Integrin β3 Phosphorylation Dictates Its Complex with the Shc Phosphotyrosine-binding (PTB) Domain*

Adaptor protein Shc plays a key role in mitogen-activated protein kinase (MAPK) signaling pathway, which can be mediated through a number of different receptors including integrins. By specifically recognizing the tyrosine-phosphorylated integrin β3, Shc has been shown to trigger integrin outside-in signaling, although the structural basis of this interaction remains nebulous. Here we present the detailed structural analysis of Shc phosphotyrosine-binding (PTB) domain in complex with the bi-phosphorylated β3integrin cytoplasmic tail (CT). We show that this complex is primarily defined by the phosphorylation state of the integrin C-terminal Tyr759, which fits neatly into the classical PTB pocket of Shc. In addition, we have identified a novel binding interface which concurrently accommodates phosphorylated Tyr747 of the highly conserved NPXY motif of β3. The structure represents the first snapshot of an integrin cytoplasmic tail bound to a target for mediating the outside-in signaling. Detailed comparison with the known Shc PTB structure bound to a target TrkA peptide revealed some significant differences, which shed new light upon the PTB domain specificity.

Integrins, a major class of non-covalent heterodimeric, glycoprotein cell surface receptors, are among the most studied and best characterized cell adhesion molecules. Integrins mediate a plethora of cell-cell, cell-extracellular matrix (ECM), 2 and cell-pathogen interactions and hence are responsible for controlling a wide array of biological processes including homeostasis, cell migration, differentiation, adhesion, immune response etc. The unique bidirectional flow of information through integrins involves inside-out signals, which allow them to interact with extracellular ligands (such as fibrinogen, von Willebrand factor, fibronectin) and ligand-dependent outside-in signals which adjust the cellular response to cell-cell adhesion (1). Although our understanding of the molecular details of inside-out integrin signaling (2,3) has grown by leaps and bounds over the past decade, the early intracellular events following the integrin-mediated ECM engagement, outside-in signal transduction, still require further clarification. With respect to the outside-in signaling, the important unanswered questions center on selective recognition of proximal effectors by integrin cytoplasmic domains at different stages of cell spreading. Phosphorylation of the integrin tails is considered to be one of the spatiotemporal mechanisms for imparting such selectivity and, indeed, phosphorylation switches are thought to be a common principle of integrin regulation. Platelet integrin ␤ 3 cytoplasmic tail (CT) is laden with various phosphorylation sites, including two tyrosines, one serine, and multiple threonines. However, only tyrosine phosphorylation is found to be specific for the outside-in signaling (4 -8) and Shc (in particular its p52 isoform) was identified as a primary signaling partner for the tyrosine-phosphorylated ␤ 3 CT (8).
Adaptor protein Shc (Src homology 2 domain) plays a key role in mitogen-activated protein kinase (MAPK) signaling pathway (9) and can be recruited through many different types of receptors, including integrins, growth factor, antigen, cytokine, G-protein-coupled, and hormone receptors (10). In the context of the present study, it is important to mention that Shc has also been coupled to the integrin controlled cell cycle progression (11). One of the three isoforms, the p52 Shc contains three distinct domains: phosphotyrosine-binding (PTB) domain, a poorly characterized glycine/proline-rich region termed as collagen homology domain (CH1), and the SH2 domain. Previous studies have shown that two of these domains, PTB and SH2, could potentially interact with ␤ 3 CT containing phosphorylated tyrosines (12,13). However, based on in vitro peptide affinity chromatography assays, Higashi et al. (14) proved that p52 Shc binds to the tyrosine-phosphorylated ␤ 3 peptide through its PTB domain.
Overall, PTB domains comprise a large family of protein binding modules, which exhibit a conserved structural architecture similar to the pleckstrin homology (PH) domains (also termed as PH domain superfold) consisting of a core ␤-sandwich made of two anti-parallel ␤ sheets flanked by a C-terminal helix. In terms of the PTB domain ligand specificity, although phosphorylated tyrosine is required for high affinity binding in case of proteins such as Shc PTB, IRS-1/IRS-2/IRS-3, Dok1, and SNT/FRS2, the PTB domains of Dab1/Dab2, ARH, Fe65, ICAP1␣, JIP-1/JIP-1b, Numb, Talin, and X11␣ exhibit similar or in some cases even higher affinity for non-phosphorylated peptides (15). For Shc PTB-integrin interaction, a bi-phosphorylated (pY 747 and pY 759 ) peptide has been shown to have greater binding affinity than a mono-phosphorylated (pY 759 ) peptide (8). However, the exact structural basis underlying the second phosphotyrosine-binding site in a canonical PTB domain is not clearly understood.
