled-coils were previously used to model dimerized integrin ␤ cytoplasmic domains (12); this strategy was employed to produce heterodimeric integrin cytoplasmic domains. Heptad repeats of identical length derived from GCN4 were used to form the nearly symmetrical, parallel coiled-coil. To favor heterodimer formation, Lys residues were placed in the in the g1 and e2 positions of the first and second heptad repeats to make a "Jun-like" helix (25). A complementary "Fos-like" helix was prepared by introducing Glu residues in these positions. These complementary helices produce preferentially heterodimerizing parallel coiled-coils (25). We also introduced a unique Cys residue N-terminal of each helix so that the formation of a cystine bridge would ensure parallel orientation and correct stagger of the coiled-coil. (Fig. 1A).
A dimer was formed in which the Jun-like helix was fused to the ␣ IIb cytoplasmic domain and a Fos-like helix was fused to the ␤ 3 cytoplasmic domain (Fig. 1B) A three-Gly spacer was inserted between the helices and the integrin ␣ or ␤ cytoplasmic domains to prevent the propagation of helical structure into the integrin tails (12). 2 The mass of the heterodimer differed by less than 0.1% from that predicted for the desired covalent structure (Fig. 1C), and SDS-PAGE confirmed that disulfide bonds covalently linked the heterodimers (Fig. 1D). Furthermore, NMR analysis confirmed formation of the symmetrical, parallel coiled-coil domain by the heptad repeats and that the helical structure of this domain did not propagate into the ␤ 3 tail as will be discussed in full detail elsewhere. 2 In particular, the coiled-coil behaved as a rigid, folded unit, whereas the motions of the ␤ 3 tail residues appear largely uncoupled from each other as judged from NMR signal intensities of the FG3␤ 3 subunit of JG3␣ IIb -FG3␤ 3 (Fig. 1E). Because of the symmetry of the coiled-coil all properties of the Fos-like heptad-repeats can be inferred to be present the Jun-like heptad repeats. The value of the 3 J HN␣ coupling constant reflects the backbone conformation (ranging from helical to extended conformations) at the residue in question (31). For the Fos-like heptad repeats, values below 5.5 Hz show distinctly helical (and folded) conformations. Thus, the recombinant JG3␣ IIb -FG3␤ 3 model protein had the expected properties.
Immunochemical Characterization of the JG3␣ IIb -FG3␤ 3 Protein-As noted above, there is substantial mutational and in vitro data to suggest that integrin ␣ IIb and ␤ 3 cytoplasmic domains interact with each other. An immunochemical approach was used to determine whether such interactions could occur within the intact receptor. Antibodies were raised against JG3␣ IIb -FG3␤ 3 protein (anti-␣ IIb ␤ 3 -Cyt). As expected, those antibodies reacted with native ␣ IIb ␤ 3 as judged by immunoprecipitation of the intact receptor. Furthermore, the capacity of these antibodies to immunoprecipitate the receptor depended on the presence of the cytoplasmic domains, because deletion of both cytoplasmic domains abolished reactivity ( Fig. 2A).
Having established that antibodies raised against JG3␣ IIb -FG3␤ 3 reacted with the cytoplasmic domain of the native receptor, we sought to analyze potential interactions between the tails. We reasoned (Fig. 2B) that if the ␣ and ␤ cytoplasmic domains interact with each other, then this interaction could create novel, combinatorial epitopes. To detect potential combinatorial epitopes, we assayed the effect of linear peptides comprising the ␣ IIb or ␤ 3 cytoplasmic domains on the binding of anti-␣ IIb ␤ 3 -Cyt to integrin ␣ IIb ␤ 3 purified from human platelets. A full-length ␣IIb cytoplasmic domain peptide produced negligible inhibition (Fig. 2C). The ␤ 3 cytoplasmic domain peptide competed to a maximum of 63 Ϯ 0.7%. Thus, antibodies reactive with intact integrin ␣ IIb ␤ 3 were resistant to competition by either of the linear peptides (Fig. 2C). These residual antibodies, however, could be competed by a mixture of the ␣ IIb and ␤ 3 peptides (Fig. 2C). Consequently, these antibodies recognize epitopes expressed in the native ␣ IIb ␤ 3 and formed by the interaction of the ␣ IIb and ␤ 3 cytoplasmic domains.
