Activation of Stat3 by the Src Family Kinase Hck Requires a Functional SH3 domain

this investigated the mechanism of Stat3 activation by the Src family of non-receptor tyrosine kinases, which have been linked to Stat activation in both normal and transformed cell types. Using Sf-9 insect cells, we demonstrate direct Stat3 tyrosine phosphorylation and stimulation of DNA binding activity by five members of the Src kinase family (Src, Hck, Lyn, Fyn, and Fgr). We also observed stable Stat3-Src family kinase complex formation in this system. Recombinant Src family kinase SH3 domains were sufficient for interaction with Stat3, suggesting a mechanistic basis for Src kinase-Stat3 interaction. To test the contribution of Src family kinase SH3 domains to the recruitment and activation of Stat3 in vivo, we used Rat-2 fibroblasts expressing activated mutants of the myeloid Src family member Hck. Transformation of fibroblasts by an activated Hck mutant lacking the negative regulatory tail tyrosine residue (Hck-YF) induced strong DNA-binding activity of endogenous Stat3. Inactivation of Hck SH3 function by Ala substitution of a conserved Trp residue (W93A mutant) completely abolished Stat3 activation by Hck-YF and reduced transforming activity by 50% without affecting Hck kinase activity. Finally, over-expression of Stat3 in Rat-2 cells transiently stimulated Hck and c-Src kinase activity in the absence of extracellular signals, an effect that was dependent upon a putative SH3-binding motif in Stat3. These results support a model in which Src family kinases Stat3 through an SH3-dependent mechanism, resulting in transient kinase activation and Stat3 phosphorylation.


(Summary)
Stat3 is a member of a family of transcription factors with SH2 domains that are activated by tyrosine phosphorylation in response to a wide variety of cytokines and growth factors. In this study, we investigated the mechanism of Stat3 activation by the Src family of non-receptor tyrosine kinases, which have been linked to Stat activation in both normal and transformed cell types. Using Sf-9 insect cells, we demonstrate direct Stat3 tyrosine 4 induce a transformed phenotype in fibroblasts, providing evidence for an essential role of this SH3-mediated interaction to overall regulation of kinase activity (18,19). Together, these SH2 and SH3-mediated interactions work through an allosteric mechanism to push the two lobes of the kinase domain together, thus preventing substrate binding and stabilizing the inactive form of the kinase domain.
The crystal structures of the down-regulated forms of Src family kinases suggest that physiological protein-protein interactions through these domains may induce transient kinase activation, leading to substrate phosphorylation. In this report, we investigated the SH3-dependent interaction of the Stat3 transcription factor with Src family kinases in vitro and in vivo. Stats are latent cytoplasmic transcription factors that become activated by tyrosine phosphorylation, which induces Stat dimerization, nuclear translocation and transcriptional regulation (20,21). A growing body of evidence strongly supports the idea that Stats are substrates for Src family kinases, particularly Stat3 (22). For example, transformation of rodent fibroblasts with constitutively activated Src kinases results in strong activation of Stat3, and activation appears to be essential for transformation (23)(24)(25)(26).
Here we show that Stat3 is a direct substrate for all members of the Src kinase family tested, and that the SH3 domains of Src kinases are sufficient for Stat3 binding in vitro. µg primary antibody and 20 µl of protein G-Sepharose (50% slurry, Amersham Biosciences) for 2 h at 4° C. The immunoprecipitates were collected by centrifugation and washed twice with cold radioimmune precipitation assay (RIPA) buffer (18,27). For immunoblotting, clarified lysates and immunoprecipitates were heated in SDS-PAGE sample buffer and resolved on SDS-polyacrylamide gels and transferred to PVDF membranes. Following incubation with primary antibody, immunoreactive proteins were detected with an appropriate secondary antibody-alkaline phosphatase conjugate and NBT/BCIP colorimetric substrate (Sigma) or CDP-Star chemiluminescent alkaline phosphatase substrate (Perkin-Elmer). For glutathione (GSH)-agarose precipitation experiments, GSH-agarose beads (20 µl of a 50% w/v slurry; Sigma) were used to precipitate the GST-Stat3/Src kinase protein complexes; beads were washed with RIPA buffer and associated Src kinases were visualized by immunoblotting.
Expression of GST-SH3 domain fusion proteins and Stat3 binding assay -For GST-SH3 fusion protein expression, PCR fragments encoding the SH3 domains of Hck, c-Src, Lyn, Fyn and Hck-W93A were subcloned into the bacterial expression vector pGEX-2T (Amersham Biosciences). GST and GST-SH3 fusion proteins were expressed in E. coli and immobilized on GSH-agaose beads as described elsewhere (30,31). Clarified lysates from Sf-9 cells expressing recombinant Stat3 were incubated with 10 µg of each GST-SH3 fusion protein or GST alone for 2 h at 4° C. Stat3-SH3 complexes were precipitated by centrifugation, washed extensively in RIPA buffer, and eluted by heating in SDS-PAGE sample buffer. The samples were resolved by SDS-PAGE and Stat3 was detected by immunoblotting.

