Two Amino Acid Residues Confer Different Binding Affinities of Abelson Family Kinase Src Homology 2 Domains for Phosphorylated Cortactin*

Background: Abl family kinases bind different targets despite highly similar sequences. Results: Two residues in the Src homology (SH) 2 domain regulate binding to phosphorylated cortactin and modulate cell protrusion. Conclusion: The Arg SH2 domain binds with higher affinity than the Abl SH2 domain to phosphorylated cortactin. Significance: Slight sequence changes can cause affinity differences, leading to important functional changes in cellular interactions. The closely related Abl family kinases, Arg and Abl, play important non-redundant roles in the regulation of cell morphogenesis and motility. Despite similar N-terminal sequences, Arg and Abl interact with different substrates and binding partners with varying affinities. This selectivity may be due to slight differences in amino acid sequence leading to differential interactions with target proteins. We report that the Arg Src homology (SH) 2 domain binds two specific phosphotyrosines on cortactin, a known Abl/Arg substrate, with over 10-fold higher affinity than the Abl SH2 domain. We show that this significant affinity difference is due to the substitution of arginine 161 and serine 187 in Abl to leucine 207 and threonine 233 in Arg, respectively. We constructed Abl SH2 domains with R161L and S187T mutations alone and in combination and find that these substitutions are sufficient to convert the low affinity Abl SH2 domain to a higher affinity “Arg-like” SH2 domain in binding to a phospho-cortactin peptide. We crystallized the Arg SH2 domain for structural comparison to existing crystal structures of the Abl SH2 domain. We show that these two residues are important determinants of Arg and Abl SH2 domain binding specificity. Finally, we expressed Arg containing an “Abl-like” low affinity mutant Arg SH2 domain (L207R/T233S) and find that this mutant, although properly localized to the cell periphery, does not support wild type levels of cell edge protrusion. Together, these observations indicate that these two amino acid positions confer different binding affinities and cellular functions on the distinct Abl family kinases.

The closely related Abl family kinases, Arg and Abl, play important non-redundant roles in the regulation of cell morphogenesis and motility. Despite similar N-terminal sequences, Arg and Abl interact with different substrates and binding partners with varying affinities. This selectivity may be due to slight differences in amino acid sequence leading to differential interactions with target proteins. We report that the Arg Src homology (SH) 2 domain binds two specific phosphotyrosines on cortactin, a known Abl/Arg substrate, with over 10-fold higher affinity than the Abl SH2 domain. We show that this significant affinity difference is due to the substitution of arginine 161 and serine 187 in Abl to leucine 207 and threonine 233 in Arg, respectively. We constructed Abl SH2 domains with R161L and S187T mutations alone and in combination and find that these substitutions are sufficient to convert the low affinity Abl SH2 domain to a higher affinity "Arg-like" SH2 domain in binding to a phospho-cortactin peptide. We crystallized the Arg SH2 domain for structural comparison to existing crystal structures of the Abl SH2 domain. We show that these two residues are important determinants of Arg and Abl SH2 domain binding specificity. Finally, we expressed Arg containing an "Abl-like" low affinity mutant Arg SH2 domain (L207R/T233S) and find that this mutant, although properly localized to the cell periphery, does not support wild type levels of cell edge protrusion. Together, these observations indicate that these two amino acid positions confer different binding affinities and cellular functions on the distinct Abl family kinases.
Abelson (Abl) 7 family non-receptor tyrosine kinases are important regulators of actin dynamics and coordinate changes in cell morphology and migration (1). The N-terminal halves of Abl and the Abl-related gene (Arg), containing their Src homology 3 (SH3), Src homology 2 (SH2), and kinase domains, are nearly 90% identical, whereas their C-terminal extensions, which contain cytoskeletal-binding domains and scaffolding functions, are much less conserved (1,2). Despite sequence similarities and known roles in actin cytoskeletal regulation, Abl and Arg regulate differential zones of actomyosin contractility in fibroblasts (3). Moreover, Abl and Arg have differential roles in neuronal dendrite maintenance and in the support of invadopodial function in invasive breast cancer cells (4,5).
