INTRODUCTION

Phosphorylation of cellular proteins on tyrosine residues is an early event that follows triggering of a variety of transmembrane receptors on leukocytes. Protein tyrosine kinases (PTKs)¹ of the Src family are thought to be key players in many of these receptor systems (1, 2). More recently, the two currently known mammalian members of the Syk family, Syk (3, 4) and Zap (5, 6), have also been found to participate in early signaling events in lymphocytes (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15) and Syk also in other leukocytes (16, 17, 18, 19) and in erythrocytes (20). In T cells, both PTKs are rapidly phosphorylated upon receptor stimulation, and at least Zap associates with the  and CD3 chains of the T cell antigen receptor (TCR) complex in a tyrosine phosphorylation-dependent manner (5, 6, 10, 21, 22, 23). Although Syk can bind to phosphorylated receptor subunits in mast cells (18, 24), the mechanism and exact site(s) of Syk binding to the TCR are less clear, since receptor association can be detected in resting T cells in the absence of CD3 or TCR phosphorylation (8).

It is clear from recent publications that the Syk and Src family PTKs interact and synergize in inducing substrate phosphorylation (6, 8, 10, 11, 25, 27, 28, 29),² but the exact mechanism(s) or sequence of events remain incompletely understood. The reported physical association between the two classes of PTKs, which occurs in T cells following stimulation of the TCR, may offer some explanation. Indeed, Lck can associate through its SH2 domain with tyrosine phosphorylated Syk and Zap (8, 27).² To further characterize this interaction, we have identified the site in Syk that binds the Lck SH2 domain with high affinity.

MATERIALS AND METHODS

The preparation and characterization of the rabbit antiserum directed against Syk (residues 253 to 365) has been described earlier (8). The anti-Tyr(P) monoclonal antibody 4G10 was from Upstate Biotechnology Inc. (Lake Placid, NY). Anti-CD3 (OKT3) was from Ortho. The H902 monoclonal antibody, recognizing the HIV-1_IIIB-derived sequence RIQRGPGRAFVTIGK (30) which is encoded as an N-terminal epitope tag by both the pTag/SR and pTag/CMV-neo expression vectors (8, 25), is available from the AIDS Research and Reference Program (Bethesda, MD). The GST-SH2 domain of Lck (residues 121-224) construct is described elsewhere.²

The Y518F, Y519F, Y539F, Y540F, Y622F, Y623F, Y624F, and K395R mutations were made by site-directed mutagenesis of syk in the pTag/SR expression vector (8, 25) using the Transformer Site-Directed mutagenesis kit (Clontech, Palo Alto, CA), according to the manufacturer's instructions. The disabling Y192E mutation of the SH2 domain of Lck is described elsewhere.² Each mutation was verified by sequencing. COS-1 cells were grown in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum and antibiotics, and were transfected by Lipofection as described previously (8, 25).

The Saccharomyces cerevisiae strain L40 (MATa, trp1, leu2, his3, LYS2::lexA-HIS3, URA3::lexA-lacZ) and the yeast expression plasmid pBTM116 were provided by A. Vojtek (Seattle, WA); the yeast expression vector pACTII was from Clontech. These cells and plasmids were described previously (31, 32). The full-length wild-type syk cDNA, as well as the Y518F/Y519F double mutant and the K395R mutant, encoding a kinase-inactive form of the enzyme, were inserted into the pBTM116 plasmid, containing a Trp⁺ selection marker, in frame with the DNA binding domain of LexA. The Lck SH2 domain constructs were obtained by polymerase chain reaction amplification of the SH2 domain (residues 121-224) from a wild-type and Y192E-mutated Lck,² using the proofreading Vent DNA polymerase (New England Biolabs). The amplified fragments were inserted in frame with the Gal4 activation domain of the pACTII vector, which contains a Leu⁺ selection marker. Simultaneous transformation of L40 cells, maintained in standard conditions (33), with pBTM116 and pACTII-based constructs was performed as described (34, 35). Cotransformants were selected on Trp, Leu plates. The transformants were finally tested for -galactosidase activity, either by a color filter assay (36) or more frequently by a quantitative assay using 8 mM chlorophenol-red--D-galactopyranoside as a substrate, essentially as described previously (34, 35). The results are expressed as units of -galactosidase activity, as defined by Miller (37).

