Tesk1 Interacts with Spry2 to Abrogate Its Inhibition of ERK Phosphorylation Downstream of Receptor Tyrosine Kinase Signaling*

The Sprouty (Spry) proteins function as inhibitors of the Ras-ERK pathway downstream of various receptor tyrosine kinases. In this study, we have identified Tesk1 (testicular protein kinase 1) as a novel regulator of Spry2 function. Endogenous Tesk1 and Spry2 exist in a complex in cell lines and mouse tissues. Tesk1 coexpression relocalizes Spry2 to vesicles including endosomes, inhibiting its translocation to membrane ruffles upon growth factor stimulation. Independent of its kinase activity, Tesk1 binding leads to a loss of Spry2 function as an inhibitor of ERK phosphorylation and reverses inhibition of basic fibroblast growth factor (bFGF)- and nerve growth factor-induced neurite outgrowth in PC12 cells by Spry2. Furthermore, depletion of endogenous Tesk1 in PC12 cells leads to a reduction in neurite outgrowth induced by bFGF. Tesk1 nullifies the inhibitory effect of Spry2 by abrogating its interaction with the adaptor protein Grb2 and interfering with its serine dephosphorylation upon bFGF and FGF receptor 1 stimulation by impeding its binding to the catalytic subunit of protein phosphatase 2A. A construct of Tesk1 that binds to Spry2 but does not localize to the vesicles does not interfere with its function, highlighting the importance of subcellular localization of Tesk1 in this context. Conversely, Tesk1 does not affect interaction of Spry2 with the E3 ubiquitin ligase, c-Cbl, and consequently, does not affect its inhibition of Cbl-mediated ubiquitination of the epidermal growth factor receptor. By selectively modulating the downstream effects of Spry2, Tesk1 may thus serve as a molecular determinant of the signaling outcome.

Sprouty (Spry) was first discovered in Drosophila as a negative feedback regulator of receptor tyrosine kinase signaling (1,2), a function that is seemingly conserved in its four mammalian homologs, Spry1-4 (3,4). Although the Spry proteins contain no known domains, they all have a unique, highly conserved Cys-rich C-terminal region. This sequence was not known to exist in any other protein until the discovery of Spred (Sproutyrelated proteins with EVH1 domain) (5), of which three mammalian isoforms have been currently identified. Various isoforms of Spry and Spred are reported to be deregulated in cancers, including those of the breast, prostate, and liver (reviewed in Ref. 6), suggesting a potential role for them as tumor suppressors. A recent report that Spry2 regulates oncogenic K-Ras signaling (7) has added strength to this argument. Understanding their regulation and mechanism of action is thus a necessary step toward validating these proteins as therapeutic targets for disease treatment.
Spry and Spred share several features, including extensive homology over their Cys-rich C terminus (8). The best characterized function for both protein families is as inhibitors of ERK 3 phosphorylation downstream of a range of stimuli including growth factors and cytokines. However, their similarity does not seem to extend to their mode of action. Although various groups have reported the point of action of Spry as being either upstream of Ras by disrupting the Grb2⅐Sos complex (9) or at the level of Raf activation (10,11), the current understanding of Spred function suggests that it may interfere with Ras-induced activation of Raf (5).
Tesk1 belongs to a novel family of cofilin kinases comprising of two members (Tesk1 and 2) that share similarity with the LIM (Lin-11, Isl-1, and Mec-3) kinases in the catalytic domain (18). Tesk1 regulates actin dynamics in the context of cell adhesion and spreading downstream of integrin signaling by phosphorylating cofilin on the Ser residue at position 3, thereby inactivating it. Spry4 has previously been reported to bind to Tesk1 and negatively regulate its kinase activity, resulting in inhibition of cell spreading (14,19). In this study, we present nal) antibody or normal rabbit immunoglobulin G (IgG) as control. The resulting immunoprecipitates were separated on SDS-PAGE and blotted with mouse anti-Tesk1 and rabbit anti-Spry2 (N-terminal) to detect the proteins.