Pathologically, Shc phosphorylation is linked to the stimulation of vascular endothelial growth factor (VEGF) production in tumors (16). Thus, deciphering the molecular details of this interaction may influence the development of new anti-cancer therapeutic strategies. Here we present the NMR-derived atomic view of how tyrosine phosphorylation affects ␤ 3 CT interaction with Shc PTB, and we show for the first time a high resolution three-dimensional structure of Shc PTB domain in complex with bi-phosphorylated integrin ␤ 3 CT peptide.

EXPERIMENTAL PROCEDURES
Expression and Purification-Cloning, expression, and purification of ␤ 3 CT have been described previously (2). Tyrosine phosphorylation was achieved in vivo by expressing ␤ 3 CT in TKB1 cell line from Stratagene. Details of this procedure and purification are described elsewhere. 3 The Shc PTB domain (residues 17-207, see Fig. 1A) containing pET15b vector, generously provided by Dr. Zhou, was expressed in Rosetta (DE3) cell line from Novagen to improve the expression levels. These cells express rare tRNAs facilitating the translation of genes that encode rare Escherichia coli codons. Purification of Shc PTB domain was performed according to the protocol from Qiagen under nondenaturing conditions followed by gel-filtration on HiLoad 16/60 Superdex 75 column in 50 mM Na 2 HPO 4 , 50 mM NaCl, 5 mM DTT buffer at pH 6.5. Short tyrosine(s)phosphorylated peptides corresponding to mono-and bi-phosphorylated ␤ 3 CT, MPN␤ 3 , MPC␤ 3 , and BP␤ 3 Peptide (Fig. 1B), were chemically synthesized (Genemed Synthesis, Inc.; NEOpeptides, Inc.).
NMR Sample Preparation-The heteronuclear NMR experiments were performed on uniformly 15 N and/or 13 C labeled, ϳ0.5 mM Shc PTB samples (unless mentioned otherwise), with or without the ligands/peptides on Varian Inova 600 MHz equipped with inverse-triple resonance cold probe at 35°C. The samples were prepared in pH 6.5 (unless mentioned otherwise) buffer containing 50 mM Na 2 HPO 4 , 50 mM NaCl, 5 mM DTT, 7% D 2 O, and 1 mM DSS acting as an internal standard. Chemical shift titration experiments between 15 N-labeled fulllength ␤ 3 CTs and non-labeled Shc PTB and 15 N-labeled Shc PTB and non-labeled full-length ␤ 3 CTs were performed at pH 6.1 to avoid precipitation of ␤ 3 CT.
NMR Spectroscopy-Chemical shifts assignments for the free form Shc PTB were obtained from BMRB Entry 5566 (17) and have been modified to match our construct and experimental conditions with the help of triple resonance NMR experiments, HNCACB and HNCO. For Shc PTB-BP␤ 3 Peptide complex (1:2 ratio), the backbone and side chain 1 H, 15 N, 13 C resonance assignments were made by using the standard triple resonance NMR experiments, namely HNCA, HNCACB, HNCO, HBHA-(CO)NH, and HNCACO. The intra-molecular NOE distance restraints were obtained from 13 C-and 15 N-edited three-dimensional NOESY-HSQC experiments and 13 C-edited aromatic three-dimensional NOESY-HSQC experiments (mixing time 150 ms). The inter-molecular NOE distance restraints between Shc PTB and BP␤ 3 Peptide were obtained from F1 13 C, 15 N-filtered, F2 13 C-edited NOESY-HSQC spectrum (18,19). Sequence-specific assignments of the non-phosphorylated and phosphorylated 3 integrin tails are described elsewhere (2) and have been modified to match changes in experimental conditions (pH, temperature and salt). The 1 H resonance assignments for BP␤ 3 Peptide in complex with Shc PTB were obtained from two-dimensional 13 C, 15 N-filtered TOCSY and two-dimensional 13 C, 15 N-filtered NOESY experiments (mixing time 65, 75 ms for TOCSY and 300, 400 ms in case of NOESY experiments). All the spectra were processed with NMRPipe (20) and/or Rnmrtk (21) and were analyzed by CCPN software suite (22). For the NMR dynamics study, 1 H-15 N NOE, 15 N T 1 and T 2 data were collected on Varian Inova 600 MHz spectrometer using the scheme adopted from L. Kay (23). 15 15 N NOE values were determined from spectra recorded with 5 s relaxation delay and the presence and absence of a proton presaturation period of 5 s. T 1 , T 2 , and NOE values were extracted by a curve-fitting subroutine included in the CCPN software suite (22). For curve fitting analysis, the spectra were processed with 10 Hz exponential broadening in direct dimension and zero-filled to 2048 ϫ 1024 data points in t2 and t1, respectively.
