Autophosphorylation of the Fes tyrosine kinase. Evidence for an intermolecular mechanism involving two kinase domain tyrosine residues.

The human c-fes proto-oncogene encodes a cytoplasmic tyrosine kinase (Fes) that is associated with multiple hematopoietic cytokine receptors. Fes tyrosine autophosphorylation sites may regulate kinase activity and recruit downstream signaling proteins with SH2 domains. To localize the Fes autophosphorylation sites, full-length Fes and deletion mutants lacking either the unique N-terminal or SH2 domain were autophosphorylated in vitro and analyzed by CNBr cleavage. Identical phosphopeptides of 10 and 4 kDa were produced with all three proteins, localizing the tyrosine autophosphorylation sites to the C-terminal kinase domain. Substitution of kinase domain tyrosine residues 713 or 811 with phenylalanine resulted in a loss of the 10- and 4-kDa phosphopeptides, respectively, identifying these tyrosines as in vitro autophosphorylation sites. CNBr cleavage analysis of Fes isolated from 32PO4-labeled 293T cells showed that Tyr-713 and Tyr-811 are also autophosphorylated in vivo. Mutagenesis of Tyr-713 reduced both autophosphorylation of Tyr-811 and transphosphorylation of Bcr, a recently identified Fes substrate, supporting a major regulatory role for Tyr-713. Wild-type Fes transphosphorylated a kinase-inactive Fes mutant on Tyr-713 and Tyr-811, suggesting that Fes autophosphorylation occurs via an intermolecular mechanism analogous to receptor tyrosine kinases.

The human c-fes proto-oncogene encodes a cytoplasmic tyrosine kinase (Fes) that is associated with multiple hematopoietic cytokine receptors. Fes tyrosine autophosphorylation sites may regulate kinase activity and recruit downstream signaling proteins with SH2 domains. To localize the Fes autophosphorylation sites, full-length Fes and deletion mutants lacking either the unique N-terminal or SH2 domain were autophosphorylated in vitro and analyzed by CNBr cleavage. Identical phosphopeptides of 10 and 4 kDa were produced with all three proteins, localizing the tyrosine autophosphorylation sites to the C-terminal kinase domain. Substitution of kinase domain tyrosine residues 713 or 811 with phenylalanine resulted in a loss of the 10-and 4-kDa phosphopeptides, respectively, identifying these tyrosines as in vitro autophosphorylation sites. CNBr cleavage analysis of Fes isolated from 32 PO 4 -labeled 293T cells showed that Tyr-713 and Tyr-811 are also autophosphorylated in vivo. Mutagenesis of Tyr-713 reduced both autophosphorylation of Tyr-811 and transphosphorylation of Bcr, a recently identified Fes substrate, supporting a major regulatory role for Tyr-713. Wild-type Fes transphosphorylated a kinase-inactive Fes mutant on Tyr-713 and Tyr-811, suggesting that Fes autophosphorylation occurs via an intermolecular mechanism analogous to receptor tyrosine kinases.
The human c-fes gene encodes a non-receptor protein-tyrosine kinase (Fes) that is homologous to the transforming oncogenes encoded by several avian (v-fps) and feline (v-fes) retroviruses (1). Fes expression is restricted primarily to hematopoietic cells of myeloid origin in adults (2)(3)(4), suggesting that it functions in normal myeloid growth regulation. Transfection of immature myeloid cells with c-fes or v-fps leads to terminal differentiation, consistent with a role for this family of tyrosine kinases in differentiation signal transduction (5,6).
Further evidence that Fes may represent a component of a myeloid differentiation signaling cascade comes from its asso-ciation with hematopoietic growth factor receptors. Several studies have linked Fes to the receptors for granulocyte-macrophage colony-stimulating factor, erythropoietin, interleukin-3, interleukin-4, and interleukin-6 (7)(8)(9)(10)(11). In each case, binding of ligand to the receptor led to stimulation of Fes tyrosine kinase activity. Fes was also demonstrated to associate with the receptors for these cytokines or with the signal transducing component gp130 in the case of IL-6.