PC12 Neurite Outgrowth Assay-PC12 cells were plated on poly-L-lysine-coated coverslips and transfected with the various constructs indicated. 24 h post transfection, the cells were quiesced in medium containing 0.2% fetal bovine serum and subsequently stimulated with 50 ng/ml basic fibroblast growth factor for 4 days or nerve growth factor for 2 days. The cells were fixed and visualized as described below.
Immunofluorescence Microscopy-Cos1 and PC12 cells were fixed, stained, and visualized essentially as described (21) with the following modifications. Permeabilized cells were blocked overnight with phosphate-buffered saline supplemented with 1 mM each of CaCl 2 and MgCl 2 , 2% bovine serum albumin, and 7% fetal bovine serum. Primary and secondary antibodies were diluted in the same solution and incubated at room temperature for 1 h each. The images were captured using a Zeiss LSM 510 META (Carl Zeiss Microimaging).

Tesk1 Is a Common Interacting Partner to Spred and Spry-
With the view that the nature of associated proteins will provide clues to the physiological function of newly discovered proteins such as Spred, we performed a yeast two-hybrid screen using an adult mouse brain library with full-length Spred1 as bait. One of the positive clones recovered from the screen encoded for the last 100 amino acid residues of Tesk1 (designated C100, Fig. 1A). When expressed in HEK293T cells, this fragment of Tesk1 coimmunoprecipitates with full-length Spred1 (Fig. 1B). Among the different regions of Spred1 (schematically shown in Fig. 1A), namely the EVH1 (residues 1-136), the kinase-binding domain (residues 137-332), and the Cysrich region (C terminus; residues 333-444), C100 interacts strongly with the Cys-rich C terminus of Spred1 (Fig. 1C).
Since the Cys-rich C terminus of Spred is conserved with that of the Spry proteins and a previous study reported that Spry4 interacts with Tesk1 through this region (19), we sought to verify whether Tesk1 may be a common interacting partner to both of these protein families. Furthermore, as only the C-terminal fragment of Tesk1 was isolated in the screen, we wanted to validate this interaction in the context of the full-length protein. When coexpressed, all isoforms of Spry and Spred bind to full-length Tesk1 ( Fig. 2A). A mutant of Tesk1 where the catalytic aspartate at position 170 is mutated to alanine (D170A) has been shown to lack kinase activity (22). We used this mutant

Tesk1 Reverses Inhibition of ERK by Spry2
(designated DA) and a truncation mutant where the last 100 residues equivalent to C100 were deleted (designated Tr528) to check whether (a) the kinase activity of Tesk1 might influence its binding to Spred and Spry and (b) the binding is mediated by the C-terminal tail of Tesk1. Although the DA mutant binds to Spred1 and Spry2 similar to WT Tesk1, Tr528 does not bind FIGURE 1. Tesk1 interacts with the Cys-rich C terminus of Spred1. A, schematic diagram showing the domain structure and constructs of Tesk1 and Spred1. Kinase, kinase domain; CR1-3, conserved regions 1-3; C100, yeast two-hybrid fragment, residues 529 -628 of Tesk1; EVH1, Ena-VASP homology domain 1; KBD, kinase-binding domain; C terminus, Cys-rich C terminus. B, lysates of 293T cells transfected with Myc-C100 and FLAG-Spred1 were immunoprecipitated (IP) using anti-FLAG antibody. The immunoprecipitates were separated on SDS-PAGE and blotted with the antibodies indicated to the left. Expression of the transfected constructs was verified by immunoblotting (IB) total cell lysates (TCL). C, the different fragments of Spred1 depicted in A were coexpressed with C100 in 293T cells. The lysates were subjected to immunoprecipitation with anti-FLAG antibody followed by immunoblotting of the immunoprecipitation and TCL samples with the antibodies indicated to the left. FL, full-length; E, Ena-VASP homology domain 1; K, kinase-binding domain; C, C terminus. FIGURE 2. Tesk1 is a common interacting partner to Sprouty and Spred that binds via its C-terminal 100 residues. A, full-length Myc-Tesk1 was coexpressed with full-length FLAG-tagged Spry1-4 and Spred1 and 2 in 293T cells. The lysates were immunoprecipitated (IP) with antibody against FLAG and immunoblotted (IB) with anti-Myc to detect presence of Tesk1 in the immunoprecipitates. Expression of the constructs was verified by immunoblotting the total cell lysates (TCL). B, 293T cells were transfected with FLAG-Spry2 or Spred1 and the indicated mutants of Myc-Tesk1. The lysates were subsequently processed as described in A. WT, wild type Tesk1; DA, kinase dead Tesk1 D170A; Tr528, truncation mutant of Tesk1 from residues 1-528. C, lysates containing equal amounts of total protein from mouse tissues and cell lines were immunoprecipitated with rabbit anti-Spry2 (N terminus) or normal rabbit IgG as control. The resulting immunoprecipitates were separated on SDS-PAGE and blotted with anti-Tesk1 and anti-Spry2 as indicated. *, IgG heavy chain. (Fig. 2B). Spry4 also showed a similar pattern of binding (data not shown). Thus, the binding seems to be mediated solely through the C-terminal 100 residues of Tesk1, independent of its kinase activity.
