Tyrosine Phosphorylation of Integrin β3 Regulates Kindlin-2 Binding and Integrin Activation*

Kindlins are essential for integrin activation in cell systems and do so by working in a cooperative fashion with talin via their direct interaction with integrin β cytoplasmic tails (CTs). Kindlins interact with the membrane-distal NxxY motif, which is distinct from the talin-binding site within the membrane-proximal NxxY motif. The Tyr residues in both motifs can be phosphorylated, and it has been suggested that this modification of the membrane-proximal NxxY motif negatively regulates interaction with the talin head domain. However, the influence of Tyr phosphorylation of the membrane-distal NxxY motif on kindlin binding is unknown. Using mutational analyses and phosphorylated peptides, we show that phosphorylation of the membrane-distal NITY759 motif in the β3 CT disrupts kindlin-2 recognition. Phosphorylation of this membrane-distal Tyr also disables the ability of kindlin-2 to coactivate the integrin. In direct binding studies, peptides corresponding to the non-phosphorylated β3 CT interacted well with kindlin-2, whereas the Tyr759-phosphorylated peptide failed to bind kindlin-2 with measurable affinity. These observations indicate that transitions between the phosphorylated and non-phosphorylated states of the integrin β3 CT determine reactivity with kindlin-2 and govern the role of kindlin-2 in regulating integrin activation.

Integrin receptors play essential roles in regulating cell adhesion, migration, survival, and differentiation by serving as bidirectional conduits for exchange of information between the intracellular and extracellular compartments (1). The initiation and propagation of the bidirectional signaling depend upon binding of intracellular constituents to the cytoplasmic tails (CTs) 2 of integrin ␤ subunits (2,3). The integrin ␤ CTs are relatively short (except ␤ 4 ), and most share several conserved regions, including two Nxx(Y/F) motifs. Among numerous ␤ CT-binding partners, the talin head domain (talin-H) has been demonstrated to be an essential regulator of integrin activation (4 -8). The activating function of talin-H depends on its direct association with the membrane-proximal NPxY motif and residues near the plasma membrane of the integrin ␤ CTs (4, 9 -12). The latter interaction leads to an unclasping of the intracellular ␣/␤ membrane-proximal complex and triggers unclasping of the transmembrane ␣/␤ association (13), which, in turn, generates conformational cues to the extracellular domain, rendering it competent to engage soluble ligands with high affinity. Recently, a solid body of evidence, ranging from analyses of model cell systems to in vivo analyses in mice and humans, has clearly demonstrated that talin alone is not sufficient to induce functionally significant integrin activation in intact cells; kindlins are required to support integrin activation and function (14 -19). In contrast to the dependence of talin binding to the membrane-proximal NxxY motif, kindlin binding to the ␤ CTs depends on the membrane-distal NxxY motif (14 -16, 20).
The two NxxY motifs in the ␤ 3 CT are NPLY 747 and NITY 759 , and they have been implicated in recognition of talin-H and kindlin, respectively (14). The tyrosine residues in these two motifs can be phosphorylated in cells, and such modifications influence integrin signaling (21). Even though tyrosine phosphorylation of the ␤ 3 CT has been clearly associated with integrin outside-in signaling (22)(23)(24)(25), its functional regulation in inside-out signaling is less pronounced. Recent studies have demonstrated that Tyr 747 phosphorylation in the ␤ 3 CT inhibited talin-H binding by both direct (decreasing the affinity of talin binding) and indirect (increasing the binding affinity of Dok1, a competitor of talin binding) effects (26,27). These observations are compatible with the prior observation that tyrosine phosphorylation of the ␤ 3 CT by v-Src kinase abolished integrin ␤ 3 -mediated cell adhesion (28). Here, we show that kindlin-2 binding to the ␤ 3 CT is abrogated by Tyr 759 phosphorylation, and our functional analyses further establish that this phosphorylation event may be an important regulator of integrin activation.

