HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 276, Issue 20, 17199-17205, May 18, 2001

This Article Abstract Full Text (PDF) All Versions of this Article: 276/20/17199    most recent M011185200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Arold, S. T. Articles by Noble, M. E. M. Search for Related Content PubMed PubMed Citation Articles by Arold, S. T. Articles by Noble, M. E. M. Social Bookmarking                      What's this?

The Role of the Src Homology 3-Src Homology 2 Interface in the Regulation of Src Kinases^*

Stefan T. Arold^§, Tobias S. Ulmer^¶, Terrence D. Mulhern, Jörn M. Werner^¶, John E. Ladbury^**, Iain D. Campbell^¶§§, and Martin E. M. Noble^¶¶

From the  Laboratory of Molecular Biophysics and ^¶ Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, United Kingdom, ^** Department of Biochemistry and Molecular Biology, University College London, Gower Street, London WC1E 6BT, United Kingdom, and  Department of Biochemistry and Molecular Biology, The University of Melbourne, Parkville, Victoria 3010, Australia

Received for publication, December 12, 2000

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The regulatory fragment of Src kinases, comprising Src homology (SH) 3 and SH2 domains, is responsible for controlled repression of kinase activity. We have used a multidisciplinary approach involving crystallography, NMR, and isothermal titration calorimetry to study the regulatory fragment of Fyn (FynSH32) and its interaction with a physiological activator: a fragment of focal adhesion kinase that contains both phosphotyrosine and polyproline motifs. Although flexible, the preferred disposition of SH3 and SH2 domains in FynSH32 resembles the inactive forms of Hck and Src, differing significantly from LckSH32. This difference, which results from variation in the SH3-SH2 linker sequences, will affect the potential of the regulatory fragments to repress kinase activity. This surprising result implies that the mechanism of repression of Src family members may vary, explaining functional distinctions between Fyn and Lck. The interaction between FynSH32 and focal adhesion kinase is restricted to the canonical SH3 and SH2 binding sites and does not affect the dynamic independence of the two domains. Consequently, the interaction shows no enhancement by an avidity effect. Such an interaction may have evolved to gain specificity through an extended recognition site while maintaining rapid dissociation after signaling.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The Src family of non-receptor tyrosine kinases to date comprises nine members (Src, Fyn, Lck, Hck, Yes, Fgr, Yrk, Blk, and Lyn), all of which play important roles in cellular signal transduction. Some Src kinases (Fyn, Src, and Yes) are found in most cell types and serve to mediate diverse signaling pathways, whereas others display a restricted tissue distribution and have more specific biochemical tasks (e.g. Lck in T lymphocytes, Hck in myeloid cells, and Blk in B cells). Because Src kinases are proto-oncoproteins, their activity is tightly controlled by intramolecular inhibitory interactions. Understanding of the diverse inhibitory and activatory mechanisms has been advanced substantially in the last 3 years, helped by the determination of crystal structures of the repressed, "closed" form of the two Src kinases Src (1, 2) and Hck (3) (Fig. 1). In the inactive form, the Src homology (SH)¹ 2 domain binds to the phosphorylated tyrosine of the C-terminal tail (Tyr-527 in Src), whereas the SH3 domain docks onto the N-terminal lobe of the kinase domain via a proline-containing motif in the SH2-kinase linker. As a consequence, helix C of the kinase domain adopts a nonproductive orientation (Fig. 1). The kinase can be activated by dephosphorylation of Tyr-527 or by ligand engagement of one or both SH domains. The viral Nef protein has been shown to activate Hck via an SH3 interaction (4), whereas the focal adhesion kinase (FAK) (5, 6), p62/Sam68 (7-9), p130^cas (10), SIN (11), and AFAP-110 (12) activate Src and/or Fyn via a two-domain interaction with their SH3 and SH2 domains. The availability of the SH domains for their cognate ligands seems to be interdependent of phosphorylation of Tyr-416 of the activation segment (13). Recent reports have stressed the importance of fine-tuning of intra- and intermolecular interactions for full functionality of the Src kinases (14).

View larger version (58K): [in this window] [in a new window]   Fig. 1.   Ribbon representation of an inactive Src kinase in its inactive conformation (human Src, Protein Data Bank accession code 2SRC). Tyr-416 (Y416) is the site of activatory phosphorylation, whereas phosphorylated Tyr-527 (Y527) interacts with the SH2 domain to cause intrasteric inhibition.

To further the understanding of the molecular mechanisms that lead to regulation and adaptation of Src kinases, we focused on the role of their SH3-SH2 tandem domain for kinase repression and ligand selection. By combining crystallography, solution-state NMR spectroscopy, and isothermal titration calorimetry (ITC), we characterized the structure and dynamics of this fragment from Fyn and explored its two-domain interaction with FAK. FAK, a central component in integrin signaling (15), binds to the SH2 and SH3 domains of the Src family kinases Fyn or Src, leading to their activation (5, 6). The FAK-Src interaction is thus responsible for the majority of FAK-mediated signaling events and has been implicated in the regulation of cancer cells (16).