Tesk2 is a closely related kinase to Tesk1 that shares about 70% similarity in the kinase domain (23). The rest of their sequences have little similarity to each other except for three conserved regions designated CR1-3 (Fig. 1A). If the C terminus of Tesk1 mediated binding to Spry/Spred, it would be expected that Tesk2 would not bind as this region is not conserved between the two isoforms. This is indeed the case since none of the Spry or Spred isoforms binds to full-length Tesk2 (supplemental Fig. S1).
Tesk1 is expressed at relatively high levels in testes and at lower levels in most other tissues and cell lines (18,24). To determine whether Tesk1 interacts endogenously with Spry and Spred, we immunoprecipitated endogenous Spry2 from lysates of cell lines and mouse tissue and assayed for the presence of Tesk1 in the immunoprecipitates. Endogenous Tesk1 coimmunoprecipitates with Spry2 in mouse testes and brain tissue, indicating that these proteins form a complex in vivo. A complex was also detected in CHO-K1 cells but not in 293T cells (Fig. 2C). The other isoforms of Spry and Spred could not be tested due to a lack of specific antibodies. To study the downstream effects of Tesk1 interaction, we used Spry2 in the subsequent experiments as the result in Fig. 2C provided evidence of an endogenous complex with this isoform in addition to its being the best characterized and reportedly the strongest inhibitor of ERK phosphorylation (9,10).
Tesk1 Alters the Subcellular Localization of Spry2-To determine whether Spry2 colocalized with Tesk1 in cells, we used indirect immunofluorescence to visualize their localization in Cos1 cells. Spry2 on its own localizes to microtubules and translocates to membrane ruffles upon EGF stimulation (Fig.  3A), as reported earlier (13,25,26). On the other hand, fulllength Tesk1 WT and DA localize to vesicles irrespective of stimulation. Interestingly, the Tr528 mutant no longer localizes to the vesicles, suggesting that the C-terminal tail of Tesk1 may play a role in its own subcellular localization. The vesicular localization of Tesk1 has been previously reported (19), but the identity of these vesicles has not been determined. To this end, we costained the cells with markers for various subcellular compartments. A subset of the Tesk1-positive vesicles overlaps with the early endosomal markers EEA1 and Rab5 and the late endosomal marker Rab7 but not the Golgi marker GM130 or the lysosomal marker LAMP1 (Fig. 3B). There was also no colocalization with the endoplasmic reticulum marker calreticulin or the mitochondrial marker cytochrome c (data not shown). We noted that a subpopulation of Tesk1 vesicles did not overlap with any of the markers used in this study, the identity of which remains to be determined.
When coexpressed, Spry2 localizes to the vesicles with Tesk1 WT, and its translocation to membrane ruffles upon EGF stimulation is greatly reduced (Fig. 4, panels a and b). The DA mutant of Tesk1 has a similar effect to the WT, whereas the Tr528 does not affect Spry2 localization (panels c-f), indicating that this effect of Tesk1 is mediated by its binding rather than its kinase activity.
