An Alternative Phosphorylation Switch in Integrin β2 (CD18) Tail for Dok1 Binding

Integrins are involved in cell migration and adhesion. A large number of proteins interact with the cytoplasmic tails of integrins. Dok1 is a negative regulator of integrin activation and it binds to the phosphorylated membrane proximal NxxY motif in a number of integrin β tails. The β tail of the β2 integrins contains a non-phosphorylatable NxxF motif. Hence it is unclear how Dok1 associates with the β2 integrins. We showed in this study using NMR and cell based analyses that residues Ser745 and Ser756 in the integrin β2 tail, which are adjacent to the NxxF motif, are required for Dok1 interaction. NMR analyses detected significant chemical shift changes and higher affinity interactions between Dok1 phospho-tyrosine binding (PTB) domain and integrin β2 tail peptide containing pSer756 compared to pSer745. The phosphorylated β2 peptide occupies the canonical ligand binding pocket of Dok1 based on the docked structure of the β2 tail-Dok1 PTB complex. Taken together, our data suggest an alternate phosphorylation switch in β2 integrins that regulates Dok1 binding. This could be important for cells of the immune system and their functions.


Integrins are involved in cell migration and adhesion. A large number of proteins interact with the cytoplasmic tails of integrins. Dok1 is a negative regulator of integrin activation and it binds to the phosphorylated membrane proximal NxxY motif in a number of integrin β tails. The β tail of the β2 integrins contains a non-phosphorylatable NxxF motif. Hence it is unclear how Dok1 associates
with the β2 integrins. We showed in this study using NMR and cell based analyses that residues Ser745 and Ser756 in the integrin β2 tail, which are adjacent to the NxxF motif, are required for Dok1 interaction. NMR analyses detected significant chemical shift changes and higher affinity interactions between Dok1 phospho-tyrosine binding (PTB) domain and integrin β2 tail peptide containing pSer756 compared to pSer745. The phosphorylated β2 peptide occupies the canonical ligand binding pocket of Dok1 based on the docked structure of the β2 tail-Dok1 PTB complex. Taken together, our data suggest an alternate phosphorylation switch in β2 integrins that regulates Dok1 binding. This could be important for cells of the immune system and their functions.
Integrins are a large family of cell surface α β heterodimers that mediate cell-cell and cell-ECM interactions necessary for many physiological processes, including hemostasis, wound healing, immunity and developmental biology 1 . Each subunit of an integrin has a large extracellular region that binds to ligands, a single-pass transmembrane domain that transduces activation signal across the plasma membrane and a short cytoplasmic tail (except integrin β 4) that binds to an expanding list of cytoplasmic proteins 2 . Except β 4 and β 8, integrin β tails contain two highly conserved NxxY/F (x: other amino acid) motifs that are docking sites for cytoplasmic proteins 2 . The membrane proximal NxxY/F motif is a binding site for talin, a well-established cytoskeletal protein that directly activates integrins 3,4 . The two isoforms of talin in vetebrates (talin 1 and 2) are 4.1-ezrin-radixin-moesin (FERM)-containing proteins 5,6 . The FERM domain lies in the head region of talin, and a phosphotyrosine binding (PTB) fold in its F3 subdomain has been shown to bind the membrane proximal NPLY747 motif of the integrin β 3 tail 7 . This form of interaction is not limited to talin because it extends to other cytoplasmic proteins containing PTB folds, including negative regulator of Notch signaling (Numb), downstream target of c-Abl (Dab) and docking protein 1 (Dok1; p62Dok) 8 .
