The tandem Src homology 2 domain of the Syk kinase: A molecular device that adapts to interphosphotyrosine distances
- Departments of *Biochemistry and Molecular Biophysics and †Psychiatry, Washington University School of Medicine, 660 South Euclid Avenue, Saint Louis, MO 63110; ‡School of Crystallography, Birkbeck College, Malet Street, London WC1E 7HX, United Kingdom; and §Department of Biochemistry and Molecular Biology, University College London, Gower Street, London WC1E 6BT, United Kingdom
-
Edited by Robert L. Baldwin, Stanford University Medical Center, Stanford, CA, and approved October 13, 2003 (received for review May 13, 2003)
Abstract
Conformational flexibility is important for protein function. However, information on the range of conformations accessible to macromolecules in the unbound state is often difficult to obtain. By using the model system of the tandem Src homology 2 domain (i.e., two adjacent Src homology 2 domains) of the Syk kinase, we report a method combining calorimetric and crystallographic measurements that reveals the preexistence of a conformational equilibrium in the unbound state, and that shows that this equilibrium is crucial for function.
Structural flexibility is an important feature of protein structures and plays an important role in processes such as substrate recognition, allosteric regulation, or molecular motions of molecular motors. Structural flexibility is particularly well documented in processes requiring ligand-induced conformational changes (1). However, structural flexibility within the unbound state is difficult to study, because it is often inaccessible to experimental structure determination. Yet, as we demonstrate here, the conformational ensemble that a protein can access in the unbound state may play important functional roles.
The protein tyrosine kinase Syk is a member of the Syk family of cytosolic protein tyrosine kinases that also includes the ZAP-70 kinase (2, 3). Syk is expressed ubiquitously in hematopoietic cells such as B cells, mast cells, polymorphonuclear leukocytes, platelet macrophages, and immature T cells. Syk is also present in nonhematopoietic cells such as epithelial cells, hepatocytes, fibroblasts, and neuronal cells. The absence of either Syk family kinase results in arrested T and B cell development and functional defects for a variety of immune receptors, including the T and B cell receptors and receptors for IgG and IgE (4–10). In addition, Syk family kinases have been implicated in activating natural killer cells (11, 12) and signaling by non-immune receptors such as cytokine, thrombin, integrin, and G protein-coupled receptors (13–18).
Syk family kinases contain two Src homology 2 (SH2) domains positioned in tandem [termed collectively as tandem SH2 (tSH2) domain] (19). Syk family kinases are recruited to cell surface receptors through the interaction of their tandem SH2 domains with tyrosine-phosphorylated sequence motifs termed immunoreceptor tyrosine-based activation motifs (ITAMs) (20). ITAMs typically have the consensus sequence Yxx(L/I)-x7/8-Yxx(L/I), i.e., two tyrosines spaced by a 10/11 residue sequence and containing a hydrophobic residue (L or I) at the third position C-terminal to the phosphotyrosines (21). Remarkably, the tSH2 domain of the Syk kinase (Syk tSH2) appears to be able to recognize a variety of doubly phosphorylated (dp) ITAMs that vary considerably not only in sequence but also in the length of the spacer region between the two phosphotyrosines (13, 22, 23). For example, the γ chain of the FcεRI receptor, an Fc receptor for IgE (referred to hereafter as FcR-γ), contains an ITAM with a spacer region of 10 residues, whereas the spacer region of the ITAM on the single-chain Fc receptor class IIA, an Fc receptor for IgG (referred to hereafter as FcRIIA), is 15 residues in length (see Fig. 1a for peptide definition) (24). Their sequences are also considerably different (Fig. 1a). Yet, dp peptides derived from these ITAMs (dpITAMs) bind equally well and with high affinity (2–4 nM) to the Syk tSH2 (23).