In addition to their differences in expression, subcellular localization, and function, Arg and Abl interact with different substrates in vivo (3, 6 -9). For example, we have previously shown that whereas both Abl and Arg phosphorylate the actin polymerization regulator cortactin in vitro, Arg alone interacts with cortactin via a set of scaffolding interactions that are essential for both proteins to promote cell edge protrusion during cell adhesion (10,11). Furthermore, Arg, but not Abl, is critical for cortactin phosphorylation during actin-based protrusion in breast cancer cells (5). This selectivity can be explained in part by the high-affinity binding of the cortactin SH3 domain to an extended Pro-X-X-Pro-X-X-Pro motif that occurs in Arg but not Abl (10). A recent crystal structure suggests that the cortactin SH3 domain engages the first three of these prolines along an extended type II polyproline helix (12).
During cell adhesion, cortactin SH3 domain binding to Arg is essential for subsequent phosphorylation of cortactin by Arg at one or more of three previously identified tyrosine phos-phory10, 13,14). This phosphorylation creates a binding site for the Arg SH2 domain, a discrete second interaction interface between the two proteins, although the exact phosphotyrosines required for binding were previously unknown (15). This finding raised the question of whether the Arg SH2 domain might also confer additional specificity to the Arg-cortactin interactions. SH2 domains bind phosphotyrosines with a high degree of selectivity, which is dictated by surrounding residues. In particular, three residues C-terminal to the phosphotyrosine determine SH2 domain binding specificity (16,17). Within the SH2 domain itself, there are three binding pockets that confer specificity for each of these binding target residues as well as several loops, which govern binding pocket accessibility (18,19).
We report here that of the three previously described cortactin tyrosine phosphorylation sites (Tyr 421 , Tyr 466 , Tyr 482 ), only peptides corresponding to the phosphorylated Tyr 421 and Tyr 466 sites bind with significant affinities to the Arg SH2 domain. We also find that despite nearly 90% domain sequence identity, the Arg SH2 domain binds the phosphorylated cortactin Tyr 421 peptide with over 10-fold greater affinity than the Abl SH2 domain. We show that this binding specificity is due to two residues that differ between Abl and Arg, arginine 161 and serine 187 in Abl, which correspond to leucine 207 and threonine 233 in Arg, respectively ( Fig. 2A). We provide the first crystal structure of the Arg SH2 domain and show that Leu 207 is located near the P ϩ 1 binding pocket and Thr 233 is on the EF loop previously shown to mediate access to the P ϩ 3 binding pocket (18). Finally, we show that although these residues do not affect Arg localization to the cell periphery, they are critical for mediating actin-based cell edge protrusion in fibroblasts.
Protein Purification for Crystallization-Initial protein expression and purification were carried out in Escherichia coli BL21(DE3) cells and induced at 20°C overnight with 0.1 mM isopropyl ␤-D-1-thiogalactopyranoside. After induction, the cells were pelleted, resuspended, and lysed using a freeze-thaw protocol and sonication with the presence of 1 mg/ml of lysozyme. Following sonication and centrifugation (20,000 ϫ g) at 4°C for 30 min, the resulting supernatant was incubated with glutathione-Sepharose 4B beads (GE Healthcare) for 4 h at 4°C. 20 bead volumes of PBS buffer were then used to wash the resin before being incubated with PreScission protease (GE Healthcare) at 4°C overnight to remove the GST tag. The Arg SH2 protein was further purified on a Superdex 75 column in 20 mM Hepes (pH 7.5) buffer containing 100 mM NaCl, 1 mM EDTA, 1 mM DTT. The protein was concentrated to 20 mg/ml for crystallization.