Production of recombinant retroviruses and transformation assays -Retroviral
expression vectors for wild-type Hck, the activated tail mutant (Hck-YF), and kinasedefective Hck (Hck-KE) have been described elsewhere (18,33,34). The SH3-inactivating mutation W93A was introduced into full-length Hck using the Gene Editor oligonucleotidedirected mutagenesis kit (Promega). This mutation was combined with Hck-YF by restriction fragment swapping to create Hck-YFW. These Hck clones as well as the coding sequences for GFP, Stat3, and Stat3-2PA were subcloned into the retroviral expression vector pSRαMSVtkneo (35). These vectors were used to generate high-titer retroviral stocks in 293T cells by co-transfection with an ecotropic packaging vector (36).
Rat-2 fibroblasts (ATCC) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS and 50 µg/ml gentamycin. Retroviral stocks were supplemented with Polybrene (hexadimethrine bromide, Sigma) to 4 µg/ml and added to 9 fibroblasts in 6-well plates (2 x 10 5 cells/well). The plates were centrifuged at 1,000 x g for 3 h at room temperature to enhance infection efficiency. For the focus-forming assay, 2 x 10 4 infected cells were plated in 60 mm culture dishes in in the presence of 800 µg/ml G-418. Transformed foci were visualized 10-14 d later by Wright-Giemsa staining. For the soft-agar assay, 1 x 10 4 infected cells were plated in 35 mm culture dishes in DMEM containing 0.3% Seaplaque agarose (FMC).
Colonies were stained with iodonitrotetrazolium violet formazan (Sigma) and counted using a BioRad Imaging Densitometer and colony-counting software (37).

Direct phosphorylation of Stat3 on Tyr 705 by Src family kinases -Recent studies
have implicated Src family kinases in Stat3 activation, both in normal and transformed cell types (22,40). To investigate whether Stat3 is a direct substrate for the Src kinase family, c-Src, Hck, Lyn, Fyn and Fgr were co-expressed with Stat3 in Sf-9 insect cells using recombinant baculovirus vectors. Sf-9 cells provide a useful system to study direct tyrosine kinase-Stat3 interaction, because they lack homologs of mammalian tyrosine kinases involved in Stat3 activation (28,41). In addition, Src family kinases are constitutively active in Sf-9 cells due to the absence of Csk. Stat3 was immunoprecipitated from co-infected cell lysates and immunoblotted with anti-phosphotyrosine antibodies. As shown in Figure   1, Stat3 was strongly phosphorylated by all of the Src family members tested.
Immunoblotting of a duplicate membrane with phosphospecific antibodies that recognize does not appear to be due to differences in Stat3 protein levels or tyrosine phosphorylation ( Figure 1), and may instead relate to less efficient release of Stat3 from these kinases following tyrosine phosphorylation. These results indicate that Stat3 is a common substrate for the Src kinase family, and show that phosphorylation and activation occur via a direct interaction.