(a) dpITAM peptides used in this study. The symbol Y* indicates a phosphotyrosine. The additional five residues in the interphosphotyrosine spacing region in the FcRIIA dpITAM peptide are indicated in red, and the spacer region is indicated. The definition of the CD3-ε, FcR-γ, and FcRIIA dpITAM peptides is in the text. (b) Electron density in and around residues 48 and 235 in the 2Cys mutant. The electron density map was calculated by using F o–F c coefficients and phases from the molecular replacement model where the region 5 Å around residues 48 and 235 was omitted. Electron density for the disulfide bond is clearly visible, indicating formation of the bond. (c) Superposition of the Syk wild-type tSH2 domain-CD3-ε dpITAM and Syk 2Cys tSH2 domain-FcR-γ dpITAM complexes. Structures were superimposed with respect to their N-terminal SH2 domains. The tSH2 domains are in ribbon representation color-coded in yellow and magenta for the wild-type and 2Cys complexes, respectively. The CD3-ε and FcR-γ dpITAM peptides are in ball-and-stick representation color-coded in gold and red, respectively. (d) Polyacrylamide gel electrophoresis of proteolytically cleaved native and 2Cys tSH2 domains. The molecular masses for the tSH2 domain (30 kDa) and the fragments (16–18 kDa) are indicated. Lane 1, molecular mass markers; lanes 2–4, native protein with lane 2 containing no protease, lane 3 including the V8 protease digest under nonreducing conditions, and lane 4 including the V8 protease under reducing conditions; lanes 5–7: 2Cys mutant protein with lane 5 containing no protease, lane 6 including the V8 protease under nonreducing conditions, and lane 7 including the V8 protease under reducing conditions.
The mechanism by which the tSH2 domain of Syk can accommodate such structural variability in the ITAM was hinted at when the crystal structure of a complex of the Syk tSH2 domain bound to a dpITAM peptide derived from the CD3-ε chain of the T cell receptor (K d = 20 nM) was determined (Fig. 1a) (25). The crystals contained six complexes in the asymmetric unit. Structural alignment of the six complexes revealed substantial flexibility in the relative orientation of the two SH2 domains. This feature, coupled with the fact that the interface between the SH2 domains is relatively small, suggested that, in the unbound state, there may even be larger fluctuations in the relative orientations of the two SH2 domains: the Syk tSH2 domain may fluctuate between an “open” state where the two SH2 domains are far apart from one another and a “closed” state where they are closer to one another. This hypothesis was supported by the results of a multifaceted biophysical study of ITAM binding to the Syk tSH2 domain in solution (26). In this study, an unusual nonlinear temperature dependence of the binding enthalpy (i.e., a temperature-dependent heat capacity change) was observed that was interpreted as the thermodynamic signature of a preexisting conformational equilibrium (between a closed and an open form) in the unbound state. These results provided a model for the ability of the Syk tSH2 domain to recognize dpITAMs with spacer regions of variable length: the two SH2 domains sample a continuum of relative orientations between an open and closed state, thus allowing the tSH2 domain to adapt and adjust to the various lengths of the interphosphotyrosine region (26). Although this model was consistent with the complexity of the binding data, direct confirmation was lacking.
One way to directly test this model is to restrict the conformational ensemble accessible to the tSH2 domain in the unbound
state. If, for example, the tSH2 domain is locked in the closed conformation (the two SH2 domains closer together), one would
expect: (i) a temperature-independent heat capacity change (
), which would reflect a single species population as opposed to the previously observed temperature-dependent 
reflecting a multiconformer equilibrium, and (ii) the affinity for a dpITAM sequence with a short spacer region to be less affected than that for a dpITAM with a longer spacer
region. In this study, we characterize the binding thermodynamics of a tSH2 domain conformationally restrained by the introduction
of a disulfide bridge between the two SH2 domains. We show that this mutant behaves as predicted for the closed form of the
domain, and that flexibility in the unbound state is crucial for tSH2 domain function.
Methods
Site-Directed Mutagenesis. Leu-48 in the N-terminal SH2 domain and Asp-235 in the C-terminal SH2 domain were mutated to Cys by using reverse PCR as implemented in the QuikChange mutagenesis kit (Stratagene).