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Protein Expression and Purification-- Amino acids 81 to 247 of human Fyn (SWISS-PROT Database accession number P06241), comprising SH domains 2 and 3, were expressed in Escherichia coli BL21 cells. All cysteine residues of human Fyn (amino acids 238, 239, and 245) were mutated into serine in this construct to avoid nonspecific disulfide cross-linking. The LckSH32 construct (comprising residues 59 to 238 of human Lck as a GST fusion vector) was a kind gift of Dr. Yves Collette (U119 INSERM, Marseille, France). Cells were grown at 30 °C in Luria-Bertani medium or minimal medium containing 100 µg ampicillin/ml culture. The cells were induced at an optical density at = 600 nm of 0.5 with 0.1 mM isopropyl-1-thio--D-galactopyranoside and grown for 6 h at 30 °C. Cells were harvested by centrifugation at 4,000 × g and stored at 80 °C. Minimal medium with ¹⁵NH₄Cl or ¹³C-enriched glucose, ¹⁵NH₄Cl, and D₂O instead of H₂O was used to produce isotopically labeled protein. For purification, the cells were thawed and then lysed by mild sonication on ice in the lysate buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 2 mM dithiothreitol, 0.4% Triton X-100, 1 mM EDTA, and one tablet of Complete protease block (Roche Molecular Biochemicals)). Cell lysates were centrifuged at 20,000 × g for 1 h, and the supernatant was loaded onto a phosphotyrosine column (Amersham Pharmacia Biotech). The column with the bound protein was then washed with 10 column volumes of 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, and 0.1% Triton, with 20 column volumes of 10 mM Hepes-NaOH, pH 8.0, and finally with 5 column volumes of 50 mM Tris-HCl, pH 8.0, 150 mM NaCl. In the case of GST-Lck, 2 mM dithiothreitol was added to all buffers. Protein was eluted with 1 M NaCl in 150 mM Tris, pH 8.0, and dialyzed against the crystallization, NMR, or ITC buffer. Purity of the sample was confirmed by SDS-polyacrylamide gel electrophoresis, concentration was assessed by optical density at 280 nm, using extinction coefficients calculated from amino acid sequence.

Peptide Synthesis-- The 36-amino acid phosphotyrosine (pY) peptide ARALPSIPKL ANNEKQGVRS HTVSVSETDD (pY)AEIID (FAKp3/2), derived from residues 367 to 402 of chicken FAK, and the pY peptide TDS(pY)AEIID were purchased from MWG AG BIOTECH. The residue Ser in position pY-1 of the pY peptide was chosen because the native FAK residue Asp gave rise to problems in synthesis. FAKp3/2 used for the NMR studies was a kind gift from Michael D. Schaller, M. D. (University of North Carolina, Chapel Hill, NC) (and contains the canonical Asp). The PXXP peptide AAAARALPSIPKL and the SM1 peptide ATEPQpYQPGEN were synthesized by the Oxford Center for Molecular Sciences peptide synthesis facility.

NMR Sample Preparation-- After dialysis into 50 mM phosphate buffer, pH 7.0, 100 mM Na₂SO₄, the buffer was exchanged again into a stock solution of 50 mM phosphate buffer, pH 7.0, 100 mM Na₂SO₄ by four ultrafiltration and dilution cycles. D₂O was added to 5%. Peptides were added in 2-fold excess, and the pH was readjusted with NaOH and HCl. For the assignment, 0.8 mM samples of ¹⁵N-labeled SH32*SM1 and ~65% ²H, ¹³C, ¹⁵N-labeled SH32*SM1 and SH32*FAKp3/2 were used. For the ¹⁵N relaxation and chemical shift studies, ¹⁵N-labeled samples of 0.3 mM SH32, 0.6 mM SH32*(pY peptide)*(PXXP peptide), and 0.6 mM SH32*FAKp3/2 were used.

NMR Spectroscopy-- All the NMR experiments were performed at 27.5 °C on a home-built spectrometer consisting of Oxford Instruments magnets operated at ¹H frequencies of 600 and 750 MHz. Backbone assignment was achieved using a combination of 2D ¹H-¹⁵N heteronuclear spin quantum coherence (17, 18), three-dimensional nuclear Overhauser effect spectroscopy-heteronuclear spin quantum coherence (19), and transverse relaxation optimized spectroscopy-H-N-C (20) spectra of ¹⁵N-labeled SH32*SM1 and ~65% ²H, ¹³C, ¹⁵N-labeled SH32*SM1 and SH32*FAK32 samples. The known chemical shifts of the single Fyn SH3 (21) and SH2 domains (22) accelerated the assignment process. Data were processed with Felix 2.3 (Biosym, Inc., San Diego, CA) and analyzed with XEasy (23).