Immunoprecipitations were carried out as described elsewhere (8, 25, 38, 39, 40). Briefly, cells were lysed in TNE buffer (20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 5 mM EDTA, 1 mM Na₃VO₄, 10 µg/ml aprotinin, and leupeptin), and the lysates were clarified by centrifugation. Syk was immunoprecipitated with 2 µl of specific rabbit antiserum followed by agarose-conjugated goat anti-rabbit IgG (Sigma). Proteins were separated by SDS-PAGE and analyzed by immunoblotting by standard techniques. Detection by the ECL technique was carried out according to the manufacturer's instructions (Amersham Corp.).

2 × 10⁸ S. cerevisiae cells expressing Lex-A or the Lex-A:Syk hybrid were washed in water and harvested by centrifugation. The pellets were frozen on dry ice and resuspended in 400 µl of 60 mM Tris/HCl, pH 6.8, 2% SDS, 100 mM dithiothreitol, and 10% glycerol. Then 400 µl of acid-washed glass beads (Sigma) were added, and the samples were vortexed four times, 30 s each. The lysates were then heated for 5 min at 95 °C. After centrifugation for 15 min at 16,000 × g, 4 °C, the lysates were diluted 1:20 in 150 mM NaCl, 50 mM Tris (pH 7.4), 5 mM NaF, 5 mM sodium pyrophosphate, 1 mM Na₃VO₄, 10 µg/ml aprotinin, 10 µg/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, immunoprecipitated using anti-Syk antibodies, and analyzed by anti-phosphotyrosine immunoblotting as above.

Phosphorylated proteins were resolved by SDS-PAGE, transferred electrophoretically to nitrocellulose filters, localized by autoradiography, and excised. The filter pieces were treated with polyvinylpyrrolidone 360 and tosylphenylalanyl chloromethyl ketone-treated trypsin as described in detail by Luo et al. (41). The resulting peptides were separated in two dimensions on cellulose thin layer plates by electrophoresis at pH 1.9 and ascending chromatography.

GST fusion proteins of the region surrounding Tyr⁵¹⁸ and Tyr⁵¹⁹ were generated by polymerase chain reaction amplification of the region in Syk located between amino acid residues 501 and 531 using Vent DNA polymerase, and either the wild-type, Y518F, Y519F, or Y518F/Y519F mutants of the porcine syk cDNA as templates. The fragments were inserted in frame into the pGEX-3T vector. The selected clones were sequenced to ensure that the expected mutations were still present. The GST fusion proteins were purified from isopropyl thio--D-galactoside-induced bacterial cells by affinity chromatography over a glutathione-Sepharose column as before (8). The fusion proteins were phosphorylated on the residues corresponding to Tyr⁵¹⁸ and/or Tyr⁵¹⁹ using purified recombinant Lck kinase (Upstate Biotechnology Inc.). Briefly, 2 µg of purified GST fusion protein were incubated with 10 units of Lck in 50 mM Hepes, pH 7.5, containing 150 mM NaCl, 10 mM MnCl₂, and 1 mM ATP. The reaction was carried out at 30 °C for 2 h, and the mixture was then subjected to SDS-PAGE and nitrocellulose transfer followed by far-Western blotting.

To measure direct binding of the GST-SH2 domain of Lck to Syk, we used either COS-1 cell lysates or anti-Syk immunoprecipitates from these lysates. The samples were subjected to SDS-PAGE and transferred to nitrocellulose. The filters were then blocked in 20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.2% Tween-20, 5% nonfat dry milk and incubated overnight with 50 nM of the GST-SH2 construct in blocking buffer. After washing the membrane extensively in four changes of blocking buffer, the interaction between Syk and the SH2 domain was visualized by immunoblotting the membrane with a monoclonal anti-GST antibody (Santa-Cruz Biotechnology, Santa Cruz, CA) followed by peroxidase-conjugated goat anti-mouse IgG (Amersham Corp.) and ECL. To demonstrate binding of the SH2 domain to the GST fusion proteins corresponding to the autophosphorylation site of Syk, the GST-SH2 domain construct was first bound to glutathione-Sepharose and saturated with anti-GST monoclonal antibody. After washing out the excess anti-GST antibody, the GST-SH2/anti-GST complex was eluted from the beads using 20 mM reduced glutathione, pH 8.5. The eluate was diluted in 20 mM Tris/HCl, pH 7.5, 150 mM NaCl, 0.2% Tween-20, 5% nonfat dry milk, and used directly to probe the membrane. Binding of the GST-SH2·anti-GST complex to the autophosphorylation site constructs was finally detected using peroxidase-conjugated anti-mouse IgG and ECL, as before (8, 25).