Peptide Synthesis and 5-Carboxyfluorescein Labeling-Peptides corresponding to phosphorylated tyrosine sites 421 (SSPIpYEDAA), 466 (SEPVpYETTE), or 482 (EDDTpYDGYE) of mouse cortactin were synthesized on a 30-mol scale with a Liberty 12-channel microwave synthesizer (CEM Corp., Matthews, NC) using standard Fmoc chemistry. Following removal of the final Fmoc protecting group, the resin was washed with dimethylformamide and methylene chloride alternatively for a total of 16 washes and dried for 20 min under nitrogen gas. The peptides were N terminally labeled overnight with a mixture comprised of 5-carboxyfluorescein, succinimidyl ester (8 mg, 0.017 mmol), and diisopropylethylamine (24 l, 17.8 mg, 0.138 mmol) in 1.5 ml of dimethylformamide.
Labeled peptide was treated with a cleavage mixture of 2.5% (v/v) 3,6-dioxa-1,8-octanedithiol, 2.5% (v/v) H 2 O, 2.5% (v/v) triisopropylsilyl in TFA (50% power at 400 W maximum, 38°C ramp for 2 min, hold for 30 min; followed by treatment with fresh mixture at 50% power, 400 W maximum, 38°C, ramp 2 min, hold 5 min). Crude peptide was lyophilized and reconstituted in acetonitrile/water (1:1). Synthesis efficiency was assessed by MALDI-TOF analysis of the crude reaction mixture followed by purification to homogeneity by reverse phase HPLC. A second round of MALDI-TOF reconfirmed peak sample homogeneity. Following purification, peptide was lyophilized and stored in the dark at Ϫ20°C.
Fluorescence Polarization Measurements-SH2 domains were dialyzed into binding buffer (50 mM Hepes, pH 7.25, 150 mM NaCl, 0.01% Nonidet P-40, 5% glycerol) and a 30-l volume serially diluted into pre-chilled 384-well low flange black flat bottom non-binding microplates (Corning) while on ice. 5 l of 35 nM fluorescent peptide dissolved in binding buffer was added to each well (final concentration 5 nM), mixed 5 times by pipetting, and plates were incubated at 4°C for 40 min. Fluorescence polarization experiments were performed with an Analyst AD (Molecular Devices, Sunnyvale, CA) spectrofluorimeter. Each well was excited using 485 nm light and emission was read at 530 nm. Fluorescence polarization was measured 1 mm from the bottom of each well with an integration time of 560 ms. All experiments were conducted in triplicate. The change in fluorescence polarization was normalized such that maximum change for each condition (B max ) was set at an arbitrary value of 100 for ease of comparison ( Figs. 1 and 2). Normalized values were graphed and fit to a single binding site hyperbola using GraphPad Prism.
Crystallization and Structure Determination-Arg SH2 domain crystals were grown in conditions containing 15-17% PEG 4000, 0.2 M sodium acetate, and 100 mM Tris-HCl, pH 8.5. Crystals were grown within 3 days at room temperature by hanging drop vapor diffusion. Before data collection, the crystals were flash frozen in liquid nitrogen using the crystallization conditions supplemented with 20% PEG 400. Crystallographic data were collected at beamline X6A of the National Synchro-tron Light Source (Table 1), and data were processed using HKL2000 (20).
The structure of the Arg SH2 domain was determined by molecular replacement using the program PHASER (21). The SH2 domain from Abl (Protein Data Bank code 3K2M (22)) was used as a search model and yielded a translation Z-score of 17.2. Automated model building using ARP/wARP (23) built 95 residues, thus avoiding potential problems with model bias. Manual model building and refinement were then conducted using COOT (24) and Refmac5 (25) from the CCP4 suite, respectively. Restrained refinement with anisotropic B-factors and automatic water placement with ARP_waters (26) were used in later rounds of refinement. A total of 100 water molecules were added during refinement. The final structure was validated using PROCHECK (27) and MolProbity (28) and the secondary structure was assigned using the computer program, DSSP (29). All residues fall within the favored or allowed regions of the Ramachandran plot. Arg SH2 maintains a typical SH2-fold with 2 major ␣-helices (␣A and ␣B) and 7 ␤-strands (␤A to ␤G). The structure is deposited in the Protein Data Bank under accession code 4EIH.