Transverse (¹⁵N-T₂) and longitudinal (¹⁵N-T₁) relaxation times and the heteronuclear steady-state ¹⁵N-{¹H} NOE were measured at 60.8 MHz using procedures as described in Refs. 24 and 25. Transverse relaxation time constants (T₂) were measured using a spin-echo sequence with a Carr-Purcell-Meiboom-Gill delay of 457 µs. Dipolar and chemical shift anisotropy cross-correlation was removed by application of proton 180° pulses every 5 ms (T₁) and in the middle of the basic Carr-Purcell-Meiboom-Gill block (T₂) (26, 27). For the T₂ and T₁ measurements, data sets were acquired using acquisition times of 70.9 ms for ¹⁵N and 41 ms for ¹H. Eight different relaxation delays were used, ranging from 8 to 112 ms for the T₂ measurements and 20 to 1000 ms for the T₁ measurements, respectively. The acquisitions times of the ¹⁵N-{¹H} NOE experiment data sets were 62.4 ms for ¹⁵N and 41 ms for ¹H. ¹H saturation in the NOE experiment was effected by means of a train of 120° flip-angle pulses at 10-ms intervals for 4 s. The T₁, T₂, and ¹⁵N-{¹H} NOE spectra were processed with mild resolution enhancement to optimize resolution while maintaining a good signal to noise ratio. Relaxation rates were derived from two-parameter exponential fits to the resonance intensities (28). The heteronuclear NOE was calculated as the ratio of resonance intensities in the spectra recorded with and without saturation. Uncertainties in the parameters were estimated from Monte Carlo simulations using the root mean square noise in the respective spectra.

For a few residues, the peak intensity was too small to reliably estimate the ¹⁵N-{¹H} NOEs. Accordingly, they were excluded from further analysis. The isotropic correlation times were estimated from the ¹⁵N T₁/T₂ ratios of the well-structured regions of SH32 (residues 86-113 and 119-139 in SH3 and residues 150-153 and 155-231 in SH2) (24). For SH32 domains, they are essentially independent of anisotropy (29).

ITC-- All experiments were performed using the MSC system (MicroCal Inc., Northampton, MA) as described elsewhere (30). For the experiments, FynSH32 protein was diluted from stock solution to the concentration required for the ITC experiment (4-10 µM) and dialyzed against the ITC buffer 7.5 (10 mM Na,K phosphate buffer, pH 7.5, 150 mM NaCl) or 8.0 (50 mM Na,K phosphate buffer, pH 8.0, 200 mM NaCl). For titrations with LckSH32, 4 mM dithiothreitol was added. Peptides were diluted from a buffered 5 mM stock solution into the ITC buffer to 10-15× the concentration of the protein in the cell. For one titration experiment, typically twenty 15 µl aliquots of peptide were injected into the 1.3-ml sample cell at 25 °C containing FynSH32 or LckSH32. The data were fit by least squares regression using ORIGIN software. Because cleavage of the fused GST from LckSH32 by factor X (PROMEGA) required large amounts of factor X (cleavage with 1 µg of factor X yielded about 80 µg of LckSH32), GST-LckSH32 was used for most ITC experiments. The binding parameters for GST-LckSH32 were comparable to those for LckSH32 (Table II).

Circular Dichroism (CD) Spectroscopy-- CD spectra were measured with a nitrogen-cooled spectropolarimeter using a 4 s time constant, a 10 nm·min¹ scan speed, and a 1 nm spectral bandwidth. A 0.1-cm path-length cell was utilized for the far-UV CD region (185-260 nm), and a 1-cm path-length cell was utilized for the near-UV CD region (240-320 nm). All spectra were recorded at 10 °C in ITC buffer.

Crystallization and Crystallographic Analysis-- For crystallization, isolated FynSH32 was kept in 10 mM Tris, pH 8.0, 1 M NaCl and concentrated to 16 mg/ml. Crystals were grown with the vapor diffusion method from hanging drops using 2 µl of protein solution and 2 µl of well solution (0.25 M sodium tartrate, 12% PEG 8000, 100 mM Tris-HCl, pH 6.0). Diffraction data were recorded at room temperature on a single crystal (700 × 300 × 100 µm, space group P21 with a = 40.0 Å, b = 90.0 Å, c = 60.3 Å, b = 101.4°) on Beamline ID14-2 at the European Synchrotron Radiation Facility (Grenoble, France). Integration (with MOSFLM), merging, and scaling (with SCALA, CCP4 suite) of the data yielded an 82% complete dataset between 37.0 and 2.6 Å resolution (R_sym = 7.4%, <I/sigI> = 6.3). The structure was solved with AMORE, using as search models the crystallographic structures of Fyn SH3 (Protein Data Bank accession code 1SHF) and Src SH2 (Protein Data Bank accession code 1SPS). Placement of two Src SH2 and two Fyn SH3 molecules yielded a correlation coefficient of 55 and an R factor of 40% between 3.0 and 12.0 Å. The SH3-SH2 linker region of the two FynSH32 molecules became visible after solvent flattening and histogram matching using DM (CCP4 suite). The FynSH32 structure was rebuilt in the electron density map after DM and refined by CNS. The final model has a crystallographic R factor of 21.4% (Rfree = 27.7%) for all reflections between 37.0 and 2.6 Å resolution and root mean square deviations of 0.012 Å for bond lengths and 1.66^o for bond angles. The coordinates have been deposited in the Protein Data Bank (accession code 1G83).