To exclude potential solubility artifacts, we examined the binding of an antibody against a linear ␤ 3 cytoplasmic domain peptide to the native ␣ IIb ␤ 3 integrin (Fig. 2D). In contrast to the result with anti-␣ IIb ␤ 3 -Cyt, the ␤ 3 peptide could completely inhibit antibody binding. Furthermore, an ␣ IIb peptide completely inhibited the binding of an anti-␣ IIb cytoplasmic domain to ␣ IIb ␤ 3 (data not shown). These combinatorial antibodies were produced by only two of the four New Zealand White rabbits immunized with JG3␣ IIb -FG3␤ 3 . They may represent an unusual specificity because immunization of several strains of inbred mice (e.g. Balb C, C57Bl 6) failed to elicit such antibodies (data not shown). Thus, these immunochemical results indicate that the ␣ IIb and ␤ 3 cytoplasmic domains can form a combinatorial epitope(s) in the intact integrin.
The Interaction of ␣ IIb and ␤ 3 Cytoplasmic Domains Alters the Antigenicity of the ␤ 3 Cytoplasmic Domain-In the presence of the ␣ IIb tail there is reduced targeting of ␤ 3 -containing integrins to cytoskeletal structures and a suppression of bidirectional signaling (15,32,33). Similar effects result from changing a Tyr in the first NPXY motif of ␤ 3 to an Ala (␤ 3 (Y747A)); Refs. 11, 34, and 35). These functional similarities prompted us to test the capacity of the ␤ 3 (Y747A) mutant to compete for the set of antibodies recognizing combinatorial epitopes in the cytoplasmic domain of ␣ IIb ␤ 3 . The ␤ 3 (Y747A) mutant competed nearly completely for the binding of anti-␣ IIb ␤ 3 -Cyt to integrin ␣ IIb ␤ 3 (Fig. 3A). Furthermore, addition of the ␣ IIb cytoplasmic domain peptide to the ␤ 3 (Y747A) produced little increase in competition. In contrast, as noted previously, a population of antibodies was resistant to inhibition by the wild type ␤ 3 cytoplasmic domain. Thus, a point mutation in the ␤ 3 tail that disrupts its signaling function and its capacity to interact with the cytoskeleton causes the ␤ 3 tail to exhibit combinatorial epitopes formed in the presence of the ␣ IIb tail.
The foregoing result suggested that the ␣ IIb cytoplasmic domain causes the ␤ 3 tail to resemble the ␤ 3 (Y747A) mutant. To test this idea, we raised polyclonal antibodies against the FG3␤ 3 (Y747A) model protein and tested the reactivity of those antibodies with native ␣ IIb ␤ 3 . The capacity of those antibodies to bind ␣ IIb ␤ 3 was completely inhibited by the immunogen, FG3␤ 3 (Y747A) (Fig. 3B). No significant increase in competition was observed in the presence of the ␣ IIb cytoplasmic domain peptide (data not shown). In sharp contrast, the wild type  (25). Cytoplasmic domains, such as that of integrin ␣ IIb depicted here, were joined to the coiled-coil. Three Gly residues were inserted between the coiled-coil and the integrin tail to disrupt induced helical structure in the cytoplasmic domain (12). At the bottom of the panel is a list of the integrin-specific sequences of all recombinant constructs used in this study. All integrin peptides correspond to published human integrin sequences in Swissprot. To preserve a HindIII site, a Val residue in the cytoplasmic domain of the human integrin ␣ IIb chain was replaced by Leu. The membrane-promixal conserved segment of the ␣ IIb and ␤ 3 cytoplasmic domains are underlined and in bold type. B, schematic of the JG3␣ IIb -FG3␤ 3 model protein. The heterodimeric model protein was formed from a JG3␣ IIb and FG3␤ 3 minisubunit. The hexahistidine tag was removed from the ␣ IIb subunit to facilitate its separation from the heterodimer during purification. C, ion spray mass spectrum of an ␣ IIb Responsible for Its Effect on the ␤ 3 Cytoplasmic Domain-The foregoing studies suggested that the ␣ IIb cytoplasmic domain could change the antigenicity of the ␤ 3 tail. We next sought to evaluate the specificity of the ␣ IIb cytoplasmic domain effect. The addition of a synthetic peptide containing the ␣ IIb cytoplasmic domain sequence to 10 nM FG3␤ 3 protein resulted in dosedependent formation of