Stat3 forms stable complexes with Src family kinases -Previous reports have shown
that Src can be co-immunoprecipitated with Stat3 from fibroblasts and other cell types, suggesting that these proteins interact in vivo (25,42). To determine if this interaction is dependent upon other mammalian proteins, Src kinases were co-expressed with a GST-Stat3 fusion protein in Sf-9 insect cells. GST-Stat3 was precipitated from Sf-9 cell lysates with GSH-agarose beads, and Stat3-bound Src kinases were detected by immunoblotting. As shown in Figure 3, all of the Src kinases tested in this assay readily formed complexes with GST-Stat3, while no interaction was observed with GST alone.
These results indicate that Src family kinases have the potential to bind directly to Stat3 in vivo. Control blots show that GST and GST-Stat3 were expressed at the same level in the infected cell lysates, and that each culture contained the same amount of each active Src family member. We also investigated the dependence of the interaction on tyrosine phosphorylation using a kinase-defective mutant of Hck. This mutant also formed a stable complex with GST-Stat3, indicating that recruitment of Stat3 is independent of kinase function (data not shown).  domain to Stat3 in Figure 4C. The Hck-YFW mutant was expressed in Rat-2 fibroblasts, and Stat3 DNA binding activity was assessed by EMSA. As shown in Figure 5A, the SH3 mutation completely blocked the ability of Hck-YF to activate endogenous Stat3. As a control, the W93A mutation was also introduced into wild-type Hck, and the effect of this SH3 mutation on endogenous Stat3 activation was assessed by gel-shift assay. Figure 5A shows that this mutant was also unable to activate endogenous Stat3, despite release of kinase activity as a result of the mutation (see below). These results provide evidence for an SH3-dependent mechanism of Stat3 recognition by Src family kinases in vivo.

Src family kinase SH3 domains bind Stat3 in vitro -
To determine whether loss of Stat3 activation correlates with reduced expression of a Stat3 target gene, we examined cyclin D1 expression in the same panel of Rat-2 fibroblasts. Previous studies have shown that activation of Stat3 is sufficient to induce cyclin D1 expression in this cell type (45). Rat-2 fibroblasts expressing each of the Hck constructs were serum-starved and cyclin D1 levels were determined by immunoblotting.
As shown in Figure 5B immunoblotting with phosphospecific antibodies for the phosphorylated activation loop. We also examined tyrosine phosphorylation of p40, an endogenous Hck substrate reported previously (18,27,33,34). Figure 6 shows that p40 is phosphorylated on tyrosine by both  Figure 7. Wild-type Hck did not demonstrate transforming activity in either assay, while Hck-YF was strongly transforming in both systems, consistent with earlier results (18,33). Introduction of the SH3 mutation reduced but did not abolish the transforming activity of Hck-YF, consistent with the idea that Stat3 activation is important for the transforming signal of Hck in fibroblasts but is not solely responsible for the transformed phenotype. Hck bearing only the SH3 point mutation also showed lower focus-and colony-forming activities relative to Hck-YF, despite readily detectable autophosphorylation ( Figure 6). This is because the SH3 domain mutation has two opposing actions on Hck transforming function. On one hand, the W93A mutation releases the SH3 domain from the SH2-kinase linker, leading to loss of negative regulation.
On the other hand, this mutation also uncouples Hck from Stat3 and other substrates that are recruited via SH3. The net effect is the partial transforming phenotype seen in Figure   7. Morphological differences in the foci formed by Rat-2 cells expressing the various Hck mutants were observed as as well. Hck-YF produced large spreading foci, while those formed by Hck-W93 and Hck-YFW were smaller with more tightly clustered cells (data not shown). These data clearly show that an intact SH3 domain and Stat3 activation are required for the full transforming activity of Hck in Rat-2 fibroblasts. However, the ability of Hck mutants to transform cells in the absence of Stat3 activation suggests that Stat3 is not absolutely required for transformation to occur (see Discussion).