EXPERIMENTAL PROCEDURES
Plasmid Construction, Protein Preparation, and Peptide Synthesis-For mammalian expression, the cDNAs of human ␣ IIb and ␤ 3 subunits were inserted into pcDNA3.1. All regulators of integrin activation were cloned into either pEGFP or pDsRedmonomer vectors. GST-fused integrin ␤ CTs and GST-kindlin-2 were expressed in Escherichia coli Rosetta TM 2(IDE3) cells (Novagen) and purified by glutathione chromatography. The purified proteins were quantified using Bio-Rad protein assays. All synthetic peptides were prepared, purified, and authenticated by tandem mass spectrometry in the Molecular Biotechnology Core of the Cleveland Clinic.
Pulldown Assays and Western Blotting-Pulldown assays were performed using GST fusion proteins. Equal amounts of GST-fused integrin ␤ CT were added together with glutathione-Sepharose 4B (GE Healthcare) to aliquots of the cell lysates. In peptide inhibition experiments, the indicated pep-tide was also added to the slurries at the selected concentrations. After overnight incubation at 4°C, the precipitates were washed and boiled in Laemmli sample buffer. The eluates were analyzed on gradient acrylamide gels under reducing conditions, and interactions of the integrin ␤ CT with kindlin-2 were determined by Western blotting. In parallel, the gels were also stained with Coomassie Blue to verify that sample loadings were similar.
Surface Plasmon Resonance (SPR)-Real-time protein-protein interactions were analyzed using a Biacore 3000 (Biacore, Uppsala, Sweden). N-terminally biotinylated peptides were bound to SA5 sensor chips (Biacore) according to the manufacturer's instructions. Experiments were performed at room temperature in 10 mM HEPES buffer (pH 7.4) at a flow rate of 25 l/min. Analyte binding to the immobilized ligand was recorded by measuring the variation of the SPR angle, and the results are expressed in resonance units (RU).
Integrin ␣ IIb ␤ 3 Activation Assays-Integrin ␣ IIb ␤ 3 activation was evaluated by a PAC1 binding assay as described previously (14,29). Mutant forms of integrin ␣ IIb ␤ 3 together with the regulators, tagged with either DsRed monomer or EGFP, were expressed in CHO-K1 cells by transient transfection using Lipofectamine 2000 (Invitrogen). PAC1 binding to the different transfectants (EGFP and DsRed double-positive cells) was analyzed by flow cytometry after incubating the transfected cells with anti-PAC1 mAb for 25 min at room temperature, followed by Alexa Fluor 633-conjugated secondary antibody for 25 min on ice. Variations in integrin expression levels on the transfected cells were normalized based on reactivity with a mAb (2G12) reactive with ␣ IIb ␤ 3 independent of its activation status. Integrin activation was expressed in terms of relative median fluorescence intensities by defining the basal PAC1 binding to WT ␣ IIb ␤ 3 cells positive for EGFP and DsRed as 1.0.

RESULTS AND DISCUSSION
The NPLY 747 and NITY 759 motifs (Fig. 1A) in the integrin ␤ 3 CT are directly involved in recognition of talin-H and kindlin, respectively (14). This specificity of kindlin-2 binding was verified in assays in which the WT ␤ 3 CT and various point mutant forms expressed as GST fusion proteins were added to lysates of endothelial cells, and their ability to pull down endogenous kindlin-2 was detected by Western blotting. In these assays, interaction of endogenous kindlin-2 was observed with the WT ␤ 3 CT, and this binding was abolished by substitution of Ala for Tyr in the NITY 759 motif but not in the NPLY 747 motif (Fig. 1B). These data document selective involvement of the membranedistal NITY 759 motif in kindlin-2 recognition. To begin to probe the possible role of tyrosine phosphorylation of the NITY 759 motif in kindlin-2 binding, we employed a phosphomimetic ␤ 3 CT mutant by substituting the tyrosine residue with a negatively charged aspartic acid. This Tyr-to-Asp substitution in the NITY 759 motif but not in the NPLY 747 motif abrogated kindlin-2 binding (Fig. 1B), confirming the selective involvement of the membrane-distal NITY 759 motif in the ␤ 3 CT in kindlin-2 binding and also suggesting that modification of Tyr 759 might be a negative regulator for kindlin-2 interaction.