Peptides. All three dp peptides used in this study were obtained from Quality Control Biochemicals (Hopkington, MA). The N-terminal end is acetylated, and the C-terminal end is amidated. Peptide concentrations were determined by using an extinction coefficient of 1,304 M–1·cm–1 at 267 nm based on an extinction coefficient for phosphotyrosine of 652 M–1·cm–1 (26).
Proteolysis. The V8 protease (Boehringer Mannheim) was used in the ratio of 1/200 (wt/wt) enzyme/substrate in a buffer containing 50 mM Hepes, pH 7.5. To generate reducing conditions, the proteins were preincubated with 0.1 M DTT for ≈1–2 h at room temperature before adding the enzyme. After incubation at 4°C for 3 h, reactions were stopped by heating to 95°C for 5 min. Aliquots of reaction samples (typically 5–10 μg) were analyzed by SDS/PAGE.
Fluorescence Titrations. Fluorescence titrations were carried out as described in refs. 23 and 26. Fluorescence titrations were used to measure binding constants and thus free energies. Binding constants were determined to be independent of protein concentration.
Calorimetric Measurements. Calorimetric measurements were carried out by using batch calorimetry as in Grucza et al. (26). Batch calorimetry is a calorimetric method that utilizes single injections in saturating conditions and thus allows accurate measurement of binding enthalpies at much lower protein concentration than isothermal titration calorimetry. The dependence of the enthalpy of binding on temperature was studied over the 5–30°C temperature range. The upper limit of this temperature range was chosen based on earlier studies (26). Each reported data point is the average of three measurements. Typical standard deviations for the reported values are 0.6 kcal/mol. Slopes reported for the low temperature data were determined by standard linear regression fit (correlation coefficient ≥0.98). The protein concentrations for both proteins were determined by using the extinction coefficient measured for the native Syk tSH2 domain (34,425 M–1·cm–1) (26).
Crystallization and Structure Determination of the 2Cys tSH2 Domain Bound to the FcR-γ dpITAM Peptide. The complex of the 2Cys mutant protein bound to the FcR-γ dpITAM peptide was formed by using a ratio of 1:1.5 protein to ligand and concentrated to 30 mg/ml. Crystals were grown at room temperature by using the hanging drop vapor diffusion method. The reservoir contained 25% (wt/vol) of polyethylene glycol 3000, 0.1 M Tris·Hcl, pH 8.5. After cryoprotection in 20% glycerol, crystals diffracted to a resolution of 2.5 Å in a laboratory setting (R axis IV image plate detector mounted on a RU200 rotating anode, Rigaku/MSC, Woodlands, TX). A complete data set to that resolution was collected by using an oscillation range of 1.0° and an exposure time of 20 min/frame. Crystals were in orthorombic space group (P212121) with cell dimensions a = 37.2 Å, b = 67.4 Å, and c = 110.9 Å and one molecule in the asymmetric unit. Data were reduced and integrated by using the programs denzo and scalepack (27). The structure was solved by using the Molecular Replacement method as implemented in the program cns (28). The search model was molecule A of the native Syk tSH2/CD3-ε complex structure (PDB ID code 1A81) with the peptide ligand removed. The rotation and translation search yielded a unique unambiguous solution. An electron density map calculated by using F o–F c coefficients and phases from this model where the region around residues 48 and 235 was removed showed clear electron density for the disulfide bond between these two residues and the peptide ligand. The ligand was built in density by using the program o (29). Progress in the refinement that included conjugate gradient minimization as well as simulated annealing in both cartesian and torsional angle spaces was assessed by monitoring the free R factor (30). Final R factor and free R factor are 24.7 and 26.6, respectively. rms deviations in bond lengths and angles are 0.012 Å and 2.2°, whereas rms deviations in temperature factors between bonded atoms is 1.5 and 1.7 Å2 for main-chain and side-chain atoms, respectively.
Results and Discussion
To restrict conformationally the Syk tSH2, we introduced a disulfide bridge between the two SH2 domains. This mutant (“2Cys”) was designed based on the structure of the CD3-ε dpITAM–bound state of the tSH2 domain (25). We identified Leu-48 in the N-terminal SH2 domain and Asp-235 in the C-terminal SH2 domain, the Cα atoms of which are located within a distance of 6.2 Å. These two residues were mutated to cysteines (L48C and D235C) by using site-directed mutagenesis.