RESULTS

Although we and others have reported that the SH2 domain of Lck could interact with either tyrosine-phosphorylated Syk or Zap, we felt that it was important to further demonstrate that this interaction was direct and did not require the participation of any lymphocyte-specific intermediary component. First, we used the yeast two-hybrid system developed by Fields and Song (42) to measure the interaction between the SH2 domain of Lck and various mutants of Syk. In this system, the first hybrid consisted of a fusion between the DNA binding domain of Lex-A and the full-length wild-type Syk, Y518F/Y519F-mutated Syk, kinase-inactive K395R-mutated Syk, or Zap. The second hybrid was a fusion between the transcriptional activation domain of Gal4 and the isolated SH2 domain (residues 121-224) derived from either wild-type or Y192E-mutated Lck. As described previously (34), interaction of the two hybrid molecules creates a functional transcription factor, allowing the transcription of reporter genes. In this experiment, we used the S. cerevisiae strain L40, which contains the LacZ reporter gene located downstream of Lex-A binding sites. The activation of Lex-A/LacZ can be monitored by the production of -galactosidase. Our data show that the SH2 domain of Lck can indeed interact with the wild-type Syk in S. cerevisiae (Fig. 1A). This was not the case with a functionally impaired SH2 construct (the Y192E mutant), which shows that the interaction required the function of the SH2 domain. In agreement with this notion, the Syk hybrid was found to contain Tyr(P) when expressed in S. cerevisiae (Fig. 1B). A kinase-deficient form of Syk (K395R mutant) was unable to interact with either the wild-type or Y192E-mutated SH2 domain. When the tyrosine residues corresponding to the autophosphorylation site of Syk, Tyr⁵¹⁸, and Tyr⁵¹⁹, were mutated to phenylalanine residues, the interaction with the wild-type SH2 domain was reduced to less than 8% compared to that observed between the wild-type Syk and the Lck SH2 domain construct. Although the interaction between the wild-type SH2 domain of Lck and the Y518F/Y519F-mutated Syk seems to be marginal, it is nonetheless reproducible and may indicate the presence of one or several other binding sites. If so, these sites may be of lower affinity or their phosphorylation may depend on the primary autophosphorylation site of Syk, Tyr⁵¹⁸, and Tyr⁵¹⁹, which affects catalytic activity (25). No interaction could be detected between the SH2 domain of Lck and Zap in this system (not shown), presumably due to the inability of Zap to autophosphorylate.

We have earlier reported that tyrosine phosphorylation of Syk (when expressed in COS cells) was reduced by 90% when both of the two tyrosines in the conserved autophosphorylation sites of PTKs, Tyr⁵¹⁸, and Tyr⁵¹⁹, were mutated to phenylalanines (25). In addition to this mutant, we also generated two other tyrosine-to-phenylalanine mutants of Syk. The first of these, Y539F/Y540F, was mutated at a second NYYK motif only 21 amino acids downstream of the NYYK motif containing Tyr⁵¹⁸ and Tyr⁵¹⁹. The second, Y622F/Y623F/Y624F, was mutated at the three tyrosines in the extreme C terminus of Syk. This site was chosen because the sequence following the tyrosines, DVVN, makes this site a possible Lck SH2 binding site. When expressed in COS cells all these Syk mutants contained Tyr(P) (25) (data not shown).

Wild-type Syk and its point-mutated versions were expressed in COS cells alone or in combination with Lck (to potentially increase phosphorylation of the binding site). After 48 h the cells were lysed, and the clarified lysates were incubated with 100 nM of recombinant GST fusion protein containing the SH2 domain of Lck (Fig. 2, upper panel) or control GST (Fig. 2, lower panel) for 1 h, followed by glutathione-Sepharose beads for 1 h. After washing the beads extensively, the bound proteins were eluted and analyzed by anti-Tyr(P) immunoblotting. In these experiments, wild-type Syk expressed alone bound well to the GST-SH2 domain construct (Fig. 2, lane 2), but no Tyr(P)-containing protein of 72-74 kDa could be precipitated from lysates of cells transfected with Y518F/Y519F-mutated syk (Fig. 2, lane 3). In contrast, Syk molecules containing the Y539F/Y540F or Y622F/Y623F/Y624F mutations bound well to the SH2 domain (lanes 4 and 5), indicating that neither Tyr⁵³⁹/Tyr⁵⁴⁰ nor the C-terminal Tyr⁶²²-Tyr⁶²⁴ residues are required for binding. Co-expression of Lck did not affect the binding (lanes 6-10). This experiment is in agreement with the data obtained in the yeast two-hybrid system and indicates that the autophosphorylation site of Syk needs to be intact for the creation of a binding site for the SH2 domain of Lck.