Cell Culture and Retroviral Expression-Wild type and arg Ϫ/Ϫ fibroblasts were maintained as previously described (30). EYFP, Arg-EYFP, and Arg L207R/T233S-EYFP were expressed via retroviral expression and 1.0 g/ml of puromycin selection as previously described (10). Expression levels of wild type and retrovirally expressed Arg proteins were determined by immunoblotting 75 g of cell lysate as determined with a BCA kit (Pierce) with a mouse anti-Arg antibody and control mouse monoclonal antibody against HSP70 (3A3) (Santa Cruz). Expression levels were quantified using Quantity One software (Bio-Rad).
Immunoprecipitation and Western Blotting-Fibroblasts were serum starved for 1 h then plated on 10 g/ml of bovine fibronectin-coated plates (Sigma), blocked with 1% BSA, and allowed to adhere for 45 min. Cells were washed with PBS and lysed in 50 mM Hepes, pH 7.5, 1% Nonidet P-40, 150 mM KCl, 5% glycerol, 0.5 mM metavanadate, 0.5 mM orthovanadate, 2 mM NaF, and complete protease inhibitors. 0.75 mg/ml of soluble lysate was incubated with protein A/G beads and anti-cortactin 4F11 antibody for 24 h at 4°C. Supernatant and precipitant samples were immunoblotted with antiphosphotyrosine 4G10 antibody and anti-cortactin 4F11 antibody. Phosphorylation levels were quantified and normalized to total cortactin levels using Quantity One software (Bio-Rad).
Immunofluorescence Microscopy-Fibroblasts were plated on 10 g/ml of bovine fibronectin-coated coverslips (Sigma), blocked with 1% BSA, and allowed to adhere for 30 min. Cells were fixed with 4% PFA for 20 min at room temperature and permeabilized with 0.5% Triton X-100 for 10 min. Cells were then stained with a mouse anti-GFP monoclonal antibody (Rockland), which cross-reacts with EYFP, followed with Alexa Fluor 647 goat anti-mouse antibody (Invitrogen), as well as Alexa Fluor 488 phalloidin (Invitrogen). Cells were imaged on a model TE2000-S Nikon microscope with a ϫ20 or 100 objective with NIS Elements (Nikon).
To control for possible bleed-through of the Arg/Arg mutant-YFP signal into the actin channel, arg Ϫ/Ϫ fibroblasts expressing EYFP were stained only with antibodies to EYFP and Alexa 647 goat anti-mouse secondary antibody or only with phalloidin-Alexa 488 alone and were imaged at both 501 and 654 nm. No bleed-through of the Arg/Arg mutant-YFP signal was observed in the 501 channel, and no actin staining was observed in the 654 nm channel. These experiments indicate that the EYFP signal from the expressed protein is disrupted during fixation and processing.
Quantification of Cell Circularity-Cells stained for phalloidin and imaged at ϫ100 were traced peripherally in ImageJ. Circularity was quantified using the analyze shape descriptors function in the ImageJ Software package, which can be described as a function of cell area and cell perimeter.

Adhesion-dependent Cell Edge Protrusion Assays and
Kymography-Adhesion-dependent fibroblast cell edge protrusion assays were performed 30 min after cells were plated on 10 g/ml of bovine fibronectin (Sigma)-coated MatTek dishes (MatTek Corp.) and imaged as described previously (10). Kymographic analysis of cell edge protrusion was performed as described previously using ImageJ software and GraphPad Prism (10,30).
Measurements of Cell Spreading-Fibroblasts were plated on 10 g/ml of bovine fibronectin (Sigma)-coated MatTek dishes (MatTek Corp.). Time-lapse images of spreading cells were recorded for 30 min. The cell area was measured at 0, 10, 20, and 30 min by peripherally tracing and quantifying each cell in ImageJ.