combinatorial epitopes (EC 50 ϭ 15 nM) (Fig. 4A). In sharp contrast, the cytoplasmic domains of integrin ␣ 4 or ␣ 5 had no such effect, even though they share a highly conserved N-terminal seven residues with that of the ␣ IIb cytoplasmic domain (6). We next analyzed a series of nested deletion mutants from the N and C termini of ␣ IIb in this assay. To increase sensitivity, these experiments were conducted in the presence of 50 nM FG3␤ 3 . Deletion of the Cterminal 4 amino acids had no effect on this interaction; however, deletion of 2 additional residues of ␣ IIb completely abolished activity (Fig. 4B). A series of N-terminal deletions were performed. The first 7 amino acids were dispensable for this activity as expected from the results with the ␣ 4 and ␣ 5 cytoplasmic domains. However, deletion of 3 additional amino acids abolished activity (Fig. 4B). Consequently, progressively smaller peptides were produced and a minimal sequence spanning Arg 997 -Asp 1003 had residual activity. Furthermore, double Ala substitutions in this sequence at Pro 998 and Pro 999 blocked activity. To further confirm our identification of this as the critical site, we synthesized a loop out peptide deleting both Arg 997 and Pro 998 (␣ IIb (⌬Lys 994 -Pro 998 )) and that loop out peptide was devoid of activity (Fig. 4B). Thus, the capacity of the ␣ IIb cytoplasmic domain to alter the antigenicity of the ␤ 3 tail exhibits structural specificity and can be mapped to a minimal heptapeptide sequence.
Cell Permeable Peptides Containing the ␤ 3 -interactive Site of the ␣ IIb Tail Block Activation of ␣ IIb ␤ 3 -The foregoing experiments identified a short peptide sequence of the ␣ IIb tail that native integrin ␣ IIb ␤ 3 . Surface biotinylated Chinese hamster ovary cells expressing recombinant integrin ␣ IIb ␤ 3 (␣IIb␤3) or ␣ IIb ␤ 3 lacking its cytoplasmic domains (␣IIb(⌬996)␤3(⌬717)) were lysed in immunoprecipitation buffer. The lysates were immunoprecipitated with anti-model protein antibody (␣IIb␤3-Cyt), an antibody (rabbit 8275) directed against the cytoplasmic domain of ␤3 (␤3-Cyt), or D57, an anti-␣ IIb ␤ 3 monoclonal antibody (␣IIb␤3). Lysates were also precipitated with normal rabbit serum (NRS). The immunoprecipitates were fractionated by SDS-PAGE, transferred to nitrocellulose filters, and developed with avidin peroxidase. B, rationale for immunochemical analysis. On the left are depicted linear integrin cytoplasmic tails, and on the right are depicted the tails in their assumed relationship in the intact receptor. The classes of epitopes expected are those present in the linear peptide and the intact receptor (1). If the cytoplasmic tails of the receptor interact, then additional classes of potential epitopes are those manifest in the linear peptide but lost in the native receptor (2) and combinatorial epitopes formed in the native receptor but absent from the linear peptides (3). Note that epitope classes 1 and 2 could be present on either the ␣ or ␤ subunit cytoplasmic domain peptides. C, immunochemical analysis of the anti-model protein antibody. A competitive ELISA was used to measure the binding of the polyclonal anti-model protein antibody (anti-␣ IIb ␤ 3 -Cyt) to purified platelet ␣ IIb ␤ 3 . The results are expressed as B/B 0 , where B ϭ A 490 in the presence of competitor and B 0 ϭ A 490 in the absence of competitor. The data depict competition with full-length ␣ IIb cytoplasmic domain peptide (␣IIb), ␤ 3 (Ile 719 -Thr 762 ) cytoplasmic domain peptide (␤3), or an equimolar mixture of the ␣ IIb and ␤ 3 peptides (␣IIbϩ␤3). Depicted are the means of triplicate determinations. The curves represent best fits of the data to B/B 0 ϭ 1 Ϫ (E*C/(K ϩ C)), where E is the fraction of antibodies that can bind the competitor, C is the concentration of the competitor, and K is the average apparent dissociation constant for the binding of the antibodies to the competitor. For the ␣ IIb peptide E ϭ 0.15 Ϯ 0.01, for ␤ 3 E ϭ 0.63 Ϯ 0.007, and for ␣ IIb ϩ ␤ 3 E ϭ 1.00 Ϯ 0.01. D, immunochemical analysis of the anti-␤ 3 cytoplasmic domain antibody. ELISA was used to measure binding of an antibody directed against a linear ␤ 3 cytoplasmic domain peptide (anti-␤ 3 -Cyt, rabbit 8275) to purified platelet ␣ IIb ␤ 3 . The data depict competition with full-length ␣ IIb cytoplasmic domain peptide (␣IIb) or ␤ 3 (Thr 720 -Thr 762 ) cytoplasmic domain peptide (␤3) .   