Although Fes is often associated with hematopoietic growth regulation, other work suggests a more general physiological role for this tyrosine kinase. Expression of an activated form of Fes in transgenic mice led to hypervascularization, suggesting that Fes may function in angiogenesis (12). Fes also exhibits widespread expression in embryonic tissues, suggestive of an essential role in early development (13). Despite these multiple biological functions, the mechanisms regulating Fes tyrosine kinase activity and the signal transduction pathways in which Fes participates have not been well characterized.
Human Fes is 822 amino acids in length and can be divided into three distinct structural regions: a unique N-terminal domain, a Src homology 2 (SH2) 1 domain, and a C-terminal kinase domain. All three domains have the potential to regulate Fes kinase activity and interaction with signaling partners. The unique N-terminal region has recently been shown to contain a recognition domain for the cellular Bcr protein (14). In the same study, phosphorylation of Bcr by the transforming Fes homolog v-Fps was shown to induce the association of Bcr with the Grb-2/Sos guanine nucleotide exchange complex, suggesting that Bcr may link Fes to the Ras signal transduction pathway (14). The SH2 domains of both c-Fes and v-Fps are critical positive regulators of kinase activity and may provide binding sites for effectors or regulatory proteins (15,16). The Fes kinase domain contains tyrosine autophosphorylation sites, which may regulate both kinase activity and substrate recognition. Mutagenesis of the homologous autophosphorylation sites shared by c-Fes (Tyr-713) and v-Fps (Tyr-1073) greatly diminishes kinase activity, indicating an important role for these sites in the regulation of enzymatic function (15,17). Autophosphorylated tyrosine residues in the kinase domain may also represent binding sites for substrates with SH2 domains, such as the Ras GTPase-activating protein (18,19). In this study, we describe the identification of Tyr-811 as the second principal Fes autophosphorylation site, and demonstrate for the first time that Tyr-811 as well as the other major autophosphorylation site (Tyr-713) are both utilized in vivo. In addition, we provide direct evidence that Fes autophosphorylation is an intermolecular event, suggesting that oligomerization may be required for Fes activation.

Expression of GST-Fes Fusion Proteins and in Vitro Phosphorylation
Reactions-Construction of plasmid vectors for the expression of Fes N-terminal amino acids 1-450, 1-347, and 1-126 as fusion proteins with GST is described elsewhere (14). To express Fes amino acids 225-540 and kinase domain residues 551-822 as GST fusion proteins, the coding sequences for these regions of Fes were amplified by PCR and cloned into the expression vector pGEX-2T (20). To generate the GST fusion protein containing the inactive Fes kinase domain, a Fes cDNA containing a point mutation in the ATP-binding site (Lys 590 to Glu) was used as the PCR template (15). The GST-Fes fusion proteins were expressed in Escherichia coli and purified with glutathione agarose as described elsewhere (14,15,19). The proteins were eluted from the beads with 50 mM Tris-HCl, pH 9.0, containing 6 M urea and dialyzed extensively against 20 mM Hepes, pH 7.4. The concentration of each protein was determined by two-dimensional laser densitometry (Molecular Dynamics). Phosphorylation reactions were conducted in 60 l of kinase buffer (20 mM Hepes, pH 7.4, 5 mM MgCl 2 , 5 mM MnCl 2 ) containing 100 nM fusion protein, 10 Ci of [␥-32 P]ATP (3,000 Ci/mmol; Du Pont NEN), and recombinant, immunopurified Fes (see below). Phosphorylated proteins were separated by SDS-PAGE and visualized by autoradiography.