Disulfide bond formation was induced in an oxidation buffer (1:4 molar ratio of oxidized vs. reduced glutathione) during the last step of purification. We first sought to check for disulfide bond formation by crystallizing the 2Cys mutant protein in a complex with a dpITAM peptide derived from the FcR-γ chain (Fig. 1a). Crystals diffracted to 2.5-Å resolution. A data set was collected, and the structure was solved by the Molecular Replacement method. The structure was refined to a free R of 26.6% and an R factor of 24.7% with good stereochemistry and restrained B factors. A difference Fourier electron density map of the region around Cys-48 and Cys-235 is shown in Fig. 1b. The bond between the two sulfur atoms from each of the residues is clearly visible. Hence, disulfide bond formation indeed occurs, indicating that our rationale for the design of a disulfide bond to conformationally constrain the tSH2 domain is correct. The relative orientations of the two SH2 domains in the FcR-γ-bound 2Cys and wild-type CD3-ε-bound structures are similar, indicating that the 2Cys mutant protein is likely locked in the closed form of the tSH2 domain (Fig. 1c).
We next used limited proteolysis to (i) provide additional evidence for the presence of the disulfide bond and (ii) provide evidence for conformational restraint in the 2Cys mutant. The V8 protease was used to probe the conformational flexibility of both the native and the 2Cys mutant proteins. The unliganded form of the native tSH2 domain is cleaved by the V8 protease to produce two fragments of 16 and 18 kDa either in the presence or absence of reducing agent (0.1 M DTT) (Fig. 1d). In contrast, the unliganded 2Cys mutant was resistant to proteolysis, and very little of the cleaved fragments under nonreducing conditions was produced. When the 2Cys mutant was incubated with reducing agent (0.1 M DTT), the mutant protein was susceptible to proteolysis and was cleaved to the same extent as the native protein (Fig. 1d) to produce the same two fragments. These two fragments were transferred to nitrocellulose paper and sequenced. Cleavage occurs in the inter-SH2 domain linker region and at the same sites in both the native as well as the reduced 2Cys mutant proteins. This result suggests that after reduction of the disulfide bond by DTT, the 2Cys protein shows the same type of conformational fluctuations as the native protein, exposing the site of cleavage to proteolytic digestion. But when the disulfide bond is formed, it restricts domain motion sufficiently to abolish susceptibility of the linker region to protease cleavage.
Typically, 
, the slope of a plot of 
vs. temperature, is constant for protein folding and protein–ligand interactions over experimental temperature ranges (31, 32). Exceptions have been noted for several processes, including calmodulin binding to peptide, IMP binding to IMP dehydrogenase,
and other dehydrogenase–ligand interactions (33–35). In each of these cases, the temperature-dependent 
is attributed to the preexistence of a temperature-dependent conformational equilibrium in one of the binding partners that
is coupled to the binding process. Note that these previous studies did not seek to test this hypothesis by introducing conformational
restraints. Previously, we have shown that the heat-capacity change on dpITAM binding to the native Syk tSH2 domain is temperature-dependent
(26). The 
vs. temperature data can be fit to a two-conformer population model such as the one presented in Fig. 2c. This model was tested and confirmed by examining the temperature dependence of (i) the on-rate constant for binding and (ii) the fluorescence of the Syk tSH2 and its CD3-ε peptide complex (26). Here, we compare the temperature dependence of the binding enthalpy of the native and 2Cys mutant proteins (Fig. 2a). The CD3-ε peptide (Fig. 1a) was used because we have shown previously that no protons are exchanged on binding of this peptide under the solution conditions
examined (26). We confirm that the binding of the native tSH2 domain to the peptide exhibits a nonlinear dependence on temperature. At
low temperature, the variation of 
with temperature is nearly linear, and 
is not large; the lower limit of 
based on the three lowest temperature points is –400 cal/mol·K. At higher temperatures, the magnitudes of 
increase dramatically to –1400 cal/mol·K at the highest temperatures studied. This behavior can be understood by examination
of the scheme and Eq. 1 in Fig. 2c (see fit in Fig. 2a). Temperature induces a conformational change in one of the binding partners from “A,” which predominates at low temperature,
to “B,” which becomes populated at higher temperature. When binding occurs at higher temperatures, the observed enthalpy is
a composite of at least two exothermic components, one from the intrinsic binding enthalpy and one from the enthalpy of the
coupled B to A transition multiplied by the fractional population of the B form. Note that, as shown in our previous report
(26), all data obtained by using various methods (fluorescence titration to derive free energies and batch calorimetry to derive
binding enthalpies) are not only self-consistent but are also consistent with fitted values.