The membrane translocation of Spry2 is known to be mediated by its Cys-rich C terminus (21,25). A fragment of Spry2 lacking this region (Spry2-N; residues 1-164) exhibits a cytoplasmic distribution both in the basal and the EGF-stimulated state in contrast to the Cys-rich region (Spry2-C; residues 165-315), which shows a clear translocation to the membrane upon The cells were fixed and stained with mouse anti-FLAG and rabbit anti-Myc antibodies followed by anti-mouse IgG (green) and anti-rabbit IgG (red). WT, wild type Tesk1; DA, kinase dead Tesk1 D170A; Tr528, truncation mutant of Tesk1 from residues 1-528. B, Cos1 cells transfected with Myc-Tesk1 WT were fixed and stained with rabbit anti-Myc and mouse anti-EEA1, Rab5, GM130, and LAMP1 followed by anti-mouse (green) and anti-rabbit (red) IgG, as labeled in the panels. For Rab7, EGFP-tagged Rab7 was coexpressed with Tesk1. The boxed areas are shown enlarged in the far right column. Bar, 20 m. stimulation (Fig. 5A). As reported previously, neither the N nor the C terminus of Spry2 alone exhibits the microtubular localization seen with the full-length protein (25). After confirming that Spry2 binds to Tesk1 through its C terminus in an immunoprecipitation experiment (Fig. 5B), we determined the effect of Tesk1 coexpression on the localization of these constructs.
Although the subcellular distribution of Spry2-N is not altered in the presence of Tesk1 (Fig. 5C, panels a and b), the change in the localization of Spry2-C reflects that of the full-length both in the basal and stimulated states (panels c and d). This is also in agreement with the earlier observation that the homologous region of Spred1 mediates the binding to Tesk1 (Fig. 1C).
Tesk1 Negatively Regulates the Role of Spry2 as an Inhibitor of the Ras/ERK Pathway-One of the best known functions of Spry2 is as an inhibitor of the Ras/ERK pathway (8). We used two different systems to determine whether its interaction with Tesk1 had any influence on this function. In the first, cells cotransfected with Spry2 and Tesk1 were stimulated with bFGF, and the levels of phosphorylated ERK were assayed. As seen in Fig. 6A, although Spry2 on its own inhibits ERK phosphorylation compared with control levels ( lanes 5 and 7), coexpression of Tesk1 reverses this effect (lane 8), whereas Tesk1 itself does not significantly affect the levels of phosphorylated ERK on its own. We further verified this effect of Tesk1 using overexpression of FGFR1, which is auto-activating, to reflect long term signaling as seen during oncogenesis and development. In this system as well, Tesk1 reverses the inhibition of ERK phosphorylation by Spry2 (Fig. 6B, lanes 8 and 9). It appears that the kinase activity of Tesk1 does not play a role in this effect, since the kinase dead DA mutant has a similar effect to the WT (Fig. 6C, lane 15), whereas the Tr528 construct that retains its kinase activity, but no longer binds to Spry2, does not influence its function (Fig. 6C, lane 17).
Tesk1 Rescues Neurite Outgrowth Inhibited by Spry2 in PC12 Cells-One of the widely studied downstream consequences of the Ras/ERK inhibitory effect of Spry2 is its inhibition of bFGFinduced neurite outgrowth in the rat pheochromocytoma cells, PC12 (9,(27)(28)(29). To establish whether the reversal of its ERK inhibition by Tesk1 translated to a cellular context, we studied the effect of Tesk1 and Spry2 coexpression in PC12 cells. As previous reports in the literature have highlighted the tight regulation on cofilin phosphorylation and dephosphorylation in neurite outgrowth (30,31), we wanted to avoid interference from the cofilin kinase activity of Tesk1 in this assay. To this end, we used the kinase dead DA mutant of Tesk1 because it still binds to Spry2 and reverses its inhibition of ERK phosphorylation similar to WT Tesk1. When expressed in PC12 cells, Spry2 almost completely abrogates bFGF-induced neurite outgrowth (Fig. 7A, panel a). Although Tesk1 DA does not affect neurite outgrowth on its own under the same conditions (panel b), it restores neurite outgrowth in these cells in the presence of Spry2 (panel c). As further evidence, we also studied the differentiation of PC12 cells in response to NGF. The result in Fig. 7B shows that similar to the case with bFGF, Spry2 also inhibits neurite outgrowth induced by NGF (panel a), which is reversed by coexpression of Tesk1 DA (panel c).