Dok1 is a member of the Dok family of adaptor proteins and it is expressed in lymphoid and myeloid cells 9,10 . Dok1 and Dok2 are negative regulators of immune cell signaling and Dok1 has been reported to bind with p120RasGAP, a negative regulator of the Ras-ERK pathway 11 . All seven members of the Dok family of proteins contain an N-terminal pleckstrin homology (PH) domain, a central PTB fold and multiple SH2 and SH3 binding sites 11 . Dok1 and talin have overlapping binding sites which include the NPLY747 motif in the integrin β 3 tail. Therefore it is unlikely that both molecules can simultaneously bind the integrin β tail. Indeed biophysical analyses have shown that phosphorylated Y747 enhances Dok1 binding over that of talin, suggesting that NPLY747 is a phosphorylation switch 12 . Unlike talin, Dok1 does not activate integrins. Dok1 is a negative regulator of integrin activation by competing with talin for binding to integrin β 1A, β 3 and β 7 tails containing the membrane proximal NxxY motif 13,14 . The leukocyte-restricted β 2 integrins comprise four members that have different α subunits but a common β 2 (CD18) subunit, namely α Lβ 2, α Mβ 2, α Xβ 2 and α Dβ 2 2 . The importance of the β 2 integrins is underscored by the rare autosomal disease Leukocyte Adhesion Deficiency (LAD) I in which afflicted individuals have a compromised immune system because of defective adhesive and migratory properties of their leukocytes. The molecular basis of LAD I is the reduced expression and/or expression of dysfunctional β 2 integrins in leukocytes as a result of mutation(s) in the ITGB2 gene 15,16 . Dok1 has been reported to bind the integrin β 2 tail 8 . However, the corresponding Dok1 binding region in β 2 contains an NPLF754 motif that does not allow phosphorylation. This begs the question if there is an alternative phosphorylation site(s) in the region that regulates Dok1 binding. Residues Ser745 and Ser756 flank the NPLF754 motif in the integrin β 2 tail. The corresponding Ser residues are absent from the β tails of β 3, β 5 and β 6 integrins (Table 1). However, the β tails of β 1A, β 1D and β 7 integrins contain a Ser residue at an equivalent position to that of Ser 756 in the β 2 tail. In vitro studies have shown that integrin β 2 Ser745 and Ser756 are phosphorylatable but only the former is dependent on PKCs in T cells 17 . These observations suggest an interesting possibility that the phosphorylation state of Ser745 and Ser756 could regulate the binding of Dok1 to integrin β 2 tail.
In this study, we showed using NMR that the integrin β 2 tail interacts with the PTB domain of Dok1. Phosphorylation at Ser756 of the integrin β 2 tail enhanced its affinity towards the PTB domain of Dok1. Similar observations were made using a 15-residue peptide fragment containing phosphorylated Ser756. 15 N-1 H HSQC spectra, 31 P NMR, tr-NOESY, saturation transfer difference (STD) NMR and HADDOCK docking provided insights for the first time into the mode of binding and conformations of the integrin β 2 tail in complex with Dok1 PTB domain. The importance of Ser745 and Ser756 in the integrin β 2 tail for Dok1 binding was verified by cell based FRET analysis. In addition, we showed that cellular activation by chemokine induced the dissociation of Dok1 from the integrin β 2 tail. The current findings therefore suggest a novel paradigm involving an alternative phosphorylation switch in the integrin β 2 tail that is required for the regulation of β 2 integrins by Dok1.

Results