Addition of the FLAG and GST Sequences to Fes-PCR was used to add the coding sequence of the 8-amino-acid FLAG epitope (DYKD-DDDK) to the C-terminal region of full-length, wild-type Fes. The PCR reaction utilized a forward primer that maps to the 5Ј end of the kinase domain and a reverse primer complementary to the C-terminal coding region of Fes. The reverse primer also contained the FLAG coding region, a stop codon, and a unique restriction site (EcoRI). Following PCR, the resulting product was digested with BamHI and EcoRI and swapped with the corresponding restriction fragment in the native Fes sequence. The nucleotide sequence of the PCR-derived fragment was confirmed by automated DNA sequence analysis (Applied Biosystems). The resulting FLAG-Fes cDNA was utilized as the template to generate the mutants described below. A similar PCR-based procedure was used to add the FLAG sequence to the Fes N terminus. In this case, a forward oligonucleotide primer was employed that encoded a unique cloning site, a translational initiation sequence, and the FLAG coding sequence, followed by Fes homologous sequences.
To create a baculovirus construct for the expression of GST fusion proteins, the GST coding sequence, polylinker region, and stop codons of pGEX-2T (Pharmacia Biotech Inc.) were amplified by PCR and subcloned into the baculovirus transfer vector pVL1393 to create the vector pVL-GST. The GST-N-terminal region of Fes was subcloned into pVL-GST from an existing pGEX-2T construct (14). To create a transfer vector containing GST fused to full-length Fes, the DraIII-EcoRI fragment of Fes was subcloned into the pVL-GST-N-Fes construct.
Fes Mutants-Generation of the Fes mutants Y713F, K590E, and ⌬SH2 is described elsewhere (14,15). These mutants were tagged with the C-terminal FLAG epitope by swapping their C-terminal BamHI-EcoRI fragments for the C-terminal FLAG-modified fragment described above. The ⌬N mutant shown in Fig. 2 was generated using PCR to amplify the coding sequence of the SH2 and kinase domains. The forward primer contained an ATG start codon and a consensus sequence for the initiation of translation, while the reverse primer contained the FLAG sequence and a stop codon as described above. The Y799F and Y811F Tyr 3 Phe point mutants were generated using a two-step PCR mutagenesis procedure. In the first step, mutagenic oligonucleotides were used to change the coding sequence from Tyr to Phe. The product of the first reaction was purified and used as a megaprimer (21) in a second PCR reaction to create a cDNA fragment that could be digested with EagI and EcoRI and swapped with the corresponding wild-type fragment to create a full-length mutant cDNA. To create the Y799F/Y811F double mutant, PCR was conducted with the Tyr-799 mutagenic oligonucleotide and the Y811F mutant cDNA as template. To create the Y713F/Y811F double mutant, the C-terminal EagI-EcoRI restriction fragment of the Y811F single mutant was swapped for the corresponding fragment in the Y713F mutant. The nucleotide sequences of all PCR-derived mutant fragments used for subcloning was verified by automated DNA sequence analysis.

Generation of Recombinant Baculoviruses and Expression of Fes Proteins in Sf-9 Insect Cells-Mutant
Fes cDNAs were subcloned into the baculovirus transfer vector pVL1392 and the resulting constructs were used to generate recombinant baculoviruses using Baculogold DNA and the manufacturer's protocol (Pharmingen, San Diego, CA). For protein expression, subconfluent monolayers of Sf-9 cells were infected with recombinant Fes baculoviruses and incubated for 48 h. Infected cells were sonicated in 1.0 ml of ice-cold lysis buffer (50 mM Tris-HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 1 mM MgCl 2 and 0.1% Triton X-100) supplemented with 25 g/ml aprotinin, 50 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 20 mM NaF, 1 mM Na 3 VO 4 , and 50 M Na 2 MoO 4 . The cell lysates were clarified by centrifugation and the Fes proteins were immunoprecipitated with the M2 anti-FLAG monoclonal antibody resin. The Fes immunoprecipitates were washed with radioimmune precipitation buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1 mM EDTA, and 1% sodium deoxycholate). For direct analysis by immunoblotting, the proteins were eluted in SDS-PAGE sample buffer. For in vitro kinase reactions, the proteins were washed with kinase buffer prior to addition of [␥-32 P]ATP. Culture and maintenance of Sf-9 insect cells is described in detail elsewhere (14,22,23).