Paramagnetic Labeling-To introduce a spin label, a new peptide, BP␤ 3 Cys, was chemically synthesized (NEO-peptides, Inc.) with an additional cysteine residue at the C terminus (Fig.  1B). A typical spin labeling reaction involved 50 mM Na 2 HPO 4 , 50 mM NaCl at pH 6.5, ϳ0.4 mM BP␤ 3 Cys, and ϳ5 mM cysteinespecific spin label, 3-maleimido-PROXYL (Sigma-Aldrich), hereafter referred to as mProxyl. The reaction was allowed to proceed for 1 h at room temperature. The unreacted mProxyl was then removed from mProxyl-BP␤ 3 Cys by using combination of gel-filtration chromatography on HiLoad 16/60 Superdex 75 column in 50 mM Na 2 HPO 4 , 50 mM NaCl at pH 6.5 and reverse phase HPLC on PROTO C4 column (The Nest Group, Inc.). 15 N HSQC spectrum was collected on sample containing 0.18 mM spin-labeled peptide (mProxyl-BP␤ 3 Cys) mixed with 0.1 mM 15 N Shc PTB in 50 mM Na 2 HPO 4 , 50 mM NaCl, 1 mM TCEP, 7% D 2 O, 1 mM DSS buffer at pH 6.5 and 35°C. An additional 15 N HSQC spectrum on the 15 N-labeled Shc PTB: BP␤ 3 Cys (same concentration and in same buffer as the paramagnetic sample) was collected for comparison. The 1 H, 15 N resonance assignments of Shc PTB were modified to match these experimental conditions. For calculating the intensity ratios, again the spectra were processed with 10 Hz exponential broadening in direct dimension and zero-filled to 2048 ϫ 1024 data points in t2 and t1, respectively.
Structure Calculation-The backbone, and , dihedral angle restraints were obtained by using Talosϩ (24). Initial structure calculations were performed by using CYANA 2.1 (25). Hydrogen bond restraints were introduced during the final stages of calculations based on secondary structure elements identified from previous rounds of structure calculations. Eighty lowest energy structures from CYANA were then subjected to molecular dynamics simulations in explicit water (26) using CNS (27). Table 1 lists detailed structural statistics of the final 15 lowest energy conformers after the water refinement. None of the structures have NOE and dihedral angle violations more than 0.5 Å and 5°, respectively. The Protein Structure Software suite (PSVS; courtesy of CABM Structural Bioinformatics Laboratory, Rutgers, State University of New Jersey) was used for structure quality assessment and validation.

RESULTS
␤ 3 CT Interaction with Shc PTB Domain Strongly Depends on the Phosphorylation State of Its Tyrosines-Previous biochemical studies have indicated that p52 isoform of Shc co-immunoprecipitates with tyrosine-phosphorylated ␣ ⌱⌱b ␤ 3 from the aggregated platelets and that Shc itself gets tyrosine phosphorylated during platelet aggregation (6). This p52 isoform of Shc binds to the tyrosine-phosphorylated ␤ 3 peptide through its PTB domain (14) and phosphorylation of Tyr 759 of ␤ 3 CT is essential to meditate this interaction as only the peptides containing pY 759 have shown affinity toward GST-fused Shc (8). To further confirm and structurally characterize these findings, we have employed Nuclear Magnetic Resonance spectroscopy (NMR). To pinpoint the residues/regions involved in the Shc PTB-␤ 3 CT interaction, we began with the chemical shift mapping experiments. Non-labeled Shc PTB domain was mixed with 15 N-labeled non-phosphorylated ␤ 3 CT (hereafter referred to as ␤ 3 NP), Tyr 747 mono-phosphorylated ␤ 3 CT (hereafter referred to as ␤ 3 MP) and 747 Y-759 Y bi-phosphorylated ␤ 3 CT (hereafter referred to as ␤ 3 BP) at the ratio 2:1 and the associated chemical shifts perturbations were monitored (expanded regions of superimposed HSQC spectra are shown in Fig. 2: (A) ␤ 3 NP; (B) ␤ 3 MP, and (C) ␤ 3 BP). As expected, Shc PTB addition had no effect on HSQC spectrum of ␤ 3 NP. In contrast, both ␤ 3 MP and ␤ 3 BP HSQC spectra show significant differential line-broadening and several peaks disappearance along with some small shifts in resonance frequencies upon addition of Shc PTB (with an exception of the very last C-terminal residue Thr 762 of ␤ 3 BP, which demonstrates substantial chemical shift, Fig. 2C). This phenomenon is probably due to the intermediate exchange between free and bound states of ␤ 3 MP and ␤ 3 BP (28) combined with relatively large molecular weight of the complex (about 33 kDa complex versus 8 kDa for ␤ 3 CT alone). The ratio of the peak intensities along with chemical shifts perturbations for ␤ 3 MP and ␤ 3 BP residues plotted as a function of residue numbers is shown in Fig. 2, D and E, respectively. Combination of the differential line broadenings and chemical shift perturbations suggests the regions involved in interaction between phosphorylated ␤ 3 CT and Shc PTB, namely (i) residues from Asp 740 to Ala 750 , surrounding 744 NPLpY 747 motif, in case of ␤ 3 MP, and (ii) almost the entire C terminus, extending from Asp 740 to Gly 762 and encompassing both 744 NPLpY 747 and 756 NITpY 759 motifs plus the region connecting them in case of ␤ 3 BP. Because the affected region for ␤ 3 BP upon Shc PTB addition is much broader than the one for ␤ 3 MP, it can be argued that the binding site around 744 NPLpY 747 motif is a complimentary one and serves to stabilize ␤ 3 CT orientation within the complex which has been defined by the primary binding site around 756 NITpY 759 motif.