Binding of the PTB domain of Dok1 with phosphorylated full-length integrin β2 tail peptides. Interactions between Dok1 PTB domain with full-length integrin β 2 tail phosphorylated either at Ser745 (pSer745-β 2) or Ser756 (pSer756-β 2) were investigated by analyzing the changes in 15 N-1 H HSQC spectra of the PTB domain. In these experiments, a series of 15 N-1 H HSQC spectra of 15 N-labelled PTB domain were acquired at various concentrations of integrin β 2 tail peptides (Fig. 1A,B). 15 N-1 H HSQC spectra of the PTB domain showed discernable chemical shift changes upon addition of phosphorylated integrin β 2 tail peptides. The chemical shifts of the backbone resonances including 15 N and 1 H of the Dok1 PTB domain have been previously assigned 12,18 . Chemical shift perturbations of a number of residues including Leu52, Arg54, Tyr56, Gly57, Val61, Phe65, Arg69 at the central region of the Dok1 PTB domain were detected when pSer745-β 2 or pSer756-β 2 peptide was added (Fig. 1C,D). However, the chemical shift changes were more pronounced in the presence of pSer756-β 2. Residues Ala11, Arg14, Val22, Leu30, Ala37, Gln40 at the N-terminus and residues Phe89, Glu93, Thr94, His97 at the C-terminus of Dok1 also showed chemical shift perturbations (Fig. 1D). The binding affinity of Dok1 PTB domain with phosphorylated integrin β 2 tail was determined from the combined chemical shift changes in the presence of increasing ligand concentrations with K d values of 98 μ M and 215 μ M for pSer756-β 2 and pSer745-β 2 tail peptides, respectively (Fig. 1E,F). These data suggest that phosphorylation at Ser756 as compared to Ser745 of the integrin β 2 tail promotes higher affinity interaction with Dok1 PTB (Table 2). Due to limited solubility of the full-length unphosphorylated β 2 tail at high concentrations, 15 N-1 H HSQC spectra of Dok1 PTB domain were obtained using a fragment of the integrin β 2 Table 1. Comparison of amino acid sequences of integrin cytoplasmic tails highlighting the NxxY/F motif (blue) and potential phosphorylation of Ser residues (red). tail KSQWNNDNPLFKSAT or KT15 that contains both serines. Few chemical shift changes of residues in Dok1 PTB domain in the presence of unphosphorylated KT15 were detected ( Fig. 2A). The extent of chemical shift changes of Dok1 in the presence of KT15 was also limited (Fig. 2B). An apparent K d value of 16 mM was determined for the interaction between Dok1 PTB domain and unphosphorylated KT15 (Fig. 2C, Table 2). Compared with pSer756-β 2 and pSer745-β 2 tails, the unphosphorylated integrin β 2 tail interacts weakly with the Dok1 PTB domain. Similar observations were made in interactions studies using unphosphorylated integrin β 3 tail with the Dok1 PTB domain 12 . Chemical shift perturbations caused by pSer756-β 2 tail were mapped onto the 3-D structure of Dok1 PTB domain (Fig. 3). Most of the binding residues are located in the C-terminal α -helix and the β 5 strand-loop-β 6 strand-loop of the seven stranded β -sheet of the PTB domain (Fig. 3). These structural elements of Dok1 PTB domain are involved in binding pTyr peptide ligands 12,19 .

Conformations and interactions of phosphorylated integrin β2 tail fragment with PTB domain.
To gain further insight into phosphorylated integrin β 2 tail-Dok1 interactions, we made use of a pSer756-KT15 (KSQWNNDNPLFKpSAT) peptide. 15 N-1 H HSQC spectra of the Dok1 PTB domain were acquired at various concentrations of pSer756-KT15 peptide. Chemical shift perturbations were detected for residues of Dok1 PTB domain located at the canonical ligand binding site akin to that detected using the full-length pSer756-β 2 tail (Fig. 4A). A K d value of 189 μ M was determined for the interaction between pSer756-KT15 the and Dok1 PTB domain (Fig. 4B, Table 2). This value is somewhat higher than the K d value of 98 μ M obtained using full-length pSer756-β 2 tail peptide. This may suggest additional interactions between full-length integrin β 2 tail with the Dok1 PTB. To probe the direct involvement of the phosphate group on Ser756, we acquired 31 P NMR spectra of pSer756-KT15 peptide in aqueous solutions in the absence or presence of different concentrations of Dok1 PTB domain (Fig. 4C). The 31 P NMR signal of pSer756-KT15 peptide resonated at ~0.9 ppm in free solution. There was a significant downfield shift (toward higher chemical shift values) of the 31 P NMR signal upon addition of Dok1 PTB domain. 31 P NMR resonance of pSer756 shifted to as much as ~2.5 ppm at the protein to peptide ratio of 3:1. This large downfield shift or deshielding effect of 31 P resonance of pSer756-KT15 peptide could be attributed to ionic and/or hydrogen bond interactions with the residues of Dok1 PTB domain.