Cyanogen Bromide Cleavage and Phosphopeptide Analysis-Wildtype and mutant Fes proteins were expressed in Sf-9 cells, autophosphorylated in vitro with [␥-32 P]ATP, and resolved by SDS-PAGE. The 32 P-labeled Fes proteins were extracted from the gel and precipitated with trichloroacetic acid as described elsewhere (24). The precipitates were dried and resuspended in 70% formic acid containing fresh CNBr (20 mg/ml), and cleavage was allowed to proceed overnight at room temperature. The samples were then diluted with water and lyophilized, this process was repeated once, and the resulting dried peptides were redissolved in SDS-Tricine gel sample buffer. CNBr fragments were separated on 10 -20% polyacrylamide Tricine gradient gels (Novex) and visualized by autoradiography or storage phosphor technology using a Molecular Dynamics PhosphorImager.
Expression of Fes in 293T Cells and in Vivo Labeling-293T human embryonic kidney cells (25) were maintained in Dulbecco's modified Eagle's medium supplemented with 5% fetal bovine serum. Cells were plated in six-well dishes (5 ϫ 10 5 cells/well) and incubated for 24 h at 37°C prior to transfection using a modified calcium phosphate method. Briefly, 30 g of the expression vector pcDNA3 (In Vitrogen) containing various Fes cDNAs was diluted in 450 l of sterile water and 60 l of 2 M CaCl 2 was added to the DNA solution. The DNA/CaCl 2 suspension was added dropwise to an 0.5-ml aliquot of 2 ϫ HEPES-buffered saline solution (42 mM Hepes, 274 mM NaCl, 10 mM KCl, 1.8 mM Na 2 HPO 4 , pH 7.1) and incubated on ice for 10 min. The DNA suspension (0.25 ml/well) was then added dropwise to the cells with gentle agitation. After incubation for 3 days at 37°C, the cells were washed and incubated in 1.0 ml of PO 4 -free Dulbecco's modified Eagle's medium for 1 h at 37°C. The cells were labeled with 2.0 mCi of 32 PO 4 /ml (ICN) for 4 h at 37°C. After incubation, the medium was aspirated and the cells were immediately frozen by floating the plate on liquid nitrogen. Fes lysis buffer (1.0 ml/well) was added to the frozen plate. After thawing at 4°C, the lysates were clarified by centrifugation and Fes proteins were immunoprecipitated from the resulting supernatants with the M2 anti-FLAG antibody resin and resolved by SDS-PAGE. The radiolabeled Fes protein was extracted from the gel and subjected to cyanogen bromide cleavage and Tricine gradient gel electrophoresis as described above. Phosphoamino acid analysis of in vivo labeled Fes peptides was performed as described elsewhere (19).

Localization of Fes Autophosphorylation Sites to the C-terminal Kinase Domain in Vitro-Previous phosphopeptide map-
ping studies of autophosphorylated c-Fes resulted in two phosphopeptides with either trypsin or Staphylococcus aureus V8 protease, suggesting that Fes can autophosphorylate at least two tyrosine residues in vitro (4,15,22). We have observed that one of the Fes autophosphorylation sites is located at Tyr-713, within the C-terminal catalytic domain (15). To determine which domain of Fes contains the additional tyrosine autophosphorylation site, we expressed a series of GST fusion proteins containing various combinations of the Fes unique N-terminal, SH2, and kinase domains in E. coli (Fig. 1A). Equimolar amounts of the purified GST-Fes fusion proteins were phosphorylated in vitro with recombinant Fes and [␥-32 P]ATP and analyzed by SDS-PAGE and autoradiography. As shown in Fig.  1B, only the fusion proteins containing sequences derived from the Fes kinase domain were strongly phosphorylated by recombinant Fes.