To confirm the hypothesis that 756 NITpY 759 motif occupies the canonical PTB site and to map the residues involved from Shc PTB side, we performed similar chemical shift mapping experiments. Non-labeled ␤ 3 NP, ␤ 3 MP, ␤ 3 BP solutions were mixed with 15 N-labeled Shc PTB at the ratio 2:1. The superimposition of expanded regions of HSQC spectra for 15 N-labeled Shc PTB mixed with ␤ 3 NP, MP, BP is depicted in Fig. 3A (Shc PTB alone is shown in black; with ␤ 3 NP, in blue; ␤ 3 MP, in red; and ␤ 3 BP, in lime; the superimposition of entire spectra is presented in supplemental Fig. S1A). As predicted, there were no changes in the 15 N-labeled Shc PTB HSQC spectrum upon addition of the full-length non-labeled ␤ 3 NP. Addition of ␤ 3 MP to 15 N-labeled Shc PTB leads to several small shifts in resonance frequencies as well as appearance of few additional (doublet) weak peaks probably representing a small population of protein in bound conformation. In contrast, there are signifi-cant changes in HSQC spectrum of Shc PTB upon addition of ␤ 3 BP indicating substantial conformational rearrangement of Shc PTB, a hallmark of Shc PTB domain (17). However, low mutual solubility of Shc PTB and full-length ␤ 3 CT constructs prevented us from conducting a full scale, thorough NMR structural investigation of Shc PTB-␤ 3 BP interaction.
To circumvent the solubility issue and to understand this interaction in atomic details, we have synthesized three phosphorylated tyrosines containing peptides representing different regions of the full-length ␤ 3 CT, which are involved in interaction with Shc PTB (Fig. 1B). Similar chemical shift mapping experiments were performed on 15 N-labeled Shc PTB mixed with non-labeled N-terminal mono-phosphorylated peptide, MPN␤ 3 , C-terminal mono-phosphorylated peptide, MPC␤ 3 , and C-terminal bi-phosphorylated peptide, BP␤ 3 Peptide. Chemical shifts perturbations were monitored and are depicted in Fig. 3B (superimposition of the full spectra is presented in supplemental Fig. S1B). The outcome appeared to be very similar to the full-length ␤ 3 CT titrations. (The complete chemical shift perturbations, i.e. 15 N-labeled Shc PTB ϩ non-labeled MPN␤ 3 , MPC␤ 3 , BP␤ 3 Peptide, are presented in supplemental Fig. S1C). Addition of MPN␤ 3 , containing pY 747 , leads to small shifts in Shc amide resonance frequencies together with an appearance of several weak doublet peaks. Based upon this observation, we can suggest that in the absence of pY 759 , pY 747 may occupy the orthodox PTB site. However, its higher dissociation rate may prevent the stabilization of the structural rearrangement in Shc PTB. In contrast, addition of MPC␤ 3 , containing pY 759 , as well as addition of BP␤ 3 Peptide, containing both pY 759 and pY 747 , to Shc PTB results in significant perturbations in Shc PTB HSQC spectrum. Overall, these perturbations resemble very closely to the ones found in Shc PTB-TrkA complex (17) representing a very well established conformational rearrangement of Shc PTB upon ligand binding. Surprisingly and on a more important note, these titration experiments indicated that although MPC␤ 3 alone is sufficient to induce the classical conformational change in Shc PTB, similar or even the greater conformational rearrangement is achieved via addition of BP␤ 3 Peptide.  that accommodation of 744 NPLpY 747 motif requires some movement of the helix ␣2 and the flexible loop connecting ␣2 to ␤3 strand.