We performed isotope filtered 3-D NOESY experiments with 15 N/ 13 C labelled samples of Dok1 PTB domain with unlabeled pSer756-KT15 peptide to determine the atomic-resolution structure of the PTB domain-β 2 peptide complex. However, it was unsuccessful because of the paucity of inter-molecular NOEs, possibly due to fast dissociation of the complex and relatively insensitive nature of filtered experiments (data not shown). tr-NOESY 20,21 and STD-NMR 22,23 methods were used to determine the bound structure and interactions of pSer756-KT15 peptide ligand and Dok1 PTB domain. Others have reported that a fast chemical exchange between the free-state and bound-state of ligands are necessary for tr-NOEs and STD experiments [20][21][22][23] . 15 N-1 H HSQC spectra of Dok1 PTB domain with pSer756-KT15 peptide indicated fast exchange regime of complex formation. tr-NOESY spectra of pSer756-KT15 peptide showed only sequential and intra-residue NOEs, suggesting that the pSer756-KT15 peptide adopts an extended or a β -strand-like conformation ( Supplementary Fig. 1). 1-D STD spectrum and the rence or off-resonance spectrum of pSer756-KT15 peptide in the presence of Dok1 PTB domain were analyzed (Fig. 4D). STD spectrum of pSer756-KT15 peptide appears to retain several signals and is largely similar to the reference spectrum, indicating intimate association of pSer756-KT15 peptide with the PTB domain. A detailed group epitope mapping analyses of the pSer756-KT15 peptide in complex with Dok1 is impeded due to overlapping resonances in the 1-D STD spectrum. However, STD effects arising from diagnostic resonances can be ascertained. In particular, downfield shifted aromatic resonances at 7.2 to 7.7 ppm observed in the STD spectrum reveal proximity of the aromatic ring of residues Trp747 and Phe754 of pSer756-KT15 peptide to the Dok1 PTB domain. The upfield resonances at 0.8-0.9 ppm and 1.25 ppm of the alkyl sidechain of single Leu753 and CH 3 group of single Thr758 can be seen in the STD spectrum, indicating their proximity with the PTB domain. pSer756-KT15 peptide did not exhibit any  Fig. 2). STD experiments were also performed with unphosphorylated KT15 peptide with Dok1 PTB domain but there was apparently no or low STD effect, indicating a lack of binding with the Dok1 PTB domain (Supplementary Fig. 3).   5). In these structures, PTB domain-peptide complexes are defined by extended conformations of the peptide ligands occupying the binding region comprising the C-terminal long helix and β 5 and β 6 strands. The peptides are also in close association with the surface of the PTB domains. In the docked structure of pSer756-KT15/ Dok1 PTB domain, there are a number of potential molecular interactions that may stabilize the complex. The phosphate group of pSer756 is in close proximity to the side-chain guanidinium groups of residues Arg54, Arg69 and Arg70 of the PTB domain, suggesting salt-bridge and/or hydrogen bond interactions (Fig. 6A). These Arg residues are also involved in the recognition of the pTyr in RET and integrin β 3 tail peptide 12,19 . Apart from these interactions, aromatic-aromatic stacking and cation-π contacts may be involved between Dok1 PTB domain Phe89 and Arg58 with Trp747 of the pSer756-KT15 peptide (Fig. 6B). Hydrophobic packing interactions may occur between residues Phe754 and Leu753 in the NPLF motif of pSer756-KT15 peptide with residue Ile96 of the Dok1 PTB domain. In addition, polar and ionic interactions appear to be sustained between residues Asn749/Glu93, Asp750/Arg55 of pSer756-KT15 peptide and Dok1 PTB domain (Fig. 6B).