Statistics-Unless otherwise noted, grouped samples were subjected to two-way analysis of variance followed by post-hoc Student's t tests where described (* ϭ p Ͻ 0.05; ** ϭ p Ͻ 0.01; *** ϭ p Ͻ 0.001; **** ϭ p Ͻ 0.0001). All graphs display mean Ϯ S.E.  1D). We found that the Arg SH2 domain binds with highest affinity to the Tyr(P) 421 peptide, binds with a 2-fold lower affinity to Tyr(P) 466 peptide, and does not bind detectably to Tyr(P) 482 peptide (Fig. 1, A and C). The Abl SH2 domain showed a similar trend in binding affinity for the three phosphorylated cortactin peptides (Fig. 1, B and C). Neither SH2 domain binds unphosphorylated cortactin peptides with any detectable affinity.

Arg and Abl Bind Cortactin
Arg and Abl SH2 Domains Differ in Only a Few Non-conserved Amino Acids-Despite nearly identical sequences, the Arg SH2 domain shows a greater than 10-fold higher affinity in comparison to the Abl SH2 domain for the cortactin Tyr(P) 421 peptide (Fig. 1). ClustalW was used to perform sequence align-ment of Arg and Abl SH2 domains. These two SH2 domain sequences are 90% identical, with two non-conserved residues, two semi-conserved residues, and six conserved residues ( Fig.  2A). Of these 10 residues, three are located in regions within or near the binding pocket region of the SH2 domain (Fig. 3A). These residues are Leu 207 , Thr 233 , and Ala 234 in Arg, which correspond to Arg 161 , Ser 187 , and Ser 188 in Abl. Leu 190 in Arg and the corresponding residue in Abl, Gly 144 , are also nonconserved but are located outside of the binding pocket and we therefore predicted this substitution would have little effect on the binding affinity.
Two Amino Acid Residues in the Arg and Abl SH2 Domains Confer Different Binding Affinities for a Phosphorylated Cortactin Peptide-We measured the binding affinity of Abl SH2 mutants with substitutions at the residues that differ from the Arg SH2 domain for the cortactin Tyr(P) 421 peptide. The Abl SH2 single mutants, G144L and S188A, showed no change in affinity from that of wild type Abl SH2 domain (Fig. 2, C, D, G, and I). However, both Abl SH2 R161L and S187T showed intermediate affinities for the Tyr(P) 421 peptide (Fig. 2, E, F, and I). In light of this partial increase in affinity, we hypothesized that the double mutation (R161L/S187T) of the Abl SH2 domain may confer the strong affinity observed in the wild type Arg SH2 domain. As hypothesized, the Abl double mutant (R161L/ S187T) SH2 domain is able to bind the cortactin Tyr(P) 421 peptide with similar affinity to that observed for the wild type Arg SH2 domain (Fig. 2, B, H, and I). These data suggest that these two residues determine the relative binding affinities of Arg and Abl SH2 domains for the Tyr(P) 421 cortactin peptide.
Structural Analysis of Arg SH2 Domain-The structure of the Arg SH2 domain has never been solved crystallographically. In light of our affinity measurements (Fig. 2) we wondered whether there were structural explanations for both Arg SH2 domain specificity for Tyr(P) 421 and Tyr(P) 466 over Tyr(P) 482 , and the higher affinity of Arg SH2 domain for cortactin when compared with Abl SH2 domain. We therefore conducted a structural analysis of the Arg SH2 domain. We first purified the Arg SH2 domain and crystallized this domain in space group C222 1 . We then determined the structure of the Arg SH2 domain to 1.2-Å resolution (Fig. 3A, Table 1). This allowed us to directly compare the Arg SH2 domain to the Abl SH2 domain (22,(31)(32)(33)(34) and other previously determined type IA SH2 domain structures (35)(36)(37)(38)(39).