To conclude, our titrations results are in good correlation with Cowan et al. (8) demonstrating that the phosphorylation of Tyr 759 ␤ 3 CT is essential to meditate direct integrin interaction with Shc PTB except that the NMR methods have allowed us to observe a weak interaction between Shc PTB and ␤ 3 peptides containing only pY 747 , which the biochemical assays could not detect.
Structural and Dynamic Characterization of the Complex-Based upon our titration experiments, we could predict the possible mode of interaction of Shc PTB with BP␤ 3 Peptide (27 residues, Fig. 1B): positively charged side-chain of Arg 104 forming a saltbridge with negatively charged phosphate group of pY 747 whereas pY 759 occupying the canonical PTB site. To fully characterize this interaction at atomic level, we have determined the three-dimensional solution structure of Shc PTB-BP␤ 3 Peptide complex using modern triple resonance NMR methods (described in details under "Experimental Procedures"). Inter-molecular NOEs were paramount in defining the orientation of BP␤ 3 Peptide within the complex, which was later independently confirmed by paramagnetic relaxation enhancement (PRE) experiments (see below). Table 1 summarizes the structural statistics for the final 15 water refined structures with lowest energies and Fig. 4, A and B depict the backbone superimposition and ribbon representation of these structures respectively (the secondary structural features are presented in supplemental Fig. S2).
As seen in case of Shc PTB-TrkA complex, the N-terminal region of Shc PTB (residues 17-35) are dynamically unstructured (12), whereas the core-structured region of Shc PTB, encompassing residues 38 -201, adopts a well known PH domain superfold in complex with BP␤ 3 Peptide: a seven-stranded ␤sandwich composed of two antiparallel ␤-sheets capped by a C-terminal ␣-helix (␣3). Moreover, it contains two additional ␣-helices, N-terminal ␣1 and ␣2 connected to strands ␤1 and ␤2 respectively. Examination of this complex reveals the details of the binding sites for both pYs (Fig. 4C). As expected, residues 754 FTNITpY 759 (representing the classical consensus xNPxpY Shc PTB recognition motif with an exception of an isoleucine replacing proline residue) sits in the canonical PTB grove, an elongated cleft located between helix ␣3 and stand ␤5, with negatively charged phosphate group forming salt-bridges with positively charged side chains/amides of the Arg 67 , Arg 175 , Lys 169 , and Gln 148 . The residues 752 STF 754 of BP␤ 3 Peptide adopt an anti-parallel ␤-strand conformation aligned to ␤5 strand of Shc PTB with hydrophobic amino acid, Phe 754 , maintaining the majority of inter-subunit contacts. In fact, we can suggest that this large hydrophobic residue, Phe 754 , is crucial for directing C-terminal pY 759 into the canonical PTB pocket: compared with Ala 742 , the  NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 JOURNAL OF BIOLOGICAL CHEMISTRY 34879 pY-5 residue in case of 744 NPLpY 747 motif, it has the ability to accommodate more essential core hydrophobic contacts. Residues 756 NITpY 759 form a type-I ␤ turn, which is further stabilized by N 756 -F 198 contacts.

Integrin-Shc Interaction
The binding site for the second pY 747 is located in the grove formed between helix ␣2 and long flexible loop connecting the strand ␤2 and helix ␣2. Similar to the 756 NITpY 759 motif, 744 NPLpY 747 motif forms a type-I ␤ turn which fits nicely into the pocket formed by residues 100 KPCSRPLS 107 . Although 97 R-R 99 containing region is highly flexible (see the relaxation data below), the prolines in 100 KPCSRPLS 107 motif give rigidity to this phosphotyrosine-binding grove where the pY 747 interacts with the positively charged side chain of Arg 104 . This interaction is further stabilized by contacts between 107 S-A 742 and 107 S-T 741 (see supplemental Fig. S3 for representative intermolecular NOEs) and salt bridges between side chain of Gln 76 and Ala 742 . The region connecting these two phosphotyrosine motifs, 749 E-T 753 is stretched across one face of Shc PTB ␤-sandwich (␤5, ␤6, ␣2) defined by the intermolecular NOEs between A 750 -R 79 and E 749 -T 75 . The N-terminal residues 736 RAKW 739 of BP␤ 3 Peptide are dynamically unstructured and are not well defined. This ensemble, the NMR data and chemical shifts table have been deposited to Protein Data Bank (PDB) and Biological Magnetic Resonance Bank (BMRB) with the access codes 2L1C and 17080, respectively.