Cell-based analyses of Dok1 binding to integrin β2 cytoplasmic tail. We examined the binding of Dok1 to integrin β 2 cytoplasmic tail in cells using FRET assay. The myeloid leukemia K562 cells that do not express endogenous β 2 integrins were transfected with expression plasmids of integrin α L and Dok1-CFP (CFP fused to the C-terminus of Dok-1) and integrin β 2-YFP wild-type or S745G/S756G or S745E/S756E. Since the integrin β 2 tail assumes extended conformation in complex with Dok1, we have chosen Ser to Gly mutations in the β 2 tail. The technical control groups were cells transfected with the indicated plasmids but the Dok1-CFP was substituted with empty CFP plasmid. The expression levels of integrin β 2-YFP constructs and Dok1-CFP were examined by immunoblotting (Fig. 7A). Transfected cells were subjected to YFP-photobleach FRET analyses. Significant FRET signal was detected in cells transfected with α L, Dok1-CFP and integrin β 2-YFP wild-type compared with control group cells transfected with α L, CFP and integrin β 2-YFP. Cells transfected with either integrin β 2 S745G/S756G or S745E/S756E had significantly reduced FRET signal. These data suggest that Dok1 binds to the integrin β 2 cytoplasmic tail and the association requires β 2 Ser745 and/or Ser756. However, poor FRET signal was detected when integrin β 2 Ser745 and Ser756 were both substituted with the acidic residue Glu to mimick phosphorylation. This is line with our NMR experiments that showed weak interactions   Supplementary Fig. 4). Ser to Glu or Asp substitution has been commonly used to interrogate the properties of phosphorylated Ser but the inability of Asp/Glu to mimic phosphorylated Ser has also been reported 25 . Hence, it is possible that the single carboxylate group on the side chain of Glu is insufficient to reconstitute the double negative charge and bulky phosphate moiety on phosphorylated Ser. We have attempted to detect the phosphorylation of integrin β 2 cytoplasmic tail on Ser745 and Ser756 by immunoprecipitation of wild-type integrin α Lβ 2 from transfected K562 cells and the T cell line Jurkat followed by immunoblotting with either an anti-phospho Ser antibody or anti-integrin β 2 phospho Ser745 obtained from commercial sources. However, we were unable to obtain definitive results possibly due to poor specificity and reactivity of these antibodies (data not shown). Dok-1 binds integrin β cytoplasmic tail and it is a negative regulator of integrin activation 13,14 . To date, however, there is a lack of cell-based evidence demonstrating Dok-1 dissociating from integrin β cytoplasmic tail in the presence of an activation signal. Hence we examined if chemokine-induced cell activation leads to the dissociation of Dok-1 from the integrin β 2 cytoplasmic tail. The chemokine RANTES and its receptor CCR5 were used in our study because RANTES has been reported to stimulate leukocyte adhesion and migration via activation of integrin α Lβ 2 26,27 . We generated stable K562 expressing Dok1-CFP by antibiotic selection. These cells were then transfected with expression plasmids of integrin α L and β 2-YFP and CCR5. The expression levels of these proteins were verified by flow cytometry analyses (Fig. 7C). Cells were then treated without or with RANTES and FRET analyses performed on these cells (Fig. 7D). FRET signal was significantly lower in cells treated with RANTES compared with untreated cells. These data suggest that Dok-1 dissociates from integrin β 2 cytoplasmic tail in the presence of RANTES/CCR5 activation signal.

Discussion
A large number of cytosolic proteins, including talins, kindlins, filamins, Dok1, 14-3-3, are known to interact with the β tail of integrins and they regulate the ligand-binding activity of these integrins [28][29][30] . Interactions with cognate proteins can be modulated by phosphorylation of the integrin β tail at specific sequence motifs 12,13,28 . The membrane proximal NxxY motif found in the β tail of a number of integrins (Table 1) interacts with the talin head domain, IgFLN domains of filamin and PTB domain of Dok1 2,28-30 . Phosphorylation of Tyr in the NPxY motif enables binding of the Dok1 PTB domain which negatively regulate the activation of integrins 12,13,31 . The PTB domain of Dok1 has also been shown to bind to a Tyr phosphorylated synthetic peptide derived from the RET receptor tyrosine kinase 19 . The membrane proximal NPxF motif in the integrin β 2 tail is not a phosphorylation site. Others have reported that Ser745 and Ser756, which are adjacent to the membrane proximal NPxF in the integrin β 2 tail, are amenable  to phosphorylation 17 . This led us to hypothesize that an alternate phosphorylation switch involving Ser745 and Ser756 regulates the interaction of Dok1 PTB with the integrin β 2 tail. Previous studies have demonstrated the functional consequences of Ser745 and Ser756 phosphorylation in integrin β 2 tail [32][33][34] . However, there has been no in depth study to our knowledge demonstrating an integrin β 2 tail binding protein that recognizes these phosphorylated serines. This study demonstrates that the PTB domain of Dok1 binds to integrin β 2 tail, which is dependent on the phosphorylation