Co-expression with Stat3 activates c-Src tyrosine kinase activity -Previous work
from our laboratory has shown that engagement of Src family kinase SH3 domains by other proteins is sufficient to induce activation of the kinase domain in vivo (18,33,34). The mechanism of activation involves displacement of the negative regulatory interaction between SH3 and the polyproline helix formed by the SH2-kinase linker region (see Introduction). Our observations that Src family kinase SH3 domains are sufficient for Stat3 binding ( Figure 4) and that SH3 function is required for Stat3 recruitment in vivo ( Figure 5) suggest that Stat3 binding might activate Src family kinases. Indeed, the β-barrel region of Stat3 has a short PxxP motif with identical spacing to HIV Nef, a well-established SH3binding protein and activator of Hck (Figures 4 and 8A) (33,34,46). To investigate the effect of Stat3 on Src family kinase activity, Rat-2 fibroblasts stably expressing c-Src or Hck were infected with Stat3 or GFP retroviruses. Forty-eight hours later, c-Src and Hck were immunoprecipitated and assayed in vitro using a Sam-68 fusion protein (p50) as substrate.
As shown in Figure 8

DISCUSSION
Stats are latent cytoplasmic transcription factors that require tyrosine phosphorylation for their dimerization and activation (21). Many families of both receptor and non-receptor protein-tyrosine kinases have been implicated in Stat activation (47). In this report, we focused on the molecular mechanisms of Src family kinase interaction with Stat3. A growing body of evidence has implicated the Src kinase family in Stat3 activation, both in normal and transformed cell types (23)(24)(25)41,42,(48)(49)(50)(51)(52)(53) In a related experiment, we found that the W93A mutation released the Hck kinase domain from negative regulation in vivo, yet this mutant was also unable to recruit and activate Stat3. Together, these findings provide a compelling argument for the SH3-dependent recruitment and activation of Stat3 by Hck and other Src family kinases, at least in this model system. An activating effect of Src family kinase SH3 domain mutations has also been reported for c-Src and Lck (57)(58)(59). Whether or not these mutations also affect Stat3 recruitment and phosphorylation has not been reported.
Previous studies have established that Stat3 activation is important for fibroblast transformation by activated mutants of Src. Constitutive Stat3 activation and Src-Stat3 complex formation is readily detected in Src-transformed fibroblasts (24,25,52). More recent studies have shown that Stat3 activation may be required for Src-induced transformation (23,26). These experiments used dominant-negative forms of Stat3, which were found to block Src-mediated transformation of fibroblasts. While our data suggest that Stat3 activation is required for maximal transformation by the Hck tail mutant (Hck-YF), uncoupling Hck-YF from Stat3 activation through SH3 mutation reduced but did not eliminate transforming activity (Figure 7). This finding suggests that Stat3 is important but not absolutely required for fibroblast transformation by Hck-YF.
The finding that the W93A SH3 mutation was able to release Hck from negative regulation in vivo provides further evidence for a dominant role for the SH3 domain in the negative regulation of Hck. The crystal structures of Hck and Src show that the SH3 domain engages the SH2-kinase linker, and that this interaction helps to maintain the kinase in the inactive conformation (see Introduction). Binding of proteins to the SH3 domain, which induces SH3-linker displacement, and mutagenesis of the linker prolines essential for intramolecular SH3 binding both lead to constitutive Hck activation and fibroblast transformation, even in the presence of tail phosphorylation and SH2 engagement (33). Work presented here shows that mutagenesis of SH3 also promotes constitutive Hck autophosphorylation in vivo, presumably by releasing the negative regulatory interaction with the SH2-kinase linker ( Figure 6). However, this mutant shows substantially lower transforming activity compared to the Hck SH2-kinase linker mutant