To investigate the potential binding of the Lck SH2 to Tyr⁵¹⁸/Tyr⁵¹⁹ of Syk in more detail and eliminate the possibility of intermediate molecules, we created the single mutants Y518F and Y519F. When expressed in COS-1 cells, and compared to the wild-type and Y518F/Y519F-mutated enzymes, these two single mutants contained intermediate levels of Tyr(P) and were only marginally active in vivo (Fig. 3A). Anti-Tyr(P) immunoblotting of anti-Syk immunoprecipitates from these cells showed that the Y518F mutant of Syk contained 52.0 ± 16.5% (n = 3) and the Y519F mutant 43.8 ± 18.3% (n = 3) as much Tyr(P) as wild-type Syk, whereas the Y518F/Y519F mutant contained only 3-10% as much Tyr(P) as the wild-type enzyme (Fig. 3B), as reported earlier (25). This suggests, but does not prove, that Syk is predominantly phosphorylated at both residues when expressed in COS cells.

Far-Western probing of the nitrocellulose filters with the GST-Lck SH2 followed by anti-GST revealed a reactive band (often a doublet) comparable in size to Syk only in the lysates containing wild-type Syk (Fig. 4A). Immunoblots of the same filter with the anti-tag monoclonal antibody H902 showed that at least the lower component of the doublet co-migrated with Syk. The upper band may be a hyperphosphorylated, but much less abundant, form of Syk or another cellular protein that is phosphorylated only in the presence of wild-type Syk. The Lck SH2 domain did not bind to any proteins in cells expressing the Y518F, Y519F, or Y518F/Y519F mutants of Syk (Fig. 4A, lanes 3-5). These proteins, however, were expressed at the same levels as wild-type Syk (lower panel), and they all contained Tyr(P) (Fig. 3A). Similar results were obtained in several experiments, and co-expression of Lck did not affect the result. The identification of this SH2 domain-reactive species as Syk was accomplished by repeating the far-Western probing experiment on anti-Syk immunoprecipitates obtained from Syk-transfected COS-1 cells. Once again, we detected a 72-kDa band in the precipitates obtained from wild-type Syk-expressing cells (Fig. 4B, lane 2), but not from cells expressing the Y518F- and/or Y519F-mutated enzymes (Fig. 4B, lanes 3-5). This confirms that the wild-type Syk molecule can interact directly with the SH2 domain of Lck and suggests that phosphorylation at both Tyr⁵¹⁸ and Tyr⁵¹⁹ is required for the binding to occur.

Although our results demonstrate that the SH2 domain of Lck can interact directly with Syk, and that this interaction depends on an intact autophosphorylation site, they do not exclude the possibility that the SH2 domain of Lck interacts with another site, the phosphorylation of which depends on the full enzymatic activity of Syk and prior phosphorylation of both Tyr⁵¹⁸ and Tyr⁵¹⁹. To directly address this question, we generated GST fusion proteins containing the Syk-derived sequence surrounding the wild-type or mutated autophosphorylation site, residues 506-531 of the porcine enzyme (Fig. 5A). In order to generate tyrosine-phosphorylated versions of these constructs, we subjected each GST fusion protein to an extended in vitro kinase reaction using a recombinant purified Lck enzyme. When the phosphorylated constructs were subjected to tryptic peptide mapping, the wild-type protein generated three phosphopeptides (Fig. 5B), one of which, peptide 2, was not present in the maps derived from any of the mutated fusion proteins. This peptide corresponds to the doubly phosphorylated autophosphorylation peptide. Peptide 1 was present in all maps, except that of the Y518F/Y519F mutant, and thus corresponds to a singly phosphorylated peptide. Peptide 3, present in all maps, represents a phosphorylation site located within the GST portion of the fusion molecules. This shows that the wild-type GST-Syk peptide can be phosphorylated on both Tyr⁵¹⁸ and Tyr⁵¹⁹ by Lck in vitro. To determine whether the SH2 domain of Lck binds to the autophosphorylation site of Syk and whether it requires phosphorylation of both Tyr⁵¹⁸ and Tyr⁵¹⁹, we performed a far-Western probing experiment on each of the constructs phosphorylated in vitro by Lck. We found that only the wild-type (doubly phosphorylated) construct was recognized by the SH2 domain of Lck (Fig. 5C, lane 1), whereas the Y518F or Y519F single mutants or the Y518F/Y519F double mutant of this construct were not (Fig. 5C, lanes 2-4).