Attempts were made to obtain crystals of the complex between FynSH32 and FAKp3/2. The complex was formed by mixing together FynSH32 and FAKp3/2 and purified by gel filtration using a size exclusion chromatography column (Superdex75; Amersham Pharmacia Biotech). Crystals of this material grew after 3 to 6 months from hanging drops of 1 µl of well solution (13% PEG 4000, 13% PEG 3350, 0.24 M lithium sulfate, 100 mM Tris-HCl, pH 8.8) mixed with 1 µl of a solution containing the complex in 10 mM Tris-HCl, pH 8.0, 200 mM NaCl at a concentration of 20 mg/ml. Native data were recorded on European Synchrotron Radiation Facility beamline ID14-2. Additional data were collected from a crystal soaked for 5 h in 0.2 mM uranyl acetate on European Synchrotron Radiation Facility beamline ID14-4. Data were integrated with MOSFLM and merged and scaled with SCALA. The crystals grew in space group R32 (a = b = 113.0 Å, c = 205.0 Å) and yielded a 98% complete native dataset to 3.5 A (R_sym = 8.3%) and a 97% complete uranyl-derivative to 3.8 A (R_sym = 7.7%). Both structures could be solved by molecular replacement using the refined FynSH32 structure as template (correlation coefficient for two positioned FynSH32 was 44%, and the R factor was 50% for the native data). Phases were calculated from the uranyl derivative using the program SHARP and used to calculate a map that confirmed the molecular replacement solution. The SH3 positioning was additionally verified by maps that were phased using only the two SH2 domains of the asymetric unit and averaged by DM. Crystal packing occluded the pY+3 binding pocked on the SH2 domain and the PXXP-peptide binding site of the SH3 domain. Unattributed electron density was found at the pY binding pocked; however, no further electron density for the FAK peptide was apparent.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Structure and Dynamics of the Isolated Regulatory Domain of Fyn (FynSH32)-- FynSH32 was expressed in E. coli and purified as described under "Experimental Procedures." Crystals of this fragment were obtained in two different crystal forms, one in space group P21 and the other in space group R32, grown under different crystallization conditions. These diffracted to maximum resolutions of 2.6 Å and 3.5 Å, respectively. Both crystal forms could be solved by molecular replacement using crystal structures of Fyn SH3 domain and Src SH2 domain as search models. The higher resolution P21 data were used to obtain the final refined model (Fig. 2). This model includes the first crystal structure of Fyn SH2. As expected, Fyn SH2 is very similar to Src SH2 (31) (their 96 Cs superimpose with root mean square of 0.96 Å) and to the NMR model of Fyn SH2 (22) (C root mean square is 1.62 Å). Both crystal forms were comprised of asymmetric units, which contained two independent copies of FynSH32. All four FynSH32 structures showed a very similar SH2-SH3 orientation, which in turn resembles the SH3-SH2 arrangement observed in the repressed, "closed" form of Hck and Src (Fig. 3). When superimposed on their SH2 domain, the SH3 domains of the crystallographic FynSH32 structures are related to each other by a rotation of maximally 10° around the vector connecting the Cs of residues Ile-144 and Gln-145. A similar rotation is sufficient to superimpose the SH3 domains of FynSH32 upon the SH3 domain of the inactive Src kinase.

View larger version (38K): [in this window] [in a new window]   Fig. 2.   Crystal structure of FynSH32, showing its molecular surface (A) and secondary structure (B). Residues that are known to interact with ligand peptides (according to Refs. 22 and 39) are shaded in gray.

View larger version (73K): [in this window] [in a new window]   Fig. 3.   Regulatory fragments of Src family kinases, superimposed on the basis of their SH2 domain. The two molecules of FynSH32 from the P21 crystal form are shown in red and yellow, the regulatory fragments of the intact Src kinases Hck and Src are shown in dark red and magenta, respectively, and the isolated structure of LckSH32 is shown in cyan.