The GST-Fes fusion proteins used in this experiment were eluted from glutathione-agarose beads in 6 M urea, followed by extensive dialysis (see "Experimental Procedures"). The GST-kinase domain fusion protein purified under these conditions showed no detectable autokinase activity (data not shown), indicating that the phosphorylation shown in Fig. 1B was likely to be the result of transphosphorylation by full-length Fes. To verify this point, an additional GST-kinase domain fusion protein was constructed which contains a Glu substitution for Lys at amino acid position 590 (KIN-KE mutant). This mutation renders the kinase domain inactive (15). The GST-Fes fusion protein containing this mutant kinase domain was also readily phosphorylated by wild-type Fes in vitro (Fig. 1B). These data show that all of the Fes autophosphorylation sites are localized to the C-terminal kinase domain. Furthermore, they indicate that Fes is capable of intermolecular trans-autophosphorylation in a manner analogous to tyrosine kinases of the receptor class (26,27). Further evidence of an intermolecular mechanism of Fes autophosphorylation is presented below.
A second approach to verify that all of the Fes tyrosine autophosphorylation sites are localized to the C-terminal region involved mutants of Fes lacking the unique N-terminal or SH2 domains (Fig. 2A). These mutants and full-length wildtype Fes were expressed as C-terminal FLAG fusion proteins in Sf-9 cells. The Fes fusion proteins were isolated from infected cell lysates with the anti-FLAG monoclonal antibody and phosphorylated in vitro with [␥-32 P]ATP. The autophosphorylated kinases were cleaved with CNBr, and the resulting phosphopeptides were resolved by SDS-Tricine gradient gel electrophoresis. The positions of possible Fes CNBr cleavage sites are illustrated in Fig. 3A. As shown in Fig. 3B, CNBr cleavage of the full-length and deletion mutants of Fes resulted in three phosphopeptides of identical electrophoretic mobility in each case. This result supports the hypothesis the Fes autophosphorylation sites are localized to the kinase domain and is in good agreement with the data shown in Fig. 1.
The 10-and 4-kDa phosphopeptides produced by CNBr cleavage of autophosphorylated Fes correspond most closely in molecular mass to predicted fragments containing Tyr-713, a known autophosphorylation site, and a C-terminal CNBr fragment containing two candidate tyrosine residues (Tyr-799 and Tyr-811), respectively. To establish that the larger fragment contains Tyr-713, a mutant of Fes containing a Tyr 3 Phe substitution at this position was autophosphorylated in vitro and subjected to CNBr cleavage. As shown in Fig. 3C, cleavage of this mutant produced only the 4-kDa phosphopeptide, identifying the 10-kDa fragment as the one containing Tyr-713. To determine whether the 4-kDa phosphopeptide corresponds to the C-terminal CNBr fragment, we tested for the presence of the FLAG epitope within this fragment. Wild-type Fes containing the C-terminal FLAG epitope was autophosphorylated in vitro, cleaved with CNBr, and the immunoreactivity of the resulting peptides was tested using the anti-FLAG monoclonal antibody. As shown in Fig. 3D, the 4-kDa CNBr phosphopeptide was immunoprecipitated by the anti-FLAG antibody, confirming its identity as the C-terminal fragment. The third phosphopeptide produced in some of these experiments is approximately 14 kDa. Experiments described below indicate that this fragment is a partial digestion product containing the 10-and 4-kDa peptides.
Identification of Tyr-713 and Tyr-811 as Fes Autophosphorylation Sites in Vitro-Data described above identify Tyr-799, Tyr-811, or both as possible candidates sites for Fes autophosphorylation in vitro. To resolve this issue, we generated mutants with individual point mutations at these sites (Y799F and Y811F mutants) and with mutations of both sites (Y799F/ Y811F mutant; see Fig. 2). Wild-type and mutant forms of Fes were autophosphorylated in vitro and subjected to CNBr cleavage analysis. As shown in Fig. 4, Phe substitution of Tyr-799 alone did not affect the observed CNBr cleavage pattern. However, substitution of Tyr-811 led to a complete loss of the 4-kDa peptide, identifying this residue as a second c-Fes autophosphorylation site in vitro. As expected, an identical result was obtained with the Y799F/Y811F double mutant. Identification of Tyr-811 as a second site of Fes autophosphorylation is consistent with previous phosphopeptide mapping studies of c-Fes, which consistently produced two phosphopeptides (4,15,22).