FIG. 2. Immunochemical analysis of the anti-␣ IIb ␤ 3 model protein antibody. A, the antibody reacts with the cytoplasmic domains of alters that of ␤ 3 , mimicking the immunochemical effects of a Tyr to Ala substitution in the first NPXY motif. The ␤ 3 (Y747A) mutation blocks the activation of integrin ␣ IIb ␤ 3 in vivo (11). Consequently, we tested the effects of addition of a peptide containing the ␣ IIb sequence on activation of integrin ␣ IIb ␤ 3 in platelets. The peptide was palmitoylated, a modification that promotes entry of peptides into the platelet cytoplasm (36). Activation of ␣ IIb ␤ 3 was suppressed by a palmitoylated decapeptide (KRNRPPLEED) that contained the minimal heptapeptide sequence (Fig. 5A). Nearly 100% inhibition was observed at a peptide concentration of 100 M. Inhibition was specific, because a scrambled peptide of the same composition was nearly devoid of activity. Furthermore, a nonpalmitoylated peptide containing the same sequence caused little inhibition (Fig. 5B). Thus, a palmitoylated ␣ IIb peptide that contains the ␤ 3 -interactive site inhibits activation of integrin ␣ IIb ␤ 3 in platelets. DISCUSSION Previous evidence suggests that interactions between integrin cytoplasmic domains regulate integrin activation. In the present work, we have constructed and validated recombinant structural mimics of the heterodimeric ␣ IIb ␤ 3 cytoplasmic domains. These mimics elicited polyclonal antibodies that recognize a combinatorial epitope(s) in the presence of a mixture of the ␣ IIb and ␤ 3 cytoplasmic domains but not present in either isolated tail. The presence of this epitope(s) within intact ␣ IIb ␤ 3 indicates that interaction between the tails can occur in the native integrin. Furthermore, the combinatorial epitope(s) is present in the activation-defective ␤ 3 (Y747A) mutant. Thus, the ␣ IIb tail causes the ␤ 3 tail to resemble that of ␤ 3 (Y747A), suggesting that the interaction of the ␣ IIb ␤ 3 tails opposes activation of the integrin. Furthermore, mapping of the site in the   FIG. 3. The ␣ IIb cytoplasmic domain alters the antigenicity of the ␤ 3 cytoplasmic domain. A, ␤ 3 (Y747A) binds combinatorial antibodies in anti-␣ IIb ␤ 3 -Cyt. A competitive ELISA was used to measure the binding of the polyclonal anti-model protein antibody (anti-␣ IIb ␤ 3 -Cyt) to purified platelet ␣ IIb ␤ 3 , and the data were analyzed as described in the legend to Fig. 2. The graphs depict competition with the FG3␤ 3 minisubunit (␤3) or FG3␤ 3 (Y747A). Also depicted is competition with an equimolar mixture of the FG3␤ 3 (Y747A) and full-length ␣ IIb cytoplasmic domain peptide (␤3(Y747A)ϩ␣IIb). Depicted are the means of triplicate determinations. B, anti-␤ 3 (Y747A) recognizes combinatorial determinants formed by the interaction of ␣ IIb and ␤ 3 tails. A competitive ELISA was used to measure the binding of the polyclonal antibody against ␤ 3 (Y747A) to purified platelet ␣ IIb ␤ 3 . The graphs depict competition with FG3␤ 3 cytoplasmic domain (␤3) or FG3␤ 3 (Y747A). Also depicted is competition with an equimolar mixture of the FG3␤ 3 protein and full-length ␣IIb cytoplasmic domain synthetic peptide (␣IIbϩ␤3). Depicted are the means of triplicate determinations. ␣ IIb tail responsible for this activity identified a membranedistal region previously implicated in regulation of integrin activation. Finally, palmitoylated peptides containing the ␤ 3interactive ␣IIb sequence specifically blocked ␣ IIb ␤ 3 activation in platelets. Thus, these studies provide direct evidence for an interaction of the ␣ IIb and ␤ 3 cytoplasmic domains that regulates activation of the integrin in platelets.