First, we compared the structure of the Arg SH2 domain with previously determined Abl crystal structures as well as an NMR structure of Arg SH2 domain (PDB codes 1OPK, 1OPL, 1AB2, 2ABL, 2ECD, 2FO0, and 3K2M) (22,(31)(32)(33)(34). The crystal structure of the Arg SH2 domain shows root mean square deviations for C␣ atoms with these structures ranging from 0.75 to 1.55 Å (Fig. 3B). Overall these structures, including the NMR structure of the Arg SH2 domain, are extremely similar, however, there is distinct conformational flexibility for the BC loop, which is observed in conformations that vary by as much as 9 Å.
We next compared the Arg SH2 domain to other type 1A SH2 domain structures that have been determined in complex with phosphotyrosine peptides. These structures (e.g. for Lck (35,40) and Src (39)) show that the peptide specificity is determined at the P ϩ 3 location by a Tyr(P)[Ϫ][Ϫ]⌿ motif, where ⌿ is a hydrophobic residue. This binding site is maintained in the Arg SH2 domain by formation of the phosphotyrosine binding pocket by residues Arg 180 , Arg 198 , Gln 206 , Ser 208 , Tyr 218 , His 219 , Tyr 220 , and Arg 221 , and formation of the P ϩ 3 hydrophobic binding pocket by residues Val 232 , Thr 233 , Gly 254 , and Leu 255 . We conducted structural modeling by superposition of the Arg SH2 domain with the structure of Lck bound to an 11-residue phosphopeptide (EPQpYEEIPIYL) derived from the hamster polyoma middle-T antigen (PDB 1LCJ) (35) (Fig. 3C). The structural superposition clearly shows that a large residue at the P ϩ 3 location would be incompatible with binding. This correlates well with the undetectable binding for the cortactin Tyr(P) 482 peptide, which has a tyrosine residue in the P ϩ 3 location as compared with the Tyr(P) 421 and Tyr(P) 466 peptides (Fig. 1), which have alanine and threonine residues, respectively.
We then analyzed the locations of the residues that diverge between the Arg and Abl SH2 domains. We found that Arg  JULY 11, 2014 • VOLUME 289 • NUMBER 28

JOURNAL OF BIOLOGICAL CHEMISTRY 19707
Leu 190 /Abl Gly 144 residues are distal from the peptide binding cleft correlating well with our data showing that mutagenesis of these residues does not alter peptide binding specificity. Thr 233 is located in the Arg kinase EF loop and is proximal to the P ϩ 3 peptide binding location, consistent with our data showing that the amino acid identity at this position affects peptide binding affinity. Notably, Leu 207 is part of the BC loop, but points away from the phosphotyrosine binding site. In structures of the Abl SH2 domain, the corresponding residue, Arg 161 , can make a salt bridge to Glu 153 , potentially impacting the conformation of the BC loop. However, in the Arg SH2 domain structure the presence of Leu 207 in this location results in a loss of the salt bridge and an orientation change in the side chain of the corresponding Glu 199 (Fig. 3D). We hypothesize that for the Abl double mutant (R161L/S187T), the loss of this salt bridge formation and potential alteration in BC loop conformation combined with an altered P ϩ 3 pocket work in concert to allow increased binding affinity toward phosphorylated cortactin. Wild Type Arg-EYFP and Arg L207R/T233S-EYFP Both Localize to the Cell Periphery-Wild type fibroblasts display dynamic cell edge protrusions as they adhere to fibronectincoated surfaces (10,30,41). We have shown that the Arg SH2 domain binding to phosphorylated cortactin is critical for this process (10). To test if this high affinity interaction supported by residues Leu 207 and Thr 233 is relevant in cells we retrovirally expressed wild type Arg-EYFP or Arg L207R/ T233S-EYFP in mouse arg Ϫ/Ϫ fibroblasts (Fig. 4, A-C). Cells fixed and stained for actin, and EYFP reveal that localization of Arg L207R/T233S-EYFP is similar to that of wild type Arg-EYFP in arg Ϫ/Ϫ fibroblasts and is enriched at the cell periphery (Fig. 4E), suggesting that these residues do not impact localization within the cell. Control experiments (described under "Experimental Procedures") indicated that channel bleed-through of the Arg/Arg mutant and actin staining was not an issue (Fig. 4F).