To assess the flexibility of the SH3-SH2 domain orientation in solution, the individual correlation times of the domains were determined from ¹⁵N NMR relaxation data (Table I). A marked difference is observed between the correlation times of the SH3 and SH2 domains, in agreement with the differences in their masses. This indicates that the individual SH domains of FynSH32 are not associated rigidly but show interdomain motions. This interpretation agrees with the analysis of local motion using ¹⁵N-{¹H} NOEs (Fig. 4). The ¹⁵N-{¹H} NOE is highly sensitive to nano- to picosecond movements of each H^N-N bond vector, where a reduction of the ¹⁵N-{¹H} NOE reflects local flexibility. The ¹⁵N-{¹H} NOEs are slightly lower than expected from the overall correlation times of the SH domains, which is likely to be due to the relatively fast amide proton exchange at pH 7.0. Apart from the N- and C-terminal residues, Gly-117, Ile-144, and Leu-154 show significantly reduced ¹⁵N-{¹H} NOEs, indicating substantial fast motions. High mobility of the loop comprising Gly-117 has been observed before in the single FynSH3 domain (21). Leu-154 lies on the surface of the SH2 domain, 13 Å away from the SH3-SH2 linker, and therefore cannot be important for the flexibility of the domain interface; its low NOE probably reflects only local conformational flexibility. Ile-144 lies in the linker region and was identified as a hinge in the crystallographic analysis. Hence, the high mobility of Ile-144 is likely to be correlated with the observed interdomain flexibility of the SH3-SH2 domains. The remaining linker residues 146-148 could not be observed, most likely because these relatively unprotected amide protons exchange rapidly with solvent.

View larger version (22K): [in this window] [in a new window]   Fig. 4.   15N-{1H} NOEs of FynSH32 and FynSH32*FAKp3/2 along the protein chain. Low values are observed for Gly-117, Ile-144, and Leu-154, as well as the N- and C-terminal residues of FynSH32. NOEs for residues around Gly-117 could not be observed in the FynSH32/FAKp3/2 complex, most likely due to the micro rate constants governing peptide binding.

The crystallographic and NMR observations are consistent with a favored relative orientation of the SH3 and SH2 domains with some degree of flexibility (Fig. 3). This characteristic is probably shared by the Src family members Src, Hck, and Fyn because they have similar domain-domain interfaces. In all three members, the interface is formed by a 10-amino acid linker region, which contains a short stretch of 3₁₀ helix (residues 144-146). The linker region contacts both SH2 and SH3 domains, burying 600 and 450 Å² of molecular surface, respectively. The surface of the linker region is formed mainly by side chains of the 3₁₀ helix, which is anchored to the SH2 domain via hydrophobic contacts between Ile-144 of the linker and Trp-149, Tyr-150, Tyr-185, and Leu-224 of the SH2 domain. The whole interface is largely hydrophobic in character, which, while favoring some rotational movements of the SH3-SH2 domains relative to each other, will discourage gross rearrangements that would expose hydrophobic surfaces. The preferred arrangement is further stabilized by an electrostatic interaction between Glu-148 of the linker region and Lys-105 of the SH3 domain (Fig. 5A).

View larger version (51K): [in this window] [in a new window]   Fig. 5.   Detail of the interface between the SH3-SH2 linker (ribbon representation) and the SH3 domain (gray surface). The crystal structure of FynSH32 (A) is compared with the modeled structure of LckSH32 (B), which has been built based on the conformation of FynSH32. The observed conformation of LckSH32 is shown in Fig. 1.

Comparison with Other Tandem SH Domains-- Molecular models of three other isolated SH3-SH2 tandem domains have been published: Abl (16), Grb2 (32), and Lck (33). The short 3₁₀ helix that stabilizes the Fyn SH3-SH2 disposition is not found in the linker regions of non-Src kinases, which explains why the SH3-SH2 arrangements of Grb2 and Abl are different from that of Fyn. The correlation time of the individual SH domains of AblSH32 (29) is similar to that of FynSH32. The isotropic correlation times of Abl SH3 and SH2 are 8.9 ± 0.1 ns and 9.5 ± 0.1 ns, respectively, indicating a slightly tighter coupling than that seen in Fyn. Lck is a Src family member, and the structure of LckSH32 has been solved in the presence and absence of a phosphorylated peptide ligand for the SH2 domain (33). In both crystal forms, LckSH32 adopts a similar conformation, which is clearly distinct from isolated FynSH32 and consequently from the SH3-SH2 arrangement of the closed form of Src and Hck. The superimposition of FynSH32 and LckSH32 by their SH2 domains reveals that the different relative position of Lck SH3 is a consequence of a conformational change of residues Asn-120 and Ser-121 (Asp-142 and Ser-143 in Fyn). Molecular modeling does not reveal any amino acid substitution in Lck that could hinder the positioning of its SH3 domain as seen in isolated FynSH32 and inactive Hck and Src. This SH3 position is, however, likely to be less stable in Lck due to a Pro-X-Pro motif (residues 146-148 in Fyn numbering). The first proline blocks formation of the 3₁₀ helix, whereas the second proline, equivalent to Glu-148 of Fyn, is unable to establish the electrostatic interaction with Lys-105 (Fig. 5B). In addition, hydrophobicity analysis indicates that the face of the SH3 domain that packs against the SH3-SH2 linker in Lck is less hydrophobic than that in Fyn.