state of Ser745 and Ser756. NMR experiments using phosphorylated full-length integrin β 2 tail peptides showed that the affinity of Dok1 PTB was higher with integrin β 2 tail having phosphorylated Ser756 compared with phosphorylated Ser745. It is noteworthy that the integrin β 1A, β 1D and β 7 tails also contain a potential phosphorylatable Ser residue at the equivalent position of Ser756 in the β 2 tail (Table 1). pSer756-β 2 tail occupies the canonical binding pocket of Tyr phosphorylated peptides at the C-terminal long helix and β 5-β 6 strands of the PTB domain. The docked structure of Dok1 PTB with the integrin β 2 pSer756-KT15 peptide provides information on their relative orientation and the molecular interactions involved. In this model, the phosphate group of pSer756-KT15 is found in a close proximity to the side chains of residues Arg54 and Arg69 of the PTB domain that allows the formation of ionic interactions and/or hydrogen bonds. Indeed 31 P NMR spectra of pSer756-KT15 peptide showed significant chemical shift changes when in complex with the Dok1 PTB domain (Fig. 4C) and 15 N-1 H HSQC spectra of the PTB domain showed high chemical shift perturbations of these corresponding Arg residues upon addition of the phosphorylated integrin β 2 tail peptide (Fig. 1D). The docked structure of the phosphorylated integrin β 2 peptide/PTB domain suggests packing interactions between residues Leu731 and Phe732 in the NPLF motif and Trp725 of the β 2 tail with residue Ile96 and residues Arg58 and Phe89 of the PTB domain, respectively (Fig. 6). STD-NMR studies demonstrated that the aromatic ring of residues Trp725 and Phe732, and the aliphatic side chains of residues Leu731 and Thr758 are in close proximity to the PTB domain of Dok1 (Fig. 4D). Alongside, Ile96, Arg58 and Phe89 of the PTB domain demarcated chemical shift perturbations when in complex with the phosphorylated β 2 tail (Fig. 1D). The binding affinity of the phosphorylated integrin β 2 tail with the PTB domain of Dok1 is rather low. We hypothesize that in a physiological system, this binding affinity can be enhanced by their membrane localization. Indeed, the PH domain of Dok1 protein has been shown to be important for Dok1 targeting to the membrane 35 . Interactions between integrin tails and their binding partners are found to be of higher affinity when the integrin tails are anchored in lipids 36,37 . It has also been reported that membrane interactions stabilize the binding of the head domain of talin to integrin tails 38,39 . Although we have not examined the interaction of di-phosphorylated (pSer745 and pSer756) integrin β 2 tail peptide with Dok1 PTB domain in this study, we conjecture that the binding affinity may be further enhanced when both Ser are phosphorylated. This will be investigated in future work. In a cellular system, the simultaneous phosphorylation of both Ser745 and Ser756 could have a synergistic effect on the recruitment of Dok1 to the integrin β 2 tail. In this regard, phosphorylation of Ser745 or Ser756 or both under different conditions may allow fine-tune regulation of Dok1 binding to integrin β 2 tail, which may in turn modulate talin activation of β 2 integrins. In addition to biophysical analyses, we showed using a cell-based FRET system that the interaction between integrin β 2 tail and Dok1 is dependent on Ser745 and Ser756, and cellular activation by chemokine stimulation triggers the dissociation of Dok1 from the integrin β 2 tail. We were unable to demonstrate the phosphorylation status of integrin β 2 Ser745 and Ser756 in cells because antibodies of high specificity and reactivity are unavailable to us. However, others have shown that these residues can be phosphorylated 17,34 .
An increasing number of cytosolic proteins have been identified to bind integrin tails. These interactions are regulated by many factors including the phosphorylation of integrin β tails. Given that there are sequence variations amongst the integrin β tails, there can be slight but important differences in which phosphorylation regulates these interactions. In this study, we have identified and characterized an alternate phosphorylation switch in the integrin β 2 tail that regulates its interaction with Dok1. These findings will be important for future studies investigating the underlying mechanisms of β 2 integrin-mediated immune responses.

Materials and Methods
Plasmids. Full-length Dok1 cDNA from a human leukocyte cDNA library (Clontech Laboratories, Mountain View, CA) was PCR amplified using relevant primers and cloned into the plasmid pcDNA3.1(+ ) (Invitrogen, Carlsbad, CA). Dok1 was PCR sub-cloned into the pECFP-N1 expression plasmid (Clontech) for FRET experiments. The pcDNA3 expression plasmids containing human integrin α L or integrin β 2-YFP (YFP fused to the C-terminal of β 2 cytoplasmic tail) have been reported previously [40][41][42] . The integrin β 2-YFP S745G/S756G expression construct was generated by mutagenesis using the Quikchange II site-directed mutagenesis kit (Stratagene, La Jolla, CA) and relevant primers. The human CCR5 pcDNA3 expression plasmid was a kind gift from Dr. R.W. Doms (MD) (University of Pennsylvania, PA).