The BP␤ 3 Peptide orientation was further confirmed and validated by introduction of the cysteine specific paramagnetic spin label, mProxyl, attached to the C terminus of modified peptide, BP␤ 3 Cys, through the formation of thioether bond and measuring the distance-dependent reduction in peaks intensities in 15 N-labeled Shc PTB HSQC spectrum (29). These PRE studies independently confirmed the orientation of the phosphorylated integrin tail. Briefly, paramagnetic spin label facilitates nuclear relaxation in a distance-dependent manner (1/r 6 ), causing significant line-broadening for nuclei in proximity (Ͻ15-20 Å) to the free radical. Analysis of the reduction in NMR peaks intensities allows mapping the direct location of the spin label with respect to the protein binding surface (30). This pattern of reductions in peaks intensities of Shc PTB HSQC spectra upon addition of mProxyl-BP␤ 3 Cys is depicted in Fig. 5D and the intensities ratios are mapped on the surface of the complex in Fig. 4D. The affected residues are shown in color gradient from orange to yellow (most affected: orange, least affected: yellow), and residues for which we have no data are shown in gray. The expanded region of superimposed HSQC spectra is shown in supplemental Fig. S4. It should be noted that the chemical shift perturbations in Shc PTB resonances are very similar/almost identical upon addition of BP␤ 3 Peptide or modified BP␤ 3 Cys at 1:2 ratio indicating that the inclusion of an additional cysteine residue at the C terminus did not alter the bound conformation and/or the affinity of binding. The specific pattern of the altered cross-peak intensities, including the significant reduction in peak intensities (I para /I dia Ͻ0.3) of the 146 H-I 150 , 168 A-V 172 , 201 R-R 207 regions indicates that the C-terminal mProxyl tag is, indeed, positioned near the canonical binding site. This confirms the occupancy of the binding pocket by pY 759 and eliminates any possibility for the distant pY 747 to be found at the same place.
Further, to better understand the backbone conformational flexibility of Shc PTB in complex with ␤ 3 integrin, we have measured the relaxation parameters T 1 , T 2 , and heteronuclear 1 H-15 N NOEs (Fig. 5, A-C). The average value of T 1 is about 710 ms; the average value of T 2 is about 50 ms; and the average NOE value is about 0.66. For most of the residues, T 1 and T 2 values do not deviate significantly beyond the experimental error from the average numbers. However, several regions, including dynamically unstructured N-(residues 17-38) and C-(residues 202-207) termini plus a stretch of residues (91-100) within loop connecting helix ␣2 with strand ␤2 (89 -111), demonstrate fast internal motion with increased T 2 and reduced NOE values. The flexible nature of this loop with the most profound motion associated with the residues Arg 97 and Arg 99 allows crucial structural changes to ensure the formation of the second phosphotyrosine-binding pocket. The motion of 100 KPCSRPLS 107 region within this loop is restricted because of the interaction with ␤ 3 744 NPLpY 747 motif.

DISCUSSION
The short cytoplasmic tails of integrins (see Fig. 6E), devoid of any intrinsic enzymatic activity and unable to connect directly to cytoskeleton, can interact with variety of adapter proteins via surprisingly few specific, highly conserved motifs. These include membrane-proximal region, HDRk/rE and/or  15 N-filtered, F2 13 C-edited NOESY-HSQC, and 15 N-edited NOESY and 13 C-edited NOESY. c Generated from Talosϩ (24). d Hydrogen bonds were introduced in the last stage of structure calculations. e Residues 36 -202 (Shc PTB), 740 -760(BP␤ 3 Peptide) calculated using PSVS. f After refinement in explicit water by using CNS (27). g All residues.
one of the two NXXY motif containing regions recently reviewed in details (31). Overall, these interactions are tightly controlled with phosphorylation as one of the possible regulatory mechanisms. The phosphorylation state of the tyrosine residues within NXXY motifs of ␤ 3 CT can differentially regulate ␤ 3 interactions with PTB domain-containing proteins. Talin, for example, serves as a major activator for non-phosphorylated ␤ 3 (2, 32, 33), while Dok1 binds to pY 747 -␤ 3 with higher affinity and, thus, replaces talin favoring the latent state of the receptor (34). Furthermore, Cowan et al. (8) have demonstrated that phosphorylation of the Tyr 759 is essential for the direct Shc binding, which mediates outside-in signaling events (35). Until now, however, the detailed structural basis of this interaction has remained elusive. The present study exposes the exact molecular mechanism underlying the Shc PTB-␤ 3 CT complex formation and takes our understanding of the nature of this interaction to a new level.