Taken together, our observations show that the preferred relative orientation between SH3-SH2, which is determined by the nature of the linker, is not the same for all Src family members. Lck alone has a Pro-X-Pro motif in the SH3-SH2 linker, indicating an atypical behavior for this Src kinase. In Blk, the residue equivalent to Lys-105 is an arginine. Disrupting the salt link described above may result in greater flexibility of the Blk SH3-SH2 arrangement as compared with that of Fyn.

The Two-domain Interaction between FynSH32 and FAK-- Restriction of the relative orientations of the SH3 and SH2 domains imposes steric constraints on a bidentate ligand. In addition, the protein surface that spans the two SH domain binding sites could be used by a ligand for specific "tertiary interactions." These two effects might be exploited to distinguish cognate ligands for Fyn and Lck tyrosine kinases.

We have explored this possibility by using ITC to obtain the thermodynamic parameters of the association between FynSH32 and an SH2/SH3-binding 36-amino acid phosphopeptide derived from FAK (FAKp3/2). This fragment is comprised of an N-terminal "+ve-orientation" SH3-binding motif (RALPSIP), a 22-amino acid spacer, and a C-terminal phosphotyrosine-containing motif (pYAEIID) that conforms to the "YXXI" class preferred by the SH2 domains of Src or Fyn. The 22-amino acid linker region between the SH binding sites is about 10 amino acids longer than necessary to simply span the distance between the binding sites of the SH2 and SH3 domain. At 25 °C and in 10 mM phosphate buffer, pH 7.5, 150 mM NaCl, the dissociation constant (K_d) of the complex was 30 nM. This affinity, which is more than 1 order of magnitude higher than that commonly observed for single SH2 domain-peptide interactions, compares well to the K_d of 20 nM for the FAKp3/ 2-SrcSH32 interaction reported by Thomas et al. (6) using surface plasmon resonance.

When compared with the single SH-ligand interactions, it became apparent that the association of FAKp3/2 with FynSH32 is "anti-cooperative" (Table II): the free energy change (G in Table II) upon binding of the FAK peptide to the regulatory domain of Fyn is less favorable than the sum of the free energy changes upon binding of the individual SH2 and SH3 binding peptides (pY and PXXP, respectively; Table II). Possible explanations for this will be given, in the light of our other results, under "Discussion."

The tightness of the interaction of FynSH32 with FAKp3/2, compared with the binding of FynSH32 to the individual SH3 and SH2 binding peptides, indicates that FAKp3/2 is able to span both SH binding sites of FynSH32. To test whether this is also possible in the context of a LckSH32 domain orientation, we used ITC to study binding of peptides to LckSH32. ITC shows that FAKp3/2 binds to LckSH32 with a 20-fold lower affinity than to FynSH32, whereas the binding of the pY peptide is only 2-fold lower. The double-domain interaction is therefore more specific than the single-domain interaction. The thermodynamic parameters of the interaction between GST-LckSH32 in 50 mM phosphate buffer, pH 8.0, 200 mM NaCl resemble those of the interaction between FynSH32 and the single pY peptide. It is more likely, however, that the FAKp3/2-LckSH32 interaction is also bidentate because the thermodynamic parameters of the pY-LckSH32 interaction are clearly distinct from those of the FAKp3/2-LckSH32 association. Thus, discrimination of Lck from Fyn is not achieved by FAK through excluding simultaneous SH domain interactions in the case of FAK-Lck binding.

Mapping the Contacts between Fyn and FAK-- To search for interactions of the FAK peptide outside the canonical SH domain binding sites, we studied the effect of FAKp3/2 binding on the correlation times and the chemical shifts of FynSH32. We compared ¹H^N and ¹⁵N chemical shifts between FynSH32 in complex with the single SH-ligand peptides ("untethered") and FynSH32 bound to FAKp3/2 ("tethered"). Large chemical shift differences between tethered and untethered states were sparse and uncorrelated and were consistent with the FAKp3/2-FynSH32 interaction being mostly confined to the canonical binding sites on the SH2 and SH3 domains (data not shown).

¹⁵N NMR relaxation data of the FynSH32 -FAKp3/2 complex were collected to investigate whether the dual interaction restrains the flexibility of FynSH32. The correlation times of the two SH domains remain different even when bound to FAKp3/2, and the difference is similar to that of the uncomplexed SH domains (Table I). The peptide spanning the two domains thus does not restrict the motion notably more than in the free state. Analysis of ¹⁵N-{¹H} NOEs confirms that the dual interaction with FAKp3/2 does not significantly change the mobility of the linker residue Ile-144 (Fig. 4). FAKp3/2 binding to the SH3 caused severe broadening of the contact residues. This shows that, unlike other SH3-binding peptides that are in fast exchange (21), the off-rates are slow enough (~10⁴ s) to cause exchange broadening. There is thus no evidence from the relaxation data or chemical shifts that there is any interaction between FynSH32 and FAKp3/2 outside the canonical binding regions or that the whole peptide adopts a rigid structure.