Thermodynamics of dpITAM binding to the native and Syk tSH2 domains. (a) Calorimetric enthalpies of binding as a function of temperature for both native (⋄) and 2Cys (•) tSH2 domains. (Inset) Experimental batch calorimetry results at four temperatures (thin line, 7°C; dashed line, 17°C; diamond, 22°C; and thick
line, 27°C). (b) Fluorescence titrations of the FcRIIA(⋄), CD3-ε (Δ), and FcR-γ (○) peptides with the 2Cys tSH2 domain. The fluorescence
signal of each titration was normalized for both protein concentrations, and initial fluorescence is shown as a function of
ligand concentration. (c) Proposed model to describe the thermodynamic cycle linking Syk tSH2 ligand binding to conformation and temperature. A and
B represent the two conformers of the Syk tSH2 domain in the unligated states, and AX and BX are the ligated forms. The conformational
equilibria favor the B forms of the Syk tSH2 domain as temperature increases, but the A form preferentially binds to ligand,
thus driving the equilibrium toward the AX form on ligand binding. In this scheme, superscripts preceding thermodynamic variables
represent values that are specific to a state of the protein; 0 and 1 for unligated and ligated, respectively, and A and B
for the respective conformers. Italicized subscripts denote the process the variable is associated with, b for binding, c for the conformational change, and l for the linkage of the two processes. Observed properties reflecting a combination of intrinsic processes are indicated by
the subscript obs. All nonlinear least-squares fitting was performed with the program scientist (Micromath, Salt Lake City, UT). Enthalpy data reported in Fig. 2a for the native protein were fit to a two-conformer binding model as described (26). The equation describing the behavior expected from this model for the temperature dependence of the binding enthalpy has
been developed in a previous report (26) and is 
where 
, and 
are the intrinsic enthalpies related to the equilibrium constants AKb(T), 0Kc(T), and Kl(T) for binding to the reference state (the A conformer), the conformational change, and the linkage of the two, respectively.
In sharp contrast to the temperature dependence of the heat-capacity change for dpITAM binding exhibited by the wild-type
Syk tSH2 domain, the 
of dpITAM binding to the 2Cys mutant is constant over the experimental range of temperature examined (Fig. 2a). At low temperature, the 
of binding of the 2Cys mutant protein (
) is much closer in value to the low-temperature 
of the native protein (
) than it is to the high-temperature 
of the native protein (
). However, although the binding enthalpy for the native protein becomes more negative with temperature, the 
remains constant for the binding of the 2Cys mutant protein. Hence, the binding thermodynamics of the 2Cys tSH2 domain is
no longer linked to a conformational equilibrium. The simplest explanation for the difference in temperature dependence of
binding between the wild-type and 2Cys mutant proteins is that disulfide bond formation in the 2Cys protein locks the tSH2
domain in one particular conformation. In the 2Cys mutant protein, the B form is no longer accessible. As the temperature
is raised, the A form still predominates, and the enthalpy associated with the B to A transition is no longer observed; the
heat-capacity change remains constant. That the 
of the 2Cys mutant is close to that of the low-temperature A form of the native protein suggests that the A form is the hypothesized
closed form. Hence, in the 2Cys mutant, a single conformer, the closed A form, populates the conformational space.