Overall morphology, which we quantified as cell circularity, shows similarity in cell shape between wild type fibroblasts and arg Ϫ/Ϫ fibroblasts expressing wild type Arg-EYFP. Cells expressing mutant Arg L207R/T233S-EYFP appear similar in circularity to arg Ϫ/Ϫ fibroblasts and arg Ϫ/Ϫ fibroblasts expressing EYFP, suggesting that despite its proper localization, the mutant Arg does not support normal cell morphology (Fig. 4D).
Arg Residues Leu 207 and Thr 233 Support Wild Type Level Cell Edge Protrusion Fibroblasts-Wild type fibroblasts undergo dynamic cell edge protrusions as they adhere and spread on fibronectin-coated glass and Arg is critical for this behavior (10,30,41). To test if residues Leu 207 and Thr 233 are required to support adhesion-dependent cell edge protrusion, we recorded time-lapse images of wild type fibroblasts and arg Ϫ/Ϫ fibroblasts expressing EYFP, wild type Arg-EYFP, and Arg L207R/ T233S-EYFP adhering to fibronectin-coated coverslips (Fig.  5). We quantified cell edge protrusive behavior using kymography as previously described (30) and find that wild type fibroblasts and arg Ϫ/Ϫ fibroblasts expressing wild type Arg-EYFP exhibit a high number of protrusions and retractions (Fig. 5, A, C, E, and F). In contrast, arg Ϫ/Ϫ fibroblasts expressing EYFP alone or expressing mutant Arg L207R/ T233S-EYFP show a significant reduction in the number of protrusions and retractions (Fig. 5, B, D, E, and F), confirming that these residues are important for cell edge protrusion in fibroblasts. We do not see a significant difference in the  (22,(31)(32)(33)(34). The superposition was conducted using Topp (25). C and D, model of phosphotyrosine peptide binding to Arg SH2 domain. The model was generated by superposition of Arg SH2 with the Lck SH2-peptide (EPQpYEEIPIYL) complex (35). C, shows only the Arg SH2 domain and the location of the superposed peptide (in yellow). Structural diagrams generated using CCP4MG (48). D, shows an orientation rotated 90°with Arg residues Leu 207 , Thr 233 , and Glu 199 shown in stick format.

Two Amino Acids Regulate Abl Family Kinase SH2 Domain Function
overall spreading area of cells, indicating that whereas dynamic cell edge protrusions are disrupted, overall area of the spread is not affected (Fig. 5H). Efficient phosphorylation of cortactin by Arg relies on the Arg PXXP-cortactin SH3 domain interaction as well as an intact Arg kinase domain. These interactions should not be compromised in the Arg L207R/T233S-EYFP mutant. We immunoprecipitated cortactin from arg Ϫ/Ϫ fibroblasts, or arg Ϫ/Ϫ fibroblasts expressing EYFP, Arg-EYFP, or mutant Arg L207R/T233S-EYFP and immunoblotted for phosphotyrosine. Cortactin phosphorylation levels are reduced in arg Ϫ/Ϫ ϩ EYFP fibroblasts, but arg Ϫ/Ϫ fibroblasts were reconstituted with Arg-YFP and Arg L207R/T233S-EYFP exhibited similar levels of cortactin phosphorylation (Fig. 5G).
We also hypothesized that a chimera of Arg containing the Abl R161L/S187T SH2 domain would rescue cell edge protrusion in a manner similar to wild type Arg. Unfortunately, these constructs did not express well in fibroblasts despite repeated attempts (data not shown).