To determine how modification of the NITY 759 motif regulates kindlin function, we performed experiments in widely used CHO cells with stable expression of integrin ␣ IIb ␤ 3 , in which the coactivator activities of talin and kindlin-2 can be readily demonstrated (14). Several mutations were introduced at Tyr 759 in the ␤ 3 subunit to determine the effects of the coactivator activity of talin and kindlin-2. As shown in Fig. 1C, WT ␤ 3 exhibited extensive activation when kindlin-2 and talin were coexpressed using the activation-specific PAC1 mAb in FACS analyses. Substitution with Asp, which mimics the charge induced by phosphorylation, blocked coactivation without affecting the level of expression of ␣ IIb ␤ 3 (Fig. 1C). As shown in Fig. 1D, substitution of Ala for Tyr 759 in the receptor also blocked coactivation. However, substitution of Phe for Tyr 759 did not affect kindlin-2-mediated coactivation. The lack of effect of the Tyr-to-Phe substitution is consistent with the ability of kindlins to control activation of the ␤ 2 integrins (17,18,30), which have NxxF motifs rather than NxxY motifs.
The strategy of using a charged residue, either Asp or Glu, to mimic a phosphorylation state has been used extensively. However, such substitutions do not entirely recapitulate the structure of a phosphotyrosine. Therefore, as a more direct approach, small ␤ 3 CT peptides containing the membrane-distal NITY 759 motif with or without phospho-Tyr 759 ( Fig. 2A) were tested as inhibitors of kindlin-2 binding to the ␤ 3 CT (14). Pulldown assays similar to those described for Fig. 1B were performed using GSTfused ␤ 3 CT to precipitate endogenous kindlin-2 from human umbilical vein endothelial cell lysates, and the bound kindlin-2 was measured by Western blotting. As shown in Fig. 2B, when added to cell lysates, the WT ␤ 3 peptide significantly inhibited kindlin-2 binding. By densitometry, the inhibition was 72%. In contrast, when the Tyr residue was phosphorylated, the peptide produced no apparent inhibition. These peptide data strongly support a negative regulatory role of Tyr 759 phosphorylation in kindlin-2 binding to the ␤ 3 CT.
To exclude the possibility that such an effect could be confounded by indirect interactions that might occur in cell lysates, we further analyzed the interactions using purified kindlin-2 protein in SPR experiments. The ␤ 3 CT was immobilized on the biosensor chips, and kindlin-2 was used as the soluble analyte in the presence or absence of the two peptides. As shown in Fig.  2C, kindlin-2 bound to the immobilized ␤ 3 CT, yielding a typical progress curve with a rapid association and a prolonged dissociation phase. Inclusion of the WT peptide prevented the interaction. However, the phosphorylated ␤ 3 peptide did not show a significant inhibitory effect under the same conditions, suggesting a direct involvement of Tyr 759 phosphorylation in negatively regulating kindlin-2 binding to the ␤ 3 CT.
As an independent approach to verify the negative regulation of Tyr 759 phosphorylation in kindlin-2 binding, we directly immobilized the ␤ 3 CT C-terminal peptide conjugated with biotin at its N terminus onto streptavidin SA5 biosensor chips for SPR analysis (Fig. 3A). As shown in Fig. 3B, kindlin-2 bound to immobilized WT peptide, and a K d of 1.8 ϫ 10 Ϫ7 M was calculated. This value was similar to that obtained with immobilized full-length ␤ 3 CT (K d ϭ 1.36 ϫ 10 Ϫ7 M), demonstrating that the ␤ 3 CT C terminus is primarily responsible for kindlin-2 recognition. As a control, talin-H did not bind to the WT ␤ 3 C-terminal peptide-coated chips under the same conditions (data not shown), indicating binding specificity for kindlin-2. When the biotinylated phosphopeptide was coated onto the chips, negligible binding of kindlin-2 was detected (Fig. 3C). The failure of the phosphopeptide to bind kindlin-2 was not due to poor coating of the biosensor chip because the WT pep-FIGURE 2. Phosphorylation of the NxxY 759 motif in the ␤ 3 CT peptides disables its ability to block interaction of kindlin-2 and the ␤ 3 CT. A, the amino acid sequences of the ␤ 3 CT C-terminal peptide containing the NITY 759 motif and a modified peptide with Tyr 759 phosphorylation (␤3C-pep-phos) are shown. B, kindlin-2 and ␤ 3 CT interaction was evaluated by pulldown assays in the presence of the indicated peptides. IB, immunoblot. C, kindlin-2 or a mixture of kindlin-2 and the indicated peptide at a 1:100 molar ratio was passed over an SA5 biosensor chip coated with biotinylated ␤ 3 CT peptides, and the binding curves were recorded over time. To block nonspecific binding of the peptide to the chips, 0.1% BSA was included in the running buffer.