CD studies showed that free FAKp3/2 has no defined secondary structure in solution, except a signal at 226 nm, which suggests a polyproline type II helix. CD also failed to detect gross secondary structure changes of the FynSH32-FAKp3/2 complex compared with the uncomplexed molecules (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The Role of the SH2-SH3 Linker in Src Kinase Regulation-- We have shown that the SH2 and SH3 domains of Fyn maintain the relative orientation observed in the closed kinase form of intact Src and Hck, even in absence of the kinase domain. The tendency to this conformation results from a short 3₁₀ helix within their SH3-SH2 linker. In the solution state, the relative domain-domain orientation possesses some degree of flexibility. Isolated LckSH32 adopts a different orientation. Our analysis indicates that this is due to a PXP amino acid sequence, which is unique to the SH linker of Lck.

Recent publications have demonstrated the mechanism by which, in the closed kinase form, the SH3 domain stabilizes inactive conformations of the kinase's C-helix and activation loop. Moreover, a molecular dynamics simulation, carried out on the closed Hck structure, has shown that phosphorylation of a tyrosine residue in the activation segment alters the dynamic properties of the SH domains. In these simulations, the SH2 and SH3 movements were communicated through the SH3-SH2 linker.² Our data indicate restrained rotational flexibility of the SH3-SH2 linker, especially in the vicinity of Ile-144, which might allow the SH3 domain to buffer small movements of the N-terminal kinase lobe in a direction perpendicular to the linker axis. Movements parallel to the SH3-SH2 linker, however, will be communicated to the SH2 domain. The resulting concerted motion of the regulatory apparatus may destabilize the closed kinase form and render the SH domains more readily accessible to ligands.

The propensity of the Lck SH32 domain to deviate from the kinase-repressing conformation may decrease its potential to repress kinase activity and abolish a cross-talk between the activation segment and the SH domains. A functional consequence, consistent with such a difference, has been described by Briggs et al. (34). In their study, Src kinases expressed in Rat-2 fibroblasts were assessed in an in vitro kinase assay for autophosphorylation and for phosphorylation of a substrate (p50, a polypeptide derived from Sam68). In the case of Src, Fyn, and Lyn, the band for autophosphorylation was stronger than the signal from substrate phosphorylation, whereas in Lck, autophosphorylation was barely detectable, although the kinase was readily able to phosphorylate p50. Substrate phosphorylation by Lck without strong autophosphorylation is, moreover, highly reproducible.³ One possible way of interpreting these data is that for Fyn, Src, and Lyn, autophosphorylation is a prerequisite for substrate phosphorylation, whereas for Lck, it is not. Regulation of Lck may result instead from its intimate co-localization with CD4 and CD8 through its N-terminal unique domain (35). In the case of T cell receptor signaling, phosphorylation of the cytoplasmic chains of the T cell receptor by Lck may be triggered by recruitment of the CD4-Lck complex to the T cell receptor via an extracellular ligand. Fyn, already present in the submembrane T cell receptor environment, would have to be tighter regulated and may serve as a "backing-up" system. More generally, regulation of Src kinases may be modulated and "customized" by their unique domain, which targets the enzyme to specific cell surface receptors or subcellular compartments, presumably as a result of direct molecular recognition events (36).

Together with structural and functional observations from other groups, our data suggest that the nature of the SH3-SH2 linker region is important for repressing kinase activity and for intramolecular communication. The SH3-SH2 linker is a major determinant of domain-domain orientation and coupling. Consequently, amino acid substitutions that influence the SH linker stability of Src kinases may contribute to the idiosyncrasies of individual Src family members. Because all Src kinases share a common domain structure, it may be the precise and particular balance of intra- and intermolecular interactions that adapts Src family members to their specific cellular role. Our results may serve as a basis for future mutational analysis of these phenomena.

The Role of the SH3-SH2 Linker in Two-domain Interactions-- Our data support a model in which the 36-amino acid FAK peptide remains mostly unstructured upon binding to FynSH32 and interacts with FynSH32 only within the known SH binding sites. The bidentate binding of FAK to FynSH32 therefore appears more as two simultaneous single binding events than as a coupled two-domain interaction. Consequently, the entropic energy gain through avidity is not exploited in this association. Moreover, the contacts between FAK and Fyn would be expected to restrict the conformational freedom of the 36-amino acid FAK peptide, leading to a small entropic penalty. In the absence of sufficient favorable contributions from "tertiary" interactions, this penalty might explain the observed anticooperativity of simultaneous SH3 and SH2 binding. Additionally, small binding-induced structure in the FAK peptide may add to the negative entropy. Together, structural and dynamic properties of the Fyn-FAK association tend to decrease the affinity of the interaction, leading to an anticooperative association. However, the bipartite interaction displays a greater specificity than the single SH2 domain interaction. FAK may have evolved to maximize specificity through use of a bidentate interaction, without reaching an affinity that is too high to allow a rapid dissociation of the complex after signal transduction.