Analysis of Fes Phosphorylation in Vivo-To investigate the phosphorylation of Fes in living cells, we expressed Fes with a C-terminal FLAG epitope tag in the human embryonic cell line 293T (25) and labeled the cells with 32 PO 4 . Fes was immunoprecipitated, resolved by SDS-PAGE, and analyzed by CNBr cleavage. As shown in Fig. 5A (left panel, C-FLAG lane), the 14-, 10-, and 4-kDa bands were observed, consistent with the in vitro result. Phosphoamino acid analysis shows that each of these peptides contain phosphotyrosine (Fig. 5B). A shorter exposure of this gel revealed that the major labeled fragment resulting from the in vivo analysis is approximately 34 kDa (Fig. 5A, right panel). Phosphoamino acid analysis of the 34-kDa fragment revealed that it contains phosphoserine (Fig.  5B). This result is in good agreement with previous studies of Fes phosphorylation in vivo, in which phosphoserine was the predominant or exclusive amino acid observed (4,28). Based on the predicted sites for CNBr cleavage of Fes, this 34-kDa serine phosphopeptide is likely to contain the SH2 domain and flanking N-and C-terminal residues (see Fig. 3A). Phosphorylation of this region by a serine/threonine kinase may contribute to the negative regulation of Fes tyrosine kinase activity. Previous studies have shown that Fes has weak transforming and tyrosine kinase activities in vivo (28,29).
To investigate whether the 4-kDa tyrosine phosphopeptide was derived from the Fes C-terminal region in vivo, we also expressed Fes as an N-terminal FLAG fusion protein in 293T cells (Fig. 5A, N-FLAG lane). Following 32 PO 4 labeling, the CNBr cleavage pattern of the N-FLAG and C-FLAG Fes proteins were compared. As shown in Fig. 5A, the 4-kDa phosphopeptide produced by CNBr cleavage of the C-FLAG Fes protein shifted down in the gel when digested from N-FLAG Fes. The size of the shifted band is consistent with the Cterminal Fes fragment without the FLAG epitope. A downward shift was also observed with the 14-kDa fragment, consistent with the assignment of this fragment as a partial digestion product containing the 10-and 4-kDa peptides.
To determine whether Tyr-713 and Tyr-811 are phosphorylated in vivo, mutant Fes proteins containing Phe substitutions at these positions were expressed in 293T cells, labeled with 32 PO 4 , and subjected to CNBr cleavage analysis as described above. As shown in Fig. 6, mutation of Tyr-713 led to the complete disappearance of the 10-kDa phosphopeptide, indicating that Tyr-713 is an autophosphorylation site in vivo. Mutagenesis of Tyr-713 also resulted in a loss of the 4-kDa peptide, indicating that phosphorylation of Tyr-713 is required for Tyr-811 phosphorylation to occur in vivo. Additional evidence supporting this hypothesis is presented below. The CNBr map of the Y713F mutant also lacked the 14-kDa band, consistent with the identity of this fragment as a partial digestion product containing Tyr-713. Mutagenesis of Tyr-811 resulted in the loss of the 4-kDa phosphopeptide, identifying Tyr-811 as an in vivo autophosphorylation site. Mutagenesis of Tyr-799 was without effect, while the Tyr-799/811 double mutant produced a cleavage pattern identical to that observed with the Tyr-811 mutant. These data support the conclusion that the same tyrosine autophosphorylation sites are utilized in vitro and in vivo, and identify for the first time in vivo autophosphorylation sites for Fes. None of these mutations affected Ser phosphorylation of Fes in vivo, as judged by the equal intensities of the 34-kDa fragments from each of the mutants (Fig. 6, peptide 3). were consistently observed and are indicated by the arrows. Additional higher molecular weight peptides were observed in some cases and correspond to incomplete digestion products. C, comparative CNBr cleavage analysis of Fes wild-type and Tyr-713 3 Phe (Y713F) autophosphorylation site mutant. D, immunoprecipitation of peptide 2 with the anti-FLAG monoclonal antibody. 32 P-Labeled CNBr fragments were prepared from wild-type Fes, lyophilized, and resuspended in immunoprecipitation buffer containing immobilized anti-FLAG monoclonal antibody (M2). Following incubation and washing, the bound peptide was eluted in gel loading buffer and resolved by Tricine gradient gel electrophoresis. An aliquot of the starting peptide mixture is shown for comparison (WT) .   FIG. 4. Fes Tyr-811 is autophosphorylated in vitro. Fes mutants with Tyr to Phe substitutions at positions 799, 811, or both were expressed in the baculovirus/Sf-9 cell system, immunoprecipitated, and incubated in vitro with [␥-32 P]ATP. The autophosphorylated proteins were subjected to CNBr cleavage analysis as described in the legend to Fig. 3. The resulting CNBr fragments were resolved on a Tricine gradient gel and visualized using storage phosphor technology. Molecular mass standards are shown on the left, and the positions of Fes phosphopeptides 1 and 2 are indicated on the right. The position of the 14-kDa partial cleavage product containing peptides 1 and 2 is also shown (1ϩ2). WT, wild type.