If Tesk1 indeed antagonizes the inhibitory effect of Spry2, it follows that endogenous Tesk1 may play a positive role in this process. To determine whether this is the case, we used shRNA to knock down endogenous Tesk1 in PC12 cells. Because the Myc-Tesk1 used in this study is the rat isoform, we first tested the silencing efficiency of the shRNAs against this construct before introducing them into PC12 cells. Of those tested, shRNA#9 was the most effective in decreasing the expression of rat Tesk1 (Fig. 7C). This or the noneffective shRNA#12 were cotransfected into PC12 cells with EGFP to visualize the transfected cells. As seen in Fig. 7D, expression of shRNA#9 causes a significant decrease in bFGF-induced neurite outgrowth, whereas shRNA#12 does not have any effect.
Tesk1 Inhibits Grb2 Interaction and Serine Dephosphorylation of Spry2-Our next aim was to delineate the molecular basis for the loss of the inhibitory function of Spry2 in the presence of Tesk1. Several studies have implicated the interaction between Spry2 and the adaptor protein Grb2 as being fundamental to its inhibitory action (2,9,29). In the currently proposed model, Spry2 derails the Ras/ERK pathway by competing with the RasGEF Sos1 for binding to Grb2 (9). Because Tesk1 abrogated the inhibitory function of Spry2, we ascertained its effect on Spry2:Grb2 interaction. As seen in Fig. 8A, Tesk1 almost completely abolishes the binding of Spry2 to Grb2 (comparing lanes 7 and 8), possibly resulting in the loss of Spry2 function.
In the course of this experiment, we also noticed that Tesk1 may affect Ser phosphorylation of Spry2. Spry2 migrates as at least two distinct bands in PAGE gels. An earlier report showed that the slower migrating band is a result of phosphorylation, as it disappears upon phosphatase treatment (13). The same study also showed that Spry1 and 2 were phosphorylated exclusively on Ser residues. In a recent report from our group, we have shown that Spry2 is basally phosphorylated on specific, conserved Ser residues in the 112-132 region, which get dephosphorylated upon FGFR1 stimulation (32), resulting in an increase in the faster migrating form in PAGE gels (Fig. 8A, compare  lanes 3 and 7). This forms an essential step in the activation of Spry2 for its interaction with Grb2 and subsequent inhibition of the Ras/ ERK pathway. Tesk1 coexpression seemingly inhibits this dephosphorylation because Spry2 still migrates as two distinct bands in the presence of Tesk1, even upon FGFR1 stimulation (Fig. 8A, lanes 7 and 8). Dephosphorylation of endogenous Spry2 in HEK293T cells upon bFGF stimulation was also inhibited by Tesk1 (Fig. 8B). We further verified that the slower migrating band in the presence of Tesk1 is indeed due to phosphorylation by treating the samples with alkaline phosphatase (supplemental Fig. S2) and using a mutant of Spry2 where a Ser phosphorylation site contributing to the slower migrating band is substituted with alanine (S115A) (32). The migration of this mutant, which runs as a single band corresponding to the faster migrating band of the WT, is not affected by Tesk1 coexpression (supplemental Fig. S3). The kinase activity of Tesk1 does not play a role in this effect as kinase dead Tesk1 DA has the same effect, whereas Tr528 does not (Fig. 8C, lanes 14 and 16). This result, while ruling out the possibility that Tesk1 directly phosphorylates Spry2, suggests that it involves the interaction between the two proteins.