Antibodies raised against the ␣ IIb ␤ 3 model protein contained a population of antibodies that recognize a combinatorial epitope formed in the presence of the ␣ IIb and ␤ 3 tails. This polyclonal antibody reacted with the cytoplasmic domain of native ␣ IIb ␤ 3 (Fig. 2). However, its binding to ␣ IIb ␤ 3 was only partially competed by either the isolated ␣ IIb tail or the ␤ 3 tail. The remaining population of antibodies was inhibited only in the presence of both cytoplasmic tails. These experiments were conducted at 25-75% saturating antibody concentrations, and similar results were observed at each concentration. They were not due to insolubility of the tails, because the isolated ␤ 3 cytoplasmic domain or ␣ IIb cytoplasmic domains could effi-ciently compete for the binding of antibodies specifically directed only against the isolated cytoplasmic domains. Furthermore, antibodies raised against the FG3␤ 3 (Y747A) protein, which had no reactivity with the ␣ IIb tail, also contained antibodies against these combinatorial determinants. Thus, immunization with the JG3␣ IIb -FG3␤ 3 model protein can elicit the formation of antibodies against combinatorial epitopes in the ␣ IIb ␤ 3 cytoplasmic domain.
The presence of combinatorial epitopes in intact ␣ IIb ␤ 3 suggests that the cytoplasmic domains can interact in the native receptor. The combinatorial antibodies react with native ␣ IIb ␤ 3 in a manner dependent on its cytoplasmic domains. However, the immunochemical assay requires the presence of the antibody to detect interactions between the ␣ IIb and ␤ 3 tails. Indeed, in detailed NMR analysis, we found that the unpaired ␤ 3 tail exhibits essentially the same structural and dynamic properties as the ␣ IIb -paired ␤ 3 tail within the context of the coiledcoil (JG3␣ IIb -FG3␤ 3 ). 2 Furthermore, constructs with different lengths of the glycine linker resulting in relative vertical tailtail shifts of Ϫ2 to ϩ2 residues did not uncover any such structural differences. Thus, it is possible that the antibody is a necessary co-factor that enhanced the interaction. Within cells, such co-factors, like the antibody, could modify the strength of the interaction. One such co-factor might be a plasma membrane. Indeed, Vinogradova et al. (37) reported that when the ␣ IIb tail was myristoylated and inserted into a phospholipid bilayer, it formed a stable structure. This is in sharp contrast to the unstructured nature of the ␣ IIb peptide in aqueous solution (22,24). 2 Furthermore, Leisner et al. (38) produced an anti-␣ IIb tail antibody (anti-LIBScyt1) that manifested reduced binding to intact integrin ␣ IIb ␤ 3 . One interpretation of this result is that the availability of the epitope for this monoclonal antibody was influenced by interaction with the ␤ 3 tail. The ␣ IIb sequence implicated here in forming the combinatorial epitope (RPPLEED) is contained in the region recognized by the anti-LIBScyt1 antibody (38). Taken together, these results suggest that interactions between the integrin ␣ and ␤ cytoplasmic domains occur within cells.
The interaction of ␣ IIb and ␤ 3 tails alter the antigenicity of the ␤ 3 tail. In particular, we found that introducing a mutation that blocks integrin activation into the ␤ 3 tail resulted in expression of the combinatorial epitope(s). The ␤ 3 tail is largely unfolded in aqueous solution; however, the ␤ 3 Tyr 747 is part of an NPXY motif that has a propensity to form ␤ turns. This propensity is abolished by the Y747A mutation. 2 The Y747A mutation also abolishes talin binding. The region of talin that binds to integrin ␤ tails contains a FERM domain (14,39), a domain that contains a lobe structurally similar to PTB domains (40). NPXY motifs, such as that present at ␤ 3 Asn 744 -Tyr 747 , are often favored binding sites of PTB domains (41). Thus, binding of talin (or other FERM or PTB domain-containing proteins) may stabilize a reverse turn at the NPXY motif in integrin ␤ tails leading to integrin activation. The presence of the ␣ IIb tail may oppose formation of this stable turn, thus antagonizing integrin activation. The ␤ turn propensity at Asn 744 -Tyr 747 is preserved in the JG3␣ IIb -FG3␤ 3 model protein. 2 However, the ␤ turn is only present in a small population of ␤ 3 molecules at any time. Consequently, the ␣ IIb tail might oppose stabilization of the turn in the presence of talin but not appreciably reduce the propensity to form a turn in the absence of talin. Thus, the ␣ IIb cytoplasmic domain alters the antigenicity of the ␤ 3 tail, causing it to resemble the structurally and functionally altered ␤ 3 (Y747A) mutant.