tide and phosphopeptide yielded similar RU values (ϳ200 RU). Taken together, these results demonstrate that phosphorylation of the NITY 759 motif in the ␤ 3 CT significantly dampens its ability to bind kindlin-2.
To further demonstrate the specific involvement of Tyr 759 in kindlin-2 binding, we introduced several structure-based mutations in the ␤ 3 CT C-terminal peptides (Fig. 3A). These mutant peptides were immobilized on the streptavidin SA5 biosensor chips individually, and their capacity to bind kindlin-2 was evaluated by SPR. As shown in Fig. 3D, a conservative mutation in ␤ 3 CT peptides, Y759F, still supported kindlin-2 binding, and the calculated binding affinity (K d ) was 1.93 ϫ 10 Ϫ7 M, similar to that obtained with the WT peptide (K d ϭ 1.8 ϫ 10 Ϫ7 M). However, all other peptides with nonconservative mutations, including Y759S, Y759K, Y759L, and Y759D, showed no capacity to bind kindlin-2 (Fig. 3D). These results demonstrate that Tyr 759 is a key residue in supporting kindlin-2 binding to the ␤ 3 CT with very precise structural requirements, suggesting that alteration of the local structural environment by Tyr 759 phosphorylation would be unfavorable for kindlin-2 recognition.
In summary, our study demonstrates that tyrosine phosphorylation of the NITY 759 motif in the ␤ 3 CT provides a regulatory mechanism for controlling the role of kindlin-2 in integrin function. Based on the conservation of the membrane-distal NxxY motif in the integrin ␤ subunits, the regulatory role of tyrosine phosphorylation might apply to other integrin family members. In fact, we did observe that a ␤ 1 CT peptide homologous to the ␤ 3 CT peptide used in Fig. 2 did inhibit ␤ 3 CT and kindlin-2 interaction (Fig. 4), and as expected, direct phosphorylation of Tyr 795 or a phosphomimetic mutation (Y795D) but not a structure conservative Y795F mutation in ␤ 1 CT membrane-distal NPKY 795 motif ablated the inhibitory activities of these ␤ 1 CT peptides (Fig. 4). In platelets and endothelial cells, it is well established that the tyrosine residues of NxxY motifs in the ␤ 3 CT are subject to phosphorylation, and when modified, integrin ␤ 3 functions are altered (22,23,31,32). Because talin-H binding to the ␤ subunits can also be negatively regulated by tyrosine phosphorylation of the membrane-proximal NxxY motif (27), the two NxxY motifs in the ␤ CTs might act as a dual switch in controlling interactions with cytoplasmic regulators via phosphorylation and dephosphorylation. It would appear that phosphorylation of Tyr 759 in the ␤ 3 CT provides a mechanism for stripping kindlin-2 from the integrin. Conversely, dephosphorylation might allow for rebinding and reactivation of the integrin. Cycles of phosphorylation/dephosphorylation of the integrin ␤ CT may be part of a complex network that encompasses many regulated events, including phosphorylation of integrin regulators them-  selves, talin, and kindlin (33)(34)(35), and the susceptibility of the ␤ 3 CT to calpain proteolysis (36), which would delete the kindlinbinding site altogether. Thus, post-translation modifications may be part of a complex network of events that occur spatially and temporally within the cell to control adhesive and migratory responses. It would appear that, once activated, talin and kindlin-2 binding to the ␤ 3 CT may not be required to sustain activation. Although we anticipate that phosphorylation of Tyr 759 would also influence kindlin-3 binding to the ␤ 3 CT, its very low expression in the same bacterial system used for kindlin-2 expression has precluded direct experimental testing of this prediction and remains an open question for future studies.