The native tSH2 domain of Syk binds equally well to dpITAMs with various interphosphotyrosine spacer lengths (23). The conformational flexibility of the tSH2 domain may be the source of this binding behavior; a closed form could easily adjust to the shorter spacing, whereas an open form could accommodate binding of dpITAM with a larger spacing. If this hypothesis is correct, the 2Cys mutant that appears to be locked in the closed A form should exhibit decreased affinity for the longer dpITAM while binding possibly better to the shorter dpITAM. We therefore examined binding of both the native and 2Cys mutant tSH2 domains to the CD3-ε, FcR-γ, and FcRIIA dpITAM peptides (Fig. 1a) by using fluorescence titration. Representative titrations used to determine the binding constants of the 2Cys tSH2 domain to the various dpITAM peptides are shown in Fig. 2b. The binding constants are reported in Table 1. We confirmed that the tSH2 domain of Syk is able to bind with similar affinity to the FcR-γ and FcRIIA ITAM peptides (2–4 nM), and that binding to the CD3-ε ITAM peptide is weaker (20 nM) (23). The 2Cys tSH2 domain exhibits a very large reduction in affinity for the FcRIIA peptide (62-fold) and a modest reduction in affinity for the FcR-γ peptide (15-fold). Affinity for the CD3-ε peptide is not affected. Under reducing conditions (15 mM DTT), the 2Cys tSH2 domain binds the various dpITAMs with affinities similar to wild type under the same reducing conditions (compare columns 3 and 4 of Table 1). Thus, as predicted, conformationally constraining the tSH2 domain in the closed form dramatically affects binding to dpITAMs with longer interphosphotyrosine spacing (FcRIIA), whereas binding to dpITAMs with shorter spacing is not affected (CD3-ε and FcR-γ). That binding of the 2Cys mutant to shorter dpITAMs is not enhanced (as one would expect) is likely due to a requirement for some structural fluctuations within the A form. The effect of the disulfide bridge would then be to not only drastically reduce access to the B form (as demonstrated here by the considerably reduced binding of the FcRIIA peptide) but also rigidify the A form, thus somewhat affecting binding to the A form.
Thus, the Syk tSH2 is a molecular device capable of evaluating the distance between phosphotyrosines and adjusting the relative orientation of its SH2 domains to fit that spacing. This observation is important, because it explains how Syk may fulfill its role as a player in signal transduction pathways initiated by a variety of receptors including B and T cell antigen receptors, Fc class I and III receptors, the FcγRIIA receptor, and the colony-stimulating factor receptor, as well as pathways mediated by integrins and G protein-coupled receptors. In some of these pathways, ITAMs, if required, have yet to be identified, but in at least two cases, those for the colony-stimulating factor receptor and FcγRIIA, the putative ITAMs possess primary structures that deviate considerably from canonical [Yxx(L/I)-x7/8-Yxx(L/I)] ITAMs in that they contain four and five additional residues, respectively, between phosphotyrosines. That such conformational flexibility can be experimentally revealed by using calorimetry in general and studying the temperature dependence of the heat-capacity change in particular is clearly demonstrated in this report. Such studies provide important dynamic information that complements the high-resolution “snapshots” provided by crystallographic studies and can result in a more complete understanding of the structural and dynamic processes that are coupled to macromolecular interactions.
Acknowledgments
We thank Drs. T. M. Lohman and C. Frieden for the use of fluorometers and Dr. J. I. Gordon for the use of the Microcal (Amherst, MA) calorimeter. This work was funded by National Institutes of Health Grant GM60231 (to G.W.).
Footnotes
-
↵ ¶ To whom correspondence should be addressed. E-mail: g.waksman{at}bbk.ac.uk.
-
This paper was submitted directly (Track II) to the PNAS office.
-
Abbreviations: SH2, Src homology 2; tSH2, tandem SH2; Syk tSH2, tSH2 domain of the Syk kinase; ITAM, immunoreceptor tyrosine-based activation motif; dp, doubly phosphorylated; FcR-γ, Fc receptor for IgE; FCR-IIA, Fc receptor for IgG.
- Copyright © 2003, The National Academy of Sciences