The NMR data presented unambiguously proves that both phosphorylated tyrosine residues are involved in interaction with Shc PTB domain, although the vital role is performed by pY 759 . As depicted in Fig. 4, the BP␤ 3 Peptide wraps itself around Shc PTB with the major focal points presented by electrostatic interactions between negatively charged phosphate groups of pY 759 /pY 747 and positively charged side chains of Shc PTB (residues Arg 67 , Arg 175 , Lys 169 , Gln 148 in case of pY 759 and Arg 104 in case of pY 747 ). As predicted, 756 NITpY 759 motif occupies the canonical PTB pocket with residues 752 STF 754 forming an anti-parallel ␤ strand against the ␤5 strand ( 150 ISFA 153 ) of Shc PTB. This complex also illuminates an additional novel binding site for pY 747 with the characteristic, perpetual type-I ␤ turn of 744 NPLpY 747 motif fitting nicely into the grove formed between helix ␣2 and long flexible loop connecting the strand ␤2 and the helix ␣2.
A direct comparison between Shc PTB-BP␤ 3 Peptide complex and Shc PTB-TrkA complex (pdb id: 1SHC), depicted on Fig. 6A, reveals major similarities with some crucial differences depicted in Fig. 6A. Overall, the C␣ atoms of the structured regions (residues  superimpose to the mean structure reasonably well with an R.M.S.D. of 2.26 Å Ϯ 0.98Å. As presented in Fig. 6B, R.M.S.D. graph most of the ␤ strands superimpose very well with an R.M.S.D. below 1 Å. The pY 759 of ␤ 3 , occupying a canonical PTB site, overlaps neatly with pY 490 of TrkA with small shifts in placement accompanied by corresponding movement of the loops connecting strands ␤4/␤5 and ␤5/␤6 (Fig. 6C). However, the regions involved in the formation of the second, pY 747 , binding site show remarkable differences. The position of the loop connecting the helix ␣2 with the strand ␤2 is significantly different, reflecting the biggest fluctuation in the graph with maximum R.M.S.D. over 6 Å. Based upon our titration data and the previous biochemical assays (8), this novel phosphotyrosine-binding groove formed between the long loop and helix ␣2, is responsible for defining the precise arrangement of this large 27 residues bi-phosphorylated integrin constituent on Shc PTB surface, thereby increasing the binding affinity as compared with the small 11 residues, mono-phosphorylated (pY 759 ) MPC␤ 3 . So far the exact structural role of this unusually long loop (ϳ24 residues), exclusively found in Shc PTB domain, has not been established except for residue Arg 112 , situated at the beginning of ␤2 strand, which has been implicated in phospholipid interaction along with residues Lys 116 and Lys 139 (36). As per our knowledge, this is the first time when residues 100 KPCSRPLS 107  NOVEMBER 5, 2010 • VOLUME 285 • NUMBER 45 of this elongated loop, commonly referred as Shc loop (37), are shown to be involved in direct interaction with a phosphorylated tyrosine residue. The actual biological significance of the proximity of these two binding sites, the phospholipid and the, pY 747 , phosphotyrosine-binding site, is yet to be understood. However, this comparison between Shc PTB-TrkA and Shc PTB-BP␤ 3 Peptide, along with the NMR relaxation data, proves the flexible nature of Shc PTB loop(s) manifesting the versatility found in PTB fold.

Integrin-Shc Interaction
To further analyze the capability of PTB domains to accommodate different fragments of integrin tails we compared the known structure of integrin ␤ 3 (chimera, (34)) bound to talin (PDB ID: 1MK7) with Shc-␤ 3 complex. Talin PTB (also known as an F3 variant of the canonical PTB) domain differs significantly from Shc PTB, it is only about half in size (ϳ100 residues long versus ϳ200) with missing analogs for helices ␣1 and ␣2 and the long Shc loop and with absolutely no sequence homology even within the canonical binding pocket. However, in terms of structural architecture, the core seven-stranded ␤-sandwich together with C-terminal ␣-helix of talin PTB-␤ 3 complex superimposes surprisingly well with Shc PTB-BP␤ 3 Peptide complex (see Fig. 6D), even though the residues defining the interaction from the ␤ 3 side are completely different. In both cases, two different NXXY motifs of ␤ 3 integrin form reverse turns which are further stabilized by contacts between N 756 -F 198 for Shc and N 744 -T 354 /I 356 for talin. An aromatic residue Trp 739 (at Y-8 position), as compared with the hydrophobic Ile 485 of TrkA or Phe 754 of ␤ 3 integrin in the complex with Shc (both found in canonical pY-5 positions), defines the antiparallel orientation of the ligand ␤ strand. Both these antiparallel ␤ strands, formed by residues 752 STF 754 in case of Shc and residues 739 WDTA 742 in case of talin, superimpose surprisingly well. However, the non-phosphorylated for Shc PTB backbone amide protons in the complex with paramagnetically (I para ) and diamagnetically (I dia ) labeled BP␤ 3 Cys plotted against residue numbers. An intensity ratio of one indicates no effect of the spin label on an amide proton. All the relaxation measurements were performed at 35°C, pH 6.5 in the buffer containing 50 mM Na 2 HPO 4 , 50 mM NaCl, 5 mM DTT, 7% D 2 O, and 1 mM DSS whereas PRE experiments were performed in the buffer containing 50 mM Na 2 HPO 4 , 50 mM NaCl, 1 mM TCEP, 7% D 2 O, and 1 mM DSS.