When the ␤ 3 -interactive region of the ␣ IIb tail was palmitoylated, it suppressed activation of ␣ IIb ␤ 3 in platelets. Platelets are not readily susceptible to genetic manipulation; however, previous work from Fitzgerald's laboratory (36) has established that palmitoylated peptides enter platelets and can regulate activation of ␣ IIb ␤ 3 . Utilizing their approach, we found that the identified ␤ 3 interactive region of ␣ IIb , when palmitoylated, suppressed ␣ IIb ␤ 3 activation in platelets. Specificity was established by the markedly reduced activity in either a nonpalmitoylated peptide, presumably because the latter fails to enter platelets. In addition, a palmitoylated-scrambled peptide also exhibited greatly reduced activity. Because both the authentic and scrambled peptides bore the same lipid modification and were of identical composition, similar distributions within the platelet seem likely. However, caution is warranted, because definitive proof of a completely identical subcellular localization for these two peptides is technically challenging. Nevertheless, a potential explanation of these results is that the palmitoylated ␣ IIb peptide interacted with the ␤ 3 tail within the cell interior, rendering it less capable of supporting platelet activation. Such a mechanism would also account for the observation that a full-length myristoylated ␣ IIb peptide inhibited ␣ IIb ␤ 3 activation in platelets (37). In the same work, an ␣ IIb tail peptide inhibited ␣ IIb ␤ 3 activation in vitro; however, it failed to inhibit the activation of a receptor lacking its ␤ 3 cytoplasmic domain. Additional evidence for the view that the ␣ IIb tail suppresses activation by interacting with the ␤ 3 tail comes from the present finding that the ␣ IIb (P998A,P999A) mutation blocked formation of the combinatorial epitope. This same mutation abolishes inhibition of activation by the myristoylated ␣ IIb peptide (37) and activates intact recombinant ␣ IIb ␤ 3 (38). Thus, taken together, these data strongly suggest that the interaction of this region of ␣ IIb cytoplasmic domain with the ␤ 3 tail opposes activation of ␣ IIb ␤ 3 in platelets.
The results presented here and in previous literature suggest a unifying hypothesis for the activation of integrins. The receptor is proposed to be restrained in a resting state through an interaction involving the membrane-proximal region of the ␣ and ␤ cytoplasmic domains (6,10,23). The removal of either of these regions results in a constitutive activated receptor, and that activation is independent of other interactions of the ␤ cytoplasmic domains or cellular energy (7,15,42). Furthermore, introduction of the lipidated membrane-proximal sequence of the ␣ IIb tail (KVGFFKR) into platelets results in ␣ IIb ␤ 3 activation, presumably by disrupting this membraneproximal interaction (36). Thus, the membrane-proximal interaction forms a structural on-off switch for the receptor. However, normal activation mechanisms require the distal region of the ␤ 3 cytoplasmic domain (7,9,17,27). An interaction of talin with the ␤ 3 cytoplasmic domain can activate the receptor, and this depends on membrane distal sequences (14). The present results indicate interactions between the membrane distal ␣ IIb sequence and the ␤ 3 tail within the native receptor that block integrin activation. Disruption of this membrane distal interaction could account for energy and ␤ tail-dependent activation caused by alterations in membrane-distal ␣ cytoplasmic domain sequences (7,42). Furthermore, the interaction with the ␣ IIb tail may oppose stabilization of the reverse turn at ␤ 3 (Asn 744 -Tyr 747 ) by cytoplasmic proteins such as talin. Thus, detailed analyses of the structural effects of interactions of proteins, such as talin, with integrin ␤ cytoplasmic domains are likely be informative in testing this model of integrin activation. In addition, the identification of physiologic mechanisms that regulate the interactions of integrin tails with each other and with proteins such as talin will be central to the understanding of this important biological process.