DISCUSSION
We show that Arg and Abl SH2 domains bind with high affinity to phosphorylated cortactin residues Tyr 421 and Tyr 466 , but not Tyr 482 . Despite nearly identical sequences, we find that the Arg and Abl SH2 domains bind phosphorylated cortactin with a nearly 10-fold difference in affinity. Interestingly, we find that two specific residues, Arg 161 and Ser 187 in Abl and Leu 207 and Thr 233 in Arg, mediate the difference in affinity of these SH2 domains for phosphorylated cortactin. We provide a crystal structure of the Arg SH2 domain, which shows that these two residues are located near the binding pocket. Finally, we demonstrate that these two SH2 domain residues are functionally important to support cell edge protrusion. Phosphorylation of cortactin residues Tyr 421 and Tyr 466 is required for actin polymerization in breast cancer cell invadopodia, whereas phosphorylation of the previously identified Tyr 482 is not (42). Furthermore, we previously showed that Arg phosphorylates cortactin on two of these tyrosines, Tyr 421 and Tyr 466 (10,11). We support these observations by demonstrating that Tyr(P) 421 and Tyr(P) 466 , but not Tyr(P) 482 , interact with Arg and Abl SH2 domains. This is likely due to the inability of the Arg and Abl SH2 domain binding pocket to accommodate the bulky tyrosine residue found in the P ϩ 3 site of the Tyr(P) 482 peptide. Similarly, only Tyr 421 and Tyr 466 are con-served in and important for regulation of the cortactin homolog HS1 (43).
We show that Arg and Abl SH2 domains bind phosphorylated cortactin with dramatically different affinities. Through biochemical and structural approaches, we identified Arg residues Leu 207 and Thr 233 as key mediators of this high affinity interaction. Although the difference in affinity does not impact Arg localization, it is important for proper cell edge protrusion. Arg localization to the cell periphery is conferred by its C-terminal cytoskeletal binding domains and its SH2 domain does not impact subcellular localization (44). Our data agree with these previous studies. These findings further reinforce the fact that despite sequence similarity, Arg and Abl have unique roles within the cell. Evolving interaction specificity through affinity is one mechanism by which kinases target specific substrates, and swapping SH2 domains between kinases has been shown to Two Amino Acids Regulate Abl Family Kinase SH2 Domain Function JULY 11, 2014 • VOLUME 289 • NUMBER 28 alter the substrate profile from the wild type kinase (45). Our research suggests that this difference in affinity between Arg and Abl SH2 domains contributes to the different roles these two kinases play in the cell.
SH2 domains also play an important role in regulating processive phosphorylation of substrates with multiple phosphorylation sites (46) and SH2 domain binding preference has been shown to correlate well with phosphorylation site preference of the associated kinase (45,47). It is possible that the Arg SH2 domain specificity for Tyr(P) 421 in cortactin potentiates efficient phosphorylation of a subsequent tyrosine residue, such as Tyr 466 , leading to increased actin polymerization in invadopodia. This increased cortactin phosphorylation would lead to the up-regulation of actin polymerization in invadopodia. This mechanism may contribute to the fact that Arg, but not Abl, localizes to invadopodia in breast cancer cells and is essential for cortactin-mediated actin polymerization at these sites (5).
Previous studies show that selective interactions between Arg and cortactin underlie the ability of Arg to serve as a scaffold to support cortactin function during cell edge protrusion. Loss of the Arg SH2 domain-cortactin phosphotyrosine interaction leads to a reduction in cell edge protrusion (10). In relationship to these studies, we show that affinity of these interactions is an important determinant of whether or not they are able to perform their intended function in the cell. By expressing the mutant Arg L207R/T233S-EYFP in cells, we decrease the affinity between the Arg SH2 domain and phosphorylated cortactin and this decrease in affinity results in a failure of Arg to support wild type levels of cell edge protrusion. We also find that the circularity of cells expressing Arg L207R/T233S-EYFP is altered compared with wild type cells, reflecting a disruption of overall cytoskeletal structure and cell shape. Together, our results demonstrate the importance of two residues in the binding pocket of Arg and Abl SH2 domains for high affinity binding to phosphorylated cortactin and further highlight the non-redundant function of these two related kinases in cellular processes.