DISCUSSION

The published investigations concerning the physical interaction between Syk or Zap and the Src-family kinase Lck all support the conclusion that the SH2 domain of Lck is involved in this interaction (8, 27, 28). However, the assumption that this interaction is direct and does not require any intermediate components needed to be addressed. In this report, we show that the interaction between Syk and the SH2 domain of Lck can occur in a system (the yeast two-hybrid system) devoid of other lymphocyte-specific components. Furthermore, the binding of the SH2 domain depended on the catalytic activity of Syk itself, indicating that the binding site is likely to be an autophosphorylation site for Syk, an assumption that is verified by our experiments. The inability of the Y518F/Y519F-mutated Syk to interact with the SH2 domain of Lck in intact cells (Fig. 1) or in solution (Fig. 2) further suggests that the conserved autophosphorylation site of Syk, Tyr⁵¹⁸, and Tyr⁵¹⁹, might be the target of the Lck SH2 domain. The sequence at this region is ADENY*Y*KAQTHG. This sequence is not a perfect match with the predicted specificity of Src family SH2 domains (43), but if the first phosphorylated tyrosine goes into the Tyr(P) binding pocket of the SH2 domain, the +1 position would be occupied by the acidic Tyr(P) and the +3 amino acid would be a hydrophobic alanine. This may create an adequate docking site for the SH2 domain of Lck.

The receptor-induced tyrosine phosphorylation sites in Syk in activated lymphocytes have not yet been identified, but Zap was recently reported to be predominantly phosphorylated in triggered T cells at the two tyrosines that correspond to Tyr⁵¹⁸ and Tyr⁵¹⁹ in Syk, namely Tyr⁴⁹² and Tyr⁴⁹³ (44, 45). In the case of Zap, however, phosphorylation of both Tyr⁴⁹² and Tyr⁴⁹³ clearly depends on the presence of Src-family kinases. Indeed, Lck can phosphorylate Tyr⁴⁹³ in vitro, an event that augments the enzymatic activity of Zap, which in turn autophosphorylates on Tyr⁴⁹² (44, 45, 46). Therefore, one expects the SH2 domain of Lck to be unable to bind Zap unless it has been first phosphorylated by Lck. This is supported by our finding that the SH2 domain of Lck did not interact with Zap in the yeast two hybrid system, or when expressed in COS-1 cells in the absence of Lck (not shown). In contrast, Syk was able to bind to the SH2 domain of Lck under the same conditions. This suggests that, in absence of Lck, Syk is capable of autophosphorylating at both Tyr⁵¹⁸ and Tyr⁵¹⁹, as reported (25), whereas Zap is unable to phosphorylate the corresponding Tyr⁴⁹² and Tyr⁴⁹³ residues (44). This may well be the reason why the activation of Syk is independent of Lck or other Src-family kinases in a variety of systems (8, 11, 25, 47). Nevertheless, the SH2 domain-dependent recruitment of Src-family kinases by Syk might be a critical event in receptor-mediated lymphocyte activation (28). In B cells, the integrity of the autophosphorylation site of Syk (or the corresponding site in Zap) is required for B cell antigen receptor-mediated signaling (48, 49). Since the enzymatic activity of the Y518F/Y519F mutant of Syk is reduced by only 40%, the inability of this mutant to activate downstream signaling events in B cells (48), T cells,³ or COS cells (25) (Fig. 2), possibly stems from the absence of Syk/Src-family kinase interactions.

In addition to Lck, it was recently reported that two other molecules could interact with Syk through SH2-dependent mechanisms. First, the C-terminal SH2 domain of PLC-1 was shown to interact with Tyr³⁴⁸ and/or Tyr³⁵² of the human Syk enzyme (corresponding to Tyr³⁴¹ and Tyr³⁴⁵ of porcine Syk), located between the C-terminal SH2 domain and the kinase domain of Syk (26, 50). Second, we have found that the SH2 domain of the Vav proto-oncogene product binds to Tyr³⁴¹ of Syk, both in vitro and in vivo.⁴ Importantly, binding of either SH2 domain to their respective binding sites was largely dependent on prior phosphorylation of Tyr⁵¹⁸ and Tyr⁵¹⁹ (or Tyr⁵²⁵ and Tyr⁵²⁶ of the human sequence), although these two residues did not participate in the intermolecular interaction per se (26).⁴ Furthermore, all three SH2 domain-containing ligands of Syk have been shown to be (25, 26),² or are likely to be (51), substrates for Syk. In the case of Lck, Syk has been shown to phosphorylate the Tyr¹⁹² residue (25),² located near the +3 amino acid binding pocket of the Lck-SH2 domain. This event regulates the function of the SH2 domain.²