PP2A is a multi-subunit Ser/threonine (Thr) phosphatase that has been implicated in the dephosphorylation of Spry2 (32,33). In the latter study we have further shown that both the regulatory A and the catalytic C subunits of PP2A bind to Spry2, with the A subunit binding directly to the 50 -60 region. To determine whether Tesk1 inhibited Spry2 dephosphorylation by interfering with its interaction with PP2A, we analyzed the binding of PP2A-A and -C subunits to Spry2 upon Tesk1 coexpression (Fig. 8D). Although PP2A-A binding to Spry2 is not affected, the binding of PP2A-C is greatly reduced in the presence of Tesk1, both in the basal and the FGFR1-stimulated conditions. Lack of the catalytic C subunit in the complex would thus abolish the dephosphorylating activity of PP2A on its substrate Spry2.
Vesicular Localization Is Essential for the Negative Effect of Tesk1-The negative effect of Tesk1 could potentially arise from two different scenarios. It could be due to steric hindrance such that it blocks the PP2A-C from accessing the A subunit bound to Spry2. Alternatively, it could be due to sequestration of Spry2 into the vesicles, where the C subunit may not be available for binding to the complex. To discriminate between these two possibilities, it was essential to identify a construct of Tesk1 that retained binding to Spry2 but did not localize to the vesicles. Our results with Tr528 showed that Tesk1 binding to Spry2 and its own localization were dictated by its last 100 residues (C100). Indirect immunofluorescence, however, revealed that C100 does not by itself localize to vesicles and instead has a cytosolic and membrane distribution (Fig. 9A, panels a and b). Thus, C100 provided us with a tool to study the contribution of binding versus localization in the negative effect of Tesk1 on Spry2 function.
Spry2 colocalizes with C100 both in unstimulated and EGFstimulated conditions when expressed in Cos1 cells (Fig. 9A,  panels c and d). We noted that a subpopulation of Spry2 localizes to the plasma membrane in the presence of C100 even in  the basal state. We next analyzed the effect of C100 on Spry2 in an experimental set-up similar to that described in Fig. 4. C100 does not affect inhibition of ERK phosphorylation by Spry2, as opposed to the full-length WT Tesk1 (Fig. 9B, lanes 11 and 13), even though it coimmunoprecipitates with Spry2 similar to the WT (Fig. 9C). C100 also does not inhibit the dephosphorylation of Spry2 induced by FGFR1 (Fig. 9B, lane 13). It is interesting to note that in the presence of C100, Spry2 seems to be mainly concentrated in the faster migrating form even in the unstimulated state (lane 7), suggesting that it may influence the basal Ser phosphorylation of Spry2. In PC12 cells, Spry2 inhibits bFGF-induced neurite outgrowth even in the presence of C100 (Fig. 9D), further validating the fact that binding of C100 does not interfere with Spry2 function. Taken together, these results show that the negative effect of Tesk1 binding on Spry2 is more likely a case of sequestration that requires its vesicular localization.