Tyr 747 of ␤ 3 chimera occupies the acidic, hydrophobic grove of talin PTB domain as compared with pY 759 utilizing the strongly basic pocket of Shc PTB. This observation led to the hypothesis regarding the reduction in binding affinity for talin upon Tyr 747 phosphorylation due to the obvious charge repulsion and some steric hindrance. Surprisingly, the actual measured reduction in affinity was found to be modest, only ϳ2-fold (38). Moreover, in a previous study (39), ␤ 3 integrin exhibited strong affinity for PTB domains of 17 different proteins. In addition to ␤ 3 CT, the cytoplasmic domains of integrin ␤ 1A , ␤ 5 , and ␤ 7 also demonstrated some affinities to several of these PTB domains, reflecting the intrinsic flexibility of both the PTB fold and ␤ integrins. Among the possible reasons for such indiscrimination is the exceptional conservation of NPXY and NXXY motifs within ␤ integrin tails (Fig. 6E), which along with other critical residues, coordinate integrin-PTB domains interactions. However, considering the specific nature of Shc PTB: ␤ 3 CT interaction, we speculate that among all the integrin tails depicted in Fig. 6E, similar interaction with Shc PTB can be expected only in case of the tyrosine-phosphorylated ␤ 6 ( 740 NVTpY 743 ) due to the presence of large hydrophobic residue, Phe 738 , at pY-5 position. This bulky, hydrophobic residue (corresponding to Phe 754 in ␤ 3 CT and Ile 485 in TrkA, Fig. 6C) occupies the non-polar pocket formed between ␣3 helix and ␤5 strand of Shc PTB. Furthermore, according to our Shc PTB-BP␤ 3 Peptide complex, the placement of a large negative group (pT 753 ) next to this pY-5 residue should cause the charge repulsion with the nearby Thr 75 (from ␣2 helix of Shc PTB) along with steric hindrance with the above mentioned non-polar pocket. This is probably the most likely cause for the decreased affinity (40) of ␤ 3 CT to Shc PTB observed upon Thr 753 phosphorylation.
Overall, this presented comparison establishes two salient features: (i) proteins containing PTB fold can fine-tune their affinity toward their targets by an introduction of additional target-specific binding sites as the second phosphotyrosinebinding site defined in Shc; and (ii) integrin cytoplasmic tails are capable of accommodating different structural features depending upon the binding partner. This remarkable dexterity may be the underlying foundation for the crucial bidirectional flow of information through integrins. Although ␤ 3 CT interaction with Shc PTB is unique as compared with its interaction with talin or Dok1 PTB domains, a low sequence homology among PTB domains makes it is very difficult to predict whether the other PTB domains will interact BP␤ 3 Peptide is shown in lime; ␤ 3 chimera (GSHM-739 WDTANNPLYKE 749 ) is shown in dark red. E, sequence alignment (produced by CLC Sequence Viewer) of human ␤ integrin CTs; highly conserved residues are depicted in pink and least conserved are in blue. ␤ 4 and ␤ 8 CTs are omitted, even though ␤ 4 CT (pY 1526 ) has been shown to interact with Shc PTB (43), because ␤ 4 CT is ϳ1090 residues long and the divergent ␤ 8 CT clearly lacks the conserved NPXY and NXXY motifs required to mediate Shc PTB interaction.
with ␤ integrin tails in a manner similar to Shc, talin, or Dok1. Indeed, such low sequence homology within the PTB domains simultaneously presents a challenge for the computational modeling and an opportunity for the comprehensive structural investigation.
To conclude, we have (i) confirmed the direct Shc PTB interaction with ␤ 3 integrin cytoplasmic tail; (ii) demonstrated that this interaction depends strongly on the tyrosine(s) phosphorylation state of the receptor; (iii) structurally characterized Shc PTB in complex with bi-phosphorylated ␤ 3 CT; and (iv) defined molecular details of the secondary non-canonical phosphotyrosine-binding site within the Shc PTB. Because Shc is involved in regulating the stimulation of VEGF production in tumor cells, our data help to understand how tyrosine phosphorylation of ␤ 3 integrin is linked to MAPK pathway and how it may play multiple roles in the regulation of integrin signal transduction.