We cannot exclude that further specific interactions occur when the SH-binding FAK fragment is presented in the context of full-length FAK. The SH-binding fragment of FAK links two independently folded domains (the band 4.1 ezrin, radixin, moesin and kinase domain) and may be partially associated with one or both of them. However, the following facts seem to support that FAKp3/2 is a functionally independent fragment: FAK depleted of its band 4.1 ezrin, radixin, moesin domain still functionally associates with Src and Fyn as does the wild type (37), and isolated FAKp3/2 is a potent activator of repressed Src (6).

In the integrin signaling complex, FAK activates either Fyn or Src. In the cell, Src is also activated by a simultaneous SH32 domain interaction with p62/Sam68 (7-9), SIN (11), and AFAP-110 (12). Although the interacting sites have not been clearly identified in these proteins, their SH2 and SH3 binding motifs are not separated by a similar number of residues as those of FAK and do not share sequence similarity. Moreover, even among FAK and the strongly related proteins CAK (38) and p130^cas (10), the sequences between PXXP and phosphotyrosine motif are poorly conserved (no residue is identical in all three sequences). This adds to a model wherein the SH32 binding fragments of proteins that activate Src family kinases do not form a defined structural motif and do not exploit generic tertiary binding sites on the regulatory fragment. The relative orientation of the SH3 and SH2 domain is therefore not likely to play a major role in selecting for specific ligands, which, being flexible, reach their binding sites independently of the SH domain orientation. The dynamic properties of the FAK-FynSH32 association, which demonstrates specificity without excessive affinity, may be a common feature of Src-activating interactions necessary to achieve the fidelity and reversibility of Src-mediated signaling.

	ACKNOWLEDGEMENTS

We thank Ronan O'Brien and Ihtshamul Haq for assistance with the ITC and CD measurements and John Kuriyan for stimulating discussions. Ives Collette kindly provided the clone for LckSH32, and, at an early stage in this project, Michael D. Schaller kindly provided some FAK peptide. The NMR facilities are supported by the Oxford Center for Molecular Sciences that is funded by Biotechnology and Biological Sciences Research Council, Medical Research Council, and Engineering and Physical Sciences Research Council. We would also like to thank scientists on beamline ID14, ESRF, France, and x-ray diffraction, Electra, Italy.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ Fellow of the Medical Research Council.

A Wellcome Trust Senior Research Fellow.

^§§ Supported by the Wellcome Trust.

^¶¶ To whom correspondence should be addressed. E-mail: martin@biop.ox.ac.uk.

Published, JBC Papers in Press, February 2, 2001, DOI 10.1074/jbc.M011185200

² J. Kuriyan, manuscript in preparation.

³ T. Smithgall, personal communication.

	ABBREVIATIONS

The abbreviations used are: SH, Src homology; FAK, focal adhesion kinase; ITC, isothermal titration calorimetry; GST, glutathione S-transferase; pY, phosphotyrosine; NOE, nuclear Overhauser effect.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				S. A. Solheim, K. M. Torgersen, K. Tasken, and T. Berge Regulation of FynT Function by Dual Domain Docking on PAG/Cbp J. Biol. Chem., February 1, 2008; 283(5): 2773 - 2783. [Abstract] [Full Text] [PDF]

				J. D. Faraldo-Gomez and B. Roux On the importance of a funneled energy landscape for the assembly and regulation of multidomain Src tyrosine kinases PNAS, August 21, 2007; 104(34): 13643 - 13648. [Abstract] [Full Text] [PDF]

				M. Franco, O. Furstoss, V. Simon, C. Benistant, W. J. Hong, and S. Roche The Adaptor Protein Tom1L1 Is a Negative Regulator of Src Mitogenic Signaling Induced by Growth Factors. Mol. Cell. Biol., March 1, 2006; 26(5): 1932 - 1947. [Abstract] [Full Text] [PDF]

				M. Karplus and J. Kuriyan Chemical Theory and Computation Special Feature: Molecular dynamics and protein function PNAS, May 10, 2005; 102(19): 6679 - 6685. [Abstract] [Full Text] [PDF]

				K.-i. Sato, T. Iwasaki, and Y. Fukami Association of c-Src with p52Shc in Mitotic NIH3T3 Cells as Revealed by Src-Shc Binding Site-Specific Antibodies J. Biochem., January 1, 2005; 137(1): 61 - 67. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) All Versions of this Article: 276/20/17199    most recent M011185200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Arold, S. T. Articles by Noble, M. E. M. Search for Related Content PubMed PubMed Citation Articles by Arold, S. T. Articles by Noble, M. E. M. Social Bookmarking                      What's this?