Tesk1 Does Not Affect c-Cbl Binding or Inhibition of EGFR Ubiquitination by Spry2-In a previous study we have reported the presence of at least two mutually exclusive pools of Spry2: one bound to PP2A that contributes to its Ras/ERK inhibitory function and another bound to the E3 ubiquitin ligase, c-Cbl (32). The interaction between Spry2 and c-Cbl has been extensively studied (12,34,35) and involves the binding of the SH2like domain of c-Cbl to a conserved tyrosine at position 55 on Spry2 (35)(36)(37). To investigate the effect of Tesk1 on the c-Cblassociated pool of Spry2, we detected the amount of c-Cbl immunoprecipitated with Spry2 in the presence and absence of Tesk1. The result in Fig. 10A shows that Tesk1 coexpression only marginally decreases the binding of c-Cbl to Spry2. Spry2 with anti-FLAG antibody to detect the presence of Spry2. Expression of the constructs was verified by immunoblotting the total cell lysates (TCL) with the indicated antibodies. B, 293T cells were transfected with mock vector or Myc-Tesk1, serum-starved for 3 h, and subsequently left unstimulated or stimulated with bFGF for 2 h. The lysates were immunoprecipitated with anti-Spry2, separated on SDS-PAGE, and immunoblotted with the same antibody to detect Spry2. TCL were immunoblotted with anti-Myc to detect expression of Tesk1. C, the various constructs of Tesk1 were transfected with Spry2 in 293T cells with or without fibroblast growth factor receptor 1 (FGFR1). Total cell lysates were subjected to immunoprecipitation with anti-FLAG, separated on SDS-PAGE, and immunoblotted for the presence of Tesk1 and Spry2. Expression of the proteins was detected by immunoblotting the TCL. WT, wild type; DA, kinase dead Tesk1 D170A; Tr528, truncation mutant of Tesk1 from residues 1-528. D, lysates of 293T cells transfected as described in A were immunoprecipitated with anti-FLAG antibody, separated on SDS-PAGE, and blotted with antibodies against PP2A-A, PP2A-C, and Myc. Expression of the constructs was verified by immunoblotting the TCL. * indicates IgG heavy chain. can potentially bind to c-Cbl and Tesk1 simultaneously, because c-Cbl binds to a region distal to the Tesk1-binding Cys-rich C terminus. Furthermore, it has been reported that the interaction between Spry2 and c-Cbl occurs both at the plasma membrane and in endosomes (36), so the relocalization of Spry2 almost exclusively to endosomes by Tesk1 may deplete only a proportion of the Spry2⅐c-Cbl complex.
The interaction of Spry2 with c-Cbl plays a role in the regulation of EGFR endocytosis by Spry2 (reviewed in Ref. 38). c-Cbl binds to and ubiquitinates EGFR upon ligand stimulation, leading to receptor internalization and proteasomal degradation. Several groups have reported the inhibitory role of Spry2 in this Cbl-mediated ubiquitination and subsequent trafficking of EGFR (36,39,40) as it competes with EGFR for binding to c-Cbl via its SH2 domain. This results in increased surface retention of the receptor upon ligand binding, and consequently, prolonged downstream signaling. Because there was a slight decrease in the amount of c-Cbl bound to Spry2, our next step was to determine whether this might reflect on the ability of Spry2 to inhibit the ubiquitination of EGFR. The result in Fig.  10B shows that while EGFR gets ubiquitinated in the presence of c-Cbl upon EGF stimulation, this is reduced by coexpression of Spry2. Spry2 itself gets degraded upon EGF stimulation, as can be seen from the reduced levels of the protein, likely because of its own ubiquitination by c-Cbl. Tesk1 coexpression does not alter this pattern (lanes 7-10), suggesting that it does not significantly affect this aspect of Spry2 cellular function.

DISCUSSION
The identification of various interacting partners to Spry2 in recent years has increased our understanding of the intricate mechanism controlling its activation and mode of action. In this study, we report the identification of Tesk1 as a novel regulator of Spry2. That Spry2 and Tesk1 form a complex in vivo in mouse tissues hints at a potential developmental and/or physiological role for this interaction, indicating that Tesk1 may be an endogenous modulator of Spry2 function.
The constitutive binding of Tesk1 to Spry2 leads to its relocalization to Tesk1-positive vesicles including endosomes. Spry2 has previously been reported to localize to endosomes in certain studies (41), whereas others have reported a microtubular localization (25,26). Differing amounts of endogenous Tesk1 may form the basis for this discrepancy, since in this study we have shown that Tesk1 does indeed relocalize it to endosomes.
This interaction abolishes the Grb2 binding of Spry2, resulting in the loss of its inhibition of ERK phosphorylation. Although the negative effect of Tesk1 could potentially arise from either steric hindrance or relocalization of Spry2, the lack of effect of the cytosol-and membrane-localized C100 suggests that sequestration of Spry2 to vesicles forms the basis for its negative effect. It is interesting to note that although the fulllength Tesk1 localizes to vesicles, both the Tr528 and the C100 constructs show a cytosolic distribution. One possible explanation is that the targeting signal on Tesk1 is located around residue 528 and is abolished when this sequence is disrupted. However, such a signal is not immediately apparent from the primary sequence in this region. Another possibility is that two or more targeting motifs may be present in the full-length protein, including the last 100 residues, which synergize to determine the localization of the full-length protein. Thus, the individual motifs may not be sufficient for the vesicular targeting of the Tr528 and C100 constructs. The significance of the endosomal localization of Tesk1 itself remains to be determined.