Protein expression and purification. Human Dok1 PTB domain (Q154 -G256; Swiss-Prot Q99704, number as Q1-G103) was sub-cloned into the pET14b vector with an N-terminal six His-tag. Escherichia coli Rosetta cells were transformed with the expression plasmid and grown at 37 °C either in LB or M9 medium containing [ 15 N] ammonium chloride (Cambridge Isotope Laboratories). Expression of recombinant protein was induced by adding 1 mM IPTG to cells (OD 600 0.6-0.7) followed by incubation for 6-12 h at 16 °C. Cells were centrifuged and re-suspended in Buffer A (20 mM Tris-HCl buffer, pH 8.0) followed by affinity purification on a Nickel-NTA column (Qiagen). His-tagged Dok1 PTB domain was eluted in Buffer A containing 500 mM imidazole and dialyzed against Buffer B (50 mM sodium phosphate buffer (pH 6.0) containing 100 mM NaCl and 2 mM DTT). Protein samples were further purified by gel filtration in Buffer B on a Hiload Superdex 75 16/26 GL preparative column that was connected to an AKTA FPLC UPC-900 system (GE Healthcare UK Ltd., England). Proteins were eluted at a flow rate of 0.5 ml/min and monitored by absorbance at λ = 276 nm. Synthetic peptides. Synthetic peptides (purity >95%) of full-length β 2 tail, phosphorylated full-length β 2 (pSer745-β 2 and pSer756β 2), phosphorylated peptide fragment of β 2 (pSer756-KT15), unphosphorylated KT15, and KT18EE (  trNOESY and STD NMR experiments. pS756-KT15 peptide at a concentration of 0.6 mM was titrated against different concentrations of Dok1 PTB domain. All signals of the pS756-KT15 peptide were broadened at protein:peptide molar ratio of ~1:30. Thus, 2D trNOESY experiments were performed with a mixing time of 150 ms. 2D TOCSY spectra were acquired at 50 ms mixing time. The spectral width was 12 ppm in both dimensions. Spectra were processed using the Topspin program suite followed by analysis using Sparky. For STD-NMR, 0.6 mM of pS756-KT15 or unphosphorylated KT15 peptide was dissolved in 600 μ l of D 2 O (pH 6.2). The Dok1 PTB domain (20 μ M) was buffer exchanged in 99.9% D 2 O. STD-NMR spectra were recorded with 512 scans and selective saturation of protein resonances at − 0.7 ppm (40 ppm for reference spectra) using a series of 40 Gaussian-shaped pulses (49 ms, 1 ms delay between pulses), for a total saturation time of 2 s. 31 P NMR experiments. 31 P NMR spectra of pS756-KT15 were recorded on a Bruker DRX 400 spectrometer at 298 K. Data acquisition and processing were performed with the Topspin software (BRUKER) suite. 1-D 31 P NMR spectra of pS756-KT15 at 0.2 mM in water (pH 6.2) were recorded in the presence of different concentrations of Dok1 PTB domain. The Dok1 PTB stock solution used was prepared in unbuffered water (pH 6.2). The pH of each sample was adjusted to 6.2 after addition of protein solutions.
Docking of Dok1 PTB domain with pS756-KT15 peptide. Based on the conformation of NAK peptide (pdb accession code, 1DDM.pdb), a structural model of pS756-KT15 peptide was built using INSIGHTII (Accelrys Inc.) software. The model structure of pS756KT15 was energy minimized by conjugate gradient energy minimization protocol. The model of pS756-KT15/Dok1 complex was derived using HADDOCK program (http://haddock.science.uu.nl/) based om chemical shift changes of the PTB domain. The model structure was validated using ADIT Validation Server from the Protein Data Bank (PDB) and coordinates of the complex are provided in supporting materials.