Autophosphorylation of Tyr-713 Is Required for Full Activation of c-Fes-Previous studies of the Fes Y713F mutant indi-
cate that the extent of autophosphorylation of this mutant is reduced by more than 90% relative to the wild-type protein (15,30). In addition, this mutant shows reduced kinase activity toward model substrates in vitro. To determine whether Tyr-811 is also involved in the regulation of kinase activity, the autophosphorylation capacity of Fes proteins carrying either individual Y713F or Y811F point mutations or both was evaluated in Sf-9 cells. As shown in Fig. 7A, both proteins with Tyr-713 mutations showed greatly reduced autophosphorylation capacity. On the other hand, the Tyr-811 single mutant still reacted strongly with the anti-phosphotyrosine antibody, suggesting that Tyr-811 does not regulate the phosphorylation of Tyr-713.
To investigate the extent of Y811F autophosphorylation in comparison to wild-type Fes more quantitatively, an equal amount of each immunopurified protein was incubated in vitro with [␥-32 P]ATP and the extent of autophosphorylation was monitored over time. In two separate experiments, the extent of Y811F autophosphorylation was observed to be 60 -70% of the wild-type level (data not shown). This result is consistent with the distribution of 32 P in the 10-kDa and 4-kDa peptides observed both in vivo and in vitro (approximately a 2:1 Tyr-713:Tyr-811 ratio). These results support the conclusion that Tyr-811 does not influence the autophosphorylation of Tyr-713.
To determine the effect of autophosphorylation site mutations on substrate phosphorylation, these mutants were coexpressed with the Bcr protein, which has recently been identified as a substrate for c-Fes and its transforming homolog, v-Fps (14). As shown in Fig. 7B, mutagenesis of Tyr-713 greatly reduced Tyr phosphorylation of Bcr, whereas mutagenesis of Tyr-811 was without effect. Taken together, these results indicate that autophosphorylation of Tyr-713 is required for autophosphorylation of Tyr-811 and for maximal substrate phosphorylation, as observed previously with the model substrate enolase in vitro (15). Autophosphorylation of Tyr-811 may create a docking site for signaling proteins with SH2 domains (see below).
Autophosphorylation of Fes Is an Intermolecular Event-A final mechanistic question that we wished to address was whether autophosphorylation occurs by an intra-or intermolecular mechanism. Demonstration that Fes autophosphorylates in an intermolecular fashion would support a model for activation similar to that proposed for growth factor receptor tyrosine kinases (26,27) and for non-receptor tyrosine kinases that associate with cytokine receptors such as those of the Jak kinase family (31,32). To answer this question, we co-expressed a kinase-inactive form of full-length Fes (K590E mutant; see Fig. 2) with a catalytically active GST/Fes fusion protein in Sf-9 cells. The GST/Fes fusion protein was used because it can be readily distinguished from the K590E mutant on Western blots (120 kDa versus 93 kDa, respectively). The GST/Fes and Fes-KE proteins were immunoprecipitated and analyzed for the presence of phosphotyrosine by immunoblotting. As shown in Fig. 8, expression of the Fes kinase-inactive mutant alone resulted in no detectable tyrosine autophosphorylation. However, co-expression of the kinase-inactive mutant with the active GST/Fes fusion protein resulted in transphosphorylation of the mutant to approximately the same extent as the wild-type protein. CNBr cleavage analysis of Fes-KE following transphosphorylation by GST-Fes produced a cleavage pattern identical to that observed with autophosphorylated wild-type Fes (data not shown). Although this experiment does not rule out the possibility of intramolecular autophosphorylation, it is consistent with a model of Fes activation by oligomerization and transphosphorylation.
Data presented here show that the Fes tyrosine kinase autophosphorylates two kinase domain Tyr residues (713 and 811) both in vitro and in vivo. By analogy to growth factor receptor tyrosine kinases, autophosphorylation of these Fes tyrosine residues may form binding sites for substrate proteins with SH2 domains. Structural studies of the SH2 domains of Src and other proteins show that SH2 binding specificity is often conferred by the amino acids immediately C-terminal to the phosphotyrosine residue (33). These findings led Songyang et al. (34,35), to predict possible SH2 domain binding specificities using a degenerate phosphopeptide library in which every amino acid (except Trp or Cys) was represented at the ϩ1, ϩ2, and ϩ3 positions relative to the Tyr(P) residue. Phosphopep-tides were selected from the library with recombinant SH2 domains, and their sequences were determined directly. Comparison of these predicted SH2-binding motifs with the amino acid sequences immediately adjacent to Fes Tyr-811 (Gln-Glu-Leu) revealed partial matches to ten predicted SH2 binding motifs. A particularly striking example was the C-terminal SH2 domain of Syk, which is a perfect match for the Fes Tyr-811 SH2 binding sequence and one predicted by Songyang et al. (35). Syk is a member of a unique tyrosine kinase family with tandem N-terminal SH2 domains that exhibits hematopoietic expression and has been implicated in integrin, cytokine, and antigen receptor signal transduction (36 -38). Recent evidence suggests that multiple cytoplasmic tyrosine kinases are recruited to antigen and cytokine receptors in response to ligand binding (32). SH2-phosphotyrosine interactions may allow for interaction among the tyrosine kinases involved in the receptor-kinase complex (39), which may include Fes in the case of several hematopoietic cytokine receptors (see Introduction).
By contrast to Tyr-811, the ϩ1, ϩ2, ϩ3 motif adjacent to Tyr-713 (Ala-Ala-Ser) showed very little resemblance to any of the predicted SH2 domain binding motifs. Thus, Tyr-713 may function primarily to regulate kinase activity rather than in substrate recruitment. Data shown in Fig. 7 are consistent with this hypothesis; mutants of Fes containing the Tyr-713 to Phe substitution showed greatly reduced autophosphorylation capacity and ability to transphosphorylate Bcr.
Figs. 1 and 8 provide evidence that Fes can autophosphorylate via an intermolecular mechanism. These results are significant in the context of possible physiological mechanisms of Fes activation by cytokines. Recent studies of cytokine receptors suggest a general mechanism of activation that involves ligand-induced receptor dimerization or oligomerization, followed by activation of multiple receptor-associated tyrosine kinases of the cytoplasmic class (31,40). Receptor oligomerization is likely to activate the associated kinases by a transphosphorylation mechanism reminiscent of growth factor receptor tyrosine kinases (26,32,41). Recruitment of Fes into an activated cytokine receptor complex could lead to Fes activation by this mechanism. Because cytokines control pleiotropic responses in myeloid cells (growth, differentiation, and function), their receptors must be able to activate a diverse array of signal transduction pathways that control these responses. The ultimate biological effect of a given cytokine is likely to be dependent upon the complement of signaling molecules present in the cell at the time of challenge with the factor. Activation of Fes in the context of a myeloid progenitor may contribute to differentiation signal transduction.