The binding of Tesk1 also inhibits the Ser dephosphorylation of Spry2 by interfering with its interaction with PP2A. Although not affecting the scaffolding A subunit from binding to the N terminus of Spry2, Tesk1 displaces the catalytic C subunit from this complex. Proteins like PyST (polyoma small T antigen) have been reported to interfere with PP2A activity by displacing a third regulatory component (subunit B) of the phosphatase (42), but interference of PP2A-A and -C binding has not been previously reported. The basis for this displacement by Tesk1 remains to be understood but may involve inaccessibility of the Spry2⅐PP2A-A complex in the endosomes or a physical masking of the C binding site on the A subunit.
The phosphorylation of Spry2 on conserved Ser residues has been reported in two recent studies to play a role in the stability of the protein and in its activation upon growth factor stimulation (32,33). To date, Mnk1 and CK2 have been proposed to be the kinases that phosphorylate Spry2 on these residues. C100 seemingly affects this basal Ser phosphorylation of Spry2, but it is currently not known whether it inhibits phosphorylation by the kinases or promotes dephosphorylation by targeting Spry2 to the membrane in the resting state. It may serve as a useful tool in future studies for delineating the molecular processes that control the balance between the phosphorylation and dephosphorylation of Spry2.
Although Drosophila Spry works as a general inhibitor of receptor tyrosine kinase signaling (43), mammalian Spry isoforms seem to have evolved more selective roles. Spry2 in particular has been regarded as a positive and negative regulator of receptor tyrosine kinase signaling, owing to its opposing roles in EGFR versus other receptor tyrosine kinase signaling (38). Even in the context of EGFR signaling, the N and C termini of Spry2 have opposing effects, with the N terminus prolonging the signaling and the C terminus inhibiting it (34). This likely involves the distinct pools of Spry2 bound to c-Cbl and Grb2, respectively. However, little is currently known about the factors that determine the balance between these populations and the subsequent outcomes. Although Tesk1 negatively regulates the function of Spry2 arising from its Grb2 interaction, i.e. inhibition of the Ras/ERK pathway, it does not affect its c-Cblassociated function with respect to regulation of EGFR ubiquitination. Thus, the presence of interacting partners such as Tesk1 that differentially regulate the downstream effects of Spry2 may contribute to fine tuning its antagonistic properties.
In this study we have shown that Tesk1 regulates the activation and function of Spry2 as an inhibitor of Ras/ERK signaling irrespective of its kinase activity. Conversely, our current understanding on the cellular role of Tesk1 pertains mainly to its enzymatic activity as a cofilin kinase. In particular, evidence from other groups shows that proteins binding to the C-terminal tail of Tesk1, such as Spry4 (19) and actopaxin (44), inhibit its kinase activity. In both cases, binding of these proteins has a negative effect on Tesk1-induced cell spreading. It is possible that the C-terminal tail of Tesk1 is an important regulatory sequence that serves to control its activity not only by dictating its localization but also by binding to its regulators. Since all of the tested isoforms of Spry and Spred bind to Tesk1, they all have the potential to influence its activity, giving these proteins in turn, an additional role as regulators of Tesk1. In this context, it is worth noting that although depletion of the endogenous protein through shRNA suggests a role for Tesk1 in cellular processes such as PC12 differentiation, this does not discriminate between its kinase and scaffolding effects, and it is likely that both of these roles contribute to the overall outcome. Thus, the physiological implications of Tesk1 interaction with Spry and Spred in its capacity as a scaffold and as a cofilin kinase could link the regulation of the Ras/ERK pathway to the modulation of the cytoskeleton, which is at the core of various cellular processes including